Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Phototrophic iron oxidation and implications for biogeochemical cycling in the Archean Eon Thompson, Katharine Jean 2020

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2020_november_thompson_katharine.pdf [ 33.64MB ]
Metadata
JSON: 24-1.0392002.json
JSON-LD: 24-1.0392002-ld.json
RDF/XML (Pretty): 24-1.0392002-rdf.xml
RDF/JSON: 24-1.0392002-rdf.json
Turtle: 24-1.0392002-turtle.txt
N-Triples: 24-1.0392002-rdf-ntriples.txt
Original Record: 24-1.0392002-source.json
Full Text
24-1.0392002-fulltext.txt
Citation
24-1.0392002.ris

Full Text

Phototrophic iron oxidation andimplications for biogeochemical cyclingin the Archean EonbyKatharine Jean ThompsonBA Microbiology, Cornell University, 2011A DISSERTATION SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinThe Faculty of Graduate and Postdoctoral Studies(Microbiology and Immunology)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)June 2020c© Katharine Jean Thompson 2020The following individuals certify that they have read, and recommend to the Faculty of Graduateand Postdoctoral Studies for acceptance, the dissertation entitled:Phototrophic iron oxidation and biogeochemical cycling in the Archean EonSubmitted by Katharine J. Thompson in partial fulfilment of the requirements for the degree ofDoctor of Philosophy in Microbiology and ImmunologyExamining Committee:Associate Professor Sean A. Crowe, Microbiology and Immunology & Earth, Ocean and Atmo-spheric SciencesSupervisorProfessor Tom Beatty, Microbiology and ImmunologySupervisory Committee MemberProfessor Susan Baldwin, Chemical and Biological EngineeringUniversity ExaminerProfessor Curtis Suttle, Microbiology and Immunology & Earth, Ocean and Atmospheric SciencesUniversity ExaminerAdditional Supervisory Committee Members:Professor Steven J. Hallam, Microbiology and ImmunologySupervisory Committee MemberProfessor Stephen Calvert, Earth, Ocean and Atmospheric SciencesSupervisory Committee MemberiiAbstractBanded iron formations (BIFs), which host the world’s largest iron ore deposits, formed pre-dominantly through the deposition of ferric iron (Fe[III]) from ferruginous oceans during theArchean Eon. Available evidence suggests that phototrophic iron oxidation (photoferrotrophy)may have played a key role in coupling the carbon and iron cycles during the Archean Eon,depositing BIFs, and, in doing so, underpinned global primary production at this time. To date,however, all known photoferrotrophs form a close association with the ferric iron metabolitesthey produce during growth. This intimate association calls into question the involvement ofphotoferrotrophs in BIF deposition, their ability to act as primary producers, and their role insustaining the biosphere for millions of years. Furthermore, a lack of quantitative knowledge onthe growth of photoferrotrophs and the interactions between them and other microorganismslimit our ability to constrain models of BIF deposition and the Archean ocean-atmosphere systemas a whole. This dissertation generates new knowledge on extant photoferrotrophy that canbe used to inform and constrain models of primary production and BIF deposition during theArchean Eon. I create new knowledge on photoferrotrophy under laboratory conditions andin natural environments through data collected on the physiology and metabolic capacity ofpelagic photoferrotroph Chlorobium phaeoferrooxidans strain KB01. I also measure process rates andanalyze the composition of the microbial community in a ferruginous lake—–Kabuno Bay–—thatis dominated by photoferrotrophy. I subsequently integrate this new knowledge into models thatexamine the antiquity of nutrient acquisition in the photoferrotrophic Chlorobia and the role ofphotoferrotrophs as primary producers during the Archean. These models provide an explanationfor the formation of BIFs as a by-product of the activity of photoferrotrophic bacteria. Additionally,I demonstrate how photoferrotrophs could have sustained the biosphere, likely fueled microbialmethanogenesis, and, therefore, helped to stabilize Earth’s climate under a dim early Sun.iiiLay SummaryPrimitive photosynthesis may have played a fundamental role in the formation of the world’slargest iron ore deposits during the first 3 billion years of Earth’s history. These iron ore deposits,however, lack evidence for photosynthetic cellular remains that should have been co-depositedwith iron. This raises questions about the growth of primitive photosynthetic bacteria in Earth’searly oceans. In my thesis, I determine the physical and chemical characteristics of primitivephotosynthesis under laboratory conditions and in modern, natural environments. I then applythis knowledge to demonstrate how these photosynthetic bacteria could have thrived in Archeanoceans, driving large-scale biogeochemical cycles. My thesis demonstrates that these primitivephotosynthetic bacteria could deposit iron ore free of cellular remains and support the growth ofother bacteria. Ultimately, this process supports the production of methane, which would havecontributed to a greenhouse atmosphere and climate stability under the dim early Sun.ivPrefaceThis work was made possible through the contributions and dedication of many collaborators.Dr. Sean Crowe, as the research advisor was involved in all aspects of this work includingexperimental design, data analysis and interpretation and writing. Sections of this work are partlyof wholly published, in press, or in review. Copyright licenses were obtained and are listed below.• Chapter 1: Katharine J. Thompson wrote the main text with editorial support from Sean A.Crowe.• Chapter 2: Katharine J. Thompson wrote the main text with editorial support from Sean A.Crowe. Katharine J. Thompson performed all the laboratory work.• Chapter 3: Katharine J. Thompson and Sean A. Crowe wrote the main text with input fromRachel L. Simister. Katharine J. Thompson performed all the laboratory work, except thebiochemical verification of assimilatory sulfate reduction, which was performed by SeanA. Crowe. Katharine J. Thompson, Rachel L. Simister, and Sean A. Crowe interpreted andanalyzed the data with bioinformatic data analysis conducted by Aria S. Hahn. Steven J.Hallam contributed to data interpretation and insights. Editorial support was received fromthe entire list of authors. The reference for the published paper can be found as follows:K. J. Thompson, Rachel L. Simister et al. Nutrient Acquisition and the Metabolic Potential ofPhotoferrotrophic Chlorobi. Frontiers in microbiology, 8:1212, July 2017.• Chapter 4: Katharine J. Thompson and Sean A. Crowe wrote the manuscript with detailedinput from Paul A. Kenward, Kurt O. Konhauser, Christopher T. Reinhard, and AndreasKappler. Sean A. Crowe conceived, designed, and directed the study. Sean A. Croweand Marc Llio´s made observations in Kabuno Bay. Katharine J. Thompson performedphotoferrotroph cell settling experiments. Tyler Warchola performed cyanobacteria settlingexperiments. Katharine J. Thompson and Paul A. Kenward conducted electron microscopyimaging of strain KB01. Tina Gauger and Andreas Kappler conducted electron microscopyimaging of strain KoFox. Katharine J. Thompson, Paul A. Kenward, Rachel L. Simister, andRaul Martinez characterized cell surface chemistries. Katharine J. Thompson conductedDVLO modeling. Katharine J. Thompson, Sean A. Crowe, Kohen W. Bauer, and Ce´line C.Michiels constructed the biogeochemical models. Editorial support was received from theentire list of authors. The reference for the published paper can be found as follows:vK.J. Thompson, P.A. Kenward et. al. Photoferrotrophy, deposition of banded iron formations,and methane production in Archean oceans. Science Advances, 5:eaav2869, November 2019• Chapter 5: Katharine J. Thompson wrote the main text with editorial support from Sean A.Crowe. Sean A. Crowe and Marc Llio´s conducted process rate measurement experimentsand made observations in Kabuno Bay. Katharine J. Thompson and Rachel L. Simisterdeveloped the bioinformatic pipeline used for metagenome analyses. Katharine J. Thompsonperformed all the metagenome analyses. Editorial support was received from the entire listof authors.• Chapter 6: Katharine J. Thompson wrote the main text.Throughout this dissertation the word ‘we’ refers to Katharine J. Thompson unless otherwisestated.None of the work encompassing this dissertation required consultation with the UBC ResearchEthics Board.viTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiLay Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiiAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xivDedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 The marine iron cycle past and present . . . . . . . . . . . . . . . . . . . . . . . . . . 21.2 Extant photoferrotrophy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81.3 Photoferrotrophy during the Archean Eon . . . . . . . . . . . . . . . . . . . . . . . . 131.4 Problem statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171.5 Dissertation overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182 Chlorobium phaeoferrooxidans strain KB01 – a pelagic, Fe(II)-oxidizing, anoxygenic,photosynthetic bacterium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222.3 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272.3.1 Strain and growth medium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272.3.2 Bioinformatics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272.3.3 Analytical techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292.3.4 Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292.3.5 Fe and light dependency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302.3.6 pH, temperature, and vitamin dependence . . . . . . . . . . . . . . . . . . . 32vii2.3.7 Alternative electron donors and acceptors . . . . . . . . . . . . . . . . . . . . 332.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342.4.1 Morphology and pigmentation . . . . . . . . . . . . . . . . . . . . . . . . . . 342.4.2 Phylogeny and metabolic potential . . . . . . . . . . . . . . . . . . . . . . . . 362.4.3 Electron donor usage, Fe(II) oxidation, and growth . . . . . . . . . . . . . . . 392.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422.5.1 Characteristic traits of strain KB01 . . . . . . . . . . . . . . . . . . . . . . . . 422.5.2 Strain KB01 as a model organism for Precambrian PP . . . . . . . . . . . . . 483 Nutrient Acquisition and the Metabolic Potential of Photoferrotrophic Chlorobi . . . 503.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513.3 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 553.3.1 Strains and growth medium . . . . . . . . . . . . . . . . . . . . . . . . . . . . 553.3.2 Analytical techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563.3.3 Bioinformatics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563.3.4 Phylogenetic trees for nitrogen fixation . . . . . . . . . . . . . . . . . . . . . . 573.3.5 Phylogenetic trees for assimilatory sulfate reduction . . . . . . . . . . . . . . 573.3.6 Phylogenetic trees for 16S rRNA . . . . . . . . . . . . . . . . . . . . . . . . . 583.4 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 583.4.1 Nitrogen fixation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 583.4.2 Assimilatory sulfate reduction (ASR) . . . . . . . . . . . . . . . . . . . . . . . 663.5 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 734 Photoferrotrophy, deposition of banded iron formations, and methane production inArchean oceans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 744.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 744.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 744.3 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 774.3.1 Separation of biomass and Fe(III) . . . . . . . . . . . . . . . . . . . . . . . . . 774.3.2 Mechanisms of cell-mineral separation . . . . . . . . . . . . . . . . . . . . . . 794.3.3 Revised Precambrian Fe budgets . . . . . . . . . . . . . . . . . . . . . . . . . 814.3.4 Modeling Archean marine iron and carbon cycles . . . . . . . . . . . . . . . 834.4 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 884.5 Supplementary materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 895 Microbial community metabolism and coupled carbon and iron cycling in ferruginousenvironments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 915.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91viii5.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 925.3 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 955.3.1 Kabuno Bay site description and sampling . . . . . . . . . . . . . . . . . . . . 955.3.2 Physio-chemical analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 965.3.3 Iron oxidation, iron reduction, and sulfate reduction rates . . . . . . . . . . 965.3.4 Analytical techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 975.3.5 Flux calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 985.3.6 DNA extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 995.3.7 Metagenome sequencing and assembly . . . . . . . . . . . . . . . . . . . . . . 995.3.8 16S rRNA reconstruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 995.3.9 Gene searches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1005.3.10 Metagenome assembled genomes . . . . . . . . . . . . . . . . . . . . . . . . . 1015.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1025.4.1 Biogeochemical and physical properties of Kabuno Bay . . . . . . . . . . . . 1025.4.2 Microbial community composition and structure . . . . . . . . . . . . . . . . 1055.4.3 Photosynthetic and oxidative metabolic potential . . . . . . . . . . . . . . . . 1085.4.4 Carbon breakdown and fermentation . . . . . . . . . . . . . . . . . . . . . . . 1145.4.5 Iron reduction, sulfate reduction, and methanogenesis . . . . . . . . . . . . . 1195.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1205.5.1 Primary production in Kabuno Bay . . . . . . . . . . . . . . . . . . . . . . . . 1205.5.2 Microbial metabolisms in Kabuno Bay . . . . . . . . . . . . . . . . . . . . . . 1235.5.3 Microbial community network and tight carbon and iron coupling . . . . . 1255.5.4 Implications for coupled carbon and iron cycling in the Precambrian Eons . 1275.6 Supplementary materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1286 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1306.1 Extant photoferrotrophy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1306.2 Photoferrotrophy during the Archean Eon . . . . . . . . . . . . . . . . . . . . . . . . 1316.3 Looking ahead . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1326.4 Closing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135AppendicesA Chapter 4: supplemental material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153A.1 Supplementary materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . 153A.1.1 Experimental and growth media . . . . . . . . . . . . . . . . . . . . . . . . . 153A.1.2 Fe(II) oxidation and cell growth . . . . . . . . . . . . . . . . . . . . . . . . . . 154ixA.1.3 Determination of cellular association to Fe(III) . . . . . . . . . . . . . . . . . 154A.1.4 Zeta potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154A.1.5 Surface contact angles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155A.1.6 Cell surface titrations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155A.1.7 Electron microscopy (SEM, TEM) . . . . . . . . . . . . . . . . . . . . . . . . . 156A.1.8 Particle size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156A.2 Supplementary text . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156A.2.1 Cell surface features and acid-base chemistry . . . . . . . . . . . . . . . . . . 156A.2.2 Cell-iron surface interaction and extended DVLO modeling . . . . . . . . . 157A.2.3 Iron concentration and supply . . . . . . . . . . . . . . . . . . . . . . . . . . . 160A.2.4 Physical separation of ferric iron oxyhydroxides and cellular biomass in anocean setting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162A.2.5 Box model of Archean marine carbon and iron cycles . . . . . . . . . . . . . 163A.2.6 Organic carbon burial and diagenesis . . . . . . . . . . . . . . . . . . . . . . . 165A.3 Supplementary figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166A.4 Supplementary tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172B Chapter 5: supplemental material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185B.1 Supplemental figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186B.2 Supplemental tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192xList of Tables1.1 A summary of some of the isolated photoferrotrophic strains. . . . . . . . . . . . . . 92.1 DNA-DNA hybridization between strain KB01 and strain KoFox. . . . . . . . . . . . 362.2 Temperature and pH dependent growth rates for strain KB01. . . . . . . . . . . . . . 422.3 Iron oxidation by a select number of Fe(II) oxidizing bacteria. . . . . . . . . . . . . . 443.1 Codon adaptation index (CAI) for Chlorobi nitrogenases. . . . . . . . . . . . . . . . 623.2 Green sulfur bacteria ASR genes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 683.3 Codon adaptation index (CAI) for Chlorobi ASR genes. . . . . . . . . . . . . . . . . 704.1 Model results for both the carbon cycle—primary production (PP), carbon burial,carbon remineralization (remin.), and methane production—and the iron cycle—ironrecycling, iron deposition (dep.)—-at 2.5Ga. . . . . . . . . . . . . . . . . . . . . . . . 865.1 Extended data from a selection of MAGs . . . . . . . . . . . . . . . . . . . . . . . . . 112A.1 Range of concentrations in the growth media used throughout experiments. . . . . 172A.2 Cell surface characteristics and cell-mineral interaction modelling. . . . . . . . . . . 173A.3 Modern and Archean Fe fluxes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174A.4 Different scenarios of the physical separation model, with each case using a differentwater velocity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175A.5 Data compilations for Figure 1 and S1. . . . . . . . . . . . . . . . . . . . . . . . . . . . 176B.1 Description of each Fe speciation extraction step . . . . . . . . . . . . . . . . . . . . . 192B.2 List of genes discussed in the main text and their primary function . . . . . . . . . . 193B.3 % of metagenomic reads that aligned to each set of MAGs . . . . . . . . . . . . . . . 194xiList of Figures1.1 Overview of the iron cycle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41.2 Overview of the iron cycle through time. . . . . . . . . . . . . . . . . . . . . . . . . . 51.3 Microscopy, SEM, and TEM images of photoferrotrophic strains. . . . . . . . . . . . 101.4 Proposed metabolic model for the microorganism present at the chemocline inKabuno Bay. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121.5 The relative abundance of BIFs over time. . . . . . . . . . . . . . . . . . . . . . . . . . 131.6 The abundance of BIFs compared to the concentration of atmospheric oxygen overtime. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152.1 Microscopy images of strain KB01 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352.2 Strain KB01 pigmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352.3 The metabolic potential of strain KB01 . . . . . . . . . . . . . . . . . . . . . . . . . . . 372.4 Strain KB01 growth rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 402.5 Strain KB01 growth kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413.1 Nitrogenase gene cassettes of the photoferrotrophic Chlorobi. . . . . . . . . . . . . . 603.2 Biochemical verification of Nitrogen fixation . . . . . . . . . . . . . . . . . . . . . . . 613.3 Nitrogenase phylogenies of the Chlorobi and Bacteroidetes. . . . . . . . . . . . . . . 633.4 Nitrogenase phylogenies of multiple phyla. . . . . . . . . . . . . . . . . . . . . . . . . 653.5 Assimilatory sulfate reduction (ASR) gene cassettes for the photoferrotrophic Chlorobi 693.6 CysH phylogenies of multiple phyla. . . . . . . . . . . . . . . . . . . . . . . . . . . . 713.7 Sat/CysD phylogenies of multiple phyla. . . . . . . . . . . . . . . . . . . . . . . . . . 734.1 The organic matter concentrations in BIFs, other Precambrian sedimentary rocks,typical modern marine sediments, and oxygen minimum zone (OMZ) sediments. . 764.2 Cell surface characteristics for strain KB01 and the relationship between ferric ironsurface charge and medium anions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 784.3 Scanning electron microscopy and transmission electron microscopy image of strainKB01. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 794.4 Model of an Archean coastal upwelling zone. . . . . . . . . . . . . . . . . . . . . . . 844.5 Iron and carbon box model sensitivity results. . . . . . . . . . . . . . . . . . . . . . . 85xii5.1 Depth profiles from Kabuno Bay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1035.2 Microbial community composition in Kabuno Bay . . . . . . . . . . . . . . . . . . . . 1075.3 Abundance of key genes predicted to play roles in the biogeochemical cycles inKabuno Bay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1095.4 Primary production in Kabuno Bay . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1115.5 A representation of the metabolic potential for three MAGs . . . . . . . . . . . . . . 1165.6 Flow of carbon compounds through the metabolic pathways found in each MAG . 1185.7 Proposed metabolic model for the primary producing microorganisms present atthe chemocline in Kabuno Bay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122A.1 The redox state of iron in BIF through time where the red bars indicate the sideriterich BIFs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166A.2 Growth curve for Chlorobium phaeoferrooxidans strain KB01. . . . . . . . . . . . . 166A.3 Additional SEM and TEM images of strains KB01 and KoFox. . . . . . . . . . . . . . 167A.4 Surface charge of strains KB01 and KoFox under two conditions. . . . . . . . . . . . 168A.5 Additional cell surface characteristics for strain KoFox and the relationship betweenthe number of planktonic cells and the Fe(III) surface charge. . . . . . . . . . . . . . 168A.6 Modeling the settling velocity of carbon and iron using a range of horizontal currentvelocities. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169A.7 Modeled weight % organic carbon in the coastal and open ocean sediments. . . . . 170A.8 Iron and carbon box model sensitivity results. . . . . . . . . . . . . . . . . . . . . . . 171B.1 The location of Kabuno Bay with the sampling site . . . . . . . . . . . . . . . . . . . 186B.2 The relative abundance of the reconstructed 16S rRNA gene for the 16S rRNA geneswith a greater than 1 % relative abundance for each depth . . . . . . . . . . . . . . . 187B.3 Reads per kilobase mapped (RPKM) values for taxonomic marker genes related toclass Chlorobia and glycosyl hydrolases . . . . . . . . . . . . . . . . . . . . . . . . . . 188B.4 Gene abundances of the key genes for several pathways . . . . . . . . . . . . . . . . 189B.5 The depth integrated absolute abundance of each phyla compared to the numberof MAGs recovered from each of those groups . . . . . . . . . . . . . . . . . . . . . . 190B.6 Comparison of the Chlorobi 01 MAG to the genome of Chlorobium phaeoferrooxidansstrain KB01 at the pathway level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190B.7 Comparison of the genomes of several key MAGs at the pathway level . . . . . . . . 191B.8 Representation of the metabolic potential for two MAGs . . . . . . . . . . . . . . . . 191xiiiAcknowledgementsThis process has been an incredible experience that began back in Grade 7 when Mr. D asked usto make an oven that could bake real cookies out of cardboard and aluminium foil–—in otherwords the day I discovered that creativity and science went hand in hand with a stomach full ofraw cookie dough. There have been numerous people along the way who have encouraged andsupported me and honestly there are too many to list! So, instead, here are a few thank yous tothose who have made this thesis possible. First, I would like to thank Sean Crowe for giving me aposition in his lab. From the first moment we spoke in your office about iron oxidation and theArchean oceans to our current discussion on which model we should construct next, your vastknowledge, experience, and enthusiasm have pushed me to be a better scientist. Thank you forcontinuing to believe in me and for helping me achieve this dream. Thank you to my ”external”committee members Kurt Konhauser, Chris Reinhard, and Andreas Kappler for your ongoingsupport and willingness to help me pursue my career! I would also like to thank the members ofmy committee Steven Hallam, Steve Calvert, Tom Beatty, and John Smit for your words of wisdomand support, as well as your diligent edits to my thesis chapters. Thank you to my ExternalExaminer, Professor Alexis Templeton from the University of Colorado Boulder, who reviewedmy thesis thoughtfully. A massive thank you to all of the members of the Crowe Lab over theyears — Ce´line Michiels, Ashley Davidson, Jenifer Spence, Julia Huggins, Kohen Bauer, RachelSimister, Amani Alsufyani, Kanchi Dave, Niko Finke, Arne Sturm, and the many undergraduates-– for your knowledge, willingness to listen, and endless discussions with me on any kind of topic(usually done with a beer in hand). Finally, thank you to all my science and sports friends whogave the outlets I needed to persevere when times were tough -– all those tea breaks and soccergames were just what I needed to get this done!xivDedicationTo my parents and sisters for your love, support, and encouragement.To Tony for your love and unerring patience in the face of “just a few more months”.xvChapter 1IntroductionLife has interacted with the Earth’s surface, effectively altering its chemistry since it first evolvedover four billion years ago [1–3]. Across the two billion years that followed life’s origin, during theArchean Eon, communities of microorganisms propagated through the use of a variety of inorganicand organic electron donors and acceptors thereby modifying Earth’s surface composition, shiftingredox balances, and driving biogeochemical cycles [3]. Some of these global-scale processeshave been recorded and preserved in the sedimentary rock record, which can be accessed todaythrough geological and geochemical studies. A rich source of information comes from chemicalsediments such as carbonates, and for the Archean Eon in particular, banded iron formations(BIFs), which were deposited predominantly between 2.9 and 1.9 billion years ago (Ga) [4]. Thequantity of iron present in BIFs, among other indicators, requires that BIFs were deposited fromanoxic ferruginous Precambrian oceans, which contained high concentrations of dissolved ferrousiron (Fe[II]) [5]. BIFs likely formed from such ferruginous oceans in response to the oxidation ofthis Fe(II) and the resulting precipitation and sedimentation of ferric iron (oxyhydr)oxides (Fe[III])[4, 6, 7]. Many of the models of BIF deposition invoke photosynthesis in the oxidation of ferrousiron from seawater to induce its subsequent precipitation and deposition as mixed valence iron(oxyhydr)oxides and carbonate phases [6–9]. BIFs thus likely record the activity of Earth’s earlyphotosynthetic biosphere [6–9].Two modes of photosynthesis have been implicated in Fe(II) oxidation in the Precambrianoceans (includes the Archean and Proterozoic Eons)–—canonical oxygenic photosynthesis bythe ancestors of modern cyanobacteria [6] and iron-dependent anoxygenic photosynthesis [7, 8],referred to as photoferrotrophy. While oxygenic phototrophs likely emerged in the Archean [10–12], evidence from studies of geochemical proxies has shown that the partial pressure of oxygenremained very low in the Archean atmosphere (less than 0.001 % of the current value) during BIF1deposition [2, 13–15]. Thus, oxygenic phototrophs at this time were likely restricted to rather smalland potentially ephemeral niches, which limited their contribution to global primary productionand the flux of O2 to the atmosphere [11, 16, 17]. Such a low O2 atmosphere-ocean system likelysupported limited rates of Fe(II) oxidation, implying that O2-driven Fe(II) oxidation was likelya minor contributor to BIF deposition at this time [2, 18, 19]. Conversely, photoferrotrophicbacteria grow using light and Fe(II) to fix CO2 into biomass and produce Fe(III) as a metabolicby-product [8, 9, 20] and they can do so in the complete absence of oxygen [Eq. 1.1] [20]. Thus,photoferrotrophs, as strict anaerobes, could have thrived in the anoxic Archean oceans, utilizingFe(II) as their electron donor. As these photoferrotrophs proliferated in nutrient-rich, illuminatedsurface waters, they could have supported high rates of Fe(II) oxidation and, therefore, generatedsufficient Fe(III) (oxyhydr)oxides to deposit BIFs [8, 21].4 Fe2+ +CO2 + 11 H2O+ hv⇒ [CH2O] + 4 Fe(OH)3 + 8 H+ (1.1)This chapter introduces the iron biogeochemical cycle as it pertains to the early Earth, thephysiology of photoferrotrophs and the molecular underpinning of this physiology, as wellas links between the coupled carbon and iron biogeochemical cycles in modern and ancientenvironments. It also highlights gaps in our knowledge of photoferrotrophs and their physiologyand the resulting uncertainties associated with their role in BIF deposition and biogeochemicalcycling in the Archean Eon. Overall, my thesis will create new knowledge on the physiologyand metabolic capacity of photoferrotrophy under laboratory conditions as well as on the roleof photoferrotrophy in modern ferruginous environments. I have further incorporated this newand quantitative knowledge into numerical models and phylogenetic analyses to test the possiblerole of photoferrotrophy in BIF deposition and the maintenance of a clement climate, throughphotoferrotrophs as primary producers, throughout the Archean Eon.1.1 The marine iron cycle past and presentArchean oceans contained high concentrations of Fe(II) sourced largely from hydrothermalcirculation through subsurface ocean crust. Thus, iron in the Archean oceans occurred in ferrous2(Fe[II]) and ferric (Fe[III]) states (Fig. 1.1) [5, 22, 23], wherein solid Fe(III) nanoparticles precipitatedfrom dissolved Fe(II), aggregated, and crystallized to form a suite of Fe minerals (Fig. 1.1) [5, 22, 23].These Fe minerals are deemed highly reactive as they can be readily reduced both chemicallyand microbially (Fig. 1.1) [5, 22, 23]. The ratio of collective highly reactive iron minerals (e.g.ferrihydrite and hematite; Fig. 1.1) to the total iron minerals in a given sediment or sedimentaryrock can discriminate between whether or not the depositional environment was oxic or anoxic(where a ratio of > 0.38 is indicative of anoxic conditions) [5, 24, 25]. Furthermore, the ratio ofhighly reactive iron minerals to the quantity of pyritized minerals present in the same sedimentor rock can discriminate between whether or not the anoxic depositional environment wasferruginous or sulfide-rich—–euxinic (where a ratio of > 0.7 is indicative of euxinic conditions)[24, 25]. Thus, the mineralogy of a given sediment or rock is one of the properties that can beused to diagnose whether the depositional environment for a particular sequence of rock wasferruginous [24]. The ratio of these minerals in samples collected from both marine deposits thatare more (e.g. BIFs) or less (e.g. shales) iron-rich, whose ages align with the Archean Eon, revealthat the Archean oceans were almost universally ferruginous [5, 23, 24]. Indeed, models basedon the mineralogy of Archean rocks suggest that Fe(II) concentrations could have accumulatedto appreciable levels in the oceans (up to approximately 100 µM) [5, 22–24, 26]. The majority ofthis Fe(II) was likely supplied by hydrothermal vents—through both vents that are concentratedat major mid-ocean ridges (on-axis) and those that extend from the mid-ocean ridges, coveringlarge areas of the ocean floor (off-axis) [5, 23, 24]. The 0.3 Tmol yr– 1 flux of Fe(II) reconstructedfor Archean hydrothermal vent systems [5], however, does not consider off-axis contributions,nor the major differences between the Archean and modern oceans, such as the concentration ofsulfate [27]. Thus, this current reconstruction of the fluxes of Fe(II) to the Archean oceans likelyunderestimates the real magnitude of Fe that accumulated due to the lack of oxidative potentialthrough elements such as oxygen and sulfide.Extensive and intensive studies of the trace element geochemistry of marine sedimentary rocks[2, 5, 13–15, 28] suggest that oxygen concentrations were as low as 0.001 % present atmosphericlevel (PAL) during the Archean Eon (Fig. 1.2). Several elements, whose speciation or isotopiccomposition depends on an oxygen-dependent reaction in their biogeochemical cycles, can be used3Fe(II)aq Fe(III)sGreen rustAnoxicOxicPhototrophic orNDFO*Circumneutral pHFe(OH)3ferrihydrite!-Fe2OOHgoethiteFe3O4magnetite!-Fe2O3hematiteFeCO3sideriteMicroaerophilicMicrobial Fe(III) reductionMineral aggregation and maturationMicrobial Fe(II) oxidationMicrobial mediated abiotic Fe(II) oxidation with O2Reduced highly reactive Fe speciesOxidized highly reactive Fe speciesFigure 1.1: Overview of the iron cycle. Key abiotic and biotic reactions in the Earth surface iron biogeochemicalcycle, including the transition from amorphous Fe(III) minerals to more crystalline Fe(III) and mixed valenceiron minerals. *NDFO = Nitrate-dependent Fe(II) oxidation.as geochemical proxies for the concentration of oxygen in the Archean oceans and atmosphere[28]. For example, the oxidation of land-surface crustal sulfide minerals by oxygen leads to theexport of soluble sulfate to the oceans and the subsequent mass-dependent fractionation of thesulfur through microbial sulfate reduction [2, 13–15, 29]. In the absence of oxygen in Earth’satmosphere during the Precambrian, sulfate concentrations in the oceans were very low [27],limiting mass-dependent sulfur isotope fractionation [2, 13–15, 27, 29]. Furthermore, withoutthe increased atmospheric oxygen concentrations needed to form ozone, reactions that result ina mass-independent fractionation of sulfur isotopes (S-MIF)–namely photochemical reactions–would have made up the majority of the sulfur isotopic signal [2, 13–15, 27, 29]. Therefore, thedisappearance of S-MIF signals between 2.4 and 2.3 Ga is one of the best known indicators thatoxygen concentrations rose in the atmosphere during that time interval [2, 13–15, 30]. Whilethe loss of the S-MIF suggests that atmospheric oxygen concentrations were as low as 0.001 %PAL during the Archean Eon, other elements with oxygen sensitive steps in their biogeochemicalcycles signal that temporary oxygen fluxes (so called ‘whiffs’ of oxygen) could have permeatedthe otherwise anoxic Archean oceans and atmosphere [16, 31, 32]. For example, the isotopicfractionation patterns of chromium have been linked to the presence of oxygen due to its reaction4with manganese oxides—–compounds that are, according to current research, solely produced inthe presence of oxygen [16, 33, 34]. The concentration of oxygen in the atmosphere and oceans ofthe Archean Eon is hotly debated amongst the scientific community [28, 35, 36], and while ‘whiffs’of oxygen are certainly plausible, the majority of the Archean oceans likely remained anoxic andferruginous [5]. Thus, low oxygen meant that Fe(II) was available as an abundant electron donorfor both microbial and chemical reactions (Fig. 1.2).Fe(II)aqFe(III)s Anoxic00.51.01.52.02.53.03.5Time (Ga)4.0OxicFe(II)aqFe(III)sFe(II)aqFe(III)sFe(II)aqArchean Eon Proterozoic Eon PhanerozoicEonFe(II)aqa.b.Microbial Fe(III) reductionPhotoferrotrophyAbiotic with O2MicroaerophilicPhotochemicalAbiotic with H2SNDFO*Figure 1.2: Overview of the iron cycle through time. Oxygen concentration (a) in the atmosphere relative to themodern oxygen concentration through time when the line represents the canonical oxygen curve and thedashed line represents alternative views, adapted from Lyons et. al., 2014 [28]. Major abiotic and bioticreactions (b) are depicted through the color of the dots and their likely abundance represented by thestrength of the color. *NDFO = Nitrate-dependent Fe(II) oxidation.There are a constrained number of abiotic and biotic reactions that are responsible for cyclingand recycling of iron in aqueous environments in the Archean Eon. The lack of atmospheric oxygenin the Archean Eon would have restricted crustal surface sulfide oxidation on the continents andthus the accumulation of soluble sulfate in the oceans [27]. Given that microbial sulfate reductionin the oceans produces sulfide, which is highly reactive with Fe(II) [18, 23], the low concentrations5of sulfate would have further enhanced the accumulation of Fe(II) [5, 15, 37]. Thus, in the absenceof oxygen, sulfate or sulfide, there are very few abiotic reactions to transform iron between itstwo main stable states. One of the few, but likely the most dominant, of these abiotic reactions isthe photochemical oxidation of Fe(II) by UV radiation (Fig. 1.1) [38–40]. During the Arhcean Eon,the UV radiation flux would have been unimpeded in the absence of an ozone layer, potentiallyleading to the anoxic oxidation of Fe(II), present in surface waters, to Fe(III) (oxyhydr)oxides(Fig. 1.2) [38]. While there were minimal chemical reactions to oxidize Fe(II), microbial reactions,such as anoxygenic photosynthetic Fe(II) oxidation, could have played a key role in the Archeanoceans (Fig. 1.1, 1.2) [1, 7, 8, 21]. In the absence of dissolved sulfides, the iron cycle couldonly have been completed through microbial Fe(III) reduction (Fig. 1.1, 1.2). Microbial Fe(III)reduction is an anaerobic process (Fig. 1.1) and, therefore, could have been found throughout theArchean ocean water column and sediments (Fig. 1.2) [41, 42]. Other microbial Fe redox reactionsthat could have evolved as early as the late Archean Eon, once oxygen concentrations began torise, including microaerophilic Fe(II) oxidation [43], which couples Fe(II) oxidation to oxygenreduction under low oxygen conditions, and nitrate-dependent Fe(II) oxidation (NDFO) [44],which couples Fe(II) oxidation to nitrate reduction under anaerobic conditions (Fig. 1.1). As bothof these microbial reactions require oxygen, with a direct role in microaerophilic Fe(II) oxidation(< 15 µM O2) and an indirect role in NDFO (wherein oxygen is required for the microbialoxidation of ammonium to nitrate), the magnitude of the role these metabolisms played in the Fecycle throughout the Archean would have depended on the oxygen flux and could have shiftedfrom minor (Paleoarchean) to appreciable (Neoarchean) (Fig. 1.2). Given the limited number ofmechanisms for iron transformation under the Archean conditions, the interplay between the rateof microbial phototrophic Fe(II) oxidation and Fe(III) reduction likely controlled the export ofsolid Fe(III) (oxyhydr)oxides from the water column.Unlike the Archean, where Fe was ubiquitous in the oceans and on the continents, theconcentration of Fe in the modern environment is limited by the concentration of oxygen [5].Modern atmospheric oxygen concentrations (21 %) ensures that the oceans are largely oxic withsmall pockets of anoxia. In the presence of this oxygen, Fe(II) weathered from the continents orsupplied through hydrothermal vents on the sea floor in the oceans oxidizes to form Fe(III) mineral6particles that are delivered to the seafloor, buried, and ultimately returned to the mantle throughsubduction [24]. Thus, dissolved Fe(II) does not accumulate and instead the concentration of Fe inmodern seawater is controlled by the solubility of Fe(III) phases [24]. Due to the low solubility ofFe(III) phases, therefore, the total Fe pool in seawater is small and Fe has a short residence time inthe modern oceans [5]. Furthermore, the oxygenated atmosphere fosters the production of sulfateon the continents, which accumulates in the global oceans with a modern concentration of 28mM. These high concentrations of sulfate mean that, in the pockets of ocean anoxia, microbialsulfate reduction to sulfide dominates, resulting in euxinic conditions rather than ferruginousconditions in these anoxic zones [45, 46]. Fe(II) can accumulate in sulfide-free anoxic oceansediments or in bottom waters of modern stratified lakes where Fe(II) is supplied by hydrothermalvents or Fe-enriched ground waters respectively [47, 48]. The modern biogeochemical iron cycleis, therefore, largely controlled by the availability of oxygen and sulfate, restricting the anaerobicmicrobial iron reactions to a subset of environments.Abiotic Fe(II) oxidation, with biologically produced oxygen, dominates the modern marine Fecycle, while the direct microbial Fe redox reactions that prevailed throughout the Archean Eonare restricted to anoxic marine benthic environments, stratified marine basins (e.g. ChesepeakeBay) and iron-rich stratified lakes. Fe photochemical reactions do occur in the modern surfaceoceans – either through the photoreduction of Fe(III) bound to organic ligands (Fig. 1.2) [49] orthrough the photooxidation of Fe(II) (Fig. 1.2) [39]. Given the high concentrations of O2 in themodern oceans compared to the Archean Eon, however, photochemical Fe(II) oxidation likelyplays a minor role in the cumulative Fe(II) oxidation (Fig. 1.2). Conversely, the high concentrationsof biologically produced oxygen in the modern oceans guarantees that microbially-mediatedabiotic Fe(II) oxidation by oxygen supports most of the marine Fe oxidation (Fig. 1.2), while thehigh concentrations of sulfide produced through sulfate reduction in anoxic zones and marinesediments contributes to Fe(III) (oxyhydr)oxide reduction (Fig. 1.2). Along with the abioticreduction of Fe(III) by sulfide, microbial Fe(III) reduction plays a key role in recycling Fe inmodern marine sediments, as it is ubiquitous in these anoxic environments [50]. Direct microbialFe(II) oxidation, whether by anoxygenic photosynthetic, microaerophilic, or nitrate-dependentbacteria, however, is restricted to three different niche environments that are dependent on the7availability of light, oxygen, and nitrate respectively (Fig. 1.1) [20, 43, 44]. These microbial reactionscan, therefore, be found in some marine settings, such as in the vicinity to hydrothermal ventsystems or in the sediments of coastal marine mud flats, or in Fe-rich freshwater environments,such as pond sediments and stratified lakes. Therefore, the controls on Fe(II) oxidation haveshifted through time from predominantly anoxygenic photosynthetic to abiotic, oxygen-dependentFe(II) oxidation that is mediated by abundant oxygenic photosynthetic organisms (Fig. 1.2). Thus,while the possible Fe reactions have not changed over time, the direct microbial Fe redox reactionshave become more restricted as oxygen concentrations have risen in the atmosphere and oceans.1.2 Extant photoferrotrophyDespite the small number of known modern environments that have the geochemical conditionsand light required for photoferrotrophy, 20 photoferrotrophs, from two different bacterial phyla,have been enriched and isolated into laboratory cultures [20, 48, 51–57]. The first photoferrotrophs,Rhodopseudomonas sp. and Thiodictyon sp. (Table 1.1), were enriched and isolated from a freshwaterpond and marine sediments in Germany in the early 1990’s [20]. These were the first bacteria thatdemonstrated that photosynthetic growth with Fe(II) as the electron donor was biologically andbiochemically possible [20]. These organisms were found to grow through both photoautotrophywith ferrous iron as the electron donor as well as photoheterotrophically with succinate or othercarbon sources [20]. Following this seminal discovery, many isolates were brought into laboratoryculture from a suite of different environments, both marine and freshwater. All but one, however,originated from a benthic source (subset of isolated photoferrotrophs shown in Table 1.1). Thesingular pelagic strain, Chlorobium phaeoferrooxidans strain KB01, was isolated from one ofthe few persistently ferruginous lakes that exist today–Kabuno Bay, a sub–basin of Lake Kivuin East Africa [48, 58]. Collectively, extant photoferrotrophs come from three different classesof bacteria (Alphaproteobacteria, Gammaproteobacteria, and Chlorobia), are closely related toother anoxygenic phototrophic members of those classes, and demonstrate polyphyly of thephotoferrotrophic trait [35]. Studies with laboratory cultures of photoferrotrophs have produced awealth of information on their physiology and growth, which ultimately provides a framework8Table 1.1: A summary of some of the isolated photoferrotrophic strains. For a complete list see Bryce et. al.2018 [57]. References: 1) [20]; 2) [51]; 3) [52]; 4) [54]; 5) [53]; 6) [59]; 7) [55, 60]; 8) [48, 58]; 9) [56]Class Strain Environment (all anoxic) Location ReferenceAlphaproteobacteria(Purple non-sulfur)Rhodopseudomonas palustris Marine sediment Germany 1Rhodobacter ferrooxidans strain SW2 Freshwater ponds Germany 2Rhodomicrobium vannielii strain BS-1 Freshwater ditch Germany 3Rhodovulum iodosum Marine mud flat Germany 4Rhodovulum robiginosum Marine mud flat Germany 4Rhodopseudomonas palustris strain TIE-1 Fe-rich mat USA 6Gammaproteobacteria(Purple sulfur)Thiodictyon sp. Marine sediment Germany 1Thiodictyon sp. strain F4 Freshwater marsh USA 7Chlorobi(Green sulfur)Chlorobium ferrooxidans strain KoFox Freshwater sediment Germany 5Chlorobium phaeoferrooxidans strain KB01 Kabuno Bay Democratic Republic of Congo 8Chlorobium sp. strain N1 Marine sediment Denmark 9References:1) Widdel, 1993; 2) Ehrenreich and Widdel, 1994, Kappler and Newman, 2004, Miot, 2009; 3) Heising and Schink, 1998; 4) Straub, 1999; 5) Heising, 1999; 6) Jiao 2005; 7) Croal 2005, Hegler 2008, 2010; 8) Lliros 2015, Crowe 2017; 9) Laufer 2017for the evaluation of the role of photoferrotrophy in the Fe cycle past and present.Studies with extant photoferrotrophs have provided a wealth of knowledge on their growth,their response to different environmental conditions, and their characteristic traits. A few studies,to date, have characterized the basic growth patterns of some of the photoferrotrophic isolates.These studies have conducted light microscopy as well as scanning and transmission electronmicroscopy to determine the internal and external structures of the different photoferrotrophs andthe relationship between cell surfaces and the Fe(III) (oxyhydr)oxides they produce as a by-productof their growth (Fig. 1.3). Collectively, these imaging techniques reveal that characterized, benthic,photoferrotrophs form a close association with the Fe(III) particles (Fig. 1.3) [9, 53, 55–57, 61–63].Indeed, the morphologies of the first isolates were obscured due to encrustation of cells by theFe(III) metabolites they produced during growth and could only be identified when the isolateswere grown with an alternative electron donor [20]. To date, the ability to grow using an alternativeelectron donor such as an organic carbon compound (e.g. succinate or lactate), H2, or H2S hassince been demonstrated in all of the characterized photoferrotrophs. Further studies with thesesame isolates also revealed several differences between the organisms, such as their growth rate inresponse to changes in the environmental conditions. For example, the photoferrotrophs isolated9from marine environments grow at a higher optimum pH of approximately 7.0-7.3, while thosefrom freshwater environments thrive at a slightly lower pH of approximately 6.5-6.8 [55–57]. Otherdifferences include the growth rates exhibited by different strains at low light intensities with thehalf-light saturation constants of individual photoferrotrophs falling within a wide range of values(0.25 µmol m– 2 s– 1 for Chlorobium ferrooxidans strain KoFox to 8 µmol m– 2 s– 1 for Thiodictyonsp.; [55]). Thus, while photoferrotrophs share the ability to utilize alternative electron donorsand known benthic isolates form a close association with Fe(III) particles, there are other notabledifferences between the individual isolates that likely represent the natural environments in whichthey originated.1 µma. b.c. d.e. f.Figure 1.3: Microscopy, SEM, and TEM images of photoferrotrophic strains. A microscopy (a) image of the firstphotoferrotrophic isolate grown with succinate [20], a microscopy image (b) of the first photoferrotrophicisolate grown with Fe(II) [20], an SEM image (c) of Chlorobium ferrooxidans strain KoFox grown with Fe(II)[55], an SEM image (d) of Thiodycton strain F4 grown with Fe(II) [55], an SEM image (e) of Rhodobacterferrooxidans strain SW2 grown with Fe(II) [55], and a TEM image (f) of Rhodobacter ferrooxidans strain SW2grown with Fe(II) [9].10Ferruginous environments are rare on the modern Earth, yet they provide natural laboratoriesin which to examine microbial communities under conditions that are similar to those thatlikely existed in the oceans during the Archean Eon. Given the scarcity of modern, illuminated,Fe(II)-rich pelagic environments, the majority of extant photoferrotrophs that have been studiedin laboratory cultures are derived from benthic environments where anoxic conditions persistand Fe(II) is supplied at a rate that supports photosynthetic growth through photoferrotrophy.These environments have included iron-rich fresh-water pond sediments and marine mud flats[20, 51–54, 56]. While these benthic settings yielded the majority of the photoferrotrophs discussedabove, their sediment-based and, sometimes (for marine environments), sulfate-rich settingsmake them poor analogs to the Archean oceans. In the last few years, persistent ferruginousconditions have been discovered in a number of modern lakes, in addition to Kabuno Bay (borderof Democratic Republic of Congo and Rwanda): Lakes Matano and Towuti (Indonesia), Lake LaCruz (Spain), Lake Pavin (France), Experimental Lakes Area (ELA; Canada). All of these lakes arestratified and rich in iron due to volcanic activity, weathering of ultramafic (iron-rich) rocks, orgroundwater seepage. Detailed studies of the microbial communities and their relationship to thechemical features present in each lake present the opportunity to expand the current knowledgeof the role of photoferrotrophs in a ferruginous water column and thus these lakes can inform onbiogeochemical cycling under ferruginous conditions, which is extensible to the Archean oceans.Biogeochemical processes in modern lakes can be interrogated through a combination ofgeochemical and microbiological analyses. Studies in the aforementioned ferruginous lakesare emerging, providing a wealth of new information on geochemical cycling and microbialcommunity profiles for anoxic, iron-rich water columns [47, 48, 64–68]. Such studies have reportedprocess rates for microbially mediated reactions that occur under ferruginous conditions suchas photoferrotrophy, iron reduction, nitrogen cycling, and methane cycling [47, 48, 64–73]. Forexample, the potential for photoferrotrophy was interrogated through light-based incubationexperiments, using a chemical inhibitor for oxygenic photosynthesis, in the chemocline of KabunoBay [48]. The lack of change in the rate with and without the inhibitor demonstrates the potentialfor photoferrotrophy this lake [48]. These process rate measurements provide estimates ofthe magnitude of key pathways in biogeochemical cycles that are supported by the microbial11community. Additional studies to identify the microbial community members under suchferruginous conditions have been conducted in known ferruginous lakes [47, 48, 64–73]. Thesedata can be further used to inform putative metabolic models of the microbial community (Fig.1.4) [48]. Most of these studies have identified a large consortium of anoxygenic phototrophs ofthe class Chlorobia that, based on laboratory experiments with Chlorobia isolates (Table 1.1), areputative photoferrotrophs. While many of these have yet to be confirmed through laboratorygrowth experiments, photoferrotrophy was confirmed in an isolate from the dominant group ofChlorobia in the Kabuno Bay chemocline (Chlorobium phaeoferrooxidans strain KB01 — Table 1.1).To date, therefore, Kabuno Bay is the only known ferruginous water column with a dominantpopulation of confirmed photoferrotrophs likely acting as primary producers. Thus, while studiesin modern ferruginous lakes are emerging [47, 48, 64–73], there is little information on the ecologyof photoferrotrophs in modern microbial communities that could inform on how communityinteractions impact iron and carbon cycling in such ferruginous environments, modern or ancient.Chemocline hvFe(II)Fe(OH)3CO2OMH2OO2SO42-H2SH2H2VFAVFAFe(II) CO2 CH4Fe(OH)3OMCO2CO2Photoferrotrophs (C. ferrooxidans)Sulfate-reducers (Desulfobacca)Fermenters (Streptococcus)Fe(III)-reducers (Rhodoferax)Methanogens (GOM_Arcl)Figure 1.4: Proposed metabolic model for the microorganism present at the chemocline in Kabuno Bay. Potentialpopulations of microorganisms are suggested for each metabolism in the white boxes, hv denotes light, OMis organic matter, and VFAs are volatile fatty acids. This figure is adapted from Lliros et. al., 2015 [48].121.3 Photoferrotrophy during the Archean EonThe sedimentary rock record hosts a wealth of information regarding past ocean environments andthe large-scale changes in Earth’s surface chemistry that have occurred through time. Sedimentaryrocks contains clues to the chemical composition of the oceans and atmosphere in the past. Notonly are these deposits useful archives of Earth’s history, but they also represent economicallyviable sources of key elements required for the maintenance and growth of human society. Forexample, banded iron formations (BIFs) are the world’s largest iron ore deposits, which support asubstantial amount of global steel production [4, 74, 75]. BIFs are found world-wide with severaldeposits concentrated in specific regions where Archean and Proterozoic strata still exist, such asthe north of Western Australia, the west of South Africa, as well as northeastern North America(Fig. 1.5) [4]. While the size, age, and preservation of each of these deposits varies, they havebeen sought after, mapped, and geochemically analyzed due, primarily, to their economic value[4, 74, 75]. As a marine sedimentary deposit, BIFs can provide information that can be usedto constrain models of Precambrian seawater and atmospheric composition, and for trackingdynamics in large-scale Earth processes such as climate change over time.4.5 4 3.5 3 2.5 2 1.5 1 0.5 0Years (Ga)BIF Abundance (relative to Hammersley Group)Hamersley Group, W. AustraliaTransvaal Supergroup, S. AfricaIsua, W. GreenlandCarajás Formation, Brazil Lake Superior Region, USARapitan Group, CanadaUrucum, BrazilYilgarn Block, W. AustraliaMinas Group, BrazilGriqualand West, S. AfricaLabrador Trough, CanadaFigure 1.5: The relative abundance of BIFs versus time (compared to the Hamersley Group BIF volume) where severalof the major BIF regions are identified. Adapted from Klein, 2005 [4], where further formations are delineated.Banded iron formations were deposited between 3.8 and 0.6 Ga with the majority beingdeposited during the Archean Eon between 3.5 and 2.5 Ga (Fig. 1.5) [4, 74, 75]. Many BIFs have13very similar bulk chemistry and features which distinguish them from other iron-rich sedimentaryrocks, such as ferruginous shales, that were deposited contemporaneously [4]. They contain highweight percent iron (20 to 40 wt. % total Fe) with the remaining balance comprising SiO2 (43 to56 wt. %) and minor amounts of other elements such as calcium or magnesium [4]. Furthermore,the iron and silica concentrations typically vary antithetically, which results in banding. BIFsare thus defined as: “a chemical sediment, typically thin bedded or laminated, whose principal chemicalcharacteristic is an anomalously high content of iron, commonly but not necessarily containing layers ofchert [76]”. While the bulk chemistry of most BIFs is similar, the individual characteristics ofeach deposit vary depending on depositional processes and/or post-depositional alteration [4].The majority of BIFs have undergone appreciable metamorphic alteration, which changes theirmineralogical composition and makes it difficult to discern the primary mineralogy. Extensivework, however, on the less altered Archean BIFs, such as the Hamersley BIF in the Pilbara regionof Western Australia, have led to several hypotheses for their mode of formations [4, 77, 78]. Whilehotly debated, the general consensus is that BIFs were deposited when large amounts of seawaterFe(II) were oxidized to ferric iron minerals of varying composition [4, 6, 7]. Other research hassuggested, however, that ferrous iron minerals, and silicates in particular, were in fact the primaryminerals that contributed to BIF deposition, with post-depositional alteration of such primaryphases to secondary Fe(III) (oxyhyr)oxides [79–81].The first attempt to account for the formation of BIFs, postulated by Preston Cloud in early1970’s, involved the oxidation of seawater Fe(II) by molecular oxygen produced through oxygenicphotosynthesis, which had accumulated in the oceans and atmosphere following the evolutionof oxygenic photosynthetic prokaryotes [6]. This hypothesis was based on analyses of a subsetof BIFs, which contained putative nannofossils of eukaryotic algae. Although the so-calledmicrofossils were misidentified as organic remains, Cloud’s initial hypotheses remain partiallyvalid as there is now abundant evidence in the rock record for the presence of molecular O2in Earth’s atmosphere as early as 3.0 Ga [16, 31, 32]. Furthermore, genomic evidence concurswith the sedimentary rock record, suggesting that oxygen-producing cyanobacteria likely evolvedduring the Archean Eon [10–12]. Despite the likely early evolution of oxygenic photosynthesis andproduction of O2 during the Archean Eon, however, further work using proxies has shown that14the atmospheric oxygen content remained very low (less than 0.001 % of current atmospheric O2)throughout the majority of BIF deposition (Fig. 1.6) [2, 13–15, 18, 19]. These low O2 concentrationssuggest that oxygenic cyanobacteria were restricted to small niche environments on land or innutrient-rich coastal regions, limiting their contribution to global primary production and theflux of O2 to the atmosphere [11, 16, 17, 82]. Furthermore, recent modeling, constrained by therock record and putative ocean and atmosphere O2 concentrations, demonstrate that the ArcheanO2 concentrations could only support a limited rate of abiotic oxidation of Fe(II) [8, 15, 22].Collectively these studies present evidence that largely rules out O2 driven Fe(II) oxidation as theprimary model for BIF deposition in the Archean Eon [8, 15, 22].GOEPO2 PALDepositional fractiona.b.00.51.01.52.02.53.03.5Time (Ga)0.10.20.30.410-410-2100Archean Eon Proterozoic Eon PhanerozoicEonFigure 1.6: The abundance of BIFs compared to the concentration of atmospheric oxygen over time. Oxygenconcentration (a) in the atmosphere relative to the modern oxygen concentration through time when theline represents the canonical oxygen curve and the dashed line represents alternative views, adapted fromLyons et. al., 2014 [28]. BIF deposition (b) through time represented as the depositional fraction of the rockrecord at the time of deposition, adapted from Isley and Abbott, 1999 [83].Three alternative mechanisms for Fe(II) oxidation have been proposed since Cloud’s seminalwork—anoxic iron oxidation through UV photolysis, the chemical precipitation and subsequentoxidation (via oxygen once it had accumulated in the atmosphere and oceans) of Fe(II) ferroansilicate phases such as greenalite, and anoxygenic photosynthesis. Both the photochemical oxida-15tion and precipitation of primary Fe(II) minerals could have occurred irrespective of biologicalactivity and have thus been proposed as alternatives to biologically mediated Fe(II) oxidation. Thephotochemical oxidation of Fe(II) has been shown to occur at rates that would indeed have beensufficient to deposit large BIFs [38, 39]. The initial laboratory experiments conducted to test thishypothesis, however, were conducted under conditions that were not analogous to those of theArchean Eon oceans. For example, unlike in modern oceans, the concentrations of bicarbonate[84] and silica [85] were likely present in high concentrations in Archean seawater. Subsequentexperiments using similar laboratory, Archean-like conditions have indicated much slower ratesof photochemical Fe(II) oxidation as the high concentrations of silica and bicarbonate would havefavoured the deposition of ferrous iron silicates and carbonates at rates that grossly exceed thoseof photochemical iron oxidation [40]. While there is petrographic evidence for primary Fe(II)minerals, such as such as greenalite, in some BIF deposits [79, 81, 86, 87], there is no evidenceto support the subsequent transformation of these Fe(II) minerals to the predominantly Fe(III)minerals observed in BIFs today. Indeed, in order to reproduce the mineralogical composition ofBIF, these primary Fe(II) minerals would have had to be oxidized by ground water, saturated inoxygen, for several billion years to result in the BIFs that we observe today [88]. It is plausible thatthis mechanism could have played a small role in transforming some primary Fe(II) minerals inBIFs to Fe(III) minerals, but the magnitude of oxidation required to deposit BIFs is probably un-reasonable [88]. Thus, while chemical oxidation could have played a minor role in BIF deposition,biologically mediated photosynthetic Fe(II) oxidation reactions are far more favorable under theanoxic conditions that persisted in the deep oceans for over three billion years of Earth’s historyduring which BIF were formed.Garrels and Perry [7] originally proposed that anoxygenic phototrophic bacteria used sunlightto directly oxidize ferrous iron (photoferrotrophy) instead of oxygen dependent Fe(II) oxidationin Earth’s early oceans. This idea gained considerable traction when Widdel et. al. [20] isolatedthe first photoferrotrophic bacterium (Table 1.1). Anoxygenic phototrophs harness energy fromsunlight to fix carbon dioxide into biomass, while an electron from an inorganic electron donor(e.g. H2, Fe(II), H2S) [1] replaces the one lost from the photosystem to fix CO2 into biomass.Photoferrotrophs acquire this electron from Fe(II), producing Fe(III) as a metabolic by-product16[Eq. 1.1], completing the oxidation of Fe(II) in the absence of molecular oxygen [8, 20, 21].Indeed, laboratory experiments and direct observations from ferruginous water bodies havedemonstrated that these photoferrotrophs are capable of oxidizing ferrous iron at high enoughrates to support the fluxes of Fe(III) (oxyhydr)oxides needed to deposit even the largest BIFs[8, 21, 47, 48]. While photoferrotrophy provides a compelling mechanism for the oxidation ofFe(II) to support BIF deposition in the complete absence of oxygen, large knowledge gaps limitour ability to conclusively diagnose photoferrotrophs as the causative agents in Fe(II) oxidationand BIF deposition from Archean seawater. For example, a defining characteristic of extantphotoferrotrophs is that they associate with the Fe(III) (oxyhydr)oxides that they produce as aby-product of their growth (Fig. 1.3). This association is difficult to reconcile with the very lowconcentration of organic carbon present in BIF facies [75, 89]. Thus, while BIF deposition mayimply photoferrotrophic activity, the known physiology of cultured photoferrotrophs contradictsthis theory.1.4 Problem statementAvailable evidence suggests that photoferrotrophs may have played a key role in coupling thecarbon and iron cycles during the Archean Eon and, in doing so, underpinned global primaryproduction at this time. To date, however, all known photoferrotrophs form a close associationwith the ferric iron metabolites they produce during growth. This association calls into questiontheir involvement in BIF deposition and, notably, their ability to act as primary producers,sustaining the biosphere for millions of years. A lack of quantitative knowledge on the growth ofphotoferrotrophs and the interactions between them and other microorganisms limits our abilityto constrain models that can be applied to the conditions from which BIFs were deposited, aswell as the Archean ocean-atmosphere system as a whole. The lack of organic matter preservedin BIF and the magnitude of these deposits suggests that they could only have been produced ifcells did not associate with the iron (oxyhydr)oxides, since this would result in either microbialiron reduction, precluding BIF deposition, or in the preservation of considerably more organiccarbon in BIFs. Such an association between cell surfaces and ferric iron metabolites would17additionally cause the bacteria to sink in the ocean water column, shuttling them out of the photiczone, which would then halt photosynthetic growth. Additionally, there are very little data onthe putative growth of photoferrotrophs under the physicochemical conditions that were likelyprevalent during the period of BIF deposition, which adds further uncertainty to models in whichphotoferrotrophs act as primary producers and were the agents of BIF deposition.Some of these unknowns can now be addressed, however, through laboratory experiments withChlorobium phaeoferrooxidans strain KB01, which can provide critical information on photoferrotro-phy, generally, and on the interactions between photoferrotrophs and the microbial consortiumthey support. Experiments with Chlorobium phaeoferrooxidans strain KB01 and studies of theArchean analogue environment from which it was isolated, Kabuno Bay, may provide new insightinto key microbe-mineral and microbial community interactions with extensibility to coupledcarbon and iron cycling in the Archean Eon. Detailed quantitative knowledge of the growth ofChlorobium phaeoferrooxidans strain KB01 and its interactions with other microorganisms are neededto constrain models that aim to reconstruct BIF deposition and the Archean ocean-atmospheresystem.1.5 Dissertation overviewThe overall goal of my thesis is to generate new knowledge on photoferrotrophy that can be usedto inform and constrain models of primary production and BIF deposition during the ArcheanEon. More specifically, I aim to:i. Explore the physiology and metabolic capacity of pelagic photoferrotroph Chlorobium phaeofer-rooxidans strain KB01.ii. Elucidate the role of photoferrotrophs, as primary producers, and determine the flow of or-ganic carbon, from primary production to terminal oxidation, within the microbial communityof ferruginous Kabuno Bay.iii. Construct models of BIF deposition and the role of photoferrotrophs as primary producersthroughout the Archean Eon.18Chapter 2: Chlorobium phaeoferrooxidans strain KB01 – a pelagic, Fe(II)-oxidizing, anoxy-genic, photosynthetic bacteriumThe research described herein defines the metabolic potential, physiology, and taxonomyof Chlorobium phaeoferrooxidans strain KB01, the first photoferrotrophic bacterium to have beenisolated from a pelagic environment. Collectively, this information will provide key informationfor all three aims (i, ii, iii).Chapter 3: Nutrient acquisition in C. phaeoferrooxidans strain KB01This chapter examines some of the key nutrient (nitrogen and sulfur) requirements for photofer-rotrophy, and the metabolic potential associated with meeting these requirements, using genomicanalyses, which were, where possible, verified biochemically. Nutrient metabolism was likely akey aspect for ancient primary production under ferruginous conditions. Therefore, the probableantiquity of these metabolisms, in photoferrotrophic Chlorobia, are tested through phylogeneticanalyses (i, iii).Chapter 4: Photoferrotrophy, deposition of banded iron formations, and methane produc-tion in Archean oceansThis chapter elucidates the cell surface chemistry of C. phaeoferrooxidans strain KB01 andother phototrophic strains to assess the interactions between these microorganisms and the Fe(III)(oxyhydr)oxides produced as a by-product of either anoxygenic or oxygenic photosynthesis. Theseresults are then combined with a re-evaluation of the Archean iron budget to model the role ofphotoferrotrophs in BIF deposition (i, iii).Chapter 5: Carbon flow through microbial interactions under ferruginous conditionsThis chapter defines the likely microbial interactions in the ferruginous water column ofKabuno Bay by examining both process rates and the genomic content of the microbial communi-ties, with a particular focus on the photoferrotrophic Chlorobia that dominate the chemocline. Thegenomic information provides a blueprint for the collective metabolic potential of the microbialcommunity in Kabuno Bay, while the process rate data quantitatively tie this metabolic potentialto geochemical outcomes. The microbial interactions through pathways of carbon and iron cycling19will then be used to construct models of microbial interactions that may have played a key role incoupled carbon on and iron cycling during the Archean Eon (ii, iii).Chapter 6: ConclusionChapter 6 presents the overarching conclusions from this body of work and addresses thefuture directions required to further refine models of carbon and iron biogeochemical cycling inthe Archean Eon and the impact on ocean and atmospheric chemistry.Supplementary MaterialsThere are supplementary materials for Chapters 4 and 5.20Chapter 2Chlorobium phaeoferrooxidans strainKB01 – a pelagic, Fe(II)-oxidizing,anoxygenic, photosynthetic bacterium2.1 SummaryAnoxygenic, photosynthetic, ferrous iron oxidizing bacteria (photoferrotrophs) likely supportedthe early biosphere through primary production (PP) prior to the emergence and proliferation ofoxygenic photosynthesis and the oxygenation of the Earth’s surface. Compelling evidence for theactivity of photoferrotrophs in the Precambrian oceans comes from the world’s largest deposits offerric iron – banded iron formations (BIFs). Several extant photoferrotrophs have been broughtinto laboratory culture and their physiology studied to constrain models of primary productionand biogeochemical cycling under ferruginous conditions. The extensibility of information fromthese model organisms to ancient oceans is limited in part due to physiologies adapted to benthiclifestyles. Here we describe the cellular composition, growth kinetics, and nutrient cyclingcapabilities of a novel photoferrotroph enriched and isolated from a modern ferruginous watercolumn. This photoferrotroph (Chlorobium phaeoferrooxidans strain KB01) is distinct from its closelyrelated benthic relative Chlorobium ferrooxidans strain KoFox. They have 99 % identity between 16SrRNA genes with only 70 % identity across entire genomes. Strain KB01 grows at a maximum rateof 48 ± 7 fmol cell– 1 h– 1 under standard laboratory growth conditions with a high specific affinityfor Fe(II) (190,000 L g– 1 wet cell h– 1). These growth rates are higher than those of previouslycharacterized photoferrotrophs and other Fe(II) oxidizing organisms more generally. Strain KB0121has adaptations for growth under low light intensities (0.03-0.05 µmol m– 2 s– 1) and has a highspecific affinity for light (∼1,410,000 m2 s g– 1 wet cells h– 1) likely related to extraordinary per cellconcentrations of photosynthetic pigments (BChl e of 1.5 ± 0.3 pg). While many similar traits canbe found in other members of the class Chlorobia, or photoferrotrophs from other lineages, thiscombination of traits in Chlorobium phaeoferrooxidans strain KB01 reveals that they are collectivelybiologically and ecologically compatible with each other and growth through photoferrotrophy ina pelagic habitat. By extension, such a suite of traits could have been critical to underpinning BIFdeposition, PP, and coupled carbon and iron cycling in Precambrian Eons.2.2 IntroductionPhotosynthesis fuels the biosphere by using light energy to transform inorganic carbon to organiccarbon in biomass and, on the modern Earth, photosynthesis is responsible for nearly all of globalprimary production (PP) [90, 91]. In modern aquatic environments, PP is conducted mainly byphotosynthetic microorganisms that are either oxygenic, like cyanobacteria and eukaroytic algae[91, 92], or anoxygenic, such as the Chlorobia, Chloroflexi, and some Proteobacteria [93]. Oxygenicphotosynthesis by such microorganisms drives a large fraction (∼50 %) of global PP today [90, 91],and has likely contributed to PP since the early Archean Eon, some 3 billion years ago [16, 92, 94].Anoxygenic photosynthesis, on the other hand, likely sustained the biosphere before the evolutionand proliferation of oxygenic photosynthesis and continued to play a key role in PP throughoutthe Precambrian Eons [1, 11, 35, 93], but is now mostly marginalized to the dispersed and oftenephemeral pockets of illuminated anoxia that occur on the Earth’s surface today [93]. Multiplemodes of photosynthesis using a diversity of electron donors have played exclusive, concurrent,and even synergistic roles in PP and thus in fueling the biosphere throughout Earth’s history[1, 47, 48, 95, 96]. In so doing, they have coupled the carbon biogeochemical cycle to the cycles ofseveral other elements, including oxygen, sulfur, and iron [1].Both oxygenic and anoxygenic phototrophs support biological production by harnessingenergy from sunlight to excite electrons, supplied mostly by inorganic electron donors, to drive theendergonic conversion of inorganic carbon dioxide (CO2) into organic carbon [91]. Unlike oxygenic22phototrophs, whose virtually unlimited supply of electrons is supplied by water, anoxygenicphototrophic bacteria use electrons from both inorganic and organic substances that have loweractivities than water in aquatic environments (e.g. sulfide (HS– ), ferrous iron (Fe[II])), and tendonly to accumulate to appreciable activities under anoxic conditions [91, 93]. The photosyntheticmachinery required to excite these electrons, and to subsequently transport them down an electronpotential gradient to generate energy, likely stems from common ancestral genes and has featuresthat are conserved across most photosynthetic lineages [11, 35, 92, 97]. For example, there isstructural homology in the photosystem I found in both oxygenic and anoxygenic phototrophs,which suggests common ancestry despite differences in the underlying gene sequences [98]. Thereis, however, considerable diversity in the nature and configuration of proteins used in harnessinglight energy and the types of proteins used in the transport of electrons from an electron donor toinorganic carbon [35], which translate to dramatic differences in the physiology across lineages.While well over 100 oxygenic and anoxygenic phototrophs have been brought into laboratorycultures (DSMZ and ATCC culture collections) – enabling detailed biochemical, genomic andphysiological studies – cultures of anoxygenic phototrophs that specifically use Fe(II) as theirelectron donor (photoferrotrophs) are fewer than 20 [57]. Thus, while studies of these culturedphotoferrotrophs have provided a wealth of information on their metabolism, our knowledge ofthe potential physiological diversity in photoferrotrophy is limited compared to the much betterstudied oxygenic and sulfur-dependent anoxygenic phototrophs.Motivation for learning more about photoferrotrophy comes, in large part, from their likelydisproportionate role in supporting PP and driving fluxes of matter and energy at global scalesthroughout Earth’s early history. Prior to the advent of oxygenic photosynthesis, PP throughanoxygenic photosynthesis would have played a crucial role, for example through climate regula-tion, in sustaining the biosphere [1, 95, 99]. Anoxygenic photosynthesis, almost certainly evolvedprior to oxygenic photosynthesis [11, 35, 93], and anoxygenic photosynthesis would likely havesupported a biosphere nearly 2 orders of magnitude more productive than a biosphere devoidof photosynthesis altogether [1]. Earliest evidence for photosynthesis comes from microfossilsand isotopic signatures for carbon fixation from as early as 3.8 Ga [100–102]. Elemental budgetsfor the Precambrian Eons imply that Fe(II) would have been the most available electron donor23for anoxygenic photosynthesis, and thus photoferrotrophy (Eq. 2.1) is widely regarded as thedominant mode of primary production through Earth’s early history [1, 7, 21, 82, 103]. Earlyestimates based on these elemental budgets implied that photoferrotrophy could have supportedPP up to about 10 % of modern PP, or 170 to 500 Tmol carbon yr– 1 [1]. Revised estimates suggest,however, that global PP would have been closer to 1 % modern [99].4 Fe2+ +CO2 + 11 H2O+ hv⇒ [CH2O] + 4 Fe(OH)3 + 8 H+ (2.1)Geological evidence for the role of photoferrotrophs in PP during the Precambrian Eons comesfrom the preservation of banded iron formations (BIFs), which are enormous sedimentary depositsof iron minerals. Deposition of BIFs was initially attributed to the oxygenation of Precambrianseawater through oxygenic photosynthesis, and subsequent abiotic or biological oxidation of Fe(II)to drive deposition Fe minerals at the seafloor [6]. More quantitative considerations, however, nowsuggest that oxygen concentrations and likely production rates were, in fact, too low to supportthe rates of Fe deposition needed to form BIFs [2, 8, 15, 22, 82]. Photoferrotrophy, therefore, isthe most likely process to support Fe oxidation at the rates needed to deposit BIF [8, 21], at leastin the Archean Eon, prior to the widespread oxygenation of the atmosphere ocean system some2.3 Ga [2, 104, 105]. Furthermore, while oxygenic photosynthesis likely emerged early in theArchean Eons [16, 92, 94], competition for phosphate between nascent oxygenic phototrophs andphotoferrotrophs would have restricted the proliferation of the former until Fe(II) availabilitydeclined, likely around 2.3 Ga [85, 96, 106]. Thus, not only did photoferrotrophy likely play a keyrole in sustaining the early biosphere, it also throttled the pace of Earth’s oxygenation throughcompetition with early oxygenic phototrophs [85, 96, 106]. While such roles for photoferrotrophyin the evolution of the Earth system are well supported by available physiological and geologicalinformation, as well as models that integrate this information into quantitative frameworks,additional information on the metabolic potential and evolutionary history of photoferrotrophywould facilitate more nuanced and robust reconstructions of Earth’s early history.A number of photoferrotrophs have been brought into laboratory culture and studies ofthese cultures have progressively improved our knowledge of photosynthetic iron oxidation and24enabled evaluation of the possible role of photoferrotrophs in BIF deposition and PP throughoutEarth’s history. Early recognition, by geologists, that the oxidation of Fe(II) from low-oxygenor even anoxic seawater was likely needed to support BIF deposition led to the original ideathat Fe(II) could be directly oxidized as the electron donor in anoxygenic photosynthesis [7].The first microorganisms that showed photosynthetic growth with Fe(II) as the electron donor,Rhodomicrobium sp. and Thiodictyon sp., were enriched and isolated from a freshwater pond andmarine sediments in Germany [20]. This was the first time that photoferrotrophy was deemedbiologically and biochemically possible [20]. At the same time, experiments with these firstphotoferrotrophic isolates revealed that Fe(II) is stoichiometrically oxidized at a 4:1 ratio withCO2 fixed (Eq. 2.1) into biomass thereby directly linking, for the first time, photosynthetic Fe(II)oxidation to biological production [20].Since this discovery of photoferrotrophy in extant bacteria, a number of additional bacteriawith capacity to grow through photosynthetic Fe(II) oxidation have been enriched and isolatedfrom both marine and freshwater environments [52, 53, 55–57]. These isolates come from threedifferent classes of bacteria (Alphaproteobacteria, Gammaproteobacteria, and Chlorobia), areclosely related to other anoxygenic phototrophic members of those classes, and demonstratepolyphyly of the photoferrotrophic trait [35]. Laboratory studies of these isolates reveal severalcharacteristic features of photoferrotrophs such as a general ability to grow using alternativeelectron donors (i.e. organic carbon compounds, H2S, and H2) and a tendency to grow throughFe(II) oxidation with close physical association between cell surfaces and the iron (oxyhydr)oxidegrowth products [9, 53, 55–57, 62, 63]. Notably, laboratory studies have also shown that someextant photoferrotrophs are capable of oxidizing Fe(II) at rates that are high enough to supportdeposition of even the largest of the Precambrian BIFs [9, 21, 42, 55]. Data from such studiesunderpins conceptual and quantitative models of BIF deposition and coupled iron and carboncycling in Earth’s early oceans [21, 42, 82, 95, 96, 99, 107–109]. All but one of the photoferrotrophsenriched and isolated to date, however, originate from benthic environments [57], and basedon analogy to other photosynthetic bacteria, this suggests that most existing information isonly partly extensible to water column processes like BIF deposition. For example, previousstudies of benthic and pelagic cyanobacteria have shown that even different species of the same25genus occupy different niche environments (benthic versus pelagic) [110], which is likely dueto physiological differences, such as extracellular and cell-surface layers and appendages thatallow the benthic microorganisms to remain in their preferred environment through attachment[111–113]. Inferences related to the biogeochemistry of the Precambrian oceans based on thephysiology of benthic photoferrotrophs may not, therefore, be entirely accurate and there is a clearneed for new knowledge from more appropriate analogue organisms derived from ferruginouspelagic environments.Chlorobium phaeoferrooxidans strain KB01 was enriched and isolated from the water columnof an Fe(II)-rich lake [48] and, since it comes from a pelagic habitat, arguably represents thebest modern analogue for the photoferrotrophs that likely underpinned BIF deposition, PP, andcoupled carbon and iron cycling in the Precambrian oceans. Strain KB01 was isolated from theferruginous waters of Kabuno Bay, a permanently stratified sub-basin of Lake Kivu, located inEast Africa on the border of the Democratic Republic of Congo (DRC) and Rwanda. On-goingvolcanic activity in the area surrounding Lake Kivu releases carbon dioxide and methane gasinto the lake and its sub-basin, maintaining anoxic bottom waters [114]. Furthermore, groundwater enriched in Fe(II) continually feeds Kabuno Bay creating a chemocline that juxtaposes thelower part of the photic zone with Fe(II)-rich water. Strain KB01 was originally referred to asChlorobium ferrooxidans strain KB01 based on analyses of its 16S rRNA gene, which is 99 % similar,and thus indistinguishable by conventional definitions of species-level classification, to benthicphotoferrotroph Chlorobium ferrooxidans strain KoFox [48]. Subsequent whole genome sequencing,however, revealed that the strains were only 70 % identical at the genome level thus justifying thedesignation of strain KB01 as its own species and its renaming as C. phaeoferrooxidans strain KB01,accordingly [58]. In this paper, we provide the detailed and systematic description of strain KB01supported by ecological, phenotypic and genotypic information. We further use this informationto support the ideas that strain KB01 is a strong analogue of Precambrian photoferrotrophs andthat its physiology and metabolic potential is extensible to conceptual and quantitative models ofBIF deposition, PP, and coupled carbon and iron cycling in the Precambrian Eons.262.3 Materials and methods2.3.1 Strain and growth mediumChlorobium phaeoferrooxidans strain KB01 was enriched and isolated from the chemocline of fer-ruginous Kabuno Bay, a sub-basin of Lake Kivu in East Africa [48] and was subsequently grownin serum bottles containing an anoxic (80:20 N2:CO2) atmosphere and standard ferrous iron(Fe[II]) – containing growth media as described by Hegler et al., 2008 [55, 115]. The standardgrowth media [55] was prepared and allocated into serum bottles (100 mL media, 160 mL totalvolume), with 0.3 g L – 1 NH4Cl, 0.5g L– 1 MgSO4·7H2O, 0.1 g L – 1 CaCl2·2H2O, 0.6 g L – 1 KH2PO4.After autoclaving, 22 mmol L – 1 bicarbonate was added along with trace elements (nonchelatedtrace element mixture), mixed vitamin solution, selenate-tungstate, vitamin B12 (after chapter183 by Widdel and Bak [116]). Finally, FeCl2 was added and the pH was adjusted to 6.8-6.9under an N2/CO2 atmosphere (80:20). 10 mmol L– 1 FeCl2 was utilized for the standard mediaconditions. The 10 mmol L – 1 media was also filtered after being made to remove any precipitatesto confirm that media precipitates did not change the overall growth rates, which resulted in afinal Fe(II) concentration of between 2 and 4 mmol L – 1 for the standard media (results from theseexperiments not shown). All experiments were conducted in triplicate and sub-samples weretaken to track the growth of strain KB01 as described below.2.3.2 BioinformaticsrRNA gene comparison The full length 16S rRNA gene for Chlorobium phaeoferrooxidans strainKB01 was retrieved from Lliro´s et. al. [48]. Other 16S rRNA sequences were retrieved frommembers of the class Chlorobia from the Silva online database — version 128 [117, 118]. Onlyfull-length (> 1400 bp) sequences were selected, and then aligned using the default parametersin the package software ClustalX2.1 [119]. To rigorously test the evolutionary history of strainKB01 within the class Chlorobia multiple tree construction methods (Maximum likelihood (ML),Maximum parsimony (MP), and Neighbour Joining (NJ)) were employed. ML, MP, and NJtrees were constructed in MEGA version 7 [120, 121] and all trees bootstrapped 500 times. Thephylogeny shown is from the most robust tree construction method, ML, with the bootstrap values27indicated at the nodes. The bootstrap values for the alternative tree construction methods are alsoshown at the nodes (MP and NJ) and the legend (0.03) delineates the nucleotide substitutions persite over the indicated distance (Fig. 2.3a).Full genome analyses The strain KB01 genome was subjected to Illumina paired-end libraryconstruction and sequenced (MiSeq platform, version 3 chemistry) to generate 16,133,652 paired250-nucelotide (nt) reads as reported by Crowe et al., 2017 [58]. The quality-filtered reads weresubsequently assembled into 116 contigs using the default settings in ABySS 1.3.5 [58, 122]. Thefull genome for Chlorobium ferrooxidans strain KoFox (90.71 % completed) was retrieved fromthe National Center for Biotechnology Information (NCBI – https://www.ncbi.nlm.nih.gov/)under the accession number NZ AASE00000000.1. To identify regions with identical nucleotidesequences in strain KB01 and strain KoFox, the contigs of both strains were aligned and com-pared using MUMmer (nucleotide comparison with nucmer [123]) and visualized using Circos(http://circos.ca/) — Fig. 2.3b. To delineate the likelihood that strain KB01 is a different speciesfrom strain KoFox, in silico DNA-DNA hybridization was conducted utilizing the Genome-to-Genome Distance Calculator (http://ggdc.dsmz.de/) [124]. To compare the similarities anddifferences between the higher level pathways in strain KB01 and strain KoFox, open readingframes (ORFs) for both strains were predicted using Prodigal [125]. The predicted ORFs, for bothstrains, were subsequently annotated using the online tool Kofam Koala [126] that uses HiddenMarkov-Models (HMMs) to compare each ORF to those found in the Kyoto Encyclopedia of Genesand Genomes database (KEGG-11-06-18) [127]. The resulting annotations were compared at thepathway level and visualized through a python script as described in Graham et. al., 2018 [128] -–Fig. 2.3c. Other genes — those that are uncharacterized in the KEGG database — were found byretrieving the relevant gene or protein sequences from the NCBI, Pfam [129], or CAZy databases[130]. The sequence or HMM of the gene/protein of interest was then used to search for thosegene or proteins within nucleotide and protein predicted ORFs of the strain KB01 genome usingBLAST (initial e-value cut-off of 1 x 10 – 18, followed by a Bit Score cut-off of 50 [131]) or HMMER’shmmserach function (e-value cut-off of 1 x 10 – 18 and length cut-off of 50 % [132]).282.3.3 Analytical techniquesTo track the growth of Chlorobium phaeoferrooxidans strain KB01 under iron-rich conditions, sub-samples were taken from the serum bottles at the time of inoculation and every day (or asotherwise indicated) for three to four weeks. The first of these sub-samples was analyzed forFe(II) and Fe(III) concentrations using spectrophotometric methods. Specifically, Fe(II) and Fe(III)concentrations were determined by the ferrozine method and samples were measured directly aswell as after being fixed in 1 N HCl — after Voillier et. al., 2000 [133]. Additional sub-sampleswere taken from these same serum bottles to measure the pigment concentrations. Pigmentscontents were assayed spectrophotometrically after 24 hour extractions of 1 mL of pelleted cells inacetone:methanol (7:2 v/v) [134]. A full pigment spectrum (from 400 to 700 nm — Fig. 2.2b) wasmeasured for strain KB01. Subsequent samples were measured at 468 nm (BChl e soret peak) and652 (BChl e Qy peak). Pigment concentrations were then used as a proxy for cell abundance. Tofurther confirm the growth of strain KB01, sub-samples were taken from the serum bottles andcells were fixed in gluteraldehyde (final concentration of 0.1 %). After the cells were fixed, theywere subsequently stained with SYBR green (0.25 % final concentration) and directly counted ina 96 well plate using a Miltenyi Biotec MACSQuant, with a flow rate of medium. Pigment andcell count sub-samples were also used to track the growth of strain KB01 in media containingalternative electron donors and acceptors.2.3.4 MicroscopyLight microscopy Chlorobium phaeoferrooxidans strain KB01 was grown in standard growth media(10 mM Fe(II), 4.41 mM phosphate), the serum bottles were gently shaken, and then the Fe(III)(oxyhydr)oxides were allowed to settle for 24 hours. Sub-samples, taken at multiple time pointsthroughout the full 200 hours of growth, were collected from the fully mixed serum bottle aftershaking and from the water column once the Fe(III) (oxyhydr)oxides had settled. Wet mounts(∼100 µL) were prepared immediately following the sub-sampling and were imaged using a ZeissAxio Observer compound microscope fitted with an Axiocam camera. A representative image ofa water sub-sample of strain KB01 is shown in Fig. 2.1a.29Scanning electron microscopy (SEM) Cells for SEM were grown up to late-log, early-stationaryphase in standard growth media (10 mM Fe(II), 4.41 mM phosphate). The cell-Fe(III) oxyhydroxidesuspensions were gently shaken and allowed to settle for 24 hours. A 1 mL sub-sample wascollected from either the upper portion of the serum bottle, avoiding settled Fe(III) oxyhydroxides,or from the mixed suspension after gentle shaking. The samples were placed on a NucleoporeTrack-Etched Membrane from Whatman c©. The cells were then fixed with 2.5 % gluteraldehydebuffered with 0.1 M PIPES at pH of 7.4 for 5 minutes. The external cellular structures werepreserved using a 1 % osmium tetroxide solution buffered with 0.1 M PIPES at pH 6.8 for 1 hour.Filters were rinsed gently with MQ water and then dried using an ethanol dehydration series. Thefilters were critical-point-dried using a Samdri795 from Toosimis Research Corporation. Finally,the filters were attached to a stub and coated with 5 nm of iridium to ensure conductivity. Thefilters were imaged on a Helios SEM. Fe(III) oxyhydroxides and cells were confirmed throughenergy-dispersive X-ray spectroscopy (EDS) determinations of carbon and iron abundances andmultiple points were measured for each surface found. SEM images are shown in Fig. 2.1b, c ofthe main text.Transmission electron microscopy (TEM) 10 mL of strain KB01 cells were concentrated bycentrifugation at 5000 rpm for 10 minutes. The concentrated 1 mL of cells was subsequentlyspun at 10 000 rpm for 5 minutes. The final 30 µL cells were then frozen rapidly using a LEICAEM HPM100. The cellular water was replaced with an alcohol mix while in liquid nitrogen;the samples were sliced as thin sections from an epoxy block and placed onto a copper grid.The samples were imaged on an Osiris S/TEM at the 4D labs imaging facility at Simon FraserUniversity. TEM images are shown in Fig. 2.1d, e, f of the main text.2.3.5 Fe and light dependencyFe(II) concentration dependence of Fe(II) oxidation rates To assess the C. phaeoferrooxidansstrain KB01 growth rate versus the Fe(II) concentration, strain KB01 was grown in standard media(as described above) under optimal conditions except that the media was amended with a range ofFe(II) concentrations (5 µM to 200 µM). The media were subsequently allocated into serum bottles30(100 mL of media, 160 mL total volume) with triplicate bottles at the same Fe(II) concentration.Strain KB01 cells were harvested at mid to late log phase, spun down, washed with 0.1 N NaCl,and inoculated into each serum bottle, with final approximate concentration of 7 x 105 cells mL – 1.Sub-samples for Fe(II)/Fe(III), pigments, and cell counts were taken every 20 minutes to trackFe(II) oxidation and cellular growth.Light dependence of Fe(II) oxidation rates To quantify the dependence of Fe(II) oxidationrates on light intensity, strain KB01 was grown in standard growth media (10 mM Fe(II) – asdescribed above) at a range of light intensities. Strain KB01 cells were harvested at mid to latelog phase, spun, down, washed with 0.1 N NaCl, and inoculated into each serum bottle with anapproximately 2 % inoculum. The serum bottles were placed at different distances, in triplicate,from a 60 W incandescent light bulb with the resulting light intensities being: 0.03, 0.1, 0.25, 0.5,0.75, 4, and 15 µmol m – 2 s – 1 (a range of 1 to 654 lux). Light intensities were measured with anLI-250A light meter equipped with a LI-COR Quantum Sensor. Two negative controls were used:one was inoculated with strain KB01 and incubated in the dark, wrapped in tin foil and the otherwas an uninoculated control that was incubated at 4 µmol m – 2 s – 1. Sub-samples for Fe(II)/Fe(III),pigments, and cell counts were taken every two to four days to track Fe(II) oxidation and cellulargrowth.Kinetic modelling and calculations To calculate the Fe(II) oxidation rate over time under boththe standard conditions (described above) and the limiting conditions, a linear regression wasapplied to the data points where Fe(II) consumption was highest (example shown in Fig. 2.4a).The subsequent rate data from each Fe(II) or light concentration (from the Fe(II) and light limitingexperiments) was plotted as a function of Fe(II) concentration or light flux (Fig. 2.5). A non-linearregression was then used to fit a Michaelis-Menten model to the data:R = Vmax ·[S]km + [S](2.2)where R is the reaction rate (µM h – 1), Vmax is the maximum reaction rate (µM h– 1), km isFe(II) concentration (µM) or light flux (µmol m – 2 s – 1) at half the maximum rate (Vmax), and [S] is31the substrate concentration (µM for Fe(II) and µmol m – 2 s – 1 for light). Cell-specific Vmax werecalculated by dividing the volume specific Vmax by the cell counts. Specific affinities for Fe(II) andlight were calculated using the following equation:Sa =(V max·SmwV·D·Cells )km · Smw(2.3)where Sa is the specific affinity, Vmax is the cell specific max reaction rate (mol L– 1 h – 1),Smw is the substrate molecular weight (e.g. gFe mol– 1), V is the cell volume (cm3 cell – 1), D isthe cell density (g cm – 1), Cells is the cell counts (cell L – 1), and km is the substrate at half theVmax (mol L– 1) [135]. The calculations were completed with an average cell density (1.09 g cm – 3[136]), while the average cell volume for strain KB01 cells was ∼0.17 µm3. Further calculationsincluded elucidating the cell-specific BChl e content of the strain KB01 cells. This was executed bytransforming the pigment absorption to concentrations by using the molar extinction coefficientfor BChl e in Acetone:Methanol (7:2), 41.4 ± 0.7 L mmoles – 1 cm – 1 [137] or 58.6 L g – 1 cm – 1 [138],and dividing the concentrations by the cell counts (cells mL – 1).2.3.6 pH, temperature, and vitamin dependencepH dependence of Fe(II) oxidation rates To quantify the dependence of Fe(II) oxidation rateson pH, strain KB01 was grown in standard growth media (10 mM Fe(II) – as described above) at arange of pH. Final pH values were controlled to 6, 6.4, 6.8, 7, and 7.3 with either 1 N HCl or 1 NNaOH. Strain KB01 cells were harvested at mid to late-log phase, spun down for 5 minutes at10000 x g, washed with 0.1 N NaCl, and inoculated into each serum bottle with an approximately2 % inoculum. Each pH value was tested in triplicate, along with an uninoculated negative controlto confirm a lack of abiotic Fe(II) oxidation. Sub-samples for Fe(II)/Fe(III), pigments, and cellcounts were taken every two to four days to track Fe(II) oxidation and cellular growth.Temperature dependence of Fe(II) oxidation rates To quantify the dependence of Fe(II) oxida-tion rates on temperature, strain KB01 was grown in standard growth media (10 mM Fe(II) – asdescribed above) at 4, 15, 23, 25, 30◦C. The cells were harvested at mid to late log phase, spun,32down, washed with 0.1 N NaCl, and inoculated into each serum bottle with an approximately2 % inoculum. All temperatures were tested in triplicate, along with an uninoculated negativecontrol to confirm a lack of abiotic Fe(II) oxidation. Sub-samples for Fe(II)/Fe(III), pigments, andcell counts were taken every two to four days to track Fe(II) oxidation and cellular growth.Vitamin dependence of Fe(II) oxidation rates To quantify the dependence of Fe(II) oxidationrates on the presence of each bio-essential vitamin, strain KB01 was grown in standard growthmedia (10 mM Fe(II) – as described above) with the regular suite of vitamins and in standardgrowth media where each of the vitamins had been individually omitted. The individual vitaminsthat were tested were: aminobenzoic acid, calcium, nicotinic acid, B12, biotin, pyridoxin. StrainKB01 was also tested in media that did not contain any of the listed vitamins. Strain KB01 cellswere harvested at mid to late log phase, spun, down, washed with 0.1 N NaCl, and inoculated intoeach serum bottle with an approximately 2 % inoculum. All media composition were tested intriplicate, along with an uninoculated negative control to confirm a lack of abiotic Fe(II) oxidation.Sub-samples for Fe(II)/Fe(III), pigments, and cell counts were taken every two to four days totrack Fe(II) oxidation and cellular growth.2.3.7 Alternative electron donors and acceptorsTo test for strain KB01’s ability to grow on alternative electron donors and acceptors, strain KB01was inoculated into standard growth media (as described above) that had been amended with adiverse set of electron donors and acceptors. All of the organic compounds tested were amendedto the media with a final concentration of 4 mM, while 10 mL of H2 gas (100 %) was added tothe headspace of the serum bottles every second day. Strain KB01 cells were harvested at midto late log phase, spun, down, washed with 0.1 N NaCl, and inoculated into each serum bottlewith an approximately 2 % inoculum. Sub-samples for pigments and cell counts were taken everytwo to four days to track cellular growth. All media composition were tested in triplicate, alongwith an uninoculated negative control to confirm growth. To test whether strain KB01 could growphototrophically with an alternative electron donor, strain KB01 was inoculated into standardmedia amended with glucose or acetate or lactate or unamended with H2 in the headspace and33these serum bottles were grown under light saturation (15 µmol m – 2 s – 1). To test whether strainKB01 could ferment organic carbon compounds, strain KB01 was inoculated into standard mediaamended with glucose or acetate or lactate and these serum bottles were wrapped in tin foil in thedark. To test whether strain KB01 was capable of aerobic respiration, strain KB01 was inoculatedinto standard media, in Erlenmeyer flasks, amended with glucose or acetate or lactate or pyruvate.2.4 Results2.4.1 Morphology and pigmentationChlorobium phaeoferrooxidans strain KB01 (hereafter referred to as strain KB01) exhibits internaland external cellular characteristics that typify anoxygenic phototrophic Chlorobia. It is relativelysmall in cell size (length of ∼600 nm, width of ∼300 nm) with a rod shape that does not visiblyexhibit any of the external machinery required for motility, such as flagella and pilli (Fig. 2.1).Strain KB01 divides through binary cell division (Fig. 2.1b) and has few distinctive cell surfacefeatures (Fig. 2.1b, c). Transmission electron microscopy (TEM) reveals internal structures such asthe chlorosomes—large oval structures bound to the inner membrane that house the pigmentsneeded to capture light and are characteristic of phototrophic members of the class Chlorobia (Fig.2.1d, e). Ribosomes are clearly visible (Fig. 2.1d, e) along with a well-defined periplasmic space(Fig. 2.1d, e, f), which is similar to those of other members of the class Chlorobia [139]. Membersof the class Chlorobia utilize a combination of pigmented carotenoids and bacteriochlorophyllsto harness light energy [140–142]. Spectrophotometric measurements indicate that strain KB01pigmentation includes a carotenoid (smaller peak at about 450 nm) and bacteriochlorophyll e(BChl e) with peaks at 468 nm – BChl e soret peak – and 652 nm – BChl e Qy peak (Fig. 2.2b;[48]), which manifests as a red-brown coloration, visible in cell suspensions (Fig. 2.2a). Understandard growth conditions (as described below), volume specific pigment concentrations roseto approximately 1060 nmol L– 1 (equivalent to 750 µg L– 1; Fig. 2.4b), which, when compared tocorresponding cell counts 2.4, yields an average per cell BChl e content of 2.1 ± 0.5 fmol or 1.5 ±0.3 pg. Strain KB01 thus exhibits many of the physical properties, internal features, and overallpigmentation that typify the phototrophic Chlorobia.34250 nm 250 nm5 µmAB CD E FThompson et al Figure 1Figure 2.1: Microscopy images of strain KB01 with optical microscopy (A), SEM (B and C), and TEM (D, E,and F). Note that panel F is a close-up view of the black box in panel E demonstrating a rare instance ofFe(III) particles near the cell wall of strain KB01.400 450 500 550 600 650 700Wavelength (nm)Qy  A BCarotenoid (450 nm)Soret (468 nm)(652 nm)Figure 2.2: Strain KB01 pigmentation with a photo of strain KB01 demonstrating its brown/purple pigments(A) and graphical representation of relative abundance of the pigments of C. phaeoferrooxidans strain KB01(B).352.4.2 Phylogeny and metabolic potentialStrain KB01 is phylogenetically closely related to Chlorobium ferrooxidans strain KoFox (grown inco-culture with Geospirillum sp. KoFum [53]), but has sufficiently different metabolic potentialcoded in its genome to define it as a separate species. The whole genome of strain KB01 has beensequenced and deposited in GenBank with the accession number MPJE00000000 [58]. Phylogeneticanalyses of the Chlorobia indicate that strain KB01 falls within the Chlorobium genus and is mostclosely related to another Chlorobia photoferrotroph: Chlorobium ferrooxidans strain KoFox (Fig.2.3a). Indeed, the 16S rRNA gene from strain KB01 was 99 % identical to that of strain KoFox.To test whether strain KB01 was sufficiently different from strain KoFox at the genome levelto be classified as its own species, the whole genomes of the two organisms were comparedin silico. At the nucleotide level, the genome of strain KB01 is 72 % identical to strain KoFox(Fig. 2.3b), while in silico DNA-DNA hybridization [124] conducted between the two strainsresolved clear differences between two genomes with only 70-79 % DNA-DNA hybridization,which suggests that the two organisms are different species with a probability of greater than 0.5(Table 2.1). The GC content difference between the two species was 0.39 %, but the resolutionof GC content differences is insufficient to distinguish between two species given that differentspecies often have GC contents that are within 1 % of each other (Table 2.1). Despite strongsimilarities in phylogenetic marker genes (16S rRNA gene), the genomes of strains KB01 andKoFox are sufficiently different that strain KB01 should be considered its own species.Table 2.1: Distance metrics using two different formulas between strain KB01 and its closest phylogeneticneighbor, Chlorobium ferrooxidans strain KoFox. Note: DDH stands for DNA-DNA hybridization; HSPstands for high-scoring segment pairs.HSP length/total length Identities/HSP length G+C differenceDDH DistanceProb. DDH >= 70%DDH DistanceProb. DDH >= 70%64.4 ± 3.8 0.2286 60.86 60.4 ± 2.8 0.0468 52.48 0.39360.03Chlorobium luteolum DSM 273Chlorobium limicola DSM 245Chloroherpeton thalassium ATCC 35110Chlorobium ferrooxidans DSM 13031Chlorobium chlorochromatii CaD3Chorobium phaeoferrooxidans KB01Chlorobaculum limnaeum DSM 1677Prosthecochloris aestuarii DSM 271OutgroupChlorobium phaeobacteroides DSM 266Chlorobaculum parvum NCIB 8327Chlorobium phaeovibrioides DSM 265Chlorobaculum tepidum TLSChlorobium phaeobacteroides BS167/ 28/7860/51/5755/42/54100/99/100100/100/10094/97/9950/40/5499/100/100100/100/10031/28/2922/35/35100/100/100Genus Family OrderChlorobiumChlorobaculumProsthecochlorisChloroherpetonChlorobiaceaeChloroheptonaceaeChlorobialesGlycolysisGluconeogenesisTCA_CycleNADH_quinone_oxidoreductaseFtype_ATPaseCytochrome_c_oxidase_cbb3_typeCytochrome_bd_complexRuBisCoCBB_CyclerTCA_CycleHydroxypropionate_3_BicycleHydroxybutyrate_4_hydroxypropionate_3Dgalacturonate_epimeraseBeta_N_acetylhexosaminidaseNitrogen_fixationSulfhydrogenaseSulfide_oxidationNiFe_hydrogenase_Hyd1Thiamin_biosynthesisRiboflavin_biosynthesisCobalamin_biosynthesisPhosphate_transporterChemotaxisCoenzyme_B_Coenzyme_M_regenerationCytochrome_b6f_complexAnoxygenic_typeI_reaction_centerRetinal_biosynthesisEntnerDoudoroff_PathwayLactate_fermentationAcetate_fermentationPEP_to_Succinate_fermentationCompetence_relatedAnaplerotic_genesType_I_SecretionSec_SRPTwin_Arginine_TargetingSerine_pathwayArsenic_reductionKB01KoFoxVar1Var20.250.500.751.00PathwayGlycolysisGluconeogenesisTCA_CycleNADH_quinone_oxidor ductaseFtype_ATPaseCytochrome_c_oxidase_cbb3_typeCytochrome_bd_complexRuBisCoCBB_CyclerTCA_CycleHydroxypropionate_3_BicycleHydroxybutyrate_4_hydroxypropionate_3Dgalacturonate_epimeraseBeta_N_acetylhexosaminidaseNitrogen_fixationSulfhydrogenaseSulfide_oxidationNiFe_hydrogenase_Hyd1Thiamin_biosynthesisRiboflavin_biosynthesisCobalamin_biosynthesiPhosphate_transporterChemotaxisCoenzyme_B_Coenzyme_M_regenerationCytochrome_b6f_complexAnoxygenic_typeI_reaction_centerRetinal_biosynthesisEntnerDoudoroff_PathwayLactate_fermentationAcetate_fermentationPEP_to_Succinate_fermentationCompetence_relatedAnaplerotic genesType_I_SecretionSec_SRPTwin_Arginine_TargetingSerine_pathwayArsenic_reductionKB01KoFoxVar1Var20.25500.751.00PathwayGlycolysisGluconeogenesisTCA_CycleNADH_quinone_oxidoreductaseFtype_ATPaseCytochrome_c_oxidase_cbb3_typeCytochrome_bd_complexRuBisCoCBB_CyclerTCA_CycleHydroxypropionate_3_BicycleHydroxybutyrate_4_hydroxypropionate_3Dgalacturonate_epimeraseBeta_N_acetylhexosaminidaseNitrogen_fixationSulfhydrogenaseSulfide_oxidationNiFe_hydrogenase_Hyd1Thiamin_biosynthesisRiboflavin_biosynthesisCobalamin_biosynthesisPhosphate_transporterChemotaxisCoenzyme_B_Coenzyme_M_regenerationCytochrome_b6f_complexAnoxygenic_typeI_reaction_centerRetinal_biosynthesisEntnerDoudoroff_PathwayLactate_fermentationAcetate_fermentationPEP_to_Succinate_fermentationCompetence_relatedAnaplerotic_genesType_I_SecretionSec_SRPTwin_Arginine_TargetingSerine_pathwayArsenic_reductionKB01KoFoxVar1Var20.00GlycolysisGluconeogenesisTCA_CycleNADH_quinone_oxidoreductaseFtype_ATPaseCytochrome_c_oxidase_cbb3_typeCytochrome_bd_complexRuBisCoCBBCyclerTCA_CycleHdroxypropionate_3_BicycleHydroxybutyrate_4_hydroxypropionate_3Dgalacturonate_epimeraseBeta_N_acetylhexosaminidaseNitrogen_fixationSulfhydrogenaseSulfide_oxidationNiFe_hydrogenase_Hyd1Thiamin_biosynthesisRiboflavin_biosynthesisCobalamin_biosynthesisPhosphate_transporterChmotaxisCoenzyme_B_Coenzyme_M_regenerationCytochrome_b6f_complexAnoxygenic_typeI_reaction_centerRetinal_biosynthesisEntnerDoudoroff_PathwayLactatefermentationAcetate_fermentationPEP_to_Succinate_fermentationCompetence_relatedAnaplerotic_genesType_I_SecretionSec_SRPTwin_Arginine_TargetingSerine_pathwayArsenic_reductionKoFoxVar1Var20.250.500.751.00PathwayGlycolysisGluconeogenesisTCA_CycleNADH_quinone_oxidoreductaseFtype_ATPaseCytochrme_c_oxidase_cbb3_typeCytochrome_bd_complexRuBisCoCBB_CyclerTCA_CycleHydroxypropionate_3_BicycleHydroxybutyrate_4_hydroxyproponate_3Dgalacturonate_epimeraseBeta_N_acetylhexosaminidaseNitrogen_fixationSulfhydrogenaseSulfide_oxidationNiFe_hydrogenase_Hyd1Thiamin_biosynthesisRiboflavin_biosynthesisCob lamin_biosynthesisPhospate_transporterChemotaxisCoenzyme_B_Coenzyme_M_regenerationCytochrome_b6f_complexAnoxygeictypeI_reaction_centerRetinal_biosynthesisEntnerDoudoroff_PathwayLactate_fermentationAcetate_fermentationPEP_t _Succinate_fermentatioCompetence_relatedAnaplerotic_g nesType_I_SecretionSec_SRPTwin_A ginine_TargetingSerine_pathwayArsenic_reductionKB01Var1Var20.250.500.751.00PathwayCBBCytoplasmPeriplasmStrain KB01GlycolysisRuBisCOPstABCSPO4-PO4-CO2PP PathwayChlorosomeBChl eNH3+Cyc2Fe(II)e-N2NifDKENHHyp H22H+H2aprB SATSO42-APSSO32-dsrABHS-PstABCSPO4-PO4-PstABCSPO4-PO4-PstABCSPO4-PO4-rTCAChlorosomeBChl eChlorosomeBChl ee-A BC DFigure 2.3: The metabolic potential of strain KB01 with a 16S rRNA gene tree (A) of the class Chlorobia (strainKB01 highlighted in blue) and the photoferrotrophic strains underlined in orange. The tree was computedusing three methods — Maximum Likelihood, Maximum Parsimony, and Neighbour Joining and wasbootstrapped 500 times for each method with the bootstrap values shown in the above order at each branch.The genome of strain KB01 (purple) compared (B) at the nucleotide level to the genome of Chlorobiumferrooxidans strain KoFox (green) where the grey links demonstrates regions that are identical betweenthe two genomes. Strain KoFox and strain KB01 overall pathway comparison (C) where the darker greenindicates a more complete pathway (value of 1.00). The major cellular pathways present in strain KB01 (D).37Strain KB01 has much of the same metabolic potential as strain KoFox, but there are some keydifferences that distinguish the two. To assess such differences in the metabolic potential of bothstrains, open reading frames (ORFs) for both strains were annotated and compared (Fig. 2.3c).There were several similarities between the two genomes, including the presence of the canonicalgenes associated with anoxygenic photosynthetic growth in the class Chlorobia (e.g. those for thereverse tricarboxylic acid cycle and electron transport genes such as the inner-membrane boundcytochrome C551). They also shared some genes that are widely distributed across the Chlorobialike those required for nutrient acquisition (e.g. nitrogen fixation) (Fig. 2.3c, d). While the majorityof the pathways found in the strain KB01 genome were the same as those found in strain KoFox,strain KB01 also possesses the genes that code for RuBisCo, as well as some of the genes thatcode for key proteins in the Calvin-Benson-Bassham (CBB) pathway (Fig. 2.3c), which, althoughmore generally known for their role in oxygenic photosynthesis [143], have also been implicatedin detoxification pathways in anoxygenic photosynthetic microorganisms [144]. Some of the genesrequired for these pathways are also found in strain KoFox; however, in strain KoFox the CBBpathway is incomplete and the RuBisCo gene has a nonsense mutation resulting in a prematurestop codon and likely an ineffective protein. Like the difference between the two species in theRuBisCo gene, the majority of the variability between the two genomes is due to nucleotidesubstitutions that do not alter the gene’s overall annotation. The two species thus possess similarpredicted metabolic potential (Fig. 2.3c) despite a 30 % difference in genomes at the nucleotidelevel. There are, nevertheless, some differences at the gene and pathway level between the twospecies that include the number of ORFs attributed to some key pathways. For example, strainKB01 has three more copies of genes associated with phosphate acquisition, such as the gene forthe phosphate transport system permease protein (pstA) and accessory protein (phoU) with fourcopies found in strain KB01 compared to a single copy of each in strain KoFox. Other differencesbetween the two genomes can be mapped to hypothetical or uncharacterized genes and thenumber of copies of these genes. Conversely, like strain KoFox, strain KB01 possesses a putativeouter-membrane cytochrome (cyc2PV – 1; Fig. 2.3d) [58] that is homologous to previously identifiediron-oxidases in the microaerophilic iron oxidizer Mariprofundus strain PV-1 [145]. Combined,these results indicate that strain KB01 is very similar to its relative strain KoFox at the pathway38level, while the differences between the two genomes can mostly be attributed to a combination ofsingle nucleotide substitutions, the number of genes in key pathways (i.e. phosphate uptake), anda number of hypothetical genes.2.4.3 Electron donor usage, Fe(II) oxidation, and growthStrain KB01 has very limited capacity to use electron donors other then Fe(II) and electronacceptors apart from CO2. To determine potential alternative electron donors and acceptors weinoculated strain KB01 into a series of media containing electron donors including glucose, acetate,and hydrogen gas. No photosynthetic growth was detected in any of the media containing organiccompounds as the only electron donors. Additionally, no non-photosynthetic growth was detectedwhen organic carbon was added as electron acceptor under both aerobic and anaerobic conditions.While strain KB01 was capable of growth under standard conditions (photoferrotrophically)following exposure to atmospheric oxygen during transfer, no growth was detected under aerobicconditions. This is likely due to abiotic Fe(II) oxidation by oxygen outcompeting the phototrophicgrowth of strain KB01, although it could also be due to toxicity effects of oxygen on strain KB01with further experimentation needed to determine if strain KB01 is a strict anaerobe. Notably,despite the presence of key genes for the formation of the hydrogenase complex (i.e. nickelinsertion protein HypA) strain KB01 did not grow when hydrogen gas was provided as the soleelectron donor for photosynthesis. Collectively these results indicate that strain KB01 growsexclusively through photoautotrophy with Fe(II) as the electron donor and is thus reliant onferruginous conditions despite the apparent metabolic potential coded in its genome for growththrough alternative modes of photosynthesis.Strain KB01 grew roughly stoichiometrically through Fe(II) oxidation at rates that were depen-dant on the concentration of Fe(II). During the exponential growth phase (between approximately70 and 170 hours) under standard conditions — substrate-rich, unfiltered media (see Methods), 10mM Fe(II), circumneutral pH of 6.8-6.9, room temperature (23 ◦C), and under light saturation (15µmol m– 2 s– 1) — strain KB01 oxidized Fe(II) at a rate of 35 ± 5.3 µM h– 1 (Fig. 2.4a). Under thesesame conditions, where 5 mM Fe(II) was oxidized to Fe(III) (oxyhydr)oxides over the course of thefull 215 hours, the initial inoculum (1 ± 0.1 x 105 cells mL– 1) rose to a final density of 7.2 ± 1.2 x3910 5 cell mL– 1 (Fig. 2.4a). During the exponential growth phase, strain KB01 oxidized Fe(II) at aper cell rate of 48 ± 7 fmol cell – 1 h– 1 with specific growth rate of 0.17 ± 0.002 h– 1. Comparingthe per cell Fe(II) oxidation rates (in carbon equivalents) to an assumed per cell average carboncontent of 1 x 10 – 13 g C, based on the estimate that 50 % of cellular biomass (∼2 x 10 – 13 g [146])is carbon, results in an Fe(II)oxidized to Cfixed ratio in strain KB01 of approximately 6:1. This ratio isclose to, and likely within error of, the theoretical and stochiometric value of 4 Fe(II) oxidized perC fixed (Eq. 2.1).Thompson et al Figure 4A BFigure 2.4: Strain KB01 growth rates with a growth curve (A) for strain KB01 under optimal growthconditions with Fe(II) oxidation (red circles) compared to a control (black squares) and cell counts (bluediamonds), while pigment concentrations and cell counts for the same growth experiment are shown in(B). The additional circles in (A) delineate the range that was utilized to calculate maximum rates of Fe(II)oxidation.Fe(II) oxidation by strain KB01 followed first-order reaction kinetics with respect to Fe(II)concentrations and these could be fit with a Michaelis-Menten model (Fig. 2.5a). The change inrates of Fe(II) oxidation as a function of Fe(II) were best fit with a maximum rate (Vmax) of 43± 7 fmol h– 1 per cell and a half-saturation constant for Fe(II) (km) of 5 ± 5 µM. To calculate thespecific affinity of strain KB01 cells with respect to Fe(II) (Eq. 2.3), the average volume of strainKB01 cells (0.17 µm3; radius of 0.15 µm and length of 0.6 µm) is combined with an average celldensity (1.09 g cm – 3 [136]) and the number of cells at the end of log phase (7.2 ± 1.2 x 105 cellmL– 1) to give a wet cellular biomass of 35 µg L – 1. The wet cellular biomass is then combinedwith the volume specific Vmax (33 µmol L– 1 h – 1) to give a biomass specific Vmax of 52 g Fe g– 140wet cell – 1 h – 1. Dividing the biomass specific Vmax of 52 g Fe g– 1 wet cell – 1 h – 1 by the km (5µM) yields a specific affinity for strain KB01 with respect to Fe(II) of 190,000 L g – 1 wet cell – 1 h – 1.Rates of photosynthetic Fe(II) oxidation by KB01, and by extension its growth rates, thus dependstrongly on Fe(II) concentrations in the low µM range.0 50 100 150 200[Fe(II)] (µM)010203040Rate (µM hr-1)0 5 10 15Light (µmol m-2 s-1)010203040Rate (µM hr-1)0 50 100 150 200[Fe(II)] (µM)010203040Rate (µM hr-1)0 5 10 15Light (µmol m-2 s-1)010203040Rate (µM hr-1)Vmax = 35 ± 3.5 µM hr-1Km = 0.75 ± 0.08 µmol m-2 s-1Vmax = 33 ± 5 µM hr-1Km = 5 ± 5 µMA BThompson et al Figure 5Figure 2.5: Strain KB01 growth kinetics with Fe(II) oxidation rates over a range of Fe(II) concentrations (A)and light intensities (B) with the resulting Michaelis-Menten model (red lines) and parameters displayed.Rates of Fe(II) oxidation and growth by strain KB01 were also dependent on light intensity,pH, and temperature. Strain KB01 was capable of growth at both 15 ◦C and 30 ◦C, albeit withslightly longer lag phases and lower growth rates than at 23 ◦C (Table 2.2). Modelling of thegrowth rates at these temperatures suggests that strain KB01 is likely capable of growth over a30 ◦C temperature range (9 ◦C to 39 ◦C – Table 2.2) and although these numbers have yet to beconfirmed with empirical data, no growth was detected at 4 ◦C. Strain KB01 also exhibited similargrowth rates (within error) across a pH range of 6 to 7.3 (Table 2.2). Modeling of these growthrates suggests that strain KB01 would be able to grow over a pH range of 4.9 to 8.4 (Table 2.2)with further experiments required to confirm this range. Fe(II) oxidation rates also depended onlight availability, which strongly limited Fe(II) oxidation rates below about 2 µmol m – 2 s – 1. Thedependence of Fe(II) oxidation rate on light intensity was fit with a Michaelis-Menten model (Fig.2.5b) where maximal cell specific Fe(II) oxidation rates (49 ± 5 fmol h – 1 cell – 1) were reached by 5µmol m – 2 s – 1 (light saturation) with a half-light saturation (HLS) value 0.75 ± 0.08 µmol m – 2 s – 1(Fig. 2.5b). Furthermore, strain KB01 oxidized Fe(II) at light intensities as low as 0.03-0.05 µmol41m – 2 s – 1 albeit at very slow rates (< 1 µM Fe(II) h – 1) (Fig. 2.5b). Rates of Fe(II) oxidation, andagain by extension growth, are, thus dependent on pH, temperature and light intensity over arange of environmentally relevant conditions.Strain KB01 is also capable of growth, albeit at slower growth rates, in the absence of exogenousvitamins, as KB01 possesses the necessary pathways for the biosynthesis of these vitamins. Toassess whether strain KB01 could meet its own vitamin requirements, we tested for growth in aseries of media prepared with a standard suite of vitamins, but with a single vitamin removed.Growth was possible under all vitamin mixtures, despite the missing vitamins, albeit with longerlag phases and slower growth rates (data not shown). This result is consistent with the complementof vitamin biosynthesis pathways in KB01’s genome, such as cobalamin, which is involved in B12biosynthesis (Fig. 2.3c). Strain KB01 is, therefore, capable of growth when biologically essentialvitamins, such as vitamin B12, are scarce or absent, by producing the entire complement ofnecessary vitamins on its own.Table 2.2: Strain KB01 bulk growth rates over a range of temperature and pH as well as the predicted rangeof temperature and pH where strain KB01 should be capable of growth.Temperature (°C) 15 23 25 30Growth rate 23 ± 4 35 ± 1 34 ± 4 30 ± 5Predicted temperature rangeMin Max R29 °C 39 °C 0.99pH 6 6.4 6.8 7 7.3Growth rate 33 ± 10 43 ± 12 40 ± 14 40 ± 13 37 ± 11Predicted pH rangeMin Max R24.9 8.4 0.732.5 Discussion2.5.1 Characteristic traits of strain KB01Strain KB01 exhibits genomic traits that differentiate it both from other members of the classChlorobia and other previously studied photoferrotrophs. Strain KB01, while taxonomically42related to Chlorobium ferrooxidans strain KoFox, 16S rRNA genes are 99 % identical between thetwo strains (Fig. 2.3a), it only shares 70 % of its genome with strain KoFox (Fig. 2.3b) and canthus be considered a separate species at a probability of greater than 0.5 based on DNA-DNAhybridization (Table 2.1). Some of the genomic differences that distinguish strain KB01’s genomefrom that of its closest relative (Fig. 2.3c), strain KoFox, are the genes associated with the CBBpathway, including the RuBisCo gene, although both are found elsewhere in the class Chlorobiaand in anoxygenic phototrophs more broadly [143]. While the RuBisCo gene is most commonlyassociated with carbon fixation in oxygenic phototrophs, previous studies have suggested that theRuBisCo homologue found in the Chlorobia increases their growth efficiency by reducing oxidativestress [144]. The lack of the RuBisCo gene in otherwise very similar genomes could contribute tothe slightly lower per cell Fe(II) oxidation rates seen in strain KoFox compared to strain KB01 (Table2.3). Given that the lack of the RuBisCo gene in strain KoFox is due to a single nonsense mutationthat is not lethal, suggests that the RuBisCo gene is not necessary for photoferrotrophy, at leastin benthic strain KoFox. The greater number and extent of redundancy in phosphate transportgenes found in strain KB01 (4x greater than in strain KoFox), may reflect its adaptation to lowphosphate concentrations in Kabuno Bay [48, 96] and those more generally perceived for modernand ancient ferruginous environments [47, 70, 85, 147]. Additional differences between the twostrains include a large number of hypothetical or uncharacterized genes in strain KB01 that do notappear in strain KoFox and further investigation is required to determine their specific functions.Strain KB01 also contains the putative outer membrane cytochrome iron oxidase (cyc2PV – 1) thatwas previous linked to Fe(II) oxidation in microaerophilic Fe(II) oxidizer Mariprofundus strainPV-1 [145]. Homologues of the cyc2 gene have been previously identified in three members ofthe class Chlorobia, two of which (C. ferrooxidans strain KoFox [53] and C. sp. N-1 [56]) arecapable of photoferrotrophic growth, suggesting its possible role as an outer membrane Fe(II)oxidase. Growth of one of these organisms through photoferrotrophy, Chlorobium luteolum DSM273, however, remains uncertain, and experiments to date [56] suggest it is unlikely. Thus, therole of the cyc2 may not be specific to Fe(II) oxidation in the Chlorobia. While many of the genesand pathways found in strain KB01’s genome are present in other photoferrotrophs and membersof the class Chlorobia, the combination and numbers of specific genes, like those involved in43Table 2.3: Physiological information of photoferrotrophic and microaerophilic strains. References: 1) [115];2) [55]; 3) [56]; 4) [8]OrganismPer cell Fe(II) oxidation rate (fmol hr-1)Association to Fe(III) mineralsEnrichment location ReferenceChlorobium phaeoferrooxidans strain KB01 48 ± 2 No Kabuno Bay This workChlorobium ferrooxidansstrain KoFox 30 ± 1 YesFreshwater ditch 1Rhodobacter ferrooxidansstrain SW2 32 YesFreshwater pond 2Chlorobium sp. N-1 0.05 ± 0.001 Yes Marine sediments 3Gallionella sp. 1 Yes Various 4phosphate acquisition, are characteristic of strain KB01.Adaptations in strain KB01, such as high concentrations of low-light harnessing pigments mayimpart ability to grow in low light pelagic environments similar to Chlorobium phaeobacteroidesstrain MN1 and strain BS-1 from the Black Sea [138, 141]. The strain KB01 pigment biosynthesispathways produce an overlapping carotenoid and bacteriochlorophyll Soret peak that are blue-light shifted compared to those of strain KoFox and many other phototrophs [52, 53, 55, 57, 148].The blue-light shifted pigments, combined with BChl e and its homologues that are found instrain KB01 (Fig. 2.2), are adaptations that promote growth at low light intensities and in deepwaters [141]. Indeed, strain KB01 is capable of growth at very low light intensities (0.03-0.05µmol m – 2 s – 1; Fig. 2.5b) albeit at slow rates (< 1 µM h – 1). These low light intensities arecomparable to those required for the growth of the community of anoxygenic photosyntheticsulfide oxidizing Chlorobia found in the chemocline of the Black Sea (0.015-0.055 µmol m – 2 s – 1)[138, 149], although the Black Sea Chlorobia are also capable of maintaining biomass under evenlower light conditions (< 0.0022 µmol m – 2 s – 1; [138]). The ability of Strain KB01 to grow at lowlight intensities contributes to its relatively low HLS of 0.75 ± 0.08 µmol m – 2 s – 1 (Fig. 2.5b) that iscomparable to that of previously characterized photoferrotrophic strain KoFox (0.25 ± 0.12 µmolm – 2 s– 1 [55]) and much lower than that of Chlorobium sp. N-1 (4.6 µmol m – 2 s– 1 [56]). Despite44having a slightly higher HLS than strain KoFox, strain KB01 has a comparable specific affinityfor light to strain KoFox (∼1410000 m2 s g – 1 wet cells h – 1 and ∼1100000 m2 s g – 1 wet cells h – 1respectively (Eq. 2.3)), which implies that these two strains have a largely similar capacity togrow under low light conditions. The specific affinities of strains KB01 and KoFox to light likelyreflect a more general disposition to growth at low light in the Chlorobia, although there remainfew detailed studies of the kinetic relationship between phototrophic growth and light to makerigorous comparisons across the class. Low light adaptation in strain KB01 is also in line withrelatively high per cell concentrations of BChl e (2.1 ± 0.5 fmol or 1.5 ± 0.3 pg), which are higherthan the concentration of BChl e and c measured in Chlorobium limicola (0.05 fmol per cell [150])and higher than the BChl a concentrations measured for aerobic anoxygenic phototrophs in theChesapeake Bay (0.55 fg per cell [151]). If we consider a range of per cell protein concentrations of2.8 x 10 – 8 to 3.5 x 10 – 7 µg [47, 152], strain KB01’s BChl e concentrations (58000 ± 12000 to 4000 ±800 µg mg – 1 protein) are also orders of magnitude higher than those reported for the Chlorobiastrain (BS-1) from the Black Sea [138]. Furthermore, the total BChl e concentrations in late-log toearly stationary phase strain KB01 cultures (60-800 µg L – 1) are similar to those found in Chlorobiadominated chemoclines of meromictic lakes [47, 140]. Adaptations to low light, like high percell BChl e concentrations, may thus be key for the growth of strain KB01 in the chemocline ofKabuno bay, and by extension may allow photoferrotrophic Chlorobia to proliferate in low-lightferruginous chemoclines more generally.Strain KB01’s growth rates under standard conditions are comparable to those of otherphotoferrotrophic Chlorobia, and higher than those of other Fe(II) oxidizing microorganismsstudied to date. At maximum rates of Fe(II) oxidation under standard growth conditions at roomtemperature (23◦C), circumneutral pH (6.8-6.9), and light saturation (15 µmol m – 2 s – 1), strainKB01 oxidized 35 ± 5.3 µM h – 1 (Fig. 2.4). These maximum bulk rates are similar to those of otherpreviously characterized photoferrotrophic members of the class Chlorobia such as Chlorobiumsp. N-1 (32 ± 0.8 µM h – 1 [56]) and are faster than those of other chemotrophic Fe(II) oxidizerssuch as Gallionella (1.2 µM h – 1 [8]). The rate of Fe(II) oxidation decreases with decreasing Fe(II)concentration, following a Michaelis-Menten curve, with a relatively low half-saturation constant(km) of 5 ± 5 µM (Fig. 2.5a) – comparable (within error) to the km of anoxygenic sulfide oxidizing45members of the class Chlorobia that have low km’s compared to other phototrophic sulfideoxidizing microorganisms [153]. Furthermore, strain KB01’s low km with respect to Fe(II) suggeststhat the cells have a high affinity for Fe(II). Indeed, strain KB01 has a high specific affinity for Fe(II)of 190,000 L g – 1 wet cell h – 1, which is comparable to microorganisms considered to have extremelyhigh specific affinities for their substrates such as the sulfide oxidizing Gammaproteobacteriaof the SUP05/ARTIC96BD lineages that drive cryptic sulfur cycling in oxygen minimum zones[154]. Notably, strain KB01’s high specific affinity for Fe(II) is reflected in the per cell phototrophicFe(II) oxidation rates (48 ± 7 fmol h – 1 for strain KB01) that are higher than those of otherphotoferrotrophic and chemotrophic Fe(II) oxidizing strains (Table 2.3). The high per cell ratesof Fe(II) oxidation result in a Fe(II)oxidized to Cf ixed ratio of 6:1, which is likely within errorof the theoretical and stochiometric ratio of 4:1 (Eq. 2.1). Thus, while many of strain KB01’sphysiological characteristics are similar to other Chlorobia and strain KoFox in particular, theirhigh cell specific rates of Fe(II) oxidation and resulting high affinity for Fe(II) set strain KB01 apartfrom other photoferrotrophs studied to date.Under nutrient limited conditions, strain KB01’s per cell Fe(II) oxidation rates are diminishedwhen compared to growth under standard conditions but are still higher than those of otherphotoferrotrophs and Fe(II) oxidizing microorganisms. For example, when dinitrogen gas is theonly available source of nitrogen, strain KB01 has a per cell Fe(II) oxidation rate of 20 ± 1 fmolh – 1 [115], which is higher than that of its marine relative C. sp. N-1 and fellow photoferrotrophstrain SW2 under standard growth conditions (Table 2.3). In addition to growth under nitrogenand sulfate limiting conditions [115], strain KB01 also grew without the addition of exogenousvitamins to its growth medium. This stands in contrast to many modern marine and freshwatermicroorganisms that do not produce a full complement of vitamins, and instead utilize vitaminsthat have been excreted by other members of the microbial community (leakage [155]) to supporttheir growth and metabolism [156]. Such metabolic streamlining often confers a competitiveadvantage at the expense of self-sufficiency. Strain KB01, on the other hand, appears to possessmetabolic potential for self-sufficiency both in terms of its ability to acquire nutrients frommultiple sources and its capacity to produce a full complement of vitamins. Strain KB01 isalso capable of growth at a range of pH (6-7.3; Table 2.2). While this range is comparable to46the range of pH required for the growth of other freshwater photoferrotrophs (5.5-7.5; [55]),strain KB01 has an inferred growth potential over an extended pH range (4.9-8.4; Table 2.2) that,importantly, encompasses the pH range of modern and ancient seawater [56, 157, 158], implyingthat photoferrotrophy is possible under such conditions.An important, and distinguishing, trait of strain KB01 is its ability to shed the Fe(III) (oxy-hydr)oxides produced as a by-product of its growth [99]. Despite high rates of Fe(II) oxidation and,as a result, Fe(III) precipitation, strain KB01 maintains cell surfaces free of Fe(III) (oxyhydr)oxides(Fig. 2.1). The lack of association allows strain KB01 to remain buoyant in sunlit waters, avoidingsedimentation due to ballasting from associated Fe(III) precipitates [99]. All previously charac-terized photoferrotrophs have shown some level of association or indeed encrustation by Fe(III)(oxyhydr)oxides, although the degree of association appears to vary depending on the compositionof the growth medium (Table 2.3) [9, 63, 99]. Strain KB01 avoids association through the natureof its cell surface functional groups [99], which lead to electrostatic repulsion between the cellsurface and Fe(III) (oxyhydr)oxides when these oxyhydroxides carry negative surface charges dueto incorporation of anions like silica or phosphate [99]. Such electrostatic repulsion sets pelagicstrain KB01 apart from its benthic photoferrotrophic relatives and other benthic photoferrotrophsmore generally.Information on the physiology and metabolic potential of strain KB01 can be further used topredict its rates of growth and metabolism under a range of environmental conditions. Generally,rates of photosynthetic growth are controlled by the availability of electron donor, light andnutrient elements like phosphorus (P) and nitrogen (N). Combining predictive equations thatdescribe the rates of Fe(II) oxidation by strain KB01 as a function of electron donor (Fe(II)), light,and the nutrient P, yields the following:R = Vmax ·[Fe]kmFe + [Fe]· [light]kmlight + [light]· [PO3−4 ]kmPO3−4 + [PO3−4 ](2.4)where R is the cell specific rate of Fe(II) oxidation (fmol cell – 1 h – 1), Vmax is the cell specificFe(II) oxidation rate (43 ± 6.6 fmol cell – 1 h – 1), kmFe is the half-saturation constant for Fe(II) (5 ±5 µM), kmlight is the HLS constant (0.75 ± 0.08 µmol m – 2 s – 1), and kmPO3 –4 is the half-saturation47constant for phosphate (0.005 µM [96]). This formulation can also be used to calculate growthrates considering the stoichiometry of Cfixed to Fe(II)oxidation (0.16:1), which in turn can be usedto calculate doubling times based on the amount of carbon content of a cell. Information onthe response of photoferrotrophy by strain KB01 to N availability, pH and temperature could,in principle, be considered similarly. In this way, physiological information can be used tomodel photosynthetic production by strain KB01 across a wide range of environmental conditionsand with further information on how the relevant physiology varies across extant and ancestralphotoferrotrophs, can be used to reconstruct primary production in the ferruginous oceans of thePrecambrian Eons.2.5.2 Strain KB01 as a model organism for Precambrian PPStrain KB01 could have deposited the world’s largest BIFs with cell numbers that are comparableto those of modern oceanic primary producers, even under nutrient poor conditions. Previouscalculations (utilizing the high rates of iron deposition of 1000 m m – 1 yr – 1 estimated for theHammersley group [8, 159, 160]) have estimated that Fe deposition rates of up to 45 mol m – 2 yr – 1were needed to deposit the world’s largest BIFs (i.e. Hammersley Basin, Australia). When theseFe deposition rates are combined with the high per cell Fe(II) oxidation rates of strain KB01 (48± 7 fmol h – 1), a maximum of 1.1 x 103 cells mL – 1 would be required in the upper 100 m of theHamersley basin water column (estimated area of 1 x 1011 m2 [8]) to deposit the world’s largestknown BIF [4]. This number is comparable to previous calculations utilizing the Fe(II) oxidationrates of the environmental community of Chlorobia from Kabuno Bay (1.7 x 103 cells mL – 1 [48]).The modest difference between the two numbers is likely due to differences between the growthrates of a pure culture and those that are part of a larger microbial community. Notably, cellconcentration is two orders of magnitude lower than the estimated 5 x 105 cells mL – 1 present inthe uppermost 200 m of the modern oceans [146] and is comparable to estimates of the numberof primary producing cells per mL in modern near shore and coastal shelf regions (103 cellsmL – 1 [161]). Standard laboratory culture media is notoriously nutrient – rich, however, andtherefore is unlikely to represent Fe(II) oxidation rates in the nutrient-poor Precambrian oceans.Under nitrogen limiting conditions (dinitrogen as the sole nitrogen source), the per cell iron48oxidation rates for strain KB01 (20 ± 1 fmol h – 1) indicate that, under the previously describedhigh iron deposition rates (45 mol m – 2 yr – 1 [8]), a maximum of 2.6 x 103 cells mL – 1 wouldhave been required in the upper 100 m of the Hammersley basin water column to deposit theHammersley BIF [115]. Furthermore, strain KB01 is capable of producing sufficient Fe(III) to havedeposited BIFs under low Fe(II) (< 400 µM) and silica-rich (0.6-1.5 mM — comparable to estimatedconcentrations of silica in the Archean oceans [85]) conditions [99]. Thus, the growth of pelagicstrain KB01, under a diverse set of conditions that better mimic those of the Precambrian oceans,supports models for the deposition of BIFs by photoferrotrophs. Additionally, the difference incell numbers under nutrient rich versus nutrient poor conditions suggests that BIF depositionrates could have scaled with nutrient availability such as biological nitrogen availability.The genotypic, phenotypic, and ecological information gleaned from Chlorobium phaeoferrooxi-dans strain KB01 suggests that it is a strong analogue of Precambrian photoferrotrophs. StrainKB01 possesses a number of traits, such as a high number of phosphate transport genes or thelack of cell-Fe(III) (oxyhydr)oxide association, that distinguish it from previously characterizedphotoferrotrophic species that likely promote the proliferation of strain KB01 in low P ferruginouswater columns like Kabuno Bay. Furthermore, stain KB01’s physical separation from its Fe(III)(oxyhydr)oxide metabolic products allows it to maintain its position within the photic zoneof Kabuno Bay’s water column. Such traits, combined with characteristics that are ubiquitousthroughout the class Chlorobia (i.e. the ability to fix dinitrogen gas under nitrogen limitingconditions), allow strain KB01 to grow efficiently under conditions where nutrients are scarce orlimiting. Such nutrient scarcity was likely prevalent under the ferruginous ocean conditions thatcharacterized much of the Precambrian Eons (nitrogen [162], sulfate [27], phosphorus [85]) andthus the ability to acquire biologically scarce essential nutrients would have been key to sustainingappreciable primary production. We argue, therefore, that the physiology and metabolic potentialof strain KB01, which underpin its capacity to proliferate in modern ferruginous basins likeKabuno Bay, is broadly extensible to conceptual and quantitative models of PP, coupled carbonand iron cycling, and BIF deposition in the Precambrian Eons. Further support for this claimcould come from information that constrains the antiquity of photoferrotrophy in stem groupChlorobia, and the corresponding history of the metabolic potential hosted in strain KB01.49Chapter 3Nutrient Acquisition and the MetabolicPotential of Photoferrotrophic Chlorobi3.1 SummaryAnoxygenic photosynthesis evolved prior to oxygenic photosynthesis and harnessed energyfrom sunlight to support biomass production on the early Earth. Models that consider theavailability of electron donors predict that anoxygenic photosynthesis using Fe(II), known asphotoferrotrophy, would have supported most global primary production before the proliferationof oxygenic phototrophs at approximately 2.3 billion years ago. These photoferrotrophs have alsobeen implicated in the deposition of banded iron formations, the world’s largest sedimentaryiron ore deposits that formed mostly in late Archean and early Proterozoic Eons. In this work wepresent new data and analyses that illuminate the metabolic capacity of photoferrotrophy in thephylum Chlorobi. Our laboratory growth experiments and biochemical analyses demonstrate thatphotoferrotrophic Chlorobi are capable of assimilatory sulfate reduction and nitrogen fixationunder sulfate and nitrogen limiting conditions, respectively. Furthermore, the evolutionaryhistories of key enzymes in both sulfur (CysH and CysD) and nitrogen fixation (NifDKH)pathways are convoluted; protein phylogenies, however, suggest that early Chlorobi could havehad the capacity to assimilate sulfur and fix nitrogen. We argue, then, that the capacity forphotoferrotrophic Chlorobi to acquire these key nutrients enabled them to support primaryproduction and underpin global biogeochemical cycles in the Precambrian.503.2 IntroductionModern global primary production is supported through oxygenic photosynthesis, which convertssunlight and CO2 into biomass, fuelling the biosphere and driving fluxes of matter and energy atglobal scales [163]. Primary production is limited by the availability of nutrients that are essentialfor growth such as phosphorus, nitrogen, and sulfur [164]. Primary producers thus expendvaluable energy to meet their nutrient quotas. In the modern oceans, for example, cyanobacteriacan fix nitrogen in the photic zone to support their nitrogen requirements [165]. This in turnprovides a competitive advantage that frequently allows nitrogen-fixing cyanobacterial species likeTrichodesmium to outcompete non-nitrogen fixing species and can lead to cyanobacterial blooms[166, 167]. In addition to their role as primary producers in the modern oceans, cyanobacteria playa key role in the acquisition and redistribution of nutrients [168], driving global biogeochemicalcycles since their evolution and proliferation in the Precambrian Eons.Oxygenic photosynthesis and cyanobacteria emerged early in the Archean Eon [16, 94], evolv-ing from anoxygenic phototrophs [101], which arose as early as 3.8 Ga [169]. Like oxygenicphototrophs, anoxygenic phototrophs fix carbon dioxide into biomass, but instead of water asthe electron donor they use a diverse set of inorganic species [e.g., H2, H2S, and Fe(II)] to replaceelectrons transferred from the photosystem to CO2 [93]. Most anoxygenic phototrophs that growin illuminated anoxic waters today use reduced sulfur species as their electron donors. Duringmuch of Earth’s early history, however, reduced sulfur species were likely scarce and the chemistryof marine sediments suggests that the oceans were overwhelmingly iron-rich (ferruginous) forlong stretches of both the Archean and Proterozoic Eons [5, 170, 171]. Under these ferruginousconditions, ferrous iron would have been the most abundant and available inorganic electrondonor [1]. Models for primary production in these ferruginous oceans suggest that anoxygenicphototrophs using Fe(II) as their electron donor–photoferrotrophs–could have supported up to10% of modern day primary production before the proliferation of cyanobacteria [1, 85]. Together,the evolutionary history of the photosystem and current knowledge on the history of ocean redoxstates imply that photoferrotrophs could have played a key role in driving global fluxes of matterand energy throughout the Precambrian Eons.51Compelling, but indirect, evidence for photoferrotrophy during Archean and Paleoproterozoictimes comes from the deposition of banded iron formations (BIFs) [7, 8, 21]. BIFs are massiveiron ore deposits that were mostly deposited toward the end of the Neoarchean, though theirdeposition spans from the Eoarchean through to the Neoproterozoic Eras [4]. Classical models forthe deposition of iron from seawater to form BIF invoke large-scale oxidation of seawater Fe(II)by oxygen produced as a by-product of cyanobacterial growth and the subsequent precipitationand sedimentation of ferric iron minerals [6, 7, 172]. Oxygen levels through the Archean, however,appear too low to support oxidation of Fe(II) at rates sufficient to sustain the rapid ferric Fedeposition needed to form even some of the apparently small BIFs like the Isua Greenstone beltin Greenland [169]. Instead, Fe(III) could have come from abiotic photochemical iron oxidationthrough UV photolysis [7, 173], but this also appears too slow to support ferric iron deposition atrates recorded in BIFs [40]. Alternatively, direct photosynthetic iron oxidation through photofer-rotrophy could supply ferric Fe to form BIFs [8, 20]. Accepting that oxygen levels were too low todrive Fe(II) oxidation and that UV photolysis appears similarly ineffective, photoferrotrophy maybe the only viable mechanism to support appreciable ferric iron deposition and BIF formation.Nevertheless, the role of photoferrotrophs in BIF deposition remains controversial since directevidence, like lipid biomarkers in BIFs, to diagnose photoferrotrophy, remain elusive. Extantcultures of photoferrotrophic bacteria are thus employed in efforts to further test the possiblerole of photoferrotrophs in BIF deposition and to identify signals that might be used to diagnosephotoferrotrophy in the rock record.A total of eight enrichments and isolates of photoferrotrophic bacteria have been broughtinto laboratory collections over the last 30 years. These cultures were largely obtained from avariety of benthic environments, such as marine mud flats and freshwater sediments [20, 51–54],with a single isolate originating from a ferruginous water column [48]. Laboratory cultures ofphotoferrotrophs are distributed across the Alphaproteobacteria, the Gammaproteobacteria, andthe Chlorobi and experiments conducted with these cultures reveal diverse physiological traits thattranslate to differential growth rates across a wide range of culture conditions [9, 55, 62]. Notably,under modest light availability, many of these cultures grow sufficiently fast to oxidize Fe(II) atrates that would support the deposition of some of the largest BIFs [8, 21]. This gives confidence52in the capacity of photoferrotrophs to deposit BIFs, but laboratory culturing media are notoriouslynutrient rich. Natural settings, on the other hand, are typically nutrient poor in comparison [174],and thus the role of photoferrotrophs in both BIF deposition and primary production wouldhave depended on their capacity to grow and acquire nutrients from Precambrian seawater atconcentrations almost certainly much lower than typical culture media.Many laboratory experiments have been conducted with photoferrotrophs from the Alphapro-teobacteria and Gammaproteobacteria [9, 55, 62, 175, 176], but the ecological relevance of thesegroups in natural ferruginous settings is uncertain. In all modern ferruginous environmentssupporting photoferrotrophy, members of the Chlorobi appear to dominate [47, 48, 67]. Further-more, most or many extant photosynthetic communities dominated by anoxygenic phototrophsare comprised mostly of Chlorobi [177]. While anoxygenic photosynthesis by the Proteobacterialikely evolved early [101], more recent phylogenomic analyses imply that the original phototrophsbelonged to the Chlorobi [142, 178, 179]. The reason for the apparent prevalence of the Chlorobiin modern environments is uncertain, but it is likely related to their ability to grow under environ-mentally relevant conditions including low nutrient availability and low light [64–66, 138, 180, 181].Thus, despite the fact that photoferrotrophy by Proteobacteria may be relevant to Precambrianecosystems, here, we focus our analyses on the Chlorobi because of their apparent ecologicalprominence in many modern systems and their deeper ancestry compared to phototrophicProteobacteria.Both phosphorus and nitrogen often limit photosynthetic activities and primary production inthe modern oceans and in freshwater environments [164]. Phosphorus is generally considered theultimate limiting nutrient on geological time scales as nitrogen can be fixed from the atmospherewhen phosphorus is available [182]. Phosphorus is essential for life and is required in phos-pholipid, nucleic acid, and adenosine tri-phosphate (ATP) biosynthesis. Phosphorus throughoutthe Precambrian Eons was scarce with seawater concentrations orders of magnitude lower thantoday [85]. This phosphorus scarcity would have led to low primary production, influencingthe ecology and elemental stoichiometry of the photosynthetic primary producers [147]. Whilephosphorus scarcity likely played an outsized role in shaping the Precambrian biosphere, nitrogenscarcity may have developed locally and transiently throughout the Precambrian Eons [73, 183].53Nitrogen is required to build essential cellular components such as DNA and amino acids. Bi-ologically available nitrogen is supplied to the oceans through rock weathering and volcanism,but ammonium uptake and ultimate burial, however, would have eventually depleted the oceanicbioavailable nitrogen reservoir [184]. In the modern ocean, biological fixation of atmosphericnitrogen keeps pace with phosphate supplies over geologic time scales [185]. Nitrogen fixation isone of the most energetically expensive processes in the metabolic repertoire of life and yet it isdistributed across distantly related groups of microorganisms [186, 187]. This underscores theimportance of nitrogen fixation to microbial growth and production, is consistent with the earlyevolution and radiation of nitrogen fixation [32, 188, 189], and exemplifies how the distributionof core metabolic machinery across diverse lineages and functional guilds ensures survival ofessential biogeochemical functions over geologic time [3]. While the genomic potential for nitrogenfixation exists within the Chlorobi [142], the capacity of photoferrotrophic Chlorobi to conductnitrogen fixation and thus support Precambrian marine nitrogen quotas remains untested andunsubstantiated. This leaves our knowledge of the possible ecological role that photoferrotrophsmay have played in the acquisition and redistribution of nitrogen and its attendant biogeochemicalcycling in the Precambrian oceans entirely unknown.In addition to phosphorus and nitrogen, sulfur is also essential for life and can limit biologicalproduction and growth when scarce [190]. Sulfur on the modern Earth is abundantly availableas the fully oxidized sulfate ion due to high concentrations of oxygen in the atmosphere andoceans, which promotes oxidative sulfur weathering and the recycling of sulfur from anoxicmarine sediments. During the Precambrian Eons, however, marine sulfate concentrations weremuch lower [27, 191] likely due to limited oxidative weathering and recycling under low O2atmospheres [27, 191–193]. Instead, sulfur was likely scarce and biologically available as lowconcentrations of sulfate, very low concentrations of sulfide, and possibly organic sulfur [27].Assimilatory sulfate reduction (ASR), therefore, would have been a key nutrient acquisitionpathway, supporting primary production under low sulfur conditions. The genomic potential forASR has been detected within two members the Chlorobi (Chlorobium ferrooxidans and Chlorobiumluteolum [194]), yet the role of ASR in photoferrotrophic growth remains uncertain. Photosyntheticgrowth of C. ferrooxidans on ferrous iron and without reduced sulfur compounds implies that the54genomic potential for ASR translates into physiological capacity to convert sulfate into biomasssulfur [53]. Given the likely low sulfate and extremely low sulfide concentrations perceivedfor the Precambrian oceans, ASR may have been absolutely critical for photoferrotrophs tooperate as primary producers and contribute to a reservoir of biologically available reducedsulfur compounds in the ocean. The evolutionary history of ASR in the photoferrotrophicChlorobi has not been explored, nor has sulfate uptake been quantitatively assessed. The role ofphotoferrotrophs in driving sulfur cycling during the Precambrian remains underappreciated anduntested creating another gap in our knowledge of nutrient acquisition and redistribution in thePrecambrian oceans.To address the response of photoferrotrophy to nitrogen and sulfur scarcity, and to createnew knowledge relevant to nitrogen and sulfur acquisition and redistribution in the Precambrianoceans, we examined two extant demonstrably photoferrotrophic Chlorobi: benthic C. ferrooxidans(grown in co-culture with Geospirillum sp. KoFum) [53], and pelagic Chlorobium phaeoferrooxidans[48, 58]. We also examined putative benthic photoferrotroph C. luteolum, postulated to growthrough photoferrotrophy because of its genomic potential for ASR [194]. We verified the capacityof photoferrotrophic Chlorobi to fix inorganic nitrogen and sulfur, and constrained the antiquityof this capacity in the Chlorobi through phylogenetic analyses.3.3 Materials and methods3.3.1 Strains and growth mediumMedia was prepared after Hegler et. al., 2008 [55], and allocated into serum bottles (100 mL media,160 mL total volume), with 0.3 g L – 1 NH4Cl, 0.5 g L– 1 MgSO4·7H2O, 0.1 g L – 1 CaCl2·2H2O, 0.6 gL – 1 KH2PO4. After autoclaving, 22 mmol L– 1 bicarbonate, trace elements, mixed vitamin solution,selenate-tungstate, vitamin B12, and FeCl2 were added and the pH was adjusted to 6.8-6.9 underan N2/CO2 atmosphere (80:20). 10 mmol L-1 FeCl2 was added to all media (regular, NH+4 deplete,and SO –4 poor) — Fe(II) concentrations from 200 µmol L– 1 to 10 mmol L – 1 have been shownto produce the same growth rates under nutrient rich conditions. The 10 mmol L – 1 media wasfiltered after being made to remove any precipitates, which resulted in a final Fe(II) concentration55of 2 mmol L – 1 for the standard media and 4 mmol L – 1 for the NH+4 deplete media. The low SO–4media was left unfiltered with an Fe(II) concentration of 10 mmol L – 1. In the ammonium freemedia, NH4Cl was replaced with 0.3 g L– 1 KCl and an additional 10 mL of N2 gas was injectedinto the headspace. In the low sulfate media, 0.0025 g L – 1 MgSO4 and 0.4 g L– 1 MgCl2 wereadded instead of the usual 0.5 g L – 1 MgSO4. Furthermore, approximately 10 kBq of carrier –free35S was added to all of the low sulfate cultures. The cultures for the N-fixation experiments weregrown in ammonium free conditions once and then transferred into the final experimental bottles.The culture for the 35S experiment was grown up in standard media, spun down and decanted toavoid adding extra sulfate, before the cells were inoculated into the final experimental bottles. Allcultures were grown under a constant light intensity of 14 µmol m – 2 s – 1.3.3.2 Analytical techniquesSpectrophotometric analysis of Fe(II) and Fe(III) concentrations were performed using the ferrozinemethod; samples were measured directly as well as after being fixed in 1 N HCl — after Voillier et.al., 2000 [133]. Pigments were measured spectrophotometrically after 24 hour extractions of 1 mLof pelleted cells in acetone:methanol (7:2 v/v) [134]. Cells numbers were then obtained using apigment to cell count conversion factor of 6.3 x 10 – 10 pigment cell – 1 mL – 1 for C. phaeoferrooxidansand 5.8 x 10 – 10 pigment cell – 1 mL – 1 for C. ferrooxidans. The cells from the 35S experiment werecollected via filtration along with a liquid sample as a background measurement. The filteredsamples were subsequently washed with 5 % Trichloroacetic acid (TCA) in order to kill, wash,and dissolve cellular material. TCA precipitates DNA and proteins, leaving only these cellularcomponents on the filter and therefore any counts associated with the filtered samples wouldindicate 35S that had been incorporated into this cellular biomass [195]. 5 mL of scintillation fluidwere added to the 35S samples (1 mL of liquid or the filter) and all samples were counted using ascintillation counter.3.3.3 BioinformaticsGenomes of Chlorobia stains used in this paper were retrieved from NCBI under the followingaccession numbers with the completion percentage of each genome in brackets after the number:56NC 008639.1 (99.45 %), NZ AASE00000000.1 (90.71 %), NC 007514.1 (97.8 %), NC 009337.1 (98.91%), NC 010803.1 (99.98 %), NC 002932.3 (97.8 %), NC 011027.1 (98.89 %), NC 007512.1(98.91 %).Genomes were analyzed using MetaPathways V2.5.1, an open source pipeline for predictingreactions and pathways using default settings [196, 197] (https://github. com/hallamlab/metap-athways2/wiki) and using the following databases: MetaCyc-v4-11-07-03 [198], Kyoto Encyclope-dia of Genes and Genomes (KEGG-11-06-18) [127], SEED-14-01-30 ( http://www.theseed.org/),Clusters of Orthologous Groups (COG-13-12-27) [199], Carbohydrate-Active enZYmes (CAZY-14-09-04) [200], and RefSeq-nr-14-01-18 [201] databases. Initially, we identified all sequences with afunctional assignment affiliated with nitrogen fixation and assimilatory sulfur reduction using theMetaPathways functional annotation table output.3.3.4 Phylogenetic trees for nitrogen fixationIndividual NifDKH gene sequences from all organisms outside of the phylum Chlorobia wereretrieved from NCBI searches from described strains, concatenated, and then aligned using thepackage software ClustalX2.1 [119]. To rigorously test the evolutionary history of nitrogen fixationmultiple tree construction methods (Maximum likelihood (ML) and Maximum parsimony (MP))were employed. ML and MP trees were constructed in MEGA version 7 [120, 121] and all treesbootstrapped 500 times. Bootstrap values are indicated at the nodes.3.3.5 Phylogenetic trees for assimilatory sulfate reductionCysH and CysD/Sat gene sequences from all organisms outside of the phylum Chlorobia wereretrieved from NCBI searches from described strains, using the package software ClustalX2.1[119]. To rigorously test the evolutionary history of ASR multiple tree construction methods(Maximum likelihood (ML) and Maximum parsimony (MP)) were employed. ML and MP treeswere constructed in MEGA version 7 [120, 121] and bootstrapped 500 times. Bootstrap values areindicated at the nodes.573.3.6 Phylogenetic trees for 16S rRNA16S rRNA sequences were retrieved from strains used in Nif and ASR gene trees from the Silvaonline database – version 128 [117, 118]. Only full-length (> 1400 bp) sequences were selected,and then aligned using the package software ClustalX2.1 [119]. To rigorously test the evolutionaryhistory of nitrogen fixation and ASR multiple tree construction methods (Maximum likelihood(ML) and Maximum parsimony (MP)) were employed. ML and MP trees were constructed inMEGA version 7 [120, 121] and all trees bootstrapped 500 times. Bootstrap values are indicated atthe nodes.3.4 Results and discussion3.4.1 Nitrogen fixationThe process of fixing dinitrogen is kinetically challenging and energetically expensive as it involvesovercoming the activation energy required in breaking the triple bond between the two nitrogenmolecules. The enzyme necessary for nitrogen fixation, nitrogenase, is a multi-subunit proteinthat is assembled and regulated by a series of other related proteins. All nitrogenases require ametal ion cofactor – molybdenum, iron, or vanadium – with each cofactor being recruited andincorporated into the nitrogenase by a different set of proteins, Nif, Anf, and Vnf, respectively.Current studies indicate that the majority of nitrogenases depend on the molybdenum ion cofactorfor their enzymatic activity (reviewed in Rubio and Ludden, 2008 [202]), while the iron andvanadium dependant nitrogenases may play a role in molybdenum limiting environments [203].Phylogenetic evidence suggests that the molybdenum-dependant version of the enzyme evolvedfirst [204], which is further supported by the observation that organisms identified as havingan iron or vanadium dependant nitrogenase all contain a copy of the molybdenum-dependantnitrogenase [187, 205]. There are up to 25 proteins, depending on the species, required to assembleand regulate the nitrogenase including three conserved structural proteins: NifD, NifK, and NifH.NifH is often used as the marker gene for nitrogen fixation in natural environments, due to itsrole in the main enzyme structure and in cofactor recruitment. Further phylogenetic information,58however, can be obtained when all three structural proteins (NifDKH) are concatenated dueto increased sequence information and the conserved nature of all three proteins. Here weexplored these key structural proteins to test for the metabolic potential for nitrogen fixationin the photoferrotrophic Chlorobi. We compare nitrogen fixation in photoferrotrophic Chlorobito the other members of the phylum Chlorobi and to representatives from all phyla capable ofnitrogen fixation to assess the evolutionary history of nitrogenase in relevant to photoferrotrophyin the Chlorobi and to place constraints on the possible role of photoferrotrophs in supplyingfixed nitrogen to the Precambrian oceans.Distribution of nitrogen fixation pathways within ChlorobiPrevious analyses of Chlorobi genomes identified that the metabolic capacity for nitrogen fixationis distributed across the phylum with the exception of the Ignavibacterium sp. [142, 206, 207].Ignavibacterium sp. is the deepest branching member of the Chlorobi and the only class of non-photosynthetic organisms in the phylum. Here we show that genes coding for the proteins requiredfor nitrogen fixation are present in the genomes of the photoferrotrophic Chlorobi, C. ferrooxidansand C. phaeoferrooxidans, putative photoferrotroph C. luteolum (Fig. 3.1), and in genomes of allother members of the Chlorobi (data not shown). Specifically, we identified one homolog of eachof the molybdenum-dependant nitrogenase proteins in all three photoferrotrophic Chlorobi. Nohomologs of the alternative vanadium or iron-only nitrogenase proteins were detected (PSI-Blast,expect threshold 10). These results indicate that the photoferrotrophic Chlorobi have the genomiccapacity to fix nitrogen. Furthermore, nitrogen fixation is wide spread among the Chlorobi, withall available Chlorobi genome sequences coding the necessary proteins apart from Ignavibacteriumalbum.Biochemical verification of nitrogen fixationTo test for the biochemical capacity to fix nitrogen during photosynthetic growth on Fe(II), nitrogenfree (below limit of detection ammonium, ammonia, nitrate, or nitrite) media was inoculatedwith C. phaeoferrooxidans or C. ferrooxidans. Both species were also grown in the standard growthmedium containing 5.6 mM ammonium [55], for comparison. Both species were able to fix59fmicb-08-01212 July 4, 2017 Time: 16:2 # 4Thompson et al. Metabolic Potential of Photoferrotrophic ChlorobiFIGURE 1 | Nitrogenase gene cassettes of the photoferrotrophic Chlorobi, detailing the position of each gene and the differences and similarities between the genecassettes.the iron and vanadium dependant nitrogenases may play a rolein molybdenum limiting environments (Joerger et al., 1988).Phylogenetic evidence suggests that the molybdenum-dependantversion of the enzyme evolved first (Boyd et al., 2011), which isfurther supported by the observation that organisms identified ashaving an iron or vanadium dependant nitrogenase all containa copy of the molybdenum-dependant nitrogenase (Raymondet al., 2004; Soboh et al., 2010). There are up to 25 proteins,depending on the species, required to assemble and regulatethe nitrogenase including three conserved structural proteins:NifD, NifK, and NifH. NifH is often used as the marker genefor nitrogen fixation in natural environments, due to its rolein the main enzyme structure and in cofactor recruitment.Further phylogenetic information, however, can be obtainedwhen all three structural proteins (NifDKH) are concatenateddue to increased sequence information and the conserved natureof all three proteins. Here we explored these key structuralproteins to test for the metabolic potential for nitrogen fixationin the photoferrotrophic Chlorobi. We compare nitrogen fixationin photoferrotrophic Chlorobi to the other members of thephylum Chlorobi and to representatives from all phyla capable ofnitrogen fixation to assess the evolutionary history of nitrogenasein relevant to photoferrotrophy in the Chlorobi and to placeconstraints on the possible role of photoferrotrophs in supplyingfixed nitrogen to the Precambrian oceans.Distribution of Nitrogen Fixation Pathways withinChlorobiPrevious analyses of Chlorobi genomes identified that themetabolic capacity for nitrogen fixation is distributed across thephylum with the exception of the Ignavibacterium sp. (Bryantet al., 2012; Liu et al., 2012; Hiras et al., 2016). Ignavibacteriumsp. is the deepest branching member of the Chlorobi and the onlyclass of non-photosynthetic organisms in the phylum. Here weshow that genes coding for the proteins required for nitrogenfixation are present in the genomes of the photoferrotrophicChlorobi C. ferrooxidans and C. phaeoferrooxidans, putativephotoferrotroph C. luteolum (Figure 1), and in genomes of allother members of the Chlorobi (data not shown). Specifically, weidentified one homolog of each of the molybdenum-dependantnitrogenase proteins in all three photoferrotrophic Chlorobi. Nohomologs of the alternative vanadium or iron-only nitrogenaseproteins were detected (PSI-Blast, expect threshold 10). Theseresults indicate that the photoferrotrophic Chlorobi have thegenomic capacity to fix nitrogen. Furthermore, nitrogen fixationis wide spread among the Chlorobi, with all available Chlorobigenome sequences coding the necessary proteins apart fromIgnavibacterium album.Biochemical Verification of Nitrogen FixationTo test for the biochemical capacity to fix nitrogen duringphotosynthetic growth on Fe(II), nitrogen free (below limitof detection ammonium, ammonia, nitrate, or nitrite) mediawas inoculated with C. phaeoferrooxidans or C. ferrooxidans.Both species were also grown in the standard growth mediumcontaining 5.6 mM ammonium (Hegler et al., 2008), forcomparison. Both species were able to fix nitrogen whilegrowing through photosynthetic Fe(II) oxidation withdoubling times of 45 and 36 h for C. phaeoferrooxidans andC. ferrooxidans, respectively (Figures 2A,B). Fe(II) oxidationFrontiers in Microbiology | www.frontiersin.org 4 July 2017 | Volume 8 | Article 1212Figure 3.1: Nitrogenase gene cassettes of the photoferrotrophic Chlorobi, detailing the position of each gene andthe differences and similarities between the gene cassettes.nitrogen while growing through photosynthetic Fe(II) oxidation with doubling times of 45 and36 hours for C. phaeof rrooxidans and C. ferrooxidans, r spectively (Fig. 3.2). Fe(II) oxid tion rates,during exponential growth phase, were 4.8 ± 0.33 µM h– 1 (C. phaeoferrooxidans) and 16 ± 0.56µM h– 1 (C. f rrooxidans) (Fig. 3.2). Growth under ammonium-rich conditions supported shorterdoubling times (15 and 27 hours) and higher rates of Fe(II) oxidation (50 ± 2.4 µM h– 1 and 23± 0.7 µM h– 1) for C. phaeoferr oxidans and C. ferrooxidans, respectively (Fig. 3.2). These resultsindicate that both pelagic C. phaeoferrooxidans and benthic C. ferrooxidans are capable of usingdinitrogen gas as their sole source of nitrogen during gr wth, b t that the need to fix N decreasesgrowth rates.To furt e explore the metabolic capacity of photofe rotrophic Chlorobi under both sets ofconditions, cell specific Fe(II) oxidation rates were calculated for each species. C. phaeoferrooxidansoxidized e(II) at 21.2 ± 1.4 fmol cell– 1 while fixing nitrogen and 47.8 ± 2.3 fmol cell– 1 underammonium-rich conditions. Conversely, C. ferrooxidans oxidized Fe(II) at 30.0 ± 0.9 fmol cell– 1and 29.4 ± 1.0 fmol cell– 1 in ammonium free and ammonium-rich media, respectively, withno appreciable difference during N-fixation. The apparent insensitivity of C. ferrooxidans toN-availability may be related to the presence of its co-culture partner, Geospirillum sp. KoFum.Further experiments with KoFum could help constrain its possible role in N metabolism within60fmicb-08-01212 July 4, 2017 Time: 16:2 # 5Thompson et al. Metabolic Potential of Photoferrotrophic ChlorobiFIGURE 2 | Fe(II) concentrations and cell counts over time for both C. phaeoferrooxidans (A,C) and C. ferrooxidans (B,D) under two sets of media: no bioavailablenitrogen – N2 as sole nitrogen source (A,B) and ammonium rich (C,D). Data points used to calculate growth rates and Fe(II) oxidation rates are highlighted in eachpanel.rates, during exponential growth phase, were 4.8± 0.33µM/hour(C. phaeoferrooxidans) and 16± 0.56 µM/hour (C. ferrooxidans)(Figures 2A,B). Growth under ammonium-rich conditionssupported shorter doubling times (15 and 27 h) and higher ratesof Fe(II) oxidation (50 ± 2.4 µM/hour and 23 ± 0.7 µM/hour)for C. phaeoferrooxidans and C. ferrooxidans, respectively(Figures 2C,D). These results indicate that both pelagicC. phaeoferrooxidans and benthic C. ferrooxidans are capableof using dinitrogen gas as their sole source of nitrogen duringgrowth, but that the need to fix N decreases growth rates.To further explore the metabolic capacity of photoferrotrophicChlorobi under both sets of conditions, cell specific Fe(II)oxidation rates were calculated for each species. C. phaeoferrooxidans oxidized Fe(II) at 21.2 ± 1.4 fmol/cell while fixingnitrogen and 47.8 ± 2.3 fmol/cell under ammonium-richconditions. Conversely, C. ferrooxidans oxidized Fe(II) at30.0 ± 0.9 fmol/cell and 29.4 ± 1.0 fmol/cell in ammonium freeand ammonium-rich media, respectively, with no appreciabledifference during N-fixation. The apparent insensitivity ofC. ferrooxidans to N-availability may be related to the presenceof its co-culture partner, Geosprillum sp. KoFum. Furtherexperiments with KoFum could help constrain its possiblerole in N metabolism within the co-culture The observationthat C. phaeoferrooxidans has lower cell specific growth ratesunder N scarcity, however, implies lower growth yields duringN fixation. Both species are ultimately capable of growthFrontiers in Microbiology | www.frontiersin.org 5 July 2017 | Volume 8 | Article 1212Figure 3.2: Biochemical verification of Nitrogen fixation with Fe(II) concentrations and cell counts over time forboth C. phaeoferrooxidans (A,C) and C. ferrooxidans (B,D) under two sets of media: no bioavailable nitrogen –N2 as sole nitrogen source (A,B) and ammonium rich (C,D). Data points used to calculate growth rates andFe(II) oxidation rates are hig lighted in each panel.the co-culture The observation that C. phaeoferrooxidans has lower cell sp cific growth rates underN scarcity, however, implies lower growth yields during N fixation. Both species are ultimatelycapable of growth and Fe(II) oxidation while fixing nitrogen but the differential response ofcell specific i on oxidation rat s to N-scarcity implies that nutrient availabi ity can influence theecology of photoferrotrophs in the environment.Evolutionary history of nitrogen fixation in the ChlorobiTo assess the evolutionary history of nitrogen fixation in the Chlorobi we tested for horizontal genetransfer (HGT) within the photoferrotrophic Chlorobi and conducted phylogenetic analyses of Nif61proteins, which we compared to small subunit 16S ribosomal RNA (SSU rRNA) genes. Deviationsin the branching orders between these phylogenies would indicate nonvertical inheritance andHGT. To test for horizontal transfer of Nif genes in the photoferrotrophic Chlorobi, we lookedfor characteristic signatures of HGT within nif gene cassettes. Codon adaptation index (CAI)values, a metric used to describe differences in codon usage between specific genes and thegenomic background, were calculated for all individual nif genes belonging to C. ferrooxidans,C. phaeoferrooxidans, and C. luteolum. All CAI values were greater than the threshold value, 0.70,below which HGT is indicated (Table 3.1). In addition, GC contents of nif genes were very similarto GC contents of genomic backgrounds providing no evidence for HGT (Table 3.1). Our analysesalso failed to identify tRNAs, transposases, or other genetic elements commonly associated withgene mobility in close proximity (within 5000 bp) to the nitrogenase gene cassette in any of thephotoferrotrophic Chlorobi. The general lack of tRNAs or transposases near the nif cassettesin the photoferrotrophic Chlorobi, combined with super threshold CAI values and nif gene GCcontents that are homogenous against genomic backgrounds, imply nif gene acquisition throughvertical decent.Table 3.1: Codon adaptation index (CAI) for Chlorobi nitrogenases. Gene length (bp), codon adaptationindex (CAI), and GC content (%) for each of the genes in the nitrogenase cassette.fmicb-08-01212 July 4, 2017 Time: 16:2 # 6Thompson et al. Metabolic Potential of Photoferrotrophic ChlorobiTABLE 1 | Gene length (bp), codon adaptation index (CAI), and GC content (%) for each of the genes in the nitrogenase cassette.Gene C. phaeoferrooxidans C. ferrooxidans C. luteolumLength (bp) CAI GC content (%) Length (bp) CAI GC content (%) Length (bp) CAI GC content (%)NifB 1275 0.80 52.63 1275 0.76 53.18 1263 0.75 60.89NifN 1353 0.79 53.22 1350 0.75 53.48 1353 0.74 59.42NifE 1362 0.79 50.07 1362 0.75 49.63 1362 0.74 57.34NifK 1383 0.81 53.51 1383 0.73 52.78 1380 0.76 60.14NifD 1635 0.79 49.54 1635 0.76 49.66 1641 0.77 57.22PII regulator 378 0.81 50.00 378 0.72 50.00 378 0.72 59.26PII regulator 357 0.76 49.30 357 0.73 48.74 357 0.70 56.30NifH 825 0.83 49.58 825 0.79 49.21 825 0.81 59.03Each parameter was calculated for all three photoferrotrophic Chlorobi, whose whole genome GC contents are: 49.72% for Chlorobium phaeoferrooxidans, 49.9% forC. ferrooxidans, and 58.1% for C. luteolum.FIGURE 3 | Phylogenies of the Chlorobi and Bacteroidetes using (A) the concatenated NifDKH proteins and (B) 16S rRNA with bootstrap values shown at eachnode (maximum likelihood/maximum parsimony). The blue colors delineate the organisms of the Phylum Chlorobi, with each shade representing a different genus,while the purple color delineates the Phylum Bacteroidetes. The orange lines indicate the position of the photoferrotrophic Chlorobi. The trees were rooted with fourcyanobacterial organisms. Note: Azobacteroides pseudotrichonymphae CFP2 is abbreviated from Candidatus Azobacteroides pseudotrichonymphae genomovar.CFP2.and Fe(II) oxidation while fixing nitrogen but the differentialresponse of cell specific iron oxidation rates to N-scarcityimplies that nutrient availability can influence the ecology ofphotoferrotrophs in the environment.Evolutionary History of Nitrogen Fixation in theChlorobiTo assess the evolutionary history of nitrogen fixation in theChlorobi we tested for horizontal gene transfer (HGT) withinthe photoferrotrophic Chlorobi and conducted phylogeneticanalyses of Nif proteins, which we compared to small subunit16S ribosomal RNA (SSU rRNA) genes. Deviations in thebranching orders between these phylogenies would indicate non-vertical inheritance and HGT. To test for horizontal transferof Nif genes in the photoferrotrophic Chlorobi, we lookedfor characteristic signatures of HGT within nif gene cassettes.Codon adaptation index (CAI) values, a metric used to describedifferences in codon usage between specific genes and thegenomic background, were calculated for all individual nifgenes belonging to C. ferrooxidans, C. phaeoferrooxidans, andC. luteolum. All CAI values were greater than the thresholdvalue, 0.70, below which HGT is indicated (Table 1). In addition,GC contents of nif genes were very similar to GC contents ofgenomic backgrounds providing no evidence for HGT (Table 1).Our analyses also failed to identify tRNAs, transposases, orother genetic elements commonly associated with gene mobilityin close proximity (within 5000 bp) to the nitrogenase genecassette in any of the photoferrotrophic Chlorobi. The generallack of tRNAs or transposases near the nif cassettes in thephotoferrotrophic Chlorobi, combined with super threshold CAIvalues and nif gene GC contents that are homogenous againstgenomic backgrounds, imply nif gene acquisition throughvertical decent.To test the evolutionary history of the nif genes in theChlorobi, we conducted phylogenetic analyses of concatenatedNifDKH proteins of all cultured and sequenced Chlorobi thathave the genomic potential to fix dinitrogen. The Chlorobisequences were aligned with selected sequences from thenext closest phylum – Bacteroidetes – and the tree wasrooted using four Cyanobacterial species as an out-group(Figure 3A). The genus Chlorobium, which includes all threephotoferrotrophic Chlorobi, the genus Chlorobaculum, andFrontiers in Microbiology | www.frontiersin.org 6 July 2017 | Volume 8 | Article 1212To test the evolutionary history of the nif genes in the Chlorobi, we conducted phylogeneticanalyses of concatenated NifDKH proteins of all cultured and sequenced Chlorobi that havethe genomic potential to fix dinitrogen. The Chlorobi sequences were aligned with selectedsequences from the next closest phylum – Bacteroidetes – and the tree was rooted using four62Cyanobacterial species as an out-group (Fig. 3.3). The genus Chlorobium, which includes allthree photoferrotrophic Chlorobi, the genus Chlorobaculum, and the genus Prosthecochloris allform monophyletic groups that are collectively part of the phylum Chlorobi clade. Likewise, theBacteroidetes form a monophyletic group and share a common ancestor with the members of thephylum Chlorobi. Furthermore, when the NifDKH tree is compared to a 16S rRNA tree of thesame organisms (Fig. 3.3), all of the genera within the phylum Chlorobi branch in an identicalorder to those in the 16S rRNA phylogeny. The phylogenetic relationship between the Chlorobiand Bacteroidetes is the same for both NifDKH and 16S rRNA sequences, indicating that thecommon ancestor to the phyla Chlorobi and Bacteroidetes likely contained a nitrogenase andtherefore the ability to fix dinitrogen. Ignavibacterium sp., the sole members of the phylum Chlorobiwho do not posses a nitrogenase, likely lost the capability to fix nitrogen as the remainder of theChlorobi and the phylum Bacteroidetes bracket the phylogenetic position of the Ignavibacteriumsp. Taken together, available data imply vertical decent.fmicb-08-01212 July 4, 2017 Time: 16:2 # 6Thompson et al. Metabolic Potential of Photoferrotrophic ChlorobiTABLE 1 | Gene length (bp), codon adaptation index (CAI), and GC content (%) for each of the genes in the nitrogenase cassette.Gene C. phaeoferrooxidans C. ferrooxidans C. luteolumLength (bp) CAI GC content (%) Length (bp) CAI GC content (%) Length (bp) CAI GC content (%)NifB 1275 0.80 52.63 1275 0.76 53.18 1263 0.75 60.89NifN 1353 0.79 53.22 1350 0.75 53.48 1353 0.74 59.42NifE 1362 0.79 50.07 1362 0.75 49.63 1362 0.74 57.34NifK 1383 0.81 53.51 1383 0.73 52.78 1380 0.76 60.14NifD 1635 0.79 49.54 1635 0.76 49.66 1641 0.77 57.22PII regulator 378 0.81 50.00 378 0.72 50.00 378 0.72 59.26PII regulator 357 0.76 49.30 357 0.73 48.74 357 0.70 56.30NifH 825 0.83 49.58 825 0.79 49.21 825 0.81 59.03Each parameter was calculated for all three photoferrotrophic Chlorobi, whose whole genome GC contents are: 49.72% for Chlorobium phaeoferrooxidans, 49.9% forC. ferrooxidans, and 58.1% for C. luteolum.FIGURE 3 | Phylogenies of the Chlorobi and Bacteroidetes using (A) the concatenated NifDKH proteins and (B) 16S rRNA with bootstrap values shown at eachnode (maximum likelihood/maximum parsimony). The blue colors delineate the organisms of the Phylum Chlorobi, with each shade representing a different genus,while the purple color delineates the Phylum Bacteroidetes. The orange lines indicate the position of the photoferrotrophic Chlorobi. The trees were rooted with fourcyanobacterial organisms. Note: Azobacteroides pseudotrichonymphae CFP2 is abbreviated from Candidatus Azobacteroides pseudotrichonymphae genomovar.CFP2.and Fe(II) oxidation while fixing nitrogen but the differentialresponse of cell specific iron oxidation rates to N-scarcityimplies that nutrient availability can influence the ecology ofphotoferrotrophs in the environment.Evolutionary History of Nitrogen Fixation in theChlorobiTo assess the evolutionary history of nitrogen fixation in theChlorobi we tested for horizontal gene transfer (HGT) withinthe photoferrotrophic Chlorobi and conducted phylogeneticanalyses of Nif proteins, which we compared to small subunit16S ribosomal RNA (SSU rRNA) genes. Deviations in thebranching orders between these phylogenies would indicate non-vertical inheritance and HGT. To test for horizontal transferof Nif genes in the photoferrotrophic Chlorobi, we lookedfor characteristic signatures of HGT within nif gene cassettes.Codon adaptation index (CAI) values, a metric used to describedifferences in codon usage between specific genes and thegenomic background, were calculated for all individual nifgenes belonging to C. ferrooxidans, C. phaeoferrooxidans, andC. luteolum. All CAI values were greater than the thresholdvalue, 0.70, below which HGT is indicated (Table 1). In addition,GC contents of nif genes were very similar to GC contents ofgenomic backgrounds providing no evidence for HGT (Table 1).Our analyses also failed to identify tRNAs, transposases, orother genetic elements commonly associated with gene mobilityin close proximity (within 5000 bp) to the nitrogenase genecassette in any of the photoferrotrophic Chlorobi. The generallack of tRNAs or transposases near the nif cassettes in thephotoferrotrophic Chlorobi, combined with super threshold CAIvalues and nif gene GC contents that are homogenous againstgenomic backgrounds, imply nif gene acquisition throughvertical decent.To test the evolutionary history of the nif genes in theChlorobi, we conducted phylogenetic analyses of concatenatedNifDKH proteins of all cultured and sequenced Chlorobi thathave the genomic potential to fix dinitrogen. The Chlorobisequences were aligned with selected sequences from thenext closest phylum – Bacteroidetes – and the tree wasrooted using four Cyanobacterial species as an out-group(Figure 3A). The genus Chlorobium, which includes all threephotoferrotrophic Chlorobi, the genus Chlorobaculum, andFrontiers in Microbiology | www.frontiersin.org 6 July 2017 | Volume 8 | Article 1212Figure 3.3: Nitrogenase phylogenies of the Chlorobi and Bacteroidetes. Phylogenies of the Chlorobi andBacteroidetes using (A) the concatenated NifDKH proteins and (B) 16S rRNA with bootstrap values shownt each node (maximum likelihoo /maximum parsimony). The blue colors delineate t e org nisms ofthe Phylum Chlorobi, with each shade representing a different genus, while the purple color delineatesthe Phylum Bacteroidetes. The orange lines indicate the position of the photoferrotrophic Chlorobi. Thetrees were rooted with four cyanobacterial organisms. Note: Azobacteroides pseudotrichonymphae CFP2 isabbreviated from Candidatus Azobacteroides pseudotrichonymphae genomovar CFP2.Accepting largely vertical descent of NifDKH from the common ancestor of the Chlorobi andBacteroidetes, NifDKH must have emerged within this line of descent before the divergence ofthe Chlorobi and Bacteroidetes. The timing of this divergence has been estimated using a whole63genome molecular clock [208] to between 3 and 1.6 Ga, which implies the capacity to fix N in theancestors of the Chlorobi before this time. Independent N isotope data from metasedimentarykerogen implies N fixation by at least 3.2 Ga [32]. Combined, the evidence for vertical inheritanceof NifDKH in the Chlorobi on the taxonomic levels of genus and phylum, the timing of divergencebetween the Chlorobi and the Bacteroidetes, and the N isotope record, imply that ancestors ofmodern Chlorobi likely had capacity to fix nitrogen in the iron-rich oceans of the paleoproterozoicand perhaps as early as the mesoarchean eras.To place N fixation in the Chlorobi, and Bacteroidetes, within the broader context of nitrogenaseevolution in general, we conducted further phylogenetic analyses using a greater diversity oforganisms. We analyzed the NifDKH phylogeny using two to four representatives from everyphylum that had a cultured and sequenced species with previously documented genomic potentialfor nitrogen fixation (Fig. 3.4). This phylogeny places the Nif proteins found in the Chlorobiand Bacteroidetes in a single clade, supporting their emergence from a common ancestor andthe vertical inheritance of NifDKH from this ancestor. The phylogeny of the NifDKH protein is,however, incongruent with that of the 16S rRNA gene from the same organisms (Fig. 3.4). Whilethe Chlorobi and Bacteroidetes group together in both phylogenies, the Spirochetes, Chloroflexi,and Firmicutes also group with the Chlorobi in the NifDKH phylogeny, but belong to distinctclades in the 16S rRNA gene phylogeny. The differences between these phylogenies confoundfurther constraints on the evolutionary history of nitrogenase within the Chlorobi based onphylogeny and add to the overwhelming evidence for horizontal transfer of NifDKH genes[3, 187, 209].Ecology of nitrogen fixation in Chlorobi, past and presentMembers of the phylum Chlorobi underpin biological production in many modern anoxicenvironments, both sulfidic [181, 210–213] and ferruginous [48, 67], through their ability toharness light energy and fix inorganic carbon into biomass, even at low light intensities [138].Chlorobi further contribute to biogeochemical cycling in these systems through the acquisition andredistribution of essential nutrients, such as nitrogen. This ecological role would have extended toglobal scales in the low oxygen Precambrian oceans. Our analyses confirm the genomic potential64fmicb-08-01212 July 4, 2017 Time: 16:2 # 7Thompson et al. Metabolic Potential of Photoferrotrophic ChlorobiFIGURE 4 | Phylogenies of (A) the concatenated NifDKH proteins and (B) 16S rRNA for two to four representatives of several nitrogen-fixing phyla with bootstrapvalues shown at each node (maximum likelihood/maximum parsimony). The green color delineates the Chlorobi/Bacteroidetes monophyletic grouping, while theorange line indicates the position of the photoferrotrophic Chlorobi. Note: Azobacteroides pseudotrichonymphae CFP2 is abbreviated from CandidatusAzobacteroides pseudotrichonymphae genomovar. CFP2.the genus Prosthecochloris all form monophyletic groups thatare collectively part of the phylum Chlorobi clade. Likewise,the Bacteroidetes form a monophyletic group and share acommon ancestor with the members of the phylum Chlorobi.Furthermore, when the NifDKH tree is compared to a 16SrRNA tree of the same organisms (Figure 3B), all of thegenera within the phylum Chlorobi branch in an identicalorder to those in the 16S rRNA phylogeny. The phylogeneticrelationship between the Chlorobi and Bacteroidetes is the samefor both NifDKH and 16S rRNA sequences, indicating thatthe common ancestor to the phyla Chlorobi and Bacteroideteslikely contained a nitrogenase and therefore the ability to fixdinitrogen. Ignavibacterium sp., the sole members of the phylumChlorobi who do not posses a nitrogenase, likely lost thecapability to fix nitrogen as the remainder of the Chlorobi andthe phylum Bacteroidetes bracket the phylogenetic position of theIgnavibacterium sp. Taken together, available data imply verticaldecent.Accepting largely vertical descent of NifDKH from thecommon ancestor of the Chlorobi and Bacteroidetes, NifDKHmust have emerged within this line of descent before thedivergence of the Chlorobi and Bacteroidetes. The timing ofthis divergence has been estimated using a whole genomemolecular clock (David and Alm, 2011) to between 3 and1.6 Gya, which implies the capacity to fix N in the ancestorsof the Chlorobi before this time. Independent N isotope datafrom metasedimentary kerogen implies N fixation by at least3.2 Gy (Stüeken et al., 2015). Combined, the evidence for verticalinheritance of NifDKH in the Chlorobi on the taxonomic levelsof genus and phylum, the timing of divergence between theChlorobi and the Bacteroidetes, and the N isotope record, implythat ancestors of modern Chlorobi likely had capacity to fixnitrogen in the iron-rich oceans of the paleoproterozoic andperhaps as early as the mesoarchean eras.To place N fixation in the Chlorobi, and Bacteroidetes, withinthe broader context of nitrogenase evolution in general, weconducted further phylogenetic analyses using a greater diversityof organisms. We analyzed the NifDKH phylogeny using twoto four representatives from every phylum that had a culturedand sequenced species with previously documented genomicpotential for nitrogen fixation (Figure 4A). This phylogeny placesthe Nif proteins found in the Chlorobi and Bacteroidetes ina single clade, supporting their emergence from a commonancestor and the vertical inheritance of NifDKH from thisancestor. The phylogeny of the NifDKH protein is, however,incongruent with that of the 16S rRNA gene from the sameorganisms (Figure 4B). While the Chlorobi and Bacteroidetesgroup together in both phylogenies, the Spirochetes, Chloroflexi,and Firmicutes also group with the Chlorobi in the NifDKHphylogeny, but belong to distinct clades in the 16S rRNA genephylogeny. The differences between these phylogenies confoundfurther constraints on the evolutionary history of nitrogenasewithin the Chlorobi based on phylogeny and add to theoverwhelming evidence for horizontal transfer of NifDKH genes(Raymond et al., 2004; Falkowski et al., 2008; Boyd and Peters,2013).Ecology of Nitrogen Fixation in Chlorobi, Past andPresentMembers of the phylum Chlorobi underpin biologicalproduction in many modern anoxic environments, bothsulfidic (Overmann, 1997; Tonolla et al., 2004; Gregersen et al.,2009; Kondo et al., 2009; Meyer et al., 2011) and ferruginous(Walter et al., 2014; Llirós et al., 2015), through their ability toFrontiers in Microbiology | www.frontiersin.org 7 July 2017 | Volume 8 | Article 1212Figure 3.4: Nitrogenase phylogenies of multiple phyla. Phylogenies of (A) the concatenated NifDKH pro-teins and (B) 16S rRNA for two to four representatives of several nitrogen-fixing phyla with bootstrapvalues shown at each node (maximum likelihood/maximum parsimony). The green color delineatesthe Chlorobi/Bacteroidetes monophyletic grouping, while the orange line indicates the position of thephotoferrotrophic Chlorobi. Note: Azobacteroides pseudotrichonymphae CFP2 is abbreviated from CandidatusAzobacte ides pseudotrichonymphae genomovar CFP2.to fix N in all but one of the Chlorobi lineages and directly demonstrate the capacity of thephotoferrotro ic Chloro i to fix dinitrogen as their sole source of nitrogen while oxi izingFe(II). Rates of Fe(II) xidation are, however, slower when photosynthetic rowth is upportedthrough N-fixati n rather t an ammonium as imilation. To test the impact of slowe rates ofFe(II) oxidation, and ther f re growth, on t e deposition of BIFs, we ran our cell counts and Fe(II)ox dati n rates through the calculation outlined by Ko hauser et. al., 2002 [8]. Our dat indi atesthat both photof rr trophic strains would be c pable of generating even the largest BIFs (i.e., theHamersley BIF) with maximum of 2.44 · 10 3 photoferrotrophic cells/mL r quired in the basin.Thus, photoferrotrophic growth coupled t N-fixation could support BIF deposition, even in theface of nitrogen scarcity.Chlorobium phaeoferrooxidans, and C. ferrooxidans exhibit differential responses to N scarcitythat manifest in different cell specific Fe(II) oxidation rates and different ratio’s between micro-bial growth (cell doubling times) and Fe(II) oxidation. C. phaeoferrooxidans has a cell doublingtime to Fe(II) oxidation ratio of 9.4 under N-fixing conditions compared to 0.3 when there is65ample ammonium, whereas C. ferrooxidans has comparable ratio’s of 2.3 and 1.2 for N-fixingand ammonium-rich conditions comparatively. This differential response indicates that underammonium-rich conditions C. phaeoferrooxidans grows more efficiently (i.e., with a higher growthyield) whereas when N-fixation is required C. ferrooxidans grows more efficiently. This createsniches for each microorganism defined by N availability. The differential response also implies thatthe stoichiometry of Fe-oxidation to biomass production and cell growth is partly decoupled anddepends on N availability. Essentially, this decoupling means that more Fe(II) is oxidized to pro-duce an individual cell during growth supported by N-fixation than by ammonium assimilation.Such a decoupling thus requires either the diversion of reducing equivalents (NADH) producedduring photosynthesis into compounds not used directly in cell growth, or that cell growth anddivision requires more fixed carbon during N-fixation. The former could include conversion ofN2 to ammines and the biosynthesis of cell exudates, and the latter might include the biosynthesisof cellular proteins needed to conduct N-fixation. Such a decoupling would influence the overallbiogeochemical functioning and ecology of ecosystems supported through primary production byphotoferrotrophy. The overall activity of the marine biosphere through the Precambrian Eons maythus have been influenced by the availability of fixed N to photoferrotrophs.3.4.2 Assimilatory sulfate reduction (ASR)Sulfate ions are biologically inert and organisms expend tremendous energy ‘activating’ sulfatefor three main functions: (1) reduction and incorporation into amino acids; (2) condensationand incorporation into sulfolipids and other small molecules; and (3) for dissimilatory sulfaterespiration. In addition to the reduction of sulfate, organisms can acquire organic sulfur com-pounds like amino acids, and hydrogen sulfide directly from the environment. Acquisition ofthese reduced sulfur compounds can considerably reduce the expenditure of energy on sulfuracquisition. Here we focus on the first two assimilatory pathways and the capacity for reductivesulfur assimilation in the photoferrotrophic Chlorobi. The proteins required to complete an entireASR pathway include: CysD, the sulfateadenyl transferase that activates sulfate to form APS;CysN which catalyzes GTP hydrolysis providing the energy needed to adenylate imported sulfate;CysC (a domain of CysN), the APS kinase that phosphorylates APS to PAPS; and CysH, the APS66reductase which reduces the sulfur in APS to sulfite. We have explored the metabolic potential forsulfate assimilation in the genomes of the photoferrotrophic Chlorobi and directly tested sulfateincorporation into biomass.Distribution of ASR pathways within ChlorobiPrevious analyses of Chlorobi genomes identified the metabolic capacity for ASR in C. ferroox-idans and C. luteolum [194]. Using all currently available genomic information we identifiedcomponents of the ASR pathways distributed throughout the Chlorobi (Table 3.2). We find thatthe photoferrotrophic Chlorobi, C. ferrooxidans and C. phaeoferrooxidans, as well as putativephotoferrotroph C. luteolum, all possess the necessary proteins for ASR – CysD, CysN/C, CysH –and therefore have the potential capacity to synthesize amino acids from exogenous sulfate (Fig.3.5). Notably, the presence of both CysD and CysN indicate that sulfate activation to APS in theseChlorobi is coupled to GTP hydrolysis. Sulfate assimilation in the Chlorobi, therefore, offsetsthe energetic expense associated with sulfate activation. The presence of CysN/C indicates themetabolic potential to phosphorylate APS to PAPS implying that these strains might have capacityto synthesize sulfate-containing compounds like sulfolipids. Finally, while components of assimi-latory sulfate metabolisms are more broadly distributed throughout the Chlorobi, genes codingfor key components of the pathway are mostly missing implying a lack of capacity for sulfateassimilation outside the photoferrotrophic Chlorobi (Table 3.2). Given the metabolic potential forASR in the photoferrotrophic Chlorobi, we sought to biochemically verify this process.Biochemical verification of ASRChlorobium phaeoferrooxidans and Chlorobium ferrooxidans are both known to grow in media wheresulfur is supplied exclusively in the form of sulfate, which directly demonstrates the physiologicalcapacity for sulfate assimilation. We quantitatively tested this capacity by measuring the uptakeof 35S labeled sulfate in low sulfate growth media. C. phaeoferrooxidans, indeed took up 35Slabeled sulfate into TCA extractable biomass, demonstrating assimilatory reduction of sulfateand its incorporation into amino acids. Over the course of these sulfate uptake experiments, C.phaeoferrooxidans oxidized 3070 µM Fe(II). This implies the fixation of 770 µM C, based on the 4:167Table 3.2: Green sulfur bacteria ASR genes. Assimilatory (CysD, CysNC, CysH) and dissimilatory (Sat,AprAB, DsrAB) sulfur proteins present in the genomes of green sulfur bacteria.fmicb-08-01212 July 4, 2017 Time: 16:2 # 9Thompson et al. Metabolic Potential of Photoferrotrophic ChlorobiTABLE 2 | Assimilatory (orange) and dissimilatory (blue) sulfur proteins present in the genomes of green sulfur bacteria.Organism CysD Sat CysNC CysH AprAb DsrABChlorobium phaeoferrooxidans KB01 + − + + − −Chlorobium ferrooxidans DSM 13101 + − + + − −Chlorobium luteolum DSM 273 + − + + − +Chlorobium phaeovibrioides DSM 265 + − + − − +Prosthecochloris aestuarii DSM 271 + − + − − +Chlorobium phaeobacteroides BS1 − + + − + +Chlorobium chlorochromatii CaD3 − + + − + +Pelodictyon phaeoclathratiforme − + − − + +Chlorobium tepidum TLS − + − − + +Chlorobium phaeobacteroides DSM 266 − − − − − +Chlorobium limicola DSM 245 − − − − − +Chlorobaculum parvum NCIB 8327 − − − − − +Chloroherpeton thalassium − − − − − −Ignavibacterium album − − − − − −FIGURE 5 | Assimilatory sulfate reduction (ASR) gene cassettes for the photoferrotrophic Chlorobi, detailing the position of each gene and the differences andsimilarities between the gene cassettes.Fe(II). This implies the fixation of 770 µM C, based onthe 4:1 stoichiometry between Fe(II) oxidation and C fixationobserved for C. phaeoferrooxidans during growth on Fe(II),and for photoferrotrophic organisms, more generally (Widdelet al., 1993). A corresponding total of 3 µM S was fixeddemonstrating a ratio of 260:1 C to S, which we take asapproximately indicative of the S content ofC. phaeoferrooxidans.There are few data to compare with, but our results suggestthat C. phaeoferrooxidans has relatively low S quotas comparedto aquatic and cultured bacteria (C:S from 10–60) (Fagerbakkeet al., 1996) and particulate organic matter from the NorthPacific (C:S of 50) (Chen et al., 1996). By analogy toC. phaeoferrooxidans, photoferrotrophic Chlorobi likely havecapacity to fix sulfate into biomass under low sulfate conditions,Frontiers in Microbiology | www.frontiersin.org 9 July 2017 | Volume 8 | Article 1212stoichiometry between Fe(II) oxidation and C fixation observed for C. phaeoferrooxidans duringgrowth on Fe(II), and for photoferrotrophic organisms, more generally [20]. A correspondingtotal of 3 µM S was fixed demonstrating a ratio of 260:1 C to S, which we take as approximatelyindicative of the S content of C. phaeoferrooxidans. There are few data to compare with, but ourresults suggest that C. phaeoferrooxidans has relatively low S quotas compared to aquatic andcultured bacteria (C:S from 10–60) [214] and particulate organic matter from the North Pacific (C:Sof 50) [215]. By analogy to C. phaeoferrooxidans, photoferrotrophic Chlorobi likely have capacity tofix sulfate into biomass under low sulfate conditions, which they appear well adapted to do basedon minimal cellular sulfur quotas in comparison to other bacteria and marine organic material.Evolutionary history of ASR in the ChlorobiTo test for horizontal transfer of ASR genes to the photoferrotrophic Chlorobi, we searched forcharacteristic sign tures of HGT within the ASR cassettes and conduct d phylog n ti a alysesof ASR genes, which we compared to 16S rRNA gene phylogenies. The CAI value for eachof the ASR genes belonging to C. ferrooxidans and C. phaeoferrooxidans were all greater than thethreshold value, 0.70, below which HGT is indicated (Table 3.3). ASR genes in C. luteolum, however,had sub-threshold CAI values, as low as 0.53, indicating p ssible ASR gene acquisition throughhorizontal transfer. The GC contents of ASR genes for all three species were very similar to GC68fmicb-08-01212 July 4, 2017 Time: 16:2 # 9Thompson et al. Metabolic Potential of Photoferrotrophic ChlorobiTABLE 2 | Assimilatory (orange) and dissimilatory (blue) sulfur proteins present in the genomes of green sulfur bacteria.Organism CysD Sat CysNC CysH AprAb DsrABChlorobium phaeoferrooxidans KB01 + − + + − −Chlorobium ferrooxidans DSM 13101 + − + + − −Chlorobium luteolum DSM 273 + − + + − +Chlorobium phaeovibrioides DSM 265 + − + − − +Prosthecochloris aestuarii DSM 271 + − + − − +Chlorobium phaeobacteroides BS1 − + + − + +Chlorobium chlorochromatii CaD3 − + + − + +Pelodictyon phaeoclathratiforme − + − − + +Chlorobium tepidum TLS − + − − + +Chlorobium phaeobacteroides DSM 266 − − − − − +Chlorobium limicola DSM 245 − − − − − +Chlorobaculum parvum NCIB 8327 − − − − − +Chloroherpeton thalassium − − − − − −Ignavibacterium album − − − − − −FIGURE 5 | Assimilatory sulfate reduction (ASR) gene cassettes for the photoferrotrophic Chlorobi, detailing the position of each gene and the differences andsimilarities between the gene cassettes.Fe(II). This implies the fixation of 770 µM C, based onthe 4:1 stoichiometry between Fe(II) oxidation and C fixationobserved for C. phaeoferrooxidans during growth on Fe(II),and for photoferrotrophic organisms, more generally (Widdelet al., 1993). A corresponding total of 3 µM S was fixeddemonstrating a ratio of 260:1 C to S, which we take asapproximately indicative of the S content ofC. phaeoferrooxidans.There are few data to compare with, but our results suggestthat C. phaeoferrooxidans has relatively low S quotas comparedto aquatic and cultured bacteria (C:S from 10–60) (Fagerbakkeet al., 1996) and particulate organic matter from the NorthPacific (C:S of 50) (Chen et al., 1996). By analogy toC. phaeoferrooxidans, photoferrotrophic Chlorobi likely havecapacity to fix sulfate into biomass under low sulfate conditions,Frontiers in Microbiology | www.frontiersin.org 9 July 2017 | Volume 8 | Article 1212Figure 3.5: Assimilatory sulfate reduction (ASR) gene cassettes for the photoferrotrophic Chlorobi, detailing theposition of each gene and the differences and similarities between the gene cassettes.contents of th ir respective gen mic b ckgrounds, providi g no evidence for HGT (Table 3.3).Collectively, these data provide little evidence for the lateral cquisition f ASR gene cassettesin the photoferrotrophic Chlorobi, although the evidence for vertical descent is greater in C.phaeofer ooxidans and C. ferrooxidans than in C. luteolum. A single transposase (Fig. 3.5) was foundon a contig adjacent to that hosting the ASR gene cassette in C. phaeoferrooxidans. The general lackof tRNAs or transposases near the ASR cassettes in the photoferrotrophic Chlorobi combined withsuper threshold CAI values and ASR gene GC contents that are homogenous against the genomicbackgrounds, implies ASR gene acquisition through vertical decent.To further test the evolutionary history of ASR, the CysH protein was analyzed to examine thephylogenetic relationship between the proteins used in the photoferrotrophic Chlorobi and ASRin other organisms. The photoferrotrophic Chlorobi grouped together forming a monophyleticclade within the CysH phylogeny (Fig. 3.6). The photoferrotrophic Chlorobi exhibit congruentphylogenies between the CysH protein and 16S rRNA gene (Fig. 3.6), providing further evidencein support of vertical inheritance of the ASR pathway in the photoferrotrophic Chlorobi. The69Table 3.3: Codon adaptation index (CAI) for Chlorobi ASR genes. Gene length (bp), CAI, and GC content(%) for each of the genes in the ASR cassette.fmicb-08-01212 July 4, 2017 Time: 16:2 # 11Thompson et al. Metabolic Potential of Photoferrotrophic ChlorobiTABLE 3 | Gene length (bp), CAI, and GC content (%) for each of the genes in the ASR cassette.Gene C. phaeoferrooxidans C. ferrooxidans C. luteolumLength (bp) CAI GC content (%) Length (bp) CAI GC content (%) Length (bp) CAI GC content (%)CysH 714 0.78 54.34 714 0.78 54.62 753 0.59 56.97CsyD 882 0.80 56.12 882 0.78 55.56 915 0.61 57.38CsyN/C 1800 0.81 54.28 1800 0.77 53.67 1800 0.69 58.06Siroheme synthase 453 0.81 54.08 453 0.71 52.98 453 0.53 57.17Uroporphyrin-IIIC-methyltransferase1287 0.73 57.96 1287 0.73 55.40 258 0.62 57.36CysA 1074 0.77 53.26 1074 0.80 52.42 1074 0.67 59.22CysW 870 0.79 53.22 870 0.79 52.76 870 0.64 58.74CysT 834 0.79 52.64 834 0.81 53.36 834 0.66 58.03Sulfate transporter 1020 0.82 53.14 1020 0.79 52.65 1008 0.75 59.52Each parameter was calculated for all three photoferrotrophic Chlorobi, whose whole genome GC contents are: 49.72% for C. phaeoferrooxidans, 49.9% forC. ferrooxidans, and 58.1% for C. luteolum.FIGURE 6 | Phylogenies of (A) the CysH protein and (B) 16S rRNA with bootstrap values shown at each node (maximum likelihood/maximum parsimony). Theorange line indicates the position of the photoferrotrophic Chlorobi. The trees are rooted with two Archaeal species.may inform evolutionary histories and should be tested in thefuture.OUTLOOKPhotoferrotrophy links the C and Fe biogeochemical cyclesthrough coupled CO2 fixation and Fe(II) oxidation and haslikely done so since the early Archean Eon. Models forphotoferrotrophic growth in the Archean oceans remain poorlyconstrained as they are extrapolated from growth rates innutrient rich laboratory culture media. Here we demonstratethat photoferrotrophic Chlorobi have the physiological capacityto fix inorganic N and S into biomass when availability ofthese nutrients is low and have likely had this capacity sincethe Archean Eon. Thus, under N and S limited ferruginousconditions, photoferrotrophy underpins biogeochemical cyclingof C, N, S, and Fe. Nutrient availability, however, influencesgrowth and Fe(II) oxidation rates and has consequences forthe stoichiometric relationships between C, N, S, and Fetransformations. Undoubtedly, these relationships should beassessed and studied in more detail with additional physiologicalexperimentation and should be applied to further constrainmodels of photoferrotrophy, biological production, and globalbiogeochemical cycling in the Archean Eon.MATERIALS AND METHODSStrains and Growth MediumMedia was prepared after Hegler et al. (2008), and allocated intoserum bottles (100 mL media and 160 mL total volume), with0.3 g/L NH4Cl, 0.5g/L MgSO4·7H2O, 0.1g/L CaCl2·2H2O, and0.6g/L KH2PO4. After autoclaving, 22 mmol L−1 bicarbonate,trace elements, mixed vitamin solution, selenate-tungstate,vitamin B12, and FeCl2 were added and the pH was adjustedFrontiers in Microbiology | www.frontiersin.org 11 July 2017 | Volume 8 | Article 1212more general evolutionary history of the CysH protein, however, is convoluted given abundantincongruences between the CysH protein and 16S rRNA gene phylogenies. Accepting verticalinheritance of the ASR pathway in the photoferrotrophic Chlorobi and the early divergence of theChlorobi from other organisms, we hypothesize that gene loss explains the lack of a completeASR pathways in other photosynthetic Chlorobi and this hypothesis is supported by the partialpresence of ASR pathway components across the phylum Chlorobi (Table 3.2).Ecology of ASR in Chlorobi, past and presentThe presence of an ASR pathway in all known photoferrotrophic Chlorobi implies that ASRis advantageous to gr wth under ferruginous conditio s. The lack f th ASR pathway in thecanonically sulfur oxidizing Chlorobi makes sense in light of the availability of reduced sulfurcompounds in their preferred habitats. The energetic expense of ASR would tend to favorassimilation of reduced compounds when available. Conversely, ferruginous environments areby definition sulfur poor and the availability of reduced sulfur compounds can be limited by thesolubility of FeS. Sulfate, therefore, is likely the most abundant and available sulfur source inmodern ferruginous environments. The rock record also demonstrates that ferruginous marineconditions persisted throughout much of the Precambrian Eons and reduced sulfur species werelikely scarce with the exception of in the apparently ephemeral developments of costal euxinia.70fmicb-08-01212 July 4, 2017 Time: 16:2 # 11Thompson et al. Metabolic Potential of Photoferrotrophic ChlorobiTABLE 3 | Gene length (bp), CAI, and GC content (%) for each of the genes in the ASR cassette.Gene C. phaeoferrooxidans C. ferrooxidans C. luteolumLength (bp) CAI GC content (%) Length (bp) CAI GC content (%) Length (bp) CAI GC content (%)CysH 714 0.78 54.34 714 0.78 54.62 753 0.59 56.97CsyD 882 0.80 56.12 882 0.78 55.56 915 0.61 57.38CsyN/C 1800 0.81 54.28 1800 0.77 53.67 1800 0.69 58.06Siroheme synthase 453 0.81 54.08 453 0.71 52.98 453 0.53 57.17Uroporphyrin-IIIC-methyltransferase1287 0.73 57.96 1287 0.73 55.40 258 0.62 57.36CysA 1074 0.77 53.26 1074 0.80 52.42 1074 0.67 59.22CysW 870 0.79 53.22 870 0.79 52.76 870 0.64 58.74CysT 834 0.79 52.64 834 0.81 53.36 834 0.66 58.03Sulfate transporter 1020 0.82 53.14 1020 0.79 52.65 1008 0.75 59.52Each parameter was calculated for all three photoferrotrophic Chlorobi, whose whole genome GC contents are: 49.72% for C. phaeoferrooxidans, 49.9% forC. ferrooxidans, and 58.1% for C. luteolum.FIGURE 6 | Phylogenies of (A) the CysH protein and (B) 16S rRNA with bootstrap values shown at each node (maximum likelihood/maximum parsimony). Theorange line indicates the position of the photoferrotrophic Chlorobi. The trees are rooted with two Archaeal species.may inform evolutionary histories and should be tested in thefuture.OUTLOOKPhotoferrotrophy links the C and Fe biogeochemical cyclesthrough coupled CO2 fixation and Fe(II) oxidation and haslikely done so since the early Archean Eon. Models forphotoferrotrophic growth in the Archean oceans remain poorlyconstrained as they are extrapolated from growth rates innutrient rich laboratory culture media. Here we demonstratethat photoferrotrophic Chlorobi have the physiological capacityto fix inorganic N and S into biomass when availability ofthese nutrients is low and have likely had this capacity sincethe Archean Eon. Thus, under N and S limited ferruginousconditions, photoferrotrophy underpins biogeochemical cyclingof C, N, S, and Fe. Nutrient availability, however, influencesgrowth and Fe(II) oxidation rates and has consequences forthe stoichiometric relationships between C, N, S, and Fetransformations. Undoubtedly, these relationships should beassessed and studied in more detail with additional physiologicalexperimentation and should be applied to further constrainmodels of photoferrotrophy, biological production, and globalbiogeochemical cycling in the Archean Eon.MATERIALS AND METHODSStrains and Growth MediumMedia was prepared after Hegler et al. (2008), and allocated intoserum bottles (100 mL media and 160 mL total volume), with0.3 g/L NH4Cl, 0.5g/L MgSO4·7H2O, 0.1g/L CaCl2·2H2O, and0.6g/L KH2PO4. After autoclaving, 22 mmol L−1 bicarbonate,trace elements, mixed vitamin solution, selenate-tungstate,vitamin B12, and FeCl2 were added and the pH was adjustedFrontiers in Microbiology | www.frontiersin.org 11 July 2017 | Volume 8 | Article 1212Figure 3.6: CysH phylogenies of multiple phyla. Phylogenies f (A) t e CysH protein an (B) 16S rRNAwith bootstrap values shown at each node (maximum likelihood/maximum parsimony). The orange lineindicates the position of the photoferrotrophic Chlorobi. The trees are rooted with two Archaeal species.ASR may thus have supported sulfur requirements of photoferrotrophic primary producers overlong stretches of Earth’s history.The apparent role of ASR in supporting primary production through photoferrotrophy impliesthat sulfate availability could have been an important control on global productivity. At 28 mM,sulfate is the principle anion in moder seawater, but sulfate concentrations could have beenas low as a few µM in the Archean oceans [27]. Nutrients like phosphorus and nitrogen areknown to become limiting at such low concentrations. The apparently low sulfur quotas of thephotoferrotrophic Chlorobi (260:1, C:S) thus seem well adapted to growth in the low sulfate oceansof the Archean, which would have nh n ed productivity in the face of sulfur scarcity.Under low sulfate conditions dissimilatory sulfate reduction (DSR) would have played a com-paratively small role in the remineralization of organic matter in Archean oceans [27]. Qualitativelythen ASR would have played an outsized role in the reduction of sulfur and the global sulfurcycle in the Archean oceans, relative to today. We therefore hypothesize that primary productionthrough photoferrotrophy was a key pathway in the production of an organic reduced sulfur pool,which would have provided an important vector for sulfur to Archean sediments. We further71hypothesize that ASR may have predated DSR. Earliest evidence for DSR comes from S-isotopefractionation recorded in 3.47Ga barites [216], whereas photoferrotrophy likely operated as earlyas 3.8Ga [169] and presumably required ASR. The idea that ASR predates DSR could be testedif homology could be established in enzymes involved in both pathways. Although ASR andDSR serve different functions – sulfate acquisition versus energy transduction, respectively – bothpathways actively transport sulfate into the cell and the first enzymes in the pathways are thusanalogous. Comparison of amino acid sequences of the first enzymes (CysD and Sat, respectively)in the two pathways indicates a strong degree of homology implying evolutionary relationshipsbetween components of ASR and DSR.To examine the phylogenetic relationships between enzymes that transport sulfate for usein ASR and DSR, we aligned CysD and Sat proteins from the majority of the Chlorobi and aselection of representative microorganisms from diverse phyla. The resulting phylogeny clearlyseparated amino acid sequences annotated as CysD from those annotated as Sat (Fig. 3.7). BothCysD and Sat appear to support sulfate transport in relation to multiple sulfur metabolisms, butthe phylogenetic relationships appear complicated and likely require more detailed analyses.Nevertheless, homology between the two proteins implies a possible evolutionary relationshipbetween ASR and DSR that may inform evolutionary histories and should be tested in the future.72fmicb-08-01212 July 4, 2017 Time: 16:2 # 12Thompson et al. Metabolic Potential of Photoferrotrophic ChlorobiFIGURE 7 | Phylogeny of the Sat/CysD protein with bootstrap values shown at each node (maximum likelihood/maximum parsimony), to compare the ASR andDSR pathway among a diverse set of organisms. The dashed line delineates the organisms with CysD versus those with Sat. The orange line indicates the positionof the photoferrotrophic Chlorobi.to 6.8–6.9 under an N2/CO2 atmosphere (80:20). 10 mmol L−1FeCl2 was added to all media (regular, NH4+ deplete, and SO4−poor) – Fe(II) concentrations from 200 µmol L−1 to 10 mmolL−1 have been shown to produce the same growth rates undernutrient rich conditions. The 10 mmol L−1 media was filteredafter being made to remove any precipitates, which resulted in afinal Fe(II) concentration of 2 mmol L−1 for the standard mediaand 4mmol L−1 for the NH4+ deplete media. The low SO4−media was left unfiltered with an Fe(II) concentration of 10 mmolL−1. In the ammonium free media, NH4Cl was replaced with0.3 g/L KCl and an additional 10 mL of N2 gas was injected intothe headspace. In the low sulfate media, 0.0025 g/L MgSO4 and0.4 g/L MgCl2 were added instead of the usual 0.5 g/L MgSO4.Furthermore, approximately 10 kBq of carrier-free 35S was addedto all of the low sulfate cultures. The cultures for the N-fixationexperiments were grown in ammonium free conditions once andthen transferred into the final experimental bottles. The culturefor the S35 experiment was grown up in standard media, spundown and decanted to avoid adding extra sulfate, before the cellswere inoculated into the final experimental bottles. All cultureswere grown under a constant light intensity of 14 µE m−2 s−1.Analytical TechniquesSpectrophotometric analysis of Fe(II) and Fe(III) concentrationswere performed using the ferrozine method; samples weremeasured directly as well as after being fixed in 1 NHCl – after Viollier et al. (2000). Pigments were measuredspectrophotometrically after 24 h extractions of 1 mL of pelletedcells in acetone:methanol (7:2 v/v) (Frigaard et al., 1996).Cells numbers were then obtained using a pigment to cellcount conversion factor of 6.3 × 10−10 pigment/cell/mL forC. phaeoferrooxidans and 5.8 × 10−10 pigment/cell/mL forC. ferrooxidans. The cells from the 35S experiment were collectedvia filtration along with a liquid sample as a backgroundmeasurement. The filtered samples were subsequently washedwith 5% Trichloroacetic acid (TCA) in order to kill, wash, anddissolve cellular material. TCA precipitates DNA and proteins,leaving only these cellular components on the filter and thereforeany counts associated with the filtered samples would indicate 35Sthat had been incorporated into this cellular biomass (Cuhel et al.,1981). Five milliliter of scintillation fluid were added to the 35Ssamples (1 mL of liquid or the filter) and all samples were countedusing a scintillation counter.BioinformaticsGenomes of Chlorobi stains used in this paper were retrievedfrom NCBI under the following accession numbers withthe completion percentage of each genome in brackets afterthe number: NC_008639.1 (99.45%), NZ_AASE00000000.1(90.71%), NC_007514.1 (97.8%), NC_009337.1 (98.91%),NC_010803.1 (99.98%), NC_002932.3 (97.8%), NC_011027.1(98.89%), and NC_007512.1(98.91%). Genomes wereanalyzed using MetaPathways V2.5.1, an open sourcepipeline for predicting reactions and pathways using defaultFrontiers in Microbiology | www.frontiersin.org 12 July 2017 | Volume 8 | Article 1212Figure 3.7: Phylogeny of the Sat/CysD protein with bootstrap values shown at each node (maximum likeli-hood/maximum arsimony), to compare the ASR and DSR pathway among a diverse set of organisms.The dashed line delineates the organisms with CysD versus those with Sat. The orange line indicates theposition of the photoferrotrophic Chlorobi.3.5 OutlookPhotoferrotrophy links the C and Fe biogeoche ical cycles thr ugh oupled CO2 fixation nd Fe(II)oxidation and has likely done s since the early Archea Eon. Models for phot ferrotrophic growthin the Archean oceans remain poorly constrained as they are extrapolated from growth rates innutrient rich laboratory culture media. Here we demonstrate that photoferrotrophic Chlorobi havethe physiological capacity to fix i organic N and S into biomass when availability of these nutrientsis low and h ve likely had this capacity since the Archean Eon. Thus, under N and S limitedferruginous conditions, photoferrotrophy underpins biogeochemical cycling of C, N, S, and Fe.Nutrient availability, however, influences growth and Fe(II) oxidation rates and has consequencesfor the stoichiometric relationships between C, N, S, and Fe transformations. Undoubtedly,these relationships should be assessed and studied in more detail with additional physiologicalexperimentation and should be applied to further constrain models of photoferrotrophy, biologicalproduction, and global biogeochemical cycling in the Archean Eon.73Chapter 4Photoferrotrophy, deposition of bandediron formations, and methaneproduction in Archean oceans4.1 SummaryBanded iron formation (BIF) deposition was the likely result of oxidation of ferrous iron in seawaterby either oxygenic photosynthesis or iron-dependent anoxygenic photosynthesis—photoferrotrophy.BIF deposition, however, remains enigmatic because the photosynthetic biomass produced duringiron oxidation is conspicuously absent from BIFs. We have addressed this enigma through experi-ments with photosynthetic bacteria and modeling of biogeochemical cycling in the Archean oceans.Our experiments reveal that, in the presence of silica, photoferrotroph cell surfaces repel iron(oxyhydr)oxides. In silica-rich Precambrian seawater, this repulsion would separate biomass fromferric iron and would lead to large-scale deposition of BIFs lean in organic matter. Excess biomassnot deposited with BIF would have deposited in coastal sediments, formed organic-rich shales,and fueled microbial methanogenesis. As a result, the deposition of BIFs by photoferrotrophswould have contributed fluxes of methane to the atmosphere and thus helped to stabilize Earth’sclimate under a dim early Sun.4.2 IntroductionBanded iron formations (BIFs) host the world’s largest iron ore deposits, and they formedpredominantly through the deposition of ferric iron (Fe[III]) from ferruginous oceans during the74Archean Eon [4, 83]. Most models for BIF deposition invoke photosynthesis in the oxidationof ferrous iron from seawater to induce its subsequent precipitation and deposition as mixedvalence iron (oxyhydr)oxides and carbonate phases [6–9] (see [79] and [81] for alternative models).BIFs thus likely record the activity of Earth’s early photosynthetic biosphere [6]. Two modes ofphotosynthesis have been implicated in Fe(II) oxidation—canonical oxygenic photosynthesis bythe ancestors of modern cyanobacteria [6] and iron-dependent anoxygenic photosynthesis [7, 8],referred to as photoferrotrophy. Photoferrotrophic bacteria grow using light and Fe(II) to fixCO2 into biomass and produce Fe(III) as a metabolic byproduct [8, 9, 20]; they can do so in thecomplete absence of oxygen [20]. Current literature suggests that such anoxygenic photosynthesisis likely the evolutionary predecessor of oxygenic photosynthesis [11, 35, 93] and Fe(II) oxidationcould thus have been driven by photoferrotrophy before the emergence and proliferation ofoxygenic photosynthesis [1, 82]. Indeed, models of nutrient cycling in the Archean ocean implicatephotoferrotrophs as key primary producers [1, 85] before the rise of atmospheric oxygen 2.4 to2.3 billion years ago (Ga) during the Great Oxidation Event (GOE) [2, 104, 105]. High methaneconcentrations in the Archean atmosphere have also been qualitatively attributed to C and Fecycling associated with BIF deposition [172]. Such an ‘upside-down’ biosphere [172] in whichthe reduced products of photosynthesis end up in the atmosphere as methane and the oxidizedproducts are buried as Fe(III) in BIF can, as we show here, be quantitatively and mechanisticallylinked to the activity of photoferrotrophs in ferruginous Archean oceans.Observations and models of extant photoferrotrophs demonstrate their capacity to both oxidizeFe(II) at rates sufficient to form even the largest BIFs and support appreciable rates of primaryproduction [1, 8, 21, 48]. Laboratory experiments to date, paradoxically, reveal a tight physicalassociation between photosynthetic ferric iron metabolic byproducts (e.g., Fe(III) (oxyhydr)oxides)and cellular biomass that leads to their co-sedimentation [8, 21, 42]. In the oceans, deposition ofFe(III) along with this photosynthetic biomass would fuel sedimentary respiration that couplesoxidation of organic carbon to microbial Fe(III) mineral reduction and converts Fe(III) to dissolvedFe(II) and secondary minerals such as siderite and magnetite [42]. Such sedimentary respirationthus closes the iron redox cycle and precludes the preservation of ferric iron in BIF. Currentobservations and models, therefore, cannot explain BIF deposition through photoferrotrophy in75light of such co-sedimentation of biomass with Fe(III) and ensuing diagenetic reactions [42, 63].By contrast, a lack of Fe(III) reduction would lead to the co-deposition of ferric iron in BIF withorganic matter at relative concentrations close to the 4:1 stoichiometry of Fe(II) oxidation to Cfixation in photoferrotrophy (e.g. 2 wt. % C based on 37 wt. % Fe(III) in Dales Gorge BIF [75]). Acompilation of organic matter concentrations and Fe redox states in BIFs, however, reveals thatthey have very low organic matter concentrations (mean of 0.27 wt. % – Fig. 4.1, A.5 [75, 89])and contain appreciable Fe(III), with an average Fe redox state of 2.6 (A.1, A.5 [75, 89]). If theiron minerals in BIFs were the product of photosynthesis, then the fate of the correspondingphotosynthetic biomass remains entirely enigmatic. Furthermore, regardless of the mode of Fe(II)oxidation, current estimates for the magnitude of hydrothermal Fe(II) fluxes to the Archean oceanappear to be deficient of the mass fluxes needed to sustain BIF deposition [217] and may implyan important role for terrestrial weathering fluxes [22, 218]. We thus combined experimentsusing modern photosynthetic bacteria with revised theoretical estimates of Fe(II) fluxes to theArchean ocean and box modeling of coupled carbon and iron cycling to show that photosyntheticFe(II) oxidation could have sustained large-scale, organic-poor, BIF deposition in ocean upwellingsystems over hundreds of millions of years.Thompson et. al.Figure 10.010.1110Weight % Organic CarbonSample type BIFShale and other Precambrian rock typesModern Marine sedimentsModern OMZ sedimentsTheoretical calculation0.010.1110Weight % Organic CarbonSample type BIFShale and other Precambrian rock typesModern Marine sedimentsModern OMZ sedimentsTheoretical calculationTheoretical 4:1 Fe:CFigure 4.1: The organic matter concentrations in BIFs, other Precambrian sedimentary rocks, typical modern marinesediments, and oxygen minimum zone (OMZ) sediments. The solid lines represent the group means whilethe dotted lines delineate one standard deviation above and below the mean. The black dashed line thatspans the figure represents the theoretical organic carbon concentration that would be expected in BIFsdeposited by photoferrotrophs (2.04 wt. % C) assuming quantitative co-sedimentation of biomass withferric Fe. References for this figure can be found in A.5.764.3 Results and discussion4.3.1 Separation of biomass and Fe(III)To explore the fate of biomass and Fe(III) during photosynthetic Fe(II) oxidation, we conductedexperiments using Chlorobium phaeoferrooxidans strain KB01, a photoferrotrophic bacterium thatwas isolated from the water column of ferruginous Kabuno Bay, a sub-basin of Lake Kivu inEast Africa [48]. We also conducted experiments with Chlorobium ferrooxidans strain KoFox—–asediment-dwelling photoferrotroph that tends to associate with Fe(III) (oxyhydr)oxides withoutbecoming encrusted [53, 63], as well as a model marine oxyphototrophic cyanobacterium of thegenus Synechococcus. Strain KB01 is the only known photoferrotroph from a pelagic environment[48] and pelagic photoferrotrophs, including those that would have populated Precambrianferruginous oceans, face a special challenge—they must maintain their position within sunlitwaters despite the precipitation of heavy Fe(III) oxyhydroxide metabolic byproducts that havea tendency to adhere to cell surfaces and cause their rapid sedimentation. Most bacteria havenegatively charged cell surfaces at neutral pH [219], while Fe(III) (oxyhydr)oxides typically have apositive surface charge at the same pH [220]. This generally leads to a strong attraction betweencell surfaces and Fe(III) (oxyhydr)oxides that often manifests as encrustation of bacterial cells andformation of Fe(III) oxyhydroxide-cell aggregates [9, 221].To test the association between photoferrotrophs and their Fe(III) metabolic byproducts, wegrew strain KB01 until late exponential growth phase, gently inverted the cell-mineral sus-pension to resuspend sedimented Fe(III) (oxyhydr)oxides and cells, allowed the heavy Fe(III)(oxyhydr)oxides to resettle, and determined the percentage of cells that remained suspended(Fig. 4.2a). Experiments conducted in media with reduced phosphate concentrations (3-6 µM),approaching those of modern seawater, but still somewhat higher than sub-µM concentrationsfound in Precambrian seawater [85], show that 50 % of strain KB01 cells associated with Fe(III)(oxyhydr)oxides, leading to co-sedimentation of biomass and Fe(III) (Fig. 4.2a). While phosphateconcentrations were very low in the seawater from which BIFs deposited [85, 222], silica concen-trations in Precambrian oceans were high (∼1 mM) [85], and this would have altered the physicaland chemical properties of Fe(III) precipitates formed in seawater [222]. In experiments with 1 mM77silica, concentrations implied for Precambrian seawater [85], strain KB01 cells did not associateto, or co-sediment with their Fe(III) oxyhydroxide byproducts (Fig. 4.2a). Instead, almost all ofthe cells remained suspended (94 ± 6 %). Likewise, experiments conducted with strain KoFoxdemonstrated that cell-mineral association was diminished relative to the silica-free experiments,with 72 ± 7 % of strain KoFox cells remaining suspended (Fig. 4.2a and A.5a). We also grew bothstrains KB01 and KoFox in their standard growth media containing 4 mM phosphate and no silica,which led to little cell-mineral association for strain KB01 (Fig. 4.2a), and modest association forstrain KoFox (A.5a). Detailed electron microscopy revealed that the surfaces of KB01 cells wereentirely free from Fe(III) oxyhydroxide precipitates (Fig. 4.3), whereas strain KoFox tended toform multicellular aggregates that variably associated with the mineral precipitates (A.3a). StrainKB01, and to a somewhat lesser extent strain KoFox, thus avoid encrustation with Fe in low P,high Si waters, as well as in standard growth media with high phosphate, remaining suspendeddespite Fe(III) oxyhydroxide precipitation.Thompson et. al.Figure 2KBKB: AKB: BKB: CKB: D 1 2KF: CCy: CSW2: D0255075100Percentage-30 -25 -20 -15 -10 -5 0 5 10Zeta potential (mV)0102030405060708090100Cells in water column (%)3.0 3.5 4.0 4.5 5.0Distance (nm)-0.10.00.10.20.30.40.50.6Interaction energy (kT)AB: KB-LW: KB-EL: KB-TOT: KB-0 1 2 3 4 5Distance (nm)-505101520Interact. ener. (kT)a b c0 1 2 3 4 5 6 7 8Si:Fe and P:Fe ratio-40-30-20-100102030Zeta potential (mV)P:FeSi:Fe0 1 2 3 4 5 6 7 8Si:Fe and P:Fe ratio-40-30-20-100102030Zeta potential (mV)P:FeSi:FeFigure 4.2: Cell surface characteristics for strain KB01 and the relationship between ferric iron surface charge andmedium anions. The fractions of planktonic (blue) versus sedimented (red) cells (a) for photoferrotrophicstrain KB01 under varying geochemical conditions: 400 µM Fe(II) with low P (3 µM) [unmarked], 400 µMFe(II), low P (3 µM), with 0.6 mM Si [A], 400 µM Fe(II), low P (3 µM), with 1.0 mM Si [B], 400 µM Fe(II),low P (3 µM), with 1.5 mM Si [C], and 10 mM Fe(II) with 4.4 mM P [D]. C. ferrooxidans and Synechococcusare also shown under the [C] conditions. Rhodobacter strain SW2 is shown under the [D] conditions. Thezeta potential (b), in mV, of the ferric iron precipitates is depicted with an increasing ratio of Si or P toFe, where the error bars are all within the data points. Finally, the extended DVLO modeling for strainKB01 (c) with the main graph depicting the interaction energies of the 3 forces (AB – Lewis acid-base; LW– Lifshitz-van der Waals; EL – electrostatic) and the total (TOT – total) for those forces from 3 nm-5 nm,while the inset depicts the forces from 0 nm-5 nm.78Thompson et. al.Figure 3100 nma b100 nmFigure 4.3: Scanning electron microscopy and transmission electron microscopy image of strain KB01. Scanningelectron photomicrograph (a) and transmission electron photomicrograph (b) of C. phaeoferrooxidans strainKB01 revealing various internal and external cell structures as well as a lack of encrustation (a, b) and arare association with Fe(III) precipitates (a).We also subjected Synechococcus cells to Fe-Si-rich growth conditions in a similar fashion (Fig.4.2a and A.5a). Synechococcus associated to a greater extent with the Fe(III) oxyhydroxide productsof Fe(II) oxidation than did the photoferrotrophs under these conditions. Such an associationwas expected, despite the presence of silica, given that oxygen effuses from Synechococcuscells, reacts rapidly with Fe(II), and causes precipitation of Fe(III) (oxyhydr)oxides on the cellsurfaces—an association likely maintained due to binding with organic ligands, as observed inmany previous environmental and laboratory studies [221, 223, 224]. Photoferrotrophs adapted topelagic lifestyles thus appear capable of avoiding co-sedimentation with Fe(III) and remain buoyantwith the potential for separation of biomass from Fe(III) at larger scales in the environment.4.3.2 Mechanisms of cell-mineral separationSurface charge often influences cell-mineral association, or lack thereof, and as solution P:Fe orSi:Fe ratios increased, the surface charge on the Fe(III) (oxyhydr)oxides formed became morenegative (Fig. 4.2b) due to the incorporation of P or Si anions into the oxyhydroxide structure[222]. Conversely, Fe(III) oxyhydroxide particles were positively charged under low P or Siconditions, which is consistent with observations of positively charged Fe(III) precipitates in manyenvironments [220]. Strain KB01 cells had strongly negative surface charges (A.4), whereas cellsof strain KoFox had near neutral surface charges (A.4), similar to some microaerophillic Fe(II)79oxidizers [225]. This difference in surface charge between strains KB01 and KoFox reflects differentcell surface chemistries (A.2), notably an abundance of anionic surface functional groups on strainKoFox with different acid-base behavior than strain KB01 (SI, A.3). Benthic microorganisms,such as strain KoFox, commonly produce surface layers rich in anionic functional groups tofacilitate attachment to solid substrates [226] and such anionic surface functional groups tend tostrongly bind Fe(III), effectively neutralizing surface charge (A.4). The role of Fe(III) in controllingsurface charge on strain KoFox was confirmed by rinsing KoFox cells with reducing agents(sodium dithionite) that liberated Fe(III) and caused a shift in the surface charge to more negativevalues (A.4). These same rinses had little effect on cells of KB01 (A.4), confirming the role ofsurface chemistry and Fe(III) binding in controlling association of cells to Fe(III) (oxyhydr)oxides.The direct role of cell surface charges in dictating mineral association and co-sedimentation isalso evident from the strong inverse relationship (A.5b) between mineral surface charges andcell-mineral separation.To assess the biophysical mechanisms that control cell-mineral associations we determinedinterfacial properties of the cell and Fe(III) oxyhydroxide surfaces and conducted extendedDerjaguin–Landau–Verwey–Overbeek (DVLO) modeling to quantify the forces that developbetween these surfaces (AppendixA) [225, 227]. DVLO modeling uses measurements of physicalsurface properties to calculate interfacial forces as a function of distance between surfaces. Ourmodel results revealed that the negative surface charge on strain KB01 indeed led to electrostaticrepulsion between its surface and negatively-charged Fe(III) (oxyhydr)oxides (Fig. 4.2c). The moreneutral charge on strain KoFox, conversely, led to a weak electrostatic attraction (A.5c). Lewisacid-base and Lifshitz-van der Waals forces were much weaker than the electrostatic forces, butwe note that the acid-base properties of strain KB01 lead to repulsion of Fe(III) (oxyhydr)oxides,even without electrostatic effects. Surface charge, therefore, controls cell-mineral association andultimately like-charged surfaces cause the physical separation of strain KB01, and by extension,other pelagic photoferrotrophs from their Fe(III) oxyhydroxide byproducts. These findings haveimportant implications for BIF deposition and the coupling of carbon and iron cycles in theArchean Eon. We note that while strain KB01 is a derived member of the phylum Chlorobi, itscell surface chemistry is typical for gram negative bacteria [219, 228] and our results are thus80likely extensible to its ancestors and most other gram negative bacteria, with the exception ofbenthic organisms, like strain KoFox, that have unusual modifications to their cell surfaces. Stemgroup, pelagic photoferrotrophs in Archean oceans, therefore, most likely possessed cell surfacechemistries and interacted with Fe(III) (oxyhydr)oxides much like stain KB01.4.3.3 Revised Precambrian Fe budgetsBefore proceeding to model coupled carbon and iron cycling, we sought to reconcile the possiblematerial fluxes of Fe(II) to the oceans and those needed to sustain BIF deposition [217, 229]. Thesematerial fluxes serve as boundary conditions for our models and help tether our results to thegeologic record. Peak BIF deposition rates imply an Fe burial flux of 45 mol m – 2 yr – 1 over areasgreater 1011 m2, or ∼4.5 Tmol Fe yr – 1 [8] and, at steady state, Fe(II) must be actively resuppliedto the oceans at this rate through a combination of hydrothermal venting and continental andseafloor weathering. Previous estimates of modern hydrothermal Fe fluxes of ∼2 Tmol yr – 1(AppendixA), on their own, are insufficient to support steady-state BIF deposition [83, 217,229] (A.3). These estimates were based on the product of hydrothermal fluid flow and Fe(II)concentrations measured in circulating on- and off-axis fluids [24]. Global Fe fluxes, however,would have been much different in the Archean Eon because of enhanced hydrothermal activity[229, 230], smaller continents [231], reduced seawater sulfate [27, 230], weathering of Earth’s crustat low oxygen [2, 16, 28], and pervasively anoxic oceans.Hydrothermal fluid fluxes to the oceans have been calculated from a variety of geochemicaland physical data resulting in a wide range of estimates for these fluxes in the literature. A recentcompilation from a number of sources [158], however, provides a synthesis of both on- and off-axisfluid flow as well as a modeled distribution of these values that yields averages (modes) of 5 x1013 and 1.5 x 1016 kg y – 1, respectively [158]. These values are greater than values considered inprevious estimates of modern hydrothermal Fe(II) budgets, implying that modern Fe(II) fluxes tothe oceans were underestimated [24]. Archean hydrothermal fluid flow was, furthermore, likelyhigher than today due to greater heat loss from the Archean lithosphere [217, 232]. Followingprevious work [217], we thus scaled the updated estimates for modern fluid flow by the ratio ofpast to modern lithospheric heat loss (AppendixA). This yields estimates for hydrothermal fluid81flow at 2.5 Ga of 1 ± 0.6 x 1014 and 4 ± 0.6 x 1016 kg yr – 1 for on- and off-axis venting, respectively(A.3). Crucially, these revised estimates for fluid flow increase the possible Fe(II) fluxes to boththe modern and Archean oceans.The concentration of Fe(II) in modern on-axis hydrothermal fluids is ∼6 mmol kg – 1 [24], butthis depends on the chemistry of seawater that circulates through seafloor basalts. Notably, highconcentrations of sulfate (28 mM) in modern seawater lead to production of hydrogen sulfidein anoxic hydrothermal fluids, and this hydrogen sulfide reacts with Fe(II) to form iron-sulfideminerals that limit the concentration of Fe(II) in efluxing fluids [230]. Sulfate concentrationsin Archean seawater were much lower than today due, in part, to the lack of oxidative pyriteweathering on the continents [27]. Models of hydrothermal fluid chemistry, therefore, predict thatFe(II) concentrations in high-temperature hydrothermal fluids were ten-fold higher when seawaterwas sulfate free versus the modern 28 mM [230], implying that high-temperature vent fluids inthe Archean Eon could have had up to 60 mmol kg – 1 Fe(II). Combining such a high-temperaturehydrothermal Fe(II) concentration [230] with revised on-axis hydrothermal fluid flows results inFe(II) fluxes of 8 ± 3 Tmol yr – 1 at 2.5 Ga (A.3) and these are similar to, but smaller than, otherrecent estimates [95]. The effect of seawater sulfate concentrations on the Fe(II) concentration inlower temperature, off-axis hydrothermal fluids appears less pronounced and so we conservativelyconsider off-axis hydrothermal Fe(II) concentrations equivalent to today (∼0.75 mmol kg – 1) [230].This results in possible off-axis Fe(II) fluxes of up to 30 ± 5 Tmol yr – 1 (A.3). Combining these newestimates for on- and off-axis Fe(II) fluxes, we redefine upper possible limits on hydrothermalFe(II) delivery to the oceans at 2.5 Ga as ∼40 Tmol yr – 1 (A.3).In addition to the Fe(II) fluxes from hydrothermal venting, modern continental weathering alsocontributes Fe to the oceans. We thus assessed the potential delivery flux of continental Fe(II) toArchean oceans taking into consideration the low oxygen atmosphere [28] and smaller continentsizes [231], resulting in a total Fe(II) weathering flux of 5 Tmol yr – 1 to the oceans (AppendixA).Summing continental weathering and hydrothermal Fe(II) fluxes yields an upper possible globalFe(II) flux of ∼45 Tmol y – 1 to the oceans 2.5 Ga (A.3). This is more than sufficient to deposit BIFat 4.5 Tmol yr – 1 and is supplied mostly through previously unconsidered off-axis hydrothermalventing. These revised Fe(II) fluxes place upper boundaries on the magnitude of global coupled82carbon and iron cycling, which we explore below using biogeochemical box models.4.3.4 Modeling Archean marine iron and carbon cyclesWe assume that biological production in our model is driven by photoferrotrophy and, whileoxygenic photosynthesis could have been active from the Mesoarchean [16, 94], its contribution toprimary production may have been small before the GOE due to competition with photoferrotrophs[85, 106]. Nevertheless, discriminating between modes of photosynthesis would have no realeffect on our model outcomes since oxygen produced would react with Fe(II) yielding the same4:1 stoichiometry between iron oxidation and carbon fixation as in photoferrotrophy. Biomassdegradation in our model was first channeled through heterotrophic Fe(III) reduction, and then,given the near absence of sulfate in the Archean ocean [27], microbial methanogenesis [27, 42].We implicitly assumed that both Fe(III) reduction and methanogenesis are preceded by thebreakdown of organic matter through hydrolysis, glycolysis, and fermentation. Rates of organicmatter breakdown were parameterized based on observations from modern anoxic marine basins[233, 234]. We note, however, that unlike today, the biological pump in the Archean oceans wouldhave operated without ballasting from fecal pellets. Biological production and nutrient cyclingin the modern oceans can be divided into three broad oceanographic provinces—open ocean,coastal zone, and upwelling regions [233]. In detail, the distribution and biological activity ofthese provinces depends on ocean circulation patterns, continental configurations, and nutrientsupply, but in the absence of robust constraints on these parameters for the Archean oceans,we assumed that the relative contributions of similar such provinces to biological productionscale with continental area and were otherwise similar to the modern. We thus distributed totalbiological production across these three provinces according to their relative productivities in themodern ocean [233] and scaled continental area from 0-100 % of the modern (Fig. 4.4; AppendixA).Our results revealed that upwelling provinces support Fe(II) oxidation and Fe(III) oxyhydroxidesedimentation at rates sufficient to deposit even the largest BIFs, such as those of the HamersleyBasin in Western Australia (Fig. 4.5, Table 4.1, AppendixA). Notably, Fe(III) (oxyhydr)oxides, witha relatively high density (3.8 g cm – 1) and a tendency to aggregate to larger particle sizes [235], hadan average settling velocity of 2 x 104 m yr – 1. This led to their deposition within a maximum of83Thompson et. al.Figure 4CH2O ! H2! CH4BIFFe2+Shalehv4Fe2+ + CO2 + 11H2O! 8H+ + 4Fe(OH)3 +  [CH2O] upper coastalupper upwellingupper open oceandeep/intermediate ocean sedimentsFe2+iron oxidationiron oxidationiron oxidationiron reduction/methanogenesiscarbon burialabFigure 4.4: Model of an Archean coastal upwelling zone. A schematic (a) depicting the cycling of iron andcarbon in the model and the boxes (b) that were used to create the model structure, accompanied by arrowsdemonstrating the fluxes of various biochemical parameters between the boxes.40 km from their locus of initial precipitation, assuming a 150 m deep water column and currentvelocities less than 650 m yr – 1 [233, 236] (A.6, A.4). Furthermore, such localized depositionoccurred despite the strong horizontal current velocities that are characteristic of upwellingprovinces [233, 236]. In stark contrast, the average settling velocity of biomass not associated withFe(III) (oxyhydr)oxides was 0.35 m yr – 1 and biomass can thus be transported distances more than6000 km—–greater than the width of the modern Pacific Ocean (A.6, A.4). Substantial biomasscan thus be exported from productive upwelling areas and broadly distributed between oceanicprovinces before its deposition. Notably, separation of biomass from Fe(III) (oxyhydr)oxides led tolittle pelagic and sedimentary Fe-recycling within the upwelling province itself. The small fractionof biomass deposited in upwelling regions fueled diagenesis resulting in both Fe(II) recyclingto the water column and the authigenesis of reduced Fe phases such as siderite and magnetite.Siderite can be an abundant component in many BIF [75, 89] and an example carbon isotope84mass balance suggests that 30 % of the siderite in BIFs can be diagenetic with the balance likelyrepresenting a primary precipitate (AppendixA). In our benchmark scenario (AppendixA), where15 % of cells are associated with their Fe(III) byproducts, organic carbon deposition rates couldhave supported conversion of 10 % of the total Fe(III) deposited to diagenetic siderite. Whenthe diagenetic siderite is combined with primary non-diagenetic siderite, as inferred from theisotopic composition of siderite, the total siderite content of BIFs in our benchmark scenariois as high as 30 %. This combination of primary and diagenetic siderite and remaining Fe(III)(oxyhydr)oxides yielded an average redox state of Fe in BIF of 2.7 and is very similar to that ofmany siderite-bearing Neoarchean BIFs (A.1, A.5).0 25 50 75 100% of Modern continents05101520253035Fe dep (mol m-2 y-1)Fe depositionMethane fluxGlobal PP0 25 50 75 100% cells associated with Fe05101520CH4/PP (Tmol y-1)0 25 50 75 100[Fe] µM0102030405060Fe dep (mol m-2 y-1)0 250 500 750 1,000Upwelling (m y-1)0510152025CH4/PP (Tmol y-1)Thompson et. al.Figure 5a bc dFigure 4.5: Iron and carbon box model sensitivity results. Model sensitivity results for varying continent size (a),percentage of cells associated with Fe(III) (oxyhydr)oxides (b), deep ocean Fe(II) (c), and varying upwellingrates in the upwelling provinces (d). Iron deposition rates in the upwelling provinces are depicted on they-axis, while global rates of primary production and methane production are both shown on the x-axis inTmol yr – 1.85Table 4.1: The low, middle, and high represent the model outputs from the low, middle, and high partsof the ranges depicted by the grey boxes in Fig. 4.5 and A.8. For reference, modern rates of primaryproduction range from 2760 to 3510 Tmol C yr – 1 [90].C PPC burialC remin.C to Fe reductionC to CH4productionFe recycling Fe dep. (upwelling province)Global Fe dep.Global CH4Tmol C/yr% of PP% of PP% of remin.% of remin.Tmol Fe/yr Tmol Fe/yr Tmol Fe/yrTmol CH4/yr2.5 Ga Low 4.5 14.8 85.2 72.6 27.4 11.1 1.0 17.9 0.52.5 Ga Middle 14.8 14.7 85.3 49.1 50.9 24.8 9.9 59.1 3.22.5 Ga High 22.5 14.6 85.4 41.0 59.0 31.4 17.5 89.9 5.7Biomass exported from upwelling provinces augmented the biomass produced in the otherprovinces and drove extensive pelagic and sedimentary Fe-recycling through heterotrophic Fe(III)reduction. As decoupling of biomass from Fe(III) (oxyhydr)oxides in upwelling regions led toexcess biomass over Fe(III) in both other provinces, it fueled methanogenesis once Fe(III) hadbeen entirely reduced or was buried as Fe-poor, organic-carbon bearing coastal and deep-seasediments (Fig. 4.4). Deposition of biomass in these sediments would have led to organic carbonconcentrations of between 0.5 and 5 wt. % depending on sedimentation rates of detrital material(SI, A.7). At low sedimentation rates characteristic of deep water environments, organic carbonconcentrations are similar to those of the organic carbon-rich shales that deposited throughout theArchean Eon [28].The globally integrated rates of these processes varied depending on continental area, biomassassociation to Fe(III) (oxyhydr)oxides, Fe(II) concentrations in the deep ocean, and rates ofupwelling (Fig. 4.5). These variables can be constrained to likely ranges based on material fluxesrecorded in sedimentary rocks [4, 83, 229], bounds for global cycling [22, 83, 229], and temperedanalogies to modern systems [48, 69]—–model outputs within these ranges are delineated by theshaded area in Fig. 4.5. Importantly, deep ocean Fe(II) concentrations are constrained to <70 µM,similar to earlier estimates of 50 µM based on siderite and calcite solubility [22], because higherFe(II) concentrations led to global Fe(II) oxidation rates that exceed the revised maximum global86Fe(II) supplies and recycling combined (A.3). Even at these relatively low Fe(II) concentrations(20-70 µM), upwelling provinces could have supported area-specific rates of BIF deposition ofup to 54 mol m – 2 yr – 1 over areas greater than 1011 m2 (Table 4.1), as needed to deposit thelargest of the Archean BIFs [8]. Notably, with decreased cell mineral association, increasedcontinent size, increased deep water Fe(II) concentrations, and increased upwelling rates, Fe(II)consumption could have eventually outpaced even upper limits for Fe(II) supply and this wouldhave led to depletion of deep ocean Fe(II). Within the most likely parameter space, rates of globalprimary production through photoferrotrophy would have been less than 1 % of modern primaryproduction (Table 4.1) and, in the absence of oxygen and sulfate, the majority of this biomass wouldhave been remineralized through Fe(III) reduction and methanogenesis in the open ocean with theremainder buried in sediments (Table 4.1). Global primary production is limited by P availabilityover geological time and marine primary production scales with P concentrations in the oceans[237]. Reconstruction of P concentrations in the Archean oceans implied 0.04-0.13 µM [85], whichis ∼1-4 % of the modern, and sufficient to support our estimates for primary production. Notably,up to 45 % of all remineralized carbon was channeled through methanogenesis and the fractionof remineralization through methanogenesis increased with decreased cell mineral association,increased continent size, increased deep water Fe(II) concentrations, and increased upwellingrates (Fig. 4.5, Table 4.1). Variations in all of these processes thus likely influenced global ratesof iron deposition and methane production and could have contributed to the development ofdiffering BIF facies.The model prediction that methanogenesis would have played a key role in carbon degra-dation in Archean ferruginous oceans is also supported by other Earth system models [95] andobservations from modern ferruginous basins. In Kabuno Bay, East Africa, for example, carbonproduced through photoferrotrophy is remineralized through methanogenesis [48] leading to anexport of ferric iron from the euphotic zone even under extremely quiescent physical conditions.Likewise, despite abundant Fe(III) (oxyhydr)oxides, as much as 50 % of the carbon degradationin ferruginous Lake Matano is channeled through methanogenesis, rather than Fe(III) reduction[69]. Empirical observations thus point generally to carbon and iron cycling under ferruginousconditions that ultimately leads to methane production even in the absence of the large-scale87advective processes that operate in the oceans. When integrated into a box model of the Archeanmarine iron and carbon cycles, our observations of cell-Fe(III) oxyhydroxide separation duringphotoferrotrophy quantitatively describe an Archean biosphere that: (1) accounts for the Fe(III)deposition rates needed to form BIFs; (2) explains the fate of biomass and its absence from BIFs;and (3) provides a strong source of methane to the biosphere.This model demonstrates that, within the geologically constrained parameter space, globalrates of methanogenesis can easily reach 3.2 Tmol yr – 1 (Table 4.1) and with the near absence ofoxygen and sulfate to fuel methane oxidation, most of this methane would be delivered to theatmosphere [29]. Based on new solutions [95] to photochemical models [238], this 3.2 Tmol yr – 1biospheric methane flux would support a 10 ppmv methane atmosphere under our benchmarkscenario with a possible range of between 1 and 20 ppmv across the likely model parameter space(Table 4.1). While such methane concentrations alone are unlikely to support a warm climate,positive feedbacks with other concurrent modes of photosynthesis, like H2-based anoxygenicphotosynthesis, dramatically increase both biospheric methane fluxes to the atmosphere andatmospheric methane concentrations [95]. Our observations of extant photoferrotrophs, revisedglobal Fe budgets, and models of coupled C and Fe cycling thus support an up-side-downArchean biosphere, similar to that originally proposed by Walker [172], in which the deposition ofBIFs leads to a methane-rich atmosphere. Our results further imply that cell-Fe(III) separationaugments marine methane production through the breakdown of photoferrotrophic biomass andmechanistically ties biospheric methane fluxes to the deposition of BIF. The role this plays inatmospheric chemistry and greenhouse warming should be tested through further biogeochemical-photochemical modeling efforts.4.4 MethodsPhotoferrotrophic strains and cyanobacteria were grown in basal media [48] and in alternativemedia (AppendixA and A.1) until late exponential phase. Sub-samples of both Fe(II)/Fe(III) andpigments were used to track the growth kinetics as well as to assess the in vitro cellular associationto Fe(III). The surface properties of each strain and their metabolic Fe(III) byproducts were assessed88by measuring their surface potential using a Particle Metrix:ZetaView c©. The acid-base propertiescell surfaces were determined through titrations. Cell surface contact angles were measuredfor DVLO modelling in three different liquids – water, glycerol, and diiodomethane [225, 227].Electron microscopies (SEM and TEM) were used to image cell-mineral interactions (AppendixA).The particle sizes of the Fe(III) (oxyhydr)oxides produced by the photoferrotrophic strains weremeasured using a Mastersizer2000, and these were then used to model the impact of horizontalocean current velocities on the settling time of both Fe(III) particles and cellular biomass (SI, A.6,A.4). The box model of the Archean marine carbon and iron cycles was designed with three boxes(each representing an oceanic province) to capture variability in fluxes through each box (SI, Fig.4.4, 4.5, A.8, and Table 4.1).4.5 Supplementary materials• Materials and Methods• Section 1: Cell surface features and acid-base chemistry• Section 2: Cell-iron surface interaction and extended DVLO modeling• Section 3: Iron concentration and supply• Section 4: Physical separation of ferric iron oxyhydroxides and cellular biomass in an oceansetting• Section 5: Box model of Archean marine carbon and iron cycles• Section 6: Organic carbon burial and diagenesis• Table A1: Media concentrations• Table A2: Cell surface characteristics• Table A3: Modern and Archean Fe fluxes• Table A4: Physical separation model results89• Table A5: Data compilations for Fig. 4.1 and A.1• Fig. A1: BIF redox state• Fig. A2: Growth curve for Chlorobium phaeoferrooxidans strain KB01• Fig. A3: Additional SEM and TEM images of strains KB01 and KoFox• Fig. A4: Surface charge of strains KB01 and KoFox• Fig. A5: Additional cell surface characteristics• Fig. A6: Fe particle aggregation and physical separation model results• Fig. A7: Modelled weight % organic carbon• Fig. A8: Iron and carbon box model sensitivity results90Chapter 5Microbial community metabolism andcoupled carbon and iron cycling inferruginous environments5.1 SummaryModels of Precambrian microbial communities and their impact on global geochemical cyclesover 3 billion years of Earth’s history remain almost entirely conceptual. This is due to a lackof information from environments with biogeochemistry extensible to iron-rich (ferruginous)oceans. Kabuno Bay, a sub-basin of Lake Kivu in East Africa, is one of the few permanentlystratified ferruginous environments on the planet today. Process rate measurements and tagsequencing suggest a tight coupling between the primary producing photoferrotrophic bacteriaand heterotrophic iron reducers. This tight coupling dictates the export of Fe(III) and thedistribution of carbon to other microbial metabolisms, such as methanogenesis. While therole of the photoferrotrophic Chlorobia in Fe(II) oxidation has been established, the pathwaysthrough which carbon is channeled from primary production to terminal oxidation, as well as theorganisms responsible for each pathway, remain largely unknown. Yet, it is the flow of carbonthrough these pathways that dictates the relative rates of iron export and methane productionin Kabuno Bay. To gain molecular level insight into coupled C and Fe cycling we conductedmetagenomic analyses to map metabolic pathways at the population and community level. Thesepathway-centric analyses highlight the abundance of key genes involved in carbon degradationand fermentation that dominate the overall metabolic potential of the microbial community. They91also elucidate the abundance of multiple modes of photosynthesis coupled to the oxidationof Fe(II), H2S, H2, or H2O. Kabuno Bay hosts largely unknown taxa, such as Candidate phylaMBNT15 and Patescibacteria that predominantly support the C degradation and fermentationpathways in the chemocline. The analyses further highlight previously identified taxa, such asmembers of the phylum Bacteriodetes, that couple the metabolic potential for C degradation,fermentation, and Fe(III) reduction in individual microorganisms. This tight coupling between Cdegradation, fermentation, and respiration pathways leads to little carbon leakage for alternativemicrobial metabolisms, such methanogenesis. Ultimately, our metagenomic analyses revealmechanistic aspects of coupled C and Fe cycling that likely supported microbial communitiesacross vast stretches of Earth’s early history. The metabolic potential and process rates of themicrobial community found in Kabuno Bay, thus lays the groundwork for the elucidation of therelationships under the naturally complex ferruginous conditions and can be used to informmodels of C and Fe biogeochemical cycling throughout the Precambrian Eons.5.2 IntroductionMicrobial community networks drive global biogeochemical cycles in modern environments andhave done so throughout Earth’s history [3, 146, 239]. Microorganisms possess the metabolicpotential to utilize organic and inorganic substrates for growth resulting in reduction-oxidation(redox) transformations [3]. These redox transformations mediate both large- and small-scaleprocesses in a number of systems on Earth’s surface [3, 240], dictating, therefore, the exportand import of key substrates to and from modern environments. Through these mechanismsmicrobial community networks play a key role in processes such as climate regulation [241–243],the availability of bio-essential nutrients [3], and mineralization to form ore bodies [221, 240, 244].While many studies have been conducted on microbial community networks under a diverse set ofmodern conditions—both oxic [245–247] and anoxic [248–252]—–very few modern environmentsare similar to those that prevailed throughout Earth’s early history. Modern environments areoften oxygen-rich, permeated by ephemeral pockets of anoxia, and this restricts the extensibility ofinformation that can be gleaned from modern environmental microbial communities to the past.92Oceans throughout much of Earth’s history were oxygen-free and were, thus, dominated byiron-rich (ferruginous) conditions for much of Earth’s first three billion years [5, 22, 23, 83, 170,171]. Prior to the oxygenation of the Earth’s atmosphere and oceans, due to the proliferationand diversification of oxygenic photosynthetic microorganisms [16, 99], atmospheric oxygenconcentrations ranged from 0.01 to 1 % of present atmospheric levels (PAL) [2, 15, 28]. Under suchreducing conditions, much of the ocean remained anoxic with small bouts of euxinia permeatinglargely ferruginous conditions [5, 23, 171]. Geologic evidence for these ferruginous conditionscomes from the world’s largest iron ore deposits—–banded iron formations (BIFs)–—as well as aplethora of iron-rich deposits, such as ferruginous shales [4, 5, 83, 253]. Data from these depositscombined with models of early Earth oceans indicate that there were high concentrations offerrous iron (20-100 µM) throughout the Archean Eon and the majority of the Proterozoic Eon[5, 23]. Furthermore, evidence from the isotopic and fossil records [100–102], as well as datafrom molecular clock reconstructions [12, 254] indicate that microbial communities would havethrived under these ferruginous conditions, thus establishing the biogeochemical cycling of majorelements such as carbon and iron [1].Little is known, however, about microbial community networks and interactions under suchferruginous conditions. Laboratory studies using pure microbial cultures have provided a wealthof information on the metabolic pathways that likely underpinned biogeochemical cycling inPrecambrian ferruginous oceans [21, 55, 109, 115]. Pure culture studies, for example, allowus to use information on the physiology of extant microorganisms to make predictions aboutthe possible rates and pathways of key biogeochemical reactions in the Precambrian oceans.These estimated rates can be compared with the material fluxes preserved in the rock record[8, 21, 55, 224]. There are limits to the data that can be collected from pure culture studies, however,as they are constrained to a single bacterium and outcomes are often dictated by the set laboratoryconditions. These studies often fail to mimic conditions that are pertinent to the Precambrian Eonsand can also misrepresent the growth of an individual microorganism due to excess essentialnutrients that accentuate growth or even allow growth where the natural conditions would limitit. Furthermore, pure culture studies fail to capture the relationships between multiple specieswithin a microbial community. To address some of the limits of pure culture studies, studies of93community level processes in modern analogue environments provide real world evidence ofbiogeochemical cycling under ferruginous conditions and, in particular, reveal coupled elementalcycling through pathways distributed across multiple taxa [48, 64, 68, 70, 71].Collectively data from both pure cultures and modern analogue environments provide aframework for model-based reconstructions of microbial community processes in the Precambrianoceans [8, 21, 95, 96, 99, 107]. These model-based reconstructions often combine process rateswith the inferred processes (from pure culture studies) associated with individual members ofthe microbial community. These combined data products thus produce conceptual models of thekey microorganisms that are integral in the cycling of major elements such as carbon and iron.The framework produced through these conceptual models, however, fails to fully capture thecomplexity of natural systems and the iterative relationships often needed by members of themicrobial community (syntrophy) to complete the biogeochemical cycling of key elements — i.e.carbon. This lack of mechanistic knowledge on microbial community functioning ultimately limitsthe extent to which models of Precambrian ocean biogeochemical cycling can resolve processesthat control key features of the Earth system, like climate. In particular, knowledge on how organiccarbon degradation is partitioned between iron reduction and methane production is needed toconstrain connections between Precambrian microbial communities and their suggested role in themaintenance of a clement climate [95, 99]. New information on microbial community structureand function under ferruginous conditions is thus needed to improve models of biogeochemicalcycling in the Precambrian oceans.To link the microbial communities in ferruginous lakes to the Precambrian oceans, previousstudies have elucidated the community members potentially involved in aspects of carbon cyclingand other metabolisms such as methanogenesis and iron reduction through taxonomic identifica-tion [48, 64, 68]. Of the ferruginous environments examined, Kabuno Bay is the only one to datewith a dominant population of anoxygenic phototrophic iron oxidizing bacteria (photoferrotrophs)acting as primary producers [48, 64, 68]. Examining a microbial community with an establishedanoxygenic phototroph producing primary carbon is crucial as photoferrotrophs likely supportedthe microbial biosphere during the Archean before the evolution and proliferation of oxygenicphotosynthetic bacteria – the ancestors to canonical cyanobacteria [16, 96, 99]. The identification of94other key taxa such as Rhodoferax (laboratory cultures are known iron reducers) and Desulfobacca(sulfate reducers) paved the way for a conceptual model highlighting the putative cycling ofcarbon and iron in Kabuno Bay [48]. While utilizing taxonomy to identify metabolic potential caneffectively resolve the abundance and presence of community members, this technique relies onthe metabolic potential of laboratory isolates to discern the metabolisms or genomic functionsof the microorganisms that are present. As such, the true metabolic potential of the communityremains enigmatic. Additionally, aspects of the Kabuno Bay conceptual model–—i.e. the micro-bial consortia responsible for carbon degradation and therefore availability–—remain woefullyunderrepresented. Thus, the overall metabolic blueprint and the presence of this potential withinkey microbial guilds has yet to be tested.Here we utilize metagenomic analyses to discern the metabolic blueprint of the microbialpopulations in Kabuno Bay with specific targeted analyses to demonstrate the mechanism for atight coupling between the carbon and iron biogeochemical cycles. Specifically, we examined themicrobial networks responsible for production of organic carbon (OC) by primary producers, OCdegradation, and the coupling of OC to key elements such as iron. Given that carbon and ironprocesses can be traced through time in the geologic record, elucidating microbial communitydynamics under ferruginous conditions can be extrapolated to the past. These analyses, combinedwith geochemical analyses and processes rate measurements, therefore, provide insight into howmicrobial populations could have thrived under Precambrian oceanic conditions.5.3 Materials and methods5.3.1 Kabuno Bay site description and samplingKabuno Bay (KB), a sub-basin of Lake Kivu, is located in the Democratic Republic of Congo (DRC)located at 1.58◦-1.70◦ S, 29.01◦-29.09◦ E in the East African Rift system (B.1). KB is permanentlystratified basin with a maximum depth of 120 m and a 10 m sill that connects it to adjacent LakeKivu. Water samples were collected from multiple depths throughout the water column, with thefocus being the oxic-anoxic transition zone, during four sampling trips in February 2012 (rainyseason) using a battery-driven peristaltic pump connected to a weighted-double conical intake95through plastic pumping [48]. Water samples were immediately processed for chemical analysesor filtered for molecular analyses [48].5.3.2 Physio-chemical analysesVertical depth profiles of conductivity, temperature, pH, and oxygen were measured in situ with aHydrolab DS5 (OTT Hydromet, Germany) multiparametric probe [48]. Water samples for ironspeciation (Fe(II)aq/Fe(III)s), sulfate (SO42 – ), sulfide (HS – ), and cell counts, were collected atmultiple depths and measured as described under analytical techniques. Cylindrical sedimenttraps with a 0.22 µm filter (diameter of 8 cm) were placed at a depth of 15 m and deployed for56 hours to capture particulate matter as it settled out of the water column. The filters werekept under anoxic conditions and frozen at -20◦ until extraction. The particles were subsequentlyextracted and measured as described in analytical techniques.5.3.3 Iron oxidation, iron reduction, and sulfate reduction ratesFe(II) oxidation rates were measured ex situ in 100 mL glass syringes that were illuminated with15 µmol m – 2 s – 1 light (supplied by 60 W incandescent light bulb) throughout the incubation.All syringes were amended with 50 µM Fe(II) and half were amended with DCMU, a knowninhibitor of photosystem II and oxygenic photosynthesis (3-(3,4-dichlorophenyl)-1,1-dimethylureaor DCMU, 0.5 mg L – 1). Sub-samples were collected throughout the course of the incubation totrack Fe(II), Fe(III), pigments, and cell counts, which were subsequently measured as describedin analytical techniques. Fe(II) oxidation rates were calculated from both the consumption ofFe(II) and production of Fe(III) using a least squares regression through the linear portion ofthe incubation with the rate of Fe(II) oxidation derived from the slope of the regression. Fe(II)oxidation and Fe(III) reduction rates were also measured through ex situ incubations where 100 mLglass syringes were subjected to alternating light and dark cycles. These syringes were amendedwith 60 µM Fe(II) and subjected to 1 to 3 µmol m – 2 s – 1 light during the light cycle. Sub-sampleswere collected throughout the course of the incubation to track Fe(II), Fe(III), pigments, andcell counts, which were subsequently measured as described in analytical techniques. Rates ofoxidation and reduction were calculated from the changes in Fe speciation during the linear96portions of each cycle.Sulfate reduction rates were measured using 35SO42 – . Water was collected in 250 ml glassserum bottles, which were overflowed 3x and capped with blue butyl rubber stoppers with noheadspace. Within six hours, ∼370 kBq of 35SO42 – was injected into each bottle. The bottleswere incubated in the dark and at approximately in situ temperature for 24 hours, after whichthe incubation was terminated with the addition of 5 ml 20 % zinc acetate. A small portion wascollected for sulfate activity measurements. The ZnS produced was filtered onto GFF glass fiberfilters, which were well washed with distilled water and frozen. Frozen filters were distilled with1 N HCl and reduced Cr solution, and the liberated sulfide was recollected in zinc acetate. Theactivities of the sulfate and ZnS were measured by scintillation counting. This method produced anegligible counting blank in the sulfide activity counts.5.3.4 Analytical techniquesSub-samples from both the KB water column and from incubation experiments were analyzed forFe(II) and Fe(III) concentrations using spectrophotometric analyses. Specifically, Fe(II) and Fe(III)concentrations were measured using the ferrozine method and samples were measured directlyas well as after being fixed in 1 N HCl – after Voillier et. al., 2000 [133]. Additional sub-sampleswere taken from the incubation experiments to measure the pigment concentrations. Pigmentswere measured spectrophotometrically after 24 hour extractions of 1 mL of pelleted cells inacetone:methanol (7:2 v/v) [134]. Further sub-samples from both the KB water column and fromincubation experiments were taken and cells were fixed in gluteraldehyde (final concentration of0.1 %). After the cells were fixed, they were subsequently stained with SYBR green (0.25 % finalconcentration) and directly counted in a 96 well plate using a Miltenyi Biotec MACSQuant, with aflow rate of medium.Sub-samples from the KB water column were analyzed for their sulfate concentrations. Sulfateconcentrations were measured using ion chromatography (Thermo Scientific Dionex, ICS-2100)with a Dionex Ion PacTM AS19 RFICTM analytical column (2 x 250 mm). Sample concentrationswere quantified by comparing the peak area samples to those of standard solutions. Sub-samplesfrom the KB water column for sulfide measurements were preserved with 5 µL of 20 % (w/v)97zinc acetate per mL of sample. Sulfide concentrations were determined using the methylene blueassay [255], which was calibrated using the standard iodometric titration to determine the specificabsorbance of the methylene blue reagent. To measure sulfide, 80 µl of the methylene blue reagentwas added was added to 1 mL of sample. The samples incubated with the reagent for 15 minutesand were measured spectrophotometrically at a wavelength of 670 nm.Fe-speciation measurements were performed on anaerobically preserved and freeze driedsediment samples following the method of Poulton and Canfield [25]. Fe-speciation was conductedon sediment trap material by applying each extraction, described in B.1 [25], directly to filterswithin the 15 ml centrifuge tube. 0.5 M HCl was substituted in place of the hydroxylaminehydrochloride leach (B.1) so that both Fe(II) and Fe(III) could be determined in this reactive(oxyhydr)oxide fraction. A sub-sample of the 0.5 N HCl extracted fraction [256, 257] was retainedand used to measure Fe-speciation on these easily extractable phases spectrophotometricallyusing the ferrozine assay [133]. The highly reactive, “FeHR” pool is defined as the sum of(oxyhydr)oxides including ferrihydrite, lepidocrocite, and siderite (FeHCl, 0.5 N HCl extractableFe), ferric (oxyhydr)oxides including hematite and goethite (FeDith, dithionite extractable Fe),and magnetite (FeOxa, oxalate extractable Fe). The non-reactive, “FeNR” pool is attributed toFe in silicate minerals (FeSil, near boiling 6 N HCl extractable Fe after removal of reactivephases). Fe concentration measurements were performed using a Flame Atomic AbsorptionSpectrophotometer (Flame AAS). Precision on triplicate measurements was < 1 % (2SD) and ourlimit of detection was ∼0.1 µg g – 1. Our extractions dissolved > 92 % of the Fe from the PACS-2international reference standard.5.3.5 Flux calculationsTo calculate water column Fe fluxes, area specific Fe sedimentation rates were determined bydividing the concentration of Fe captured by the sediment trap in each operationally definedmineral phase (mmol), by the area of the sediment trap (0.005 m2) and the deployment time toyield Fe fluxes in units of mmol m – 2 yr – 1.985.3.6 DNA extraction250 mL of water was collected from each of the five depths throughout the chemocline (10 m, 11 m,11.25 m, 11.5 m, and 12 m), was filtered through a 0.22 µm filter, and frozen at -20◦C . DNA wasextracted from the biomass collected on these filters by enzymatic lysis and a phenol:chloroformpurification [258]. At the end of the extraction the DNA was washed twice with TE buffer [258]and quality checked through polymerase chain reaction (PCR). The quantity and purity of thefinal DNA products were verified using a NanoDrop 1000 spectrophotometer (Thermo Scientific)and the Picogreen c© (Invitrogen) assay according to manufacturer’s instructions (Quant-iTTMdsDNA Assay Kit, 2018). TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 7.5) was used for dilutingthe Quant-iTTM PicoGreen c© reagent, and for diluting DNA samples.5.3.7 Metagenome sequencing and assemblyEach water column DNA sample (5 total) was sequenced at the DOE Joint Genome Institute(JGI) where it was subjected to Illumina paired-end library construction and sequenced (HiSeqplatform, version 3 chemistry) to generate a total of 95 Gb of paired 150-nucelotide (nt) reads. JGIquality trimmed sequences were assembled de novo using the MEGAHIT version 1.2.7 assemblyalgorithm [259]. All samples were assembled individually. To assess the abundance of eachassembled contig within the metagenomic sample, a redefined version of the RPKM (reads perkilobase mapped) calculation was applied to each contig [196]. With each contig’s abundancebeing affected by sequencing depth (may vary between samples) and contig length (longer contigsnaturally encompass more reads), this bwa-based version of the RPKM [260] aims to normalizethese differences. The RPKM for a given contig is thus:RPKM =ReadsmappedtocontigContiglength(kb)Readsmappedtowholemetagenomesample106(5.1)5.3.8 16S rRNA reconstructionTo reconstruct community profiles from the metagenomic sequences, EMIRGE [261, 262] was runon each assembled metagenome utilizing the baseline parameters [261, 262]. The final set of 16S99rRNA gene sequences reconstructed through EMIRGE were subsequently annotated through acomparison with the DADA2 16S rRNA database (Fig. 5.2a; [263]). Other 16S rRNA sequenceswere retrieved from members of the class Chlorobia from the Silva online database -– version128 [117, 118]. Only full-length (>1400 bp) sequences were selected, and then aligned using thepackage software ClustalX2.1 [119]. To rigorously test the evolutionary history of the KabunoBay 16S rRNA sequences annotated as class Chlorobia with the database of sequences previouslyannotated as class Chlorobia, multiple tree construction methods were employed (Bayesian,Maximum likelihood (ML), Maximum parsimony (MP)). ML and MP trees were constructed inMEGA version 7 [120, 121], while the Bayesian tree was constructed in Geneious [264] and alltrees bootstrapped 500 times. Bootstrap values are indicated at the nodes and the legend (2.0)delineates the nucleotide substitutions per site over the indicated distance (Fig. 5.2c). To assessthe absolute abundance of the members of the microbial community, the relative abundance ofeach 16S rRNA gene at each depth, generated through EMIRGE [261, 262], was multiplied by thetotal cell counts for that depth. 16S rRNA genes from the same taxonomic units (e.g. phyla orclass) were summed to delineate their representation within the whole community (Fig. 5.2a).5.3.9 Gene searchesTo search for specific genes, open reading frames (ORFs) were predicted for each individuallyassembled metagenome (5 depths – 10 m, 11 m, 11.25 m, 11.5 m, and 12 m) using Prodigal[125]. To assess the abundance of each of these ORFs, the RPKM value (see above) for the contigthat contained each ORF was pulled and assigned to that ORF. These ORFs were subsequentlysearched for the sequences of genes of interest using either BLAST (e-value cut-off of 1 x 10 – 18,followed by a Bit Score cut-off of 50 [131]) or HMMER’s hmmsearch function (e-value cut-off of 1 x 10 – 18 and length cut-off of 50 % [132]). The described cutoffs were used for all ofthe genes, with the exception of the ‘Paired CXXCH’ which had length cut-off of 100 %. Thesequence or HMM model for each gene of interest was pulled from the Chlorobium phaeoferrooxidansstrain KB01 genome [58], the National Center for Biotechnology Information database (NCBI— https://www.ncbi.nlm.nih.gov/), the Fungenes repository [265], the Pfam database [129], orthe CAZy databases [130] The genes of interest, their general function, and the database used to100obtained their sequence or HMM model are listed in B.2. To assess the abundance of a specificgene relative to the overall metabolic potential of the microbial community, the RPKM of a gene ofinterest was summed for a particular depth, compared to the summed RPKM of the single copytaxonomic marker gene rpoB, and reported as a percent.5.3.10 Metagenome assembled genomesTo reconstruct population genomes or metagenome assembled genomes (MAGs) the assembledmetagenomic contigs were recruited into MAGs using MaxBin version 2.0 [266], with the baselineparameters [266]. The completeness and level of contamination of each MAG was calculatedwith CheckM version 1.1.0 [267], which uses a set of specific and conserved marker genes toanalyze each MAG. The Q-score was calculated for each MAG by subtracting the contaminationmultiplied by five from the completeness. Only MAGs with a Q-score of greater than 50 werekept for downstream processing. Additionally, the taxonomy of each MAG was determined bycomparing the marker genes in each MAG to a genome taxonomy database (GTDB-Tk; [268]).To assess how well the MAGs represented the overall microbial community, the metagenomictrimmed reads from each depth were mapped (bwa-based mapping) to the MAGs recovered fromeach depth. The read coverage of the MAGs ranged from 3 to 36 % (B.3). The number of MAGsrecovered per phyla or class was also compared to the depth integrated absolute abundances ofthe microbial community (B.6) — MAGs from each of the most abundant groups were recoveredalbeit not at the same abundances as the microbial community would dictate.To partition specific metabolisms across the relevant members of the microbial community thesame gene searches as described above were conducted within the individual reconstructed MAGs.The abundance of a specific gene or process within the MAGs was determined by summingthe RPKMs for the gene of interest from the MAGs alone and comparing it to the sum of thesingle copy rpoB gene from the MAGs. To identify regions with identical nucleotide sequences ingenome of Chlorobium phaeoferrooxidans strain KB01 and the Chlorobium sp. MAG, the contigs ofboth the strain and MAG were aligned and compared using MUMmer (nucleotide comparisonwith nucmer – [123]) and visualized using Circos (http://circos.ca/). The whole genome of strainKB01 was retrieved from GenBank with the accession number MPJE00000000 [58]. To examine101the relationship between these pathways in representative, relatively abundant members of themicrobial community we selected 30 near or mostly complete MAGs (Table 5.1) some of whichwere previously implicated as key members of the community [48] and conducted further analyseson the specific pathways present (Fig. 5.4a, 5.4c, 5.5, B.6, B.7, B.8). To compare the similarities anddifferences between the higher level pathways in 30 of key MAGs from Kabuno Bay, open readingframes (ORFs) for each MAG were predicted using Prodigal [125]. The predicted ORFs weresubsequently annotated using the online tool Kofam Koala [126] that uses Hidden Markov-Models(HMMs) to compare each ORF those found in the Kyoto Encyclopedia of Genes and Genomesdatabase (KEGG-11-06-18) [127]. The resulting annotations were compared at the pathway leveland visualized through a python script as described in Graham et. al., 2018 [128] -– B.6, B.7.5.4 Results5.4.1 Biogeochemical and physical properties of Kabuno BayKabuno Bay is an anoxic and ferruginous (Fe-rich) sub-basin of Lake Kivu located on the borderof the Democratic Republic of Congo and Rwanda (B.1). Kabuno Bay is permanently stratified(Fig. 5.1a) with anoxic deep waters (Fig. 5.1b) and this stratification is sustained by volcanic seepswith high concentrations of salts and geogenic gases, as reflected in the high conductivity below10 m (Fig. 5.1a). These lead to strong density differences between the surface and deep watersthat preclude wind or temperature driven mixing. This physical stratification causes chemicalstratification, which is revealed, for example, by the disappearance of oxygen to undetectablelevels (detection limit of 0.6 µM) below approximately 10 m (Fig. 5.1b) and the accumulation ofhigh concentrations of ferrous iron (Fe[II]), up to 1.2 mM (Fig. 5.1b), and some sulfide (< 10 µM —Fig. 5.1c) in deeper waters. Particulate ferric iron (Fe[III]), accumulates in the upper reaches of thechemocline with concentrations up to 100 µM (Fig. 5.1b), but Fe(III) concentrations decline sharplybelow 12 m depth (Fig. 5.1b). The strong gradient in Fe(II) and corresponding accumulation ofFe(III) demonstrates net Fe(II) oxidation to Fe(III) between 10 and 12 m, while the decrease inFe(III) concentrations between 11 and 12 m demonstrates Fe(III) reduction and a partial to closingof the iron cycle in the chemocline. Modest sulfate concentrations persist throughout the water102column (< 600 µM) with a small peak in the chemocline (Fig. 5.1c). Thus, the combination ofphysical stratification driven by groundwater seeps and the high concentrations of Fe and lowsulfide give rise to persistently ferruginous conditions below ∼10 m depth and the concentrationprofiles of Fe(II) and Fe(III) in the upper reaches of the chemocline reveal spatially coupled Feoxidation and reduction reactions.0 200 400 600Fe(II)aq (µM)0 100 200 300DO (µM)Fe(II)aqDO0 200 400 600SO42- (µM)0 100 200 300SSR (nmol L-1 d-1)SO42-SSR0 4E+05 8E+05Cell counts (cells ml-1)0.0 0.2 0.4 0.6GSB/CelltotCell totalsGSB/Celltot22 23 24 25Temperature (°C)051015Depth (m)1,000 3,000 5,000Conductivity (µS cm-1)TempCond0 2 4 6 8HS- (µM)HS-5.5 6.0 6.5 7.0 7.5 8.0pH (units)pH0 25 50 75 100Fe(III)s (µM)Fe(III)s0 3 6 9Time (hrs)-30-20-100102030[Fe] (µM)Fe(II)aqFe(III)sFe(II)aq DCMUFe(III)s DCMU0 200 400 600Fe(II)aq (µM)0 100 200 300DO (µM)Fe(II)aqDO0 200 400 600SO42- (µM)0 100 200 300SSR (nmol L-1 d-1)SO42-SSR0 4E+05 8E+05Cell counts (cells ml-1)0.0 0.2 0.4 0.6GSB/CelltotCell totalsGSB/Celltot22 23 24 25Temperature (°C)051015Depth (m)1,000 3,000 5,000Conductivity (µS cm-1)TempCond0 2 4 6 8HS- (µM)HS-5.5 6.0 6.5 7.0 7.5 8.0pH (units)pH0 25 50 75 100Fe(III)s (µM)Fe(III)s0 3 6 9Time (hrs)-30-20-100102030[Fe] (µM)Fe(II)aqFe(III)sFe(II)aq DCMUFe(III)s DCMU0 200 400 600Fe(II)aq (µM)0 100 200 300DO (µM)Fe(II)aqDO0 200 400 600SO42- (µM)0 100 200 300SSR (nmol L-1 d-1)SO42-SSR0 4E+05 8E+05Cell counts (cells ml-1)0.0 0.2 0.4 0.6GSB/CelltotCell totalsGSB/Celltot22 23 24 25Temperature (°C)051015Depth (m)1,000 3,000 5,000Conductivity (µS cm-1)TempCond0 2 4 6 8HS- (µM)HS-5.5 6.0 6.5 7.0 7.5 8.0pH (units)pH0 25 50 75 100Fe(III)s (µM)Fe(III)s0 3 6 9Time (hrs)-30-20-100102030[Fe] (µM)Fe(II)aqFe(III)sFe(II)aq DCMUFe(III)s DCMU0 200 400 600Fe(II)aq (µM)0 100 200 300DO (µM)Fe(II)aqDO0 200 400 600SO42- (µM)0 100 200 300SSR (nmol L-1 d-1)SO42-SSR0 4E+05 8E+05Cell counts (cells ml-1)0.0 0.2 0.4 0.6GSB/CelltotCell totalsGSB/Celltot22 23 24 25Temperature (°C)051015Depth (m)1,000 3,000 5,000Conductivity (µS cm-1)TempCond0 2 4 6 8HS- (µM)HS-5.5 6.0 6.5 7.0 7.5 8.0pH (units)pH0 25 50 75 100Fe(III)s (µM)Fe(III)s0 3 6 9Time (hrs)-30-20-100102030[Fe] (µM)Fe(II)aqFe(III)sFe(II)aq DCMUFe(III)s DCMU0 200 400 600Fe(II)aq (µM)0 100 200 300DO (µM)Fe(II)aqDO0 200 400 600SO42- (µM)0 100 200 300SSR (nmol L-1 d-1)SO42-SSR0 4E+05 8E+05Cell counts (cells ml-1)0.0 0.2 0.4 0.6GSB/CelltotCell totalsGSB/Celltot22 23 24 25Temperature (°C)051015Depth (m)1,000 3,000 5,000Conductivity (µS cm-1)TempCond0 2 4 6 8HS- (µM)HS-5.5 6.0 6.5 7.0 7.5 8.0pH (units)pH0 25 50 75 100Fe(III)s (µM)Fe(III)s0 3 6 9Time (hrs)-30-20-100102030[Fe] (µM)Fe(II)aqFe(III)sFe(II)aq DCMUFe(III)s DCMU0 200 400 600Fe(II)aq (µM)0 100 200 300DO (µM)Fe(II)aqDO0 200 400 600SO42- (µM)0 100 200 300SSR (nmol L-1 d-1)SO42-SSR0 4E+05 8E+05Cell counts (cells ml-1)0.0 0. 0. 0.GSB/CelltotCell totalsGSB/Celltot22 23 24 25Temperature (°C)051015Depth (m)1,000 3,000 5,000Conductivity (µS cm-1)TempCond0 2 4 6 8HS- (µM)HS-5.5 6.0 6.5 7.0 7.5 8.0pH (units)pH0 25 50 75 100Fe(III)s (µM)Fe(III)s0 3 6 9Time (hrs)-30-20-100102030[Fe] (µM)Fe(II)aqFe(III)sFe(II)aq DCMUFe(III)s 0 200 400 600Fe(II)aq (µM)0 100 200 300DO (µM)Fe(II)aqDO0 200 400 600SO42- (µM)0 100 200 300SSR (nmol L-1 d-1)SO42-SSR0 4E+05 8E+05Cell counts (cells ml-1)0.0 0.2 0.4 0.6GSB/CelltotCell totalsGSB/Celltot22 23 24 25Temperature (°C)051015Depth (m)1,000 3,000 5,000Conductivity (µS cm-1)TempCond0 2 4 6 8HS- (µM)HS-5.5 6.0 6.5 7.0 7.5 8.0pH (units)pH0 25 50 75 100Fe(III)s ( )Fe(III)s0 3 6 9Time (hrs)-30-20-100102030[Fe] (µM)Fe(II)aqFe(III)sFe(II)aq DCMUFe(III)s DCMU0 2 0 4 0 6 0Fe(II)aq (µM)0 1 0 2 0 3 0DO (µM)Fe(II)aqDO0 2 0 4 0 6 0SO42- (µM)0 1 0 2 0 3 0SR (nmol L-1 d-1)SO42-SR0 4E+05 8E+05Cell counts (cells ml-1)0.0 0.2 0.4 0.6GSB/CelltotCell totalsGSB/Celltot2 23 24 25Temperature (°C)051015Depth (m)1, 0 3, 0 5, 0Conductivity (µS cm-1)TempCond0 2 4 6 8HS- (µM)HS-5.5 6.0 6.5 7.0 7.5 8.0pH (units)pH0 25 50 75 1 0Fe(III)s (µM)Fe(III)s0 3 6 9Time (hrs)-30-20-100102030[Fe] (µM)Fe(II)aqFe(III)sFe(II)aq DCMUFe(III)s DCMU0 0 0(II)aq ( )0 0 0 ( )Fe(II)aqD0 0 042- ( )0 0 0S  ( l -1 -1)S 42-SSRll c ts (c lls l-1). . . ./ lltotCell totalsSB/Celltot2r t r  (° )Depth (m),00 ,00 ,00ctivity (  c -1)Te pCond- ( )HS-. . . . . . ( its)pH0(III)s ( )Fe(III)s0 3 6 9i e (hrs)-30-20-10010[Fe] (µM)Fe(II)aqFe(III)sFe(II)aq DC UFe(III)s DC U0 200 400 600Fe(II)aq (µM)0 100 200 300DO (µM)Fe(II)aqDO0 200 400 600SO42- (µM)0 100 200 300SSR (nmol L-1 d-1)SO42-SSR0 4E+05 8E+05Cell counts (cells ml-1)0.0 0.2 0.4 0.6GSB/CelltotCell totalsGSB/Celltot22 23 24 25Temperature (°C)051015Depth (m)1,000 3,000 5,000Conductivity (µS cm-1)TempCond0 2 4 6 8HS- (µM)HS-5.5 6.0 6.5 7.0 7.5 8.0pH (units)pH0 25 50 75 100Fe(III)s (µM)Fe(III)s0 3 6 9Time (hrs)-30-20-100102030[Fe] (µM)Fe(II)aqFe(III)sFe(II)aq DCMUFe(III)s DCMU0 200 400 600Fe(II)aq (µM)0 100 200 300DO (µM)Fe(II)aqDO0 200 400 600SO42- (µM)0 100 200 300SSR (nmol L-1 d-1)SO42-SSR0 4E+05 8E+05Cell counts (cells ml-1)0.0 0.2 0.4 0.6GSB/CelltotCell totalsGSB/Celltot22 23 24 25Temperature (°C)051015Depth (m)1,000 3,000 5,000Conductivity (µS cm-1)TempCond0 2 4 6 8HS- (µM)HS-5.5 6.0 6.5 7.0 7.5 8.0pH (units)pH0 25 50 75 100Fe(III)s (µM)Fe(III)s0 3 6 9Time (hrs)-30-20-100102030[Fe] (µM)Fe(II)aqFe(III)sFe(II)aq DCMUFe(III)s DCMU0 200 400 600Fe(II)aq (µM)0 100 200 300DO (µM)Fe(II)aqDO0 200 400 600SO42- (µM)0 100 200 300SSR (nmol L-1 d-1)SO42-SSR0 4E+05 8E+05ll counts (cells ml-1)0.0 0.2 0.4 0.6GSB/CelltotCell totalsGSB/Celltot22 23 24 25Temperature (°C)051015Depth (m)1,000 3,000 5,000Conductivity (µS cm-1)TempCond0 2 4 6 8HS- (µM)HS-5.5 6.0 6.5 7.0 7.5 8.0pH (units)pH0 25 50 75 100Fe(III)s (µM)Fe(III)s0 3 6 9Time (hrs)-30-20-100102030[Fe] (µM)Fe(II)aqFe(III)sFe(II)aq DCMUFe(III)s DCMU0 200 4 0 6 0Fe(II)aq (µM)100 2 0 3 0DO (µM)Fe(II)aqDO200 4 0 6 0SO42- (µM)100 2 0 3 0SSR (nmol L-1 d-1)SO42-SSR0 4E+05 8E+05Cell counts (cells ml-1). .2 0.4 0.6GSB/CelltotCell totalsGSB/Celltot22 3 24 25Temp rature (°C)051015Depth (m)1,000 3,000 5,000Conductivity (µS cm-1)TempCond2 4 6 8HS- (µM)HS-5.5 6.0 6.5 7.0 7.5 8.0pH (units)pH25 50 75 100Fe(III)s (µM)Fe(III)s0 3 6 9Time (hrs)-30-20-100102030[Fe] (µM)Fe(II)aqFe(III)sFe(II)aq DCMUFe(III)s DCMU0 200 400 600Fe(II)aq (µM)100 200 300DO (µM)Fe(II)aqDO0 200 400 600SO42- (µM)0 100 200 300SR (nmol L-1 d-1)SO42-SSR0 4E+05 8E+05Cell counts (cells ml-1).0 0.2 0.4 0.6GSB/CelltotCell totalsGSB/Celltot22 23 24 25Temperature (°C)051015Depth (m)1,000 3,000 5,000Conductivity (µS cm-1)TempCond0 2 4 6 8HS- (µM)HS-5.5 6.0 6.5 7.0 7.5 8.0pH (units)pH0 25 0 75 100Fe(III)s (µM)Fe(III)s0 3 6 9Time (hrs)-30-20-100102030[Fe] (µM)Fe(II)aqFe(III)sFe(II)aq DCMUFe(III)s DCMUa b c d eFigure 5.1: Depth profiles from Kabuno Bay delineating conductivity, temperature, and pH (a); dissolvedoxygen (DO), aqueous ferrous iron (Fe(II)aq), and solid ferric iron (Fe(III)s) (b); sulfate (SO42 – ), sulfide(HS – ), and sulfate reduction rates (SSR) (c); cell counts and green sulfur bacteria specific cell countscompared to total cell counts (GSB/Celltot) (d); and phototrophic Fe(II) oxidation over time with andwithout the oxygenic photosynthesis inhibiter DCMU (e). Note: Light was attenuated to 0.1 % PAR at 10 mduring other sampling trips to Kabuno Bay [48]Much of the Fe(II) oxidation in Kabuno Bay is driven by anoxygenic photosynthesis and istightly coupled to Fe reduction. Net Fe(II) oxidation rates calculated from both Fe(II) consumptionand Fe(III) production agreed well and were ∼3 µmol L – 1 h – 1 (Fig. 5.1e). Importantly, net Fe(II)oxidation rates were similar in the presence and absence of DCMU (2.4 ± 0.2 µmol L – 1 h – 1without DCMU compared to 3.1 ± 0.4 µmol L – 1 h – 1 with DCMU; Fig. 5.1e), a specific inhibitorof photosystem II [269], demonstrating the Fe(II) oxidation is driven entirely by photosystemI through anoxygenic photosynthesis. Comparisons between illuminated and dark phases inparallel incubations revealed both that Fe(II) oxidation is entirely photosynthetic and is also tightlycoupled to Fe(III) reduction. In these incubations, gross Fe(II) oxidation rates were 0.5 ± 0.05µmol L – 1 h – 1, while Fe(III) reduction rates were 0.4 ± 0.04 µmol L – 1 h – 1. Thus approximately 80103% of the Fe(III) produced during oxidation is immediately reduced demonstrating tight couplingbetween iron oxidation and reduction in Kabuno Bay.Appreciable particulate Fe(III) and Fe(II) is exported from Kabuno Bay’s chemocline, despitethe tight coupling between Fe oxidation and reduction in the upper chemocline. Fe speciationmeasurements of particulate matter exported from the chemocline demonstrate that 97 % of theFe is present in forms generally considered reactive towards microbial metabolisms and lowtemperature biogeochemical reactions, and this fraction had an overall redox state of 2.8. Anappreciable fraction of this reactive exported Fe (20 %) was thus in Fe(II) phases. The majority ofthe exported Fe(III) (63 %) was extractable in 1 N HCl, a fraction generally considered abundantlyavailable to microorganisms and often comprised of phases like ferrihydrite [270–272]. Notably,however, the HCl extractable fraction also contained appreciable Fe(II) with an average redox stateof 2.7, revealing export of mixed-valence or Fe(II) phases like green-rust or siderite. Additionalhighly reactive Fe(III) was present as dithionite extractable phases (27 %) like goethite or hematiteand oxalate extractable phases (5 %), essentially magnetite [273]. Based on these data, the flux ofhighly reactive Fe(III) mineral phases from the water column is 0.4 mol m – 2 yr – 1, which is withinerror of the difference between the Fe(II) oxidation (4.4 ± 0.4 mol m – 2 yr – 1) and Fe(III) reduction(3.5 ± 0.4 mol m – 2 yr – 1) fluxes calculated from directly determined rates. These data thereforeconfirm appreciable export of reactive Fe(III) from the Kabuno Bay euphotic water column, despitethe fact that photosynthetic Fe oxidation is tightly coupled to reduction reactions.Sulfate reduction was also prevalent throughout Kabuno Bay’s chemocline, occurring concomi-tantly with appreciable Fe-reduction, albeit at much lower rates. The rate of sulfate reductionpeaked at approximately 11.5 m with a rate of 300 nmol L – 1 d – 1 (Fig. 5.1c), which is 40 timeslower than the corresponding Fe(III) reduction rates (12000 nmol L – 1 d – 1). The much lower ratesof sulfate reduction demonstrate that it can only play a minor role in Fe-reduction. Fe-reduction,therefore, must be directly coupled to microbial respiration. Export of pyrite from the chemoclineis 6 mmol m – 2 yr – 1 and, compared to depth integrated sulfate reduction rates of 90 mmol m – 2yr – 1, reveals appreciable S recycling. Thus, sulfate reduction and cycling is active throughoutKabuno Bay’s chemocline, where it influences Fe speciation by driving pyrite formation, but itplays a quantitatively minor role in anaerobic respiration compared to Fe-reduction.1045.4.2 Microbial community composition and structureThe microbial community in Kabuno Bay’s chemocline is overwhelmingly dominated by anoxy-genic photosynthetic bacteria of the class Chlorobia and includes representation from a broadsuite of bacterial phyla as well as a limited diversity of Archaea. Microbial community sizes arerelatively constant at 104 to 105 as a function of depth throughout the chemocline (Fig. 5.1d).The population of Chlorobia is well delineated by profiles of BChl e, a photosynthetic pigmentproduced exclusively by members of the class Chlorobia. The distribution of BChl e is abruptlytruncated by the oxygenated surface waters, peaks at 11 m depth, and decreases in size withincreasing depth below this peak as light intensity falls [48]. The taxonomic composition of themicrobial community, based on the small subunit ribosomal RNA gene (16s rRNA) (Fig. 5.2a, B.2),reflects the distribution of BChl e with the Chlorobia peaking at 11 m (47 % of the whole com-munity; Fig. 5.2a). This is similar to reconstructions of microbial community compositions fromamplicon-based community profiling [48]. Likewise, the high relative and absolute abundance,up to 50 % of the community, was also reflected in reconstructions based on other phylogeneticmarker genes, such as recA and rpoB (Fig. B.3a). The rest of the community is mostly comprised ofmembers of 11 other bacterial phyla and Archaea (based on 16S rRNA gene-based reconstructions;Fig. 5.2a). The most striking vertical differences in community composition occur between 10m and the deeper depths combined, notably the higher abundances of Alphaproteobacteria,Gammaproteobacteria, Actinobacteria, and Cyanobacteria at 10 m (Fig. 5.2a, B.2). Indeed, at10 m the Gammaproteobacteria make up 60 % of the microbial community compared to only5 % in the deeper depths (Fig. 5.2a). Below 10 m there is very little depth-dependent structurein the overall community composition, as three main phyla — Bacteriodetes (broken into itsthree classes — Bacteroidia, Chlorobia, and Ignavibacteria), Candidate phyla MBNT15, and thePatescibacteria dominate at each depth (Fig. 5.2a, B.2). The only exception, in the deeper depths,is the progressive decrease in the relative abundance of the class Chlorobia with increasing depth.The phylum Bacteriodetes is the most abundant phylum (53 %) in the deeper chemocline depths(11 m to 12 m; Fig. 5.2a) and of the three classes that comprise the phylum Bacteriodetes, themembers of class Chlorobia were most abundant—–representing 36 % of the overall community105in the deeper depths (Fig. 5.2b). Notably, members of the candidate phylum MBNT15 and therelatively unknown Patescibateria made up another 22 % of the deeper chemocline (9 % and 13 %respectively; Fig. 5.2a). Conversely, Archaea represent a maximum of 2 % of the whole microbialcommunity (at 11 m), and 1.6 % of deeper chemocline community (Fig. 5.2a). Thus, the microbialcommunity in the Kabuno Bay chemocline, and in particular the community that populates thedeeper chemocline depths, is dominated by few groups, including the abundant Chlorobia.The dominant organisms representing the class Chlorobia in Kabuno Bay are comprisedof 6 distinct taxonomic lineages, most of which are closely related to known photoferrotrophs.Collectively, the 6 distinct lineages of Chlorobia made up to 50 % of the whole community(approximately 2.7 x 104 cells mL – 1) at 11 m and between 11 and 40 % at the other depths (Fig.5.2c, B.2). The 5 most abundant lineages form a closely related, but distinct, clade that containsboth known photoferrotrophic Chlorobia, Chlorobium ferrooxidans str. KoFox and Chlorobiumphaeoferrooxidans str. KB01, the latter of which was enriched and isolated from Kabuno Bay [48].This clade is defined by divergence of 0.4 % and these organisms, therefore, would collectively beclassified as strains of the same species according to canonical definitions based on identity in16S rRNA gene sequences. The microbial community in Kabuno Bay is thus largely made up ofhighly abundant and closely related, but distinct, lineages of Chlorobia with strong phylogeneticaffinity to known photoferrotrophs.106Depth (m)101111.2511.51212m11.5m11.25m11m10mArchaeactinobacteriotaAlphaprote bacteriaG mmaproteobacteriaBacteroidiaChlorobiaIgn vibacteriaChloroflexotaCyanobacteriotaD sulfobacterotaMBNT15Omni rophotaPatescibacteriaUnclass fiedOtherPhyla2Depth2● 120000Phyla2●●●●●●●●●ArchaeaActinobacteriotaAlphaproteobacteriaGammaproteobacteriaBacteroidiaChlorobiaIgnavibacteriaChloroflexotaCyanobacteriota12m11.5m11.25m11m10mArchaeactinobacteriotaAlphaprote bacteriaG mmaproteobacteriaBacteroidiaChlorobiaIgn vibacteriaChloroflexotaCyanobacteriotaD sulfobacterotaMBNT15Omni rophotaPatescibacteriaUnclass fiedOtherPhyla2Depth2● 120000Phyla2●●●●●●●●●ArchaeaActinobacteriotaAlphaproteobacteriaGammaproteobacteriaBacteroidiaChlorobiaIgnavibacteriaChloroflexotaCyanobacteriota12m11.5m11.25m11m10mArchaeactinobacteriotaAlphaprote bacteriaG mmaproteobacteriaBacteroidiaChlorobiaIgn vibacteriaChlorof exotaCyanobacteriotaD sulfobacterotaMBNT15Omni rophotaPatescibacteriaUnclass fiedOtherPhyla2Depth2● 120000Phyla2●●●●●●●●●ArchaeaActinobacteriotaAlphaproteobacteriaGammaproteobacteriaBacteroidiaChlorobiaIgnavibacteriaChloroflexotaCyanobacteriotai il i ii il ii il l il i i il iDepth2 i il iii il ii il liDepth2Cells mL-1012m11.5m11.25m11m10mArchaeaActinobacteriotalphapro obacteriaGammap oteobacteriaBacteroidiaChlorobiaIgn v acteriaChloroflexotaCy nobac eriotaDesulf bacterotaMBNT15Omnit photaPatescibacteriaUncl ssif edOth rPhyla2Depth2alpha0.8Abundance●●●●●120120012000120000Phyla2●●●●●●●●●●●●●●●ArchaeaActinobacteriotaAlphaproteobacteriaGammaproteobacteriaBacteroidiaChlorobiaIgnavibacteriaChloroflexotaCyanobacteriotaDesulfobacterotaMBNT15OmnitrophotaPatescibacteriaUnclassifiedOther1.  x 1021.  x 1031.2 x 1041.  x 1 512m11.5m11.25m11m10mArchaeaActinobacteriotalphapro obacteriaGammap oteobacteriaBacteroidiaChlorobiaIgn v acteriaChloroflexotaCy nobac eriotaDesulf bacterotaMBNT15Omnit photaPatescibacteriaUncl ssif edOth rPhyla2Depth2alpha0.8Abundance●●●●●0120120012000120000Phyla2●●●●●●●●●●●●●●●ArchaeaActinobacteriotaAlphaproteobacteriaGammaproteobacteriaBacteroidiaChlorobiaIgnavibacteriaChloroflexotaCyanobacteriotaDesulfobacterotaMBNT15OmnitrophotaPatescibacteriaUnclassifiedOtherPhylum*Chlorobia36%Other64%% Chlorobia 11 - 12 ma b2.0Chlorobaculum tepidumChlorobium sp N1Prosthecochloris aestuarii DSM 271Chlorobium phaeobacteroides DSM 266Feb11m Chlorobium (phaeo)ferrooxidansChlorobium luteolumChlorobium sp. ShCl03Chlorobium sp. sy9Chlorobium sp. KB01Feb10m Chlorobaculum limnaeumFeb11.25m Chlorobium (phaeo)ferrooxidansChlorobaculum limnaeumChlorobaculum parvum NCIB 8327Chlorobium limicolaFeb10m Chlorobium (phaeo)ferrooxidansChlorobium sp. ChlvPS10Chlorobaculum thiosulfatiphilumChlorobium phaeovibrioides DSM 265Chlorobium clathratiformeProsthecochloris vibrioformisChlorobium tepidum TLSChloroherpeton thalassium ATCC 35110Chlorobium sp. PrPS10Feb12m Chlorobium (phaeo)ferrooxidansFeb11.5m Chlorobium (phaeo)ferrooxidansChlorobium ferrooxidans DSM 13031Chlorobium limicola DSM 245Ignavibacterium outgroupChlorobium chlorochromatii CaD3Chlorobium phaeobacteroides BS12.0Chlorobium phaeovibrioides DSM 265Chlorobium sp. PrPS10Feb10m Chlorobium ferrooxidansChlorobium sp. KB01Chlorobium luteolumChloroherpeton thalassium ATCC 35110Feb1125m Chlorobium ferrooxidansChlorobaculum tepidumChlorobium phaeobacteroides DSM 266Chlorobaculum limnaeumFeb115m Chlorobiu  ferrooxidansProsthecochloris vibrioformisChlorobium sp. sy9Chlorobium limicola DSM 245Feb11m Chlorobium ferrooxidansChlorobium tepidum TLSChlorobium sp N1Chlorobium phaeobacteroides BS1Chlorobium ferrooxidans DSM 13031Chlorobium limicolaFeb10m Chlorobaculum limnaeumFeb12m Chlorobium ferrooxidansChlorobium sp. ShCl03Chlorobium chlorochro atii CaD3Chlorobaculum thiosulfatiphilumProsthecochloris aestuarii DSM 271Chlorobaculum parvum NCIB 8327Chlorobium clathratifor eI navib cterium outgroupChlorobium sp. ChlvPS10100/99/10024/23/9097/93/100100/100/10063/66/10097/91/10066/77/10031/13/10011/18/10052/70/10038/12/10099/99/100100/89/9544/89/9565/72/10088/77/10071/76/10091/55/93100/98/10052/45/6032/38/9334/27/8577/45/9048/27/5996/30/90100/92/9979/53/92100/56/100Genus Family OrderChlorobiumChlorobaculumProsthecochlorisChloroherpetonChlorobiaceaeChlorobiales2.0cFigure 5.2: Microbial community composition in Kabuno Bay. The abundance (cells mL – 1) of each phylum (a)based on the reconstruction of the 16S rRNA sequences in the metagenomes. The abundance of the classChlorobia compared to the rest of the community in the deeper chemocline (11 m to 12 m) (b). Phylogeniesof the class Chlorobia 16S rRNA genes and the Chlorobia metagenome reconstructed 16S rRNA genes withbootstrap values shown at each node (maximum parsimony/maximum likelihood/bayesian) (c). Pie chartsillustrate the absolute abundance of that individual 16S rRNA gene (blue) compared the whole communityfor each depth. Photoferrotrophic Chlorobia are underlined in orange. *Phylum Bacteriodetes is shown asits three main classes (Bacteroidia, Chlorobia, Ignavibacteria) and phylum Proteobacteria is shown as twoof its classes (Alphaproteobacteria, Gammaproteobacteria).1075.4.3 Photosynthetic and oxidative metabolic potentialThe metabolic potential for different modes of photosynthesis exhibited strong vertical structure,with dominance by anoxygenic photosynthesis throughout the chemocline. Genes that are specificto three different modes of photosynthesis — BChl biosynthesis genes (Chlorobia, BChl A),photosystem II (specific to oxygenic photosynthesis, PSII), and reaction center genes (specificto anoxygenic photosynthetic Proteobacteria, pufM/L) (Fig. 5.3, 5.4a) – differed in their relativeabundance as a function of depth. The collective metabolic potential for photosynthesis, based onthe combined relative abundance of these genes, was greatest at 11.25 m (Fig. 5.4a) and was largelydue to high relative abundances of the bacteriochlorophyll biosynthesis genes (Fig. 5.3). Indeed,among the photosynthetic genes investigated, the bacteriochlorophyll biosynthesis genes werethe most abundant, representing almost 10 % of the total metabolic potential (summed RPKMvalues for each gene compared to the single copy marker gene rpoB — see methods), and up to20 % of the depth-specific metabolic potential in the mid-chemocline (Fig. 5.3, 5.4a). The depthdistribution of bacteriochlorophyll biosynthesis genes was similar to that of the phylogeneticmarker genes of the class Chlorobia (Fig. 5.2, B.3a). In contrast, the abundance of the PSII andpuf genes were very low (< 1 % of the metabolic potential at each depth; Fig. 5.3, 5.4a), butnotably, were highest at 10 m and PSII was present throughout the chemocline. Distributionsof key photosynthetic genes thus reveal a dominance of photosynthetic metabolic potential bybacteriochlorophyll dependent anoxygenic Chlorobia.The metabolic potential for electron donor use in anoxygenic photosynthesis was distributedacross multiple possible pathways, which were all present at comparable abundances throughoutthe chemocline. Genes implicated in electron transfer from some inorganic primary electrondonors (Fe2+ or S2O32 – ) to the photosystem include: cyc2 — putative iron oxidase; soxB -–thiosulfate oxidase; hypA -– hydrogenase (Fig. 5.3, 5.4a). All three of these genes were foundthroughout the water column and at comparable abundances, with the abundance of the cyc2 andhypA exhibiting depth distributions (Fig. 5.3, 5.4a) similar to that of the class Chlorobia (Fig. 5.2,B.3a), while the distribution of soxB (Fig. 5.3, 5.4a) better reflected that of sulfide concentrations(Fig. 5.1c). Specifically, the cyc2 genes represented 7 % of the metabolic potential of the whole108Depth (m)101111.2511.512Genes12m11.5m11.25m11m10mPSIIBChl ack buk butMM−CoA mutasehypAcyc2dsrAdsrBsoxBmcrAPai ed CXXCHNAGene3Depth2●●●●●●●●●●●●●BChlackbukbutMM−CoA mutasehypAcyc2dsrAdsrBsoxBmcrAPaired CXXCHNARPKM●●07512m11.5m11.25m11m10mPSIIBChl ack buk butMM−CoA mutasehypAcyc2dsrAdsrBsoxBmcrAPai ed CXXCHNAGene3Depth2●●●●●●●●●●●●●BChlackbukbutMM−CoA mutasehypAcyc2dsrAdsrBsoxBmcrAPaired CXXCHNARPKM●●075..II l ck k t  t sy cyc sr sr s x cri  Depth2●●●●●●●●●●●●●lckkt t sycycsrsrs xcrir  ●12m11.5m11.25m11m10mPSIIBChl ack buk butMM−CoA mutasehypAcyc2dsrAdsrBsoxBmcrAPai ed CXXCHNAGene3Depth2●●●●●●●●●●●●●BChlackbukbutMM−CoA mutasehypAcyc2dsrAdsrBsoxBmcrAPaired CXXCHNARPKM● 07512m11.5m11.25m11m10mPSIIBChl ack buk butMM−CoA mutasehypAcyc2dsrAdsrBsoxBmcrAPai ed CXXCHNAGene3Depth2●●BChlackbukbutMM−CoA mutasehypAcyc2dsrdsrBsoxBmcrAPaired CXXCHNARPKM●●07512m11.5m11.25m11m10mPSIIBChlackbukbutMM−CoA mutasehypAcyc2dsrAdsrBsoxBmcrAPaired CXXCHNAGene3Depth2alpha0.8Gene3●●●●●●●●●●●●●●PSIIBChlackbukbutMM−CoA mutasehypAcyc2dsrAdsrBsoxBmcrAPaired C XCHNARPKM●●●●●07515030060012m11.5m11.25m11m10mPSIIBChlackbukbutMM−CoA mutasehypAcyc2dsrAdsrBsoxBmcrAPaired CXXCHNAGene3Depth2alpha0.8Gene3●●●●●●●●●●●●●●PSIIBChlackbukbutMM−CoA mutasehypAcyc2dsrAdsrBsoxBmcrAPaired CXXCHNARPKM●●●●●075150300600RPKMFigure 5.3: Abundance of key genes pr dicted to play roles in the biogeochemical cycles in Kabuno Bay where thesize of each bubble represents the total RPKM (Reads per kilobase mapped) for that gene at the specifieddepth, while the colors delineate each gene.community, with a peak of 12 % at 11 m (B.4), while soxB genes peaked at 12 m and represented16 % of the community metabolic potential (B.4). The hypA genes were the most abundant ofthe three genes linked to electron donors with a peak of 30 % at 11 m (B.4). The distributions ofthese genes and those linked to photosynthesis, as described above, resolve metabolic potentialfor multiple modes of anoxygenic photosynthesis, largely affiliated with the class Chlorobia,throughout the chemocline.Metabolic potential for multiple modes of photosynthesis is distributed across the genomesof key high- and low-abundance lineages. A large number of high quality population-basedmetagenome assembled genomes (MAGs) were recovered (Table 5.1) including six individualMAGs that contained the pathways for either anoxygenic or oxygenic photosynthesis. Theseincluded three MAGs of Chlorobia, one near-complete MAG closely related to Chlorobaculumsp. (Chlorobac 01; 100 % complete, 4.7 % contamination), as well as two partially-completeMAGs (Chlorobium ferrooxidans -– Chlorobi 01; 52.8 % complete with no contamination -– andChlorobaculum sp. — 68.3 % complete with 2.2 % contamination) (Table 5.1). All three MAGscontain metabolic potential for anoxygenic photosynthesis including bacteriochlorophyll biosyn-thesis and carbon fixation genes (Fig. 5.4a, B.7). Both Chlorobaculum sp. MAGs contain genesassociated with sulfide, thiosulfate, and hydrogen oxidation similar to those identified from109anoxygenic photosynthetic pathways in other members of the Chlorobaculum genus (Fig. 5.4a,B.7). The other three, non-Chlorobia, photosynthetic MAGs contained phylogenetic marker genesof three different members the phyla Cyanobacteria (Fig. 5.4a, B.7, Table 5.1; Neosynechoccaceae,Phormidesmiaceae, and Pseudanabaena sp.(Cyano 02)). Like most Cyanobacteria, these photosyn-thetic MAGs contained necessary genes and pathways for oxygenic photosynthesis (Fig. 5.4a,B.7). Notably, all three of these Cyanobacteria MAGs also contained the genes and pathways foreither sulfide/thiosulfate oxidation or hydrogen oxidation (Fig. 5.4a), which indicates that theseoxygenic phototrophs also have the metabolic potential for anoxygenic photosynthesis. Metabolicpotential for anoxygenic photosynthetic sulfide and hydrogen oxidation in Kabuno Bay is thuspresent across low-abundance lineages of the Chlorobia and cyanobacteria, whereas the potentialfor Fe oxidation is absent from all photosynthetic MAGs (Fig. 5.4a).110bCBBCytoplasmPeriplasmChlorobi_01GH3GH13GlycolysisRuBisCOPstABCSPO4-PO4-CO2PP PathwayChlorosomeBChl ecChlorobium sp.Chlorobaculum sp. 01Chlorobaculum sp. 02NeosynechoccaceaePhormidesmiaceaePseudanabaena sp.Primary ProductionPhotosynthesistot PhotosynthesisanoxPhotosynthesisoxaFigure 5.4: Primary production in Kabuno Bay with the RPKM of each gene in the bar graph and the MAGsthat contain those genes underneath (a). Chlorobium sp. MAG (Chlorobi 01) from 10 m (blue) aligned to C.phaeoferrooxidans strain KB01 genome (green) where the grey links represent nucleotide sequences greaterthan 20 nt’s with 100 % identity (b). The major pathways found in the Chlorobi 01 MAG (c). Note: the starabove the all MAGs for soxB is due to the fact that the RPKM of the soxB genes in the MAGs extends to 356.111Table 5.1: Extended data from a selection of MAGs including completeness, contamination, and taxonomy.Note: Near-complete genomes (Near) have a completeness ≥ 90 % and contamination ≤ 5 %; Medium-complete genomes (Medium) have a completeness ≥ 70 % and contamination ≤ 10 %; Partially-completegenomes (Partial) have a completeness ≥ 50 % and contamination ≤ 4 %.Name Depth (m)MAG #Compl-etenessContami-nationQ-scoreMAG TypeNetwork tiersAverage RPKM per MAGFull taxonomyChlorobac_01 10 16 100 4.7 76.5 Near N/A 4.6 Bacteria;Bacteroidetes;Chlorobia;Chlorobiales;Chlorobiaceae;ChlorobaculumChlorobi_01 10 6 52.8 0 52.8 Partial N/A 27.7 Bacteria;Bacteroidetes;Chlorobia;Chlorobiales;Chlorobiaceae;Chlorobium;Chlorobium ferrooxidansCyano_01 10 58 94 2.3 82.5 Near N/A 0.7 Bacteria;Cyanobacteria;Melainabacteria;CaenarcaniphilalesCyano_02 10 124 89.8 6.1 59.3 Medium N/A 0.2 Bacteria;Cyanobacteria;Oxyphotobacteria;Pseudanabaenales;Pseudanabaenaceae;PseudanabaenaBact_04 11.5 28 81.9 0.7 78.4 Medium 2,3 3.2 Bacteria;Bacteroidetes;Bacteroidia;Bacteroidales;GWA2-32-17;GWA2-32-17Bact_05 11.25 32 78.2 0.2 77.2 Medium 2,3 2.9 Bacteria;Bacteroidetes;Bacteroidia;Bacteroidales;GWA2-32-17;GWA2-32-17Bact_07 11.25 22 85.5 0.2 84.5 Medium 1,2,3 4.8 Bacteria;Bacteroidetes;Bacteroidia;Bacteroidales;TTA-H9Bact_08 11.5 20 91.3 0.9 86.8 Near 1,2,3 5.0 Bacteria;Bacteroidetes;Bacteroidia;Bacteroidales;TTA-H9Igna_03 11.25 14 80.7 1.6 72.7 Medium 1,2 8.1 Bacteria;Bacteroidetes;Ignavibacteria;Ignavibacteriales;IgnavibacteriaceaeIgna_06 11.25 7 66.3 1.3 59.8 Partial 1,2 26.2 Bacteria;Bacteroidetes;Ignavibacteria;IgnavibacterialesIgna_07 12 7 73.6 1.3 67.1 Medium 1,2 8.8 Bacteria;Bacteroidetes;Ignavibacteria;IgnavibacterialesLenti_01 11.25 54 96.2 1.6 88.2 Near 3 1.3 Bacteria;Bacteroidetes;Bacteroidia;Bacteroidales;Lentimicrobiaceae;LentimicrobiumLenti_02 11.25 12 84 3.1 68.5 Medium 1,2,3 14.7 Bacteria;Bacteroidetes;Bacteroidia;Bacteroidales;LentimicrobiaceaeLenti_03 11.5 8 92.7 3.5 75.2 Near 1,2,3 14.8 Bacteria;Bacteroidetes;Bacteroidia;Bacteroidales;LentimicrobiaceaeMBNT15_01 11.25 8 85.7 1.3 79.2 Medium 1,2 25.1 Bacteria;MBNT15;MBNT15;MBNT15;MBNT15;CG2-30-66-27MBNT15_02 11.5 4 86.5 0.8 82.5 Medium 1,2 26.1 Bacteria;MBNT15;MBNT15;MBNT15;MBNT15;CG2-30-66-27MBNT15_04 11.25 18 82.1 4.3 60.6 Medium 1,2 6.0 Bacteria;MBNT15;MBNT15;MBNT15;MBNT15;CG2-30-66-27MBNT15_07 10 13 92.3 2.5 79.8 Near N/A 5.8 Bacteria;MBNT15;MBNT15;MBNT15;MBNT15;CG2-30-66-27Pates_10 11.25 21 59 0 59.0 Partial 2 5.0 Bacteria;Patescibacteria;Gracilibacteria;Absconditabacterales;X112;undefinedPates_17 12 3 69.3 0.5 66.8 Partial 1 19.4 Bacteria;Patescibacteria;Paceibacteria;Moranbacterales;UBA1568Pates_20 10 9 69.8 0 69.8 Partial N/A 13.9 Bacteria;Patescibacteria;Paceibacteria;Moranbacterales;UBA1568Desulfo_03 11.25 20 97.1 1.3 90.6 Near 1,2,3 5.4 Bacteria;Desulfobacterota;Desulfomonilia;Desulfomonilales;DesulfomonilaceaeDesulfo_04 11.5 19 97.1 1.9 87.6 Near 2,3 5.3 Bacteria;Desulfobacterota;Desulfomonilia;Desulfomonilales;DesulfomonilaceaeGamma_01 11.25 30 98.1 1.5 90.6 Near 2,3 3.0 Bacteria;Proteobacteria;Gammaproteobacteria;Betaproteobacteriales;SG8-39;2-12-FULL-64-23Gamma_02 11.5 42 98.3 0.3 96.8 Near 3 1.9 Bacteria;Proteobacteria;Gammaproteobacteria;Betaproteobacteriales;SG8-39;2-12-FULL-64-23Gamma_03 10 14 77.4 3.5 59.9 Medium N/A 5.7 Bacteria;Proteobacteria;Gammaproteobacteria;Betaproteobacteriales;Methylophilaceae;Methylopumilus_ANitros_01 11.5 38 90 2.7 76.5 Near 3 2.2 Bacteria;Nitrospirota;Thermodesulfovibrionia;Thermodesulfovibrionales;UBA1546Actino_01 11.5 50 93.6 3 78.6 Near 3 1.8 Bacteria;Actinobacteria;Acidimicrobiia;IMCC26256Arch_01 10 34 98.9 1.3 92.4 Near 3 1.3 Archaea;Halobacterota;Methanomicrobia;Methanomicrobiales;Methanoregulaceae;MethanoregulaArch_02 11 16 88 4.7 64.5 Medium N/A 0.3 Archaea;Microarchaeota;Micrarchaeia;UBA10214The most abundant photosynthetic MAG, identified as a Chlorobium sp., is very similar toChlorobium phaeoferrooxidans strain KB01, previously enriched and isolated from the Kabuno Baywater column. Indeed, the genome of the Chlorobium MAG (Chlorobi 01; Table 5.1) is 97 %identical, at the nucleotide level, to the genome of Chlorobium phaeoferrooxidans strain KB01 (Fig.5.4b). Strain KB01 is a photoferrotroph [58, 115] whose genome contains genes (cyc2) previouslylinked to photoferrotrophy [58, 115, 142, 194, 274]. Included in the genes that are identical betweenthe two genomes are some components for nutrient cycling such as the nitrogenase subunitsrequired for nitrogen fixation and phosphate transport genes (Fig. 5.4c, B.6) [115] and componentsof the pathways associated with autotrophic growth, specifically those which facilitate inorganic112carbon (CO2) fixation such as those for the Calvin-Benson-Bassham (CBB) cycle (Fig. 5.4c, B.6).The 3 % of the Chlorobi 01 genome that does not match the strain KB01 genome identically ismostly comprised of small contigs with between one and three predicted genes (open readingframes — ORFs). The majority of these ORFs are homologous to hypothetical gene sequencesthat have been predicted from a variety of microorganisms, none of which are closely relatedto strain KB01. The Chlorobi 01 genome did not contain the genes for phototrophic sulfide orhydrogen oxidation, nor did it contain the outer membrane cytochrome (cyc2). Strain KB01 doeshost the cyc2 gene, linked to Fe(II) oxidation, and does have the metabolic potential for thiosulfateoxidation (soxB) and hydrogen oxidation (hypA) (Fig. B.6). The similarities between the genomesequences of strain KB01 and Chlorobi 01, therefore, indirectly links the metabolic potential forphotoferrotrophy and anoxygenic hydrogen oxidation to the dominant Chlorobia lineages in theKabuno Bay water column.The metabolic potential for iron, sulfide and hydrogen oxidation was distributed across bothphotosynthetic and non-photosynthetic lineages, not all of which were well represented in theMAGs. All three functional marker genes (cyc2, soxB, and hypA) were found in greater abundancein the non-photosynthetic MAGs than in the photosynthetic MAGs (Fig. 5.4a). The cyc2 genewas only found in non-photosynthetic MAGs, but analyses of these sequences revealed highdegree of identity to the cyc2 gene from strain KB01 and these sequences were thus likely presentas contaminants in the non-photosynthetic MAGs. While all soxB genes were associated withMAGs—found in 28 % of the recovered MAGs—most of these MAGs were non-photosynthetic(Fig. 5.4a). Additionally, the majority of soxB genes found in MAGs represented lesser abun-dant members of the microbial community such as those of the phylum Omnitrophota or classAlphaproteobacteria (Fig. 5.2). Only a fraction of the hypA genes were associated with MAGs,and like soxB, most of these MAGs were non-photosynthetic (Fig. 5.4a). Thus, while most ofthe metabolic potential for S oxidation is clearly non-photosynthetic, the metabolic potential forhydrogen oxidation is not fully resolved though MAGs and the distribution between photosyn-thetic and non-photosynthetic taxa remains somewhat uncertain. The metabolic potential for ironoxidation is entirely unresolved through MAGs but based on cyc2 sequence analyses is likelyhosted almost entirely by the photoferrotrophic Chlorobia.1135.4.4 Carbon breakdown and fermentationThe abundance of genes associated with carbon degradation dominate the overall metabolicpotential of the microbial community in Kabuno Bay. Glycosyl hydrolases (GHs), some of whichhave capacity to hydrolyze complex carbon sources such as cellulose, like GH5, while othersare capable of breaking the bond between the sugars and amino acids found in peptidoglycan(GH23) had higher relative abundance than any other functional gene examined (Fig. 5.3, B.3b,B.4). GHs were ubiquitously abundant across all depths in Kabuno Bay (B.3b, B.4). Glycosyltransferases (GTs) were also similarly abundant throughout the chemocline (data not shown). TheGHs present were linked to a diverse set of carbon degradation pathways with the most abundantGHs implicated in the degradation of oligosaccharides (GH 2 — 35 % of the total communitymetabolic potential based on a comparison to the single copy gene rpoB), cellobiose (GH 3 -– 57%), cellulose (GH 5 -– 38 %), starch (GH 13 -– 150 % and GH 57 -– 33 %), peptidoglycan (GH 23-– 78 %), and other polysaccharides (GH 74 -– 77 %). The majority of the most abundant GHsexhibited some depth-dependent structure with a peak in abundance at 11 m (Fig. B.3b, B.4),where the most abundant GH (GH13) reached almost 200 % (twice as many copies of GH 13 ascopies of the single copy gene rpoB) (Fig. B.4). GHs, and by extension the metabolic potential forcarbon breakdown, thus represent an extremely abundant component of microbial communitymetabolism in Kabuno Bay.The metabolic potential for fermentation pathways was prevalent throughout the chemocline,albeit at lower relative abundances than those of the complex carbon breakdown pathways. Of thefermentation pathway genes, MM-CoA mutase, required for fermentation of glucose to succinateand propionate (converting phosphoenolpyruvate—–PEP—–to oxaloacetate) had the highestdepth-specific abundance (up to 95 % at 11 m depth) as well as the highest overall abundance(75 %) (Fig. B.4). The abundance of pyruvate kinase (pk – converts PEP to pyruvate) is roughly amirror image of the MM-CoA mutase (Fig. B.4), with its lowest abundance (30 %) at 11 m depth.Pyruvate dehydrogenase (converts pyruvate to acetyl-coA, releasing CO2), in contrast, had verylittle depth-dependent structure, with about 50 % of the overall metabolic potential representedby this pathway at every depth. The other fermentation pathways queried represented smaller114portions of the overall metabolic potential (Fig. 5.3) with acetate production (acetate kinase — ack)and butyrate production (butyryl-CoA transferase — but) at 25 and 20 percent, respectively (Fig.B.4). While the metabolic potential for diverse fermentation pathways is abundant and widelydistributed across depth in Kabuno Bay, none of the described genes or pathways are as abundantas those required for complex carbon breakdown as described above.The metabolic potential for complex carbon degradation and fermentation pathways is ubiqui-tous across the microbial community and well represented in the MAGs. Genes for both hydrolysisand fermentation were found in the majority of the high-quality MAGs (Q-score of greater than 50— see methods) with GH’s present in 93 % and fermentation genes present in 68 % of the recoveredMAGs, respectively. Furthermore, the taxonomy of these MAGs matched the taxonomy of someof the most abundant members of the microbial community including members of the candidatephyla MBNT15 and phylum Bacteroidetes–—namely those of the family Lentimicrobiaceae (Fig.5.2a, 5.5, Table 5.1). A subset of MAGs that represents all the most abundant classes within themicrobial community (Fig. 5.2a, Table 5.1) and includes the most abundant MAGs recovered(relative abundance based on the average RPKM of a MAG compared to the total RPKM of theMAGs; Table 5.1 — see methods) exemplify the distribution of the major carbon breakdownand fermentation pathways across prevalent microbial community members (Fig. 5.5, B.7). Forexample, the starch and other polysaccharide degradation pathways (identified through genessuch as the d-galacturonate epimerase) are found in highly abundant community members suchas members of the phylum MBNT15 (MBNT15 01; Fig. 5.5, B.7), whereas the peptidoglycan break-down pathways (identified through genes such as the beta-N-acetylhexosaminidase) are foundmostly in the Lentimicrobiaceae MAGs (Lenti 03; Fig. 5.5, B.7). A detailed schematic representationof the characteristic pathways found in two examples of these relatively abundant near completeMAGs (≥ 90 % complete and ≤ 5 % contaminated), MBNT15 01 and Lenti 03, is shown in Fig. 5.5.While some carbon breakdown pathways appear to be taxonomically restricted, many genes froma diverse set of fermentation pathways were present in almost all the MAGs selected (Fig. 5.5,B.7). A notable exception are the Patescibacteria MAGs, which contain limited metabolic potentialand either possessed the genes for complex carbon breakdown or fermentation (B.7). Collectively,these data demonstrate ubiquity of the complex carbon breakdown and fermentation pathways115across diverse lineages and highlight the partitioning of some metabolic potential within specifictaxa.CytoplasmPeriplasmMBNT15_01PhnDECGlycolysisTCAMixed acid fermentationGH1GH13GH23 CoxABCDHyp H22H+H2O ½O2CPO(OH)2PstABCSPO4-PO4- CPO(OH)2H2CytoplasmPeriplasmLenti_03GH2GH3GH5PstABCSPO4-GlycolysisTCAMixed acid fermentationGH13GH74GH23FccABS0; SO4-HS-; H2SCytCCytCFe(III)e-Fe(III)e-PO4-PP pathwayCoxABCDH2O ½O2CydABXCytoplasmPeriplasmGamma_01NarGHNO2-NxrABCoxABCDH2O ½O2CydABXSorBSO4-SO32-GH3GH13GH15GH23CytCCytCFe(III)e-Fe(III)e-GlycolysisTCAMixed acid fermentation PP pathwayPhnDECCPO(OH)2PstABCSPO4-PO4- CPO(OH)2NO2-NO3-Figure 5.5: A representation of the metabolic potential for three MAGs including an abundant fermenter(MBNT15 01), a potential Fe(III) reducer with the metabolic potential for multiple carbon degradation path-ways (Lenti 03), and a potential Fe(III) reducer with the metabolic potential for fewer carbon degradationpathways (Gamma 01). The description of each gene listed in found in B.2.The microbial community in Kabuno Bay has the metabolic potential to channel degradationof complex carbon compounds through fermentation products to CO2 and CH4 via a metabolicnetwork with a suite of non-unique, apparently redundant pathways (Fig. 5.6). The upper tier ofthe network contains a subset of abundant MAGs that possess the genes for the acquisition andbreakdown of complex carbon compounds such as disaccharides or larger more complex carbonmolecules (e.g. cellulose). This tier comprised 11 different taxa (including six taxonomicallydistinct MBNT15 MAGs and five Patescibacteria MAGs; Fig. 5.6) whose collective abundancerepresented 8 % of the total microbial community (average MAG RPKM compared to the RPKMof single copy gene rpoB for the whole community). The hydrolytic pathways hosted principally,but not exclusively, in these organisms ultimately produce glucose through the breakdown ofcomplex carbon compounds. The middle tier of the network hosts the MAGs with the metabolic116potential to perform mixed acid fermentation, channelling the glucose produced in the upper tierinto a number of fermentation pathways. This tier was comprised of seven different phylogeneticgroups, with the majority of the same MAGs from tier 1 present (25 out of 30) and with a singleGammaproteobacteria MAG representing the only new class distinct from the top tier (collectiveabundance of 7 % of the total microbial community; Fig. 5.6). Notably, many of the same MAGswere present in both tiers with metabolic potential for processing more than one substrate orthe production of more than one product (Igna 06, Lenti 02, Lenti 03; Fig. 5.6). For example,the Lentimicrobiaceae MAGs had the metabolic potential to degrade all four types of complexcarbon compounds examined as well as all four major fermentation pathways (Fig. 5.6). OtherMAGs were present on both tiers but contained fewer potential pathways (e.g. MBNT15 01,MBNT15 02; Fig. 5.6). For example, the MBNT15 MAGs contained the pathways required forcellobiose and peptidoglycan degradation as well as those for glucose fermentation to propionateand CO2 but lacked the other carbon degradation and fermentation pathways present in othercommunity members (Fig. 5.6). Thus, while members of the phylum MBNT15 are abundantthroughout the water column (Figure 2), their growth options are restricted in comparison tomembers of the phylum Bacteroidetes (e.g. Lentimicrobiaceae). A broad suite of taxa, notablyincluding Lentimicrobiaceae, Ignavibacteriaceae, MBNT15, and Patescibacteria thus have metabolicpotential to drive conversion of complex carbon compounds to glucose, and some, Lentimicrobiaceaeand Ignavibacteriaceae for example, to continue this conversion to fermentation products. Anothersuite of taxa, notably MBNT15 and Patescibacteria have metabolic potential to exclusively fermentglucose to select VFAs and hydrogen.117Acetate Butyrate H2/CO2PropionateOligosaccharides Cellobiose CellulosePeptidoglycanDesulfo_03Desulfo_04Desulfo_05MBNT15_01MBNT15_02MBNT15_03MBNT15_04MBNT15_05MBNT15_06Bact_03Bact_07Bact_08Desulfo_01Desulfo_10Inga_03Igna_04Igna_05Igna_06Igna_07Lenti_02Lenti_03Lenti_04Lenti_05Lenti_06Lenti_07Pates_01Pates_04Pates_17Pates_18Pates_19GlucoseIron reduction Sulfate reductionMethanogenesisBact_03Bact_04Bact_05Bact_07Bact_08Desulfo_01Desulfo_03Desulfo_04Desulfo_05Desulfo_10Gamma_01Igna_03Igna_04Igna_05Igna_06Igna_07Lenti_02Lenti_03Lent_04Lenti_05Lenti_06Lenti_07Lenti_08MBNT15_01MBNT15_02MBNT15_03MBNT15_04MBNT15_05MBNT15_06Pates_09Actino_01Arma_01Arma_02Bact_04Bact_05Bact_06Bact_07Bact_08Desulfo_01Desulfo_08Desulfo_10Gamma_01Gamma_02Igna_02Lenti_01Lenti_02Lenti_03Lenti_05Lenti_08Lenti_09MBNT15_03MBNT15_06Arch_01Cflexi_05Desulfo_02Desulfo_03Desulfo_04Desulfo_05Nitros_01Nitros_02Pates_10Nitros_03Figure 5.6: Flow of carbon compounds through the metabolic pathways found in each MAG. Tier 1 have thepotential to degrade complex carbon molecules to glucose, tier 2 encompasses fermenters, and tier 3 hasthose with the metabolic potential for terminal electron processes. Key MAGs discussed in the text arehighlighted with the white circles (e.g. Lenti 03).1185.4.5 Iron reduction, sulfate reduction, and methanogenesisThe metabolic potential for iron reduction in Kabuno Bay was greater than that of any of theother modes of respiration. Outer membrane heme binding cytochrome genes linked to ironreduction represented the largest portion of the microbial community respiratory metabolicpotential (roughly 30 % throughout the mid-chemocline depths of 11, 11.25, and 11.5 m – B.4).The sulfate reduction genes were much less abundant (peaking at 6 %) and, notably, their geneabundances follow the same depth-dependent distribution as sulfide concentrations in Kabuno Bay(Fig. 5.1c, B.4). The metabolic potential for methanogenesis, however, was very low (<1 % relativeabundance) throughout much of the chemocline with a small peak (2 %) at 11 m (B.4). Overall,respiratory metabolic potential linked to iron reduction (total of 25 % of the whole microbialcommunity) dominated over the other anaerobic microbial respiration pathways throughout thewater column.The metabolic potential for iron reduction and sulfur metabolism was found in a subset ofspecific taxa that were well represented in the MAGs. Putative iron reduction genes were found in15 % of the high quality MAGs, whereas those of sulfate reduction were found in 7 % of the MAGs.Notably, there was very little overlap in the taxonomy of the MAGs that housed the metabolicpotential for each of these pathways. For example, iron reduction genes were found in MAGsaffiliated with the most abundant classes (e.g. Bacteriodia, Ignavibacteria, and MBNT15; Fig. 5.5,5.6, B.7) as well as some classes that are known to contain iron reducers (i.e. Gammaproteobacteria;Fig. 5.5, 5.6, B.7), whereas the majority of the metabolic potential for sulfate reduction was foundin MAGs representing three classes of the phylum Desulfobacterota (Fig. 5.6, B.7, B.8). Themetabolic potential for methanogenesis was only found in one MAG (Arch 01; Table 5.1, Fig. 5.6,B.7, B.8), which was classified as a Methanoregula sp. (Table 5.1). The overall discrete allocationof redox pathways to several exclusive taxa differs from the comparably generic distribution ofcarbon degradation and fermentation pathways. Key and relatively abundant members of themicrobial community are thus responsible for iron respiration, while less abundant members areimplicated in sulfate respiration and methanogenesis.Metabolic potential for both respiration of iron and sulfate was distributed across taxa with119metabolic potential for the use of a broad range of electron donors including both complexcarbon compounds and simple volatile fatty acids, as well as hydrogen. A subset of MAGs withmetabolic potential for iron and sulfate respiration make up the final tier of the microbial network(Fig. 5.6) and represent a relatively smaller portion of the overall microbial community thanthe organisms comprising the upper tiers (4 % compared to 8 % and 7 % for tiers one and tworespectively). While MAGs in this lower tier were obligate with respect to their terminal electronacceptor, many had metabolic potential to use a suite of different electron donors including thefermentation products produced by the second tier (Fig. 5.6). Notably, some MAGs (e.g. Lenti 03,Desulfo 03; Fig. 5.5, 5.6, B.8, Table 5.1) exhibited metabolic potential to acquire and degrademultiple fermentation products or even all four of the main fermentation products (acetate,butyrate, propionate, and H2). Furthermore, many of these MAGs were also found in tier 1 andthus contain the metabolic potential to directly degrade complex carbon compounds (Fig. 5.6).These MAGs, with multiple pathways, were often the more abundant members of the microbialcommunity (e.g. members of the phylum Bacteroidetes—–Ignavibacteria and Bacteroidia, whichincludes the Lentimicrobiaceae; Fig. 5.2a). Other, lesser abundant, MAGs were generally restrictedto the use of only one fermentation product, such as the sulfate reducing Nitrospirota MAG(Nitros 01; Fig. 5.6, Table 5.1). Metabolic potential for the use of a broad suite of electrons is thusa key trait for many of the anaerobic iron and sulfate respiring microorganisms in Kabuno Bay.5.5 Discussion5.5.1 Primary production in Kabuno BayPhototrophic Fe(II) oxidation, supported by phototrophic members of the class Chlorobia andindependent of oxygen, is the dominant form of photosynthesis and primary production inthe Kabuno Bay chemocline. Net phototrophic iron oxidation rates of 3.1 ± 0.4 µmol L – 1 h – 1(Fig. 5.1e), in the presence of an inhibitor of oxygenic photosynthesis (3-(3,4-dichlorophenyl)-1,1-dimethylurea; DCMU), suggest that anoxygenic photosynthetic bacteria are active at 11 m.High concentrations of BChl e pigments [48], ratios of GSB specific cell counts versus total cellcounts (Fig. 5.1d), and rates of anoxygenic photosynthetic carbon fixation [72] further indicate that120there is a large population of anoxygenic phototrophs throughout the Kabuno Bay chemocline.Indeed, members of the class Chlorobia, known anoxygenic phototrophs [141, 142, 275], are themost abundant taxa throughout the deeper chemocline depths (36 % of the whole community;Fig. 5.2b). The most abundant members of the class Chlorobia identified form a clade within thegenus Chlorobium with two other known photoferrotrophic bacteria (Fig. 5.2c; [48, 53, 55, 58])–—Chlorobium phaeoferrooxidans strain KB01 and Chlorobium ferrooxidans strain KoFox—–one ofwhich (strain KB01) was previously isolated from the Kabuno Bay water column [48, 58]. A partialMAG, taxonomically identified as Chlorobium sp., was 97 % identical at the nucleotide level tothe genome of strain KB01 (Fig. 5.4b) and possessed several genes and pathways indicative ofanoxygenic photosynthesis (Fig. 5.4c, B.6), although it lacked the outer membrane iron oxidizingcytochrome (cyc2) previously attributed to photoferrotrophy. The conspicuous absence of thecyc2 gene from the MAG Chlorobi 01 genome thus likely reflects its limited completion (52 %,Table 5.1). The strain KB01 specific cyc2, however, was found throughout the water column (Fig.5.3) and peaked at 12 % of the overall community’s metabolic potential at 11 m (Fig. 5.3, B.4).This suggests that the contigs containing the cyc2 gene did not assemble into MAGs, which isconfirmed by the taxonomy of the two cyc2 genes that were found in non-Chlorobia MAGs beingidentical to the strain KB01 cyc2 gene and implies that they represent contamination in thoseMAGs. Collectively, these data provide evidence that photoferrotrophy is likely the dominantform of primary production and thus provides a significant amount of primary organic carbon tosupport the growth of the microbial community as a whole.Other modes of photosynthesis exist at each depth of the Kabuno Bay chemocline, albeitin lower abundances than the those indicated for photoferrotrophy. Functional anchors, orgenes that are required for specific processes to occur, that represent photosynthetic pathwayssuch as oxygenic photosynthesis (PSII -– photosystem II) and anoxygenic photosynthesis in thephylum Proteobacteria (pufM/L -– photosynthetic reaction center genes specific to Proteobacteria)are present throughout the chemocline (Fig. 5.3, 5.4a). PSII genes found in MAGs that aretaxonomically identified as canonical members of the phototrophic phylum Cyanobacteria (e.g.Pseudanabaena sp. -– Cyano 02; Fig. 5.4a, B.7, Table 5.1) provide further evidence for oxygenicphototrophs in the chemocline, implying metabolic potential for oxygenic photosynthesis under121hvFe(II)Fe(OH)3CO2Organic carbonH2OO2SO42-H2SChlorobium sp.CyanobacteriaChlorobaculum sp.? H2H2OChlorobiaProteobacteriaaSO42- H2SFe(II)CH4CO2Fe(OH)3Organic carbonH2VFACO2MNBT15; Ignavibacteria; LentimicrobiacaeaDesulfomonilaceae;NitrospirotaLentimicrobium;BetaproteobacterialesMethanoregulaCO2bFigure 5.7: Proposed metabolic model for the primary producing microorganisms present at the chemocline in KabunoBay where hv denotes light (a). Proposed metabolic model for the non-primary producing microorganismspresent at the chemocline in Kabuno Bay where VFAs are volatile fatty acids (b).ferruginous conditions. Their presence throughout the water column, even at deeper depths,might infer a lack of the toxicity effects commonly invoked to exclude oxygenic phototrophs fromFe-rich waters [224]. Their less than 1 % abundance implies they are minor members of the totalcommunity, while the lack of inhibition of photosynthesis and Fe(II) oxidation by DCMU (Fig. 5.1e)indicates that oxygenic photosynthesis plays little role in Fe(II) oxidation and primary productionin Kabuno Bay’s chemocline. These cyanobacteria, however, may produce oxygen at low rates,driving a cryptic oxygen cycle or, alternatively, could grow through anoxygenic photosynthesiswith electron donors such as H2S or H2. Indeed, genes required to oxidize these alternativeelectron donors are present in the cyanobacterial MAGs (H2S oxidation genes in Pseudanabaena sp.and H2 in the Neosynechoccaceae; Fig. 5.4a). Laboratory strains isolated from anoxic environments,and closely related to the cyanobacteria MAGs from Kabuno Bay, have also been shown to growthrough both oxygenic and anoxygenic photosynthesis with their photosynthetic mode controlledby the availability of anoxygenic photosynthetic electron donors: H2S or H2 [276–278]. Together,these data provide further evidence for concurrent modes of photosynthesis, including possiblyoxygenic photosynthesis (Fig. 5.7a) under ferruginous conditions.1225.5.2 Microbial metabolisms in Kabuno BayThe majority of the metabolic potential in Kabuno Bay is centered around carbon degradationand fermentation, resulting in an abundant and diverse group of microorganisms who are likelyresponsible for the partitioning of carbon metabolism across distinct microbial taxa. Glycosylhydrolases and functional anchors for fermentation pathways are highly abundant throughoutthe water column (between 40 and 200 % of the overall community metabolic potential; Fig.5.3, B.4). These same genes, found in most MAGs (93 % contained GH’s and 68 % containedfermentation genes and pathways; Fig. 5.6, B.4, B.7), define a group of highly abundant microor-ganisms that degrade and ferment organic carbon–namely the phyla Bacteroidetes, MBNT15, andPatescibacteria (Fig. 5.2, 5.5, 5.6, 5.7b). Prior studies have shown that members of the phylumBacteroidetes—–specifically members of the class Bacteroidia and Ignavibacteria—–can play anactive role (as measured through transcriptomics and proteomics) in carbon degradation andfermentation in mineral and organic-rich environments such as the fen regions of peatlands [247].Members of the phylum Patescibacteria have also been shown to play a role in carbon degradation[279], although their small genomes and general lack of many canonical microbial biosyntheticpathways (e.g. amino acid formation) suggest that these bacteria largely rely on other membersof the microbial community to support their growth [279]. Little is known at this stage aboutthe MBNT15 phylum; but our data suggest, however, that they could play an important rolein carbon compound hydrolysis and fermentation, at least under ferruginous conditions. Thepartitioning of carbon degradation and fermentation pathways across these abundant classes ofbacteria ultimately supplies small carbon compounds and VFAs to other microbial metabolismspresent throughout Kabuno Bay.Notable, non-iron-based, metabolisms in Kabuno Bay include those involved in sulfur cy-cling—sulfide oxidation and sulfate reduction. Sulfate reduction, supported by organic carboncompounds produced through carbon degradation and fermentation, peaked at 11.5 m witha rate of 300 nmol L – 1 d – 1 (Fig. 5.1c). These rates were accompanied by sulfate reductiongenes (dsrAB), which were found throughout the water column (Fig. 5.3), albeit at relatively lowabundances (peaked at 6 % of the overall metabolic potential at 12 m; B.4). The low abundance123of sulfate reduction genes is supported by the restriction of the sulfate reduction pathway tolesser abundant microorganisms (Desulfobacterota; Fig. 5.6, 5.7b, B.7, B.8). Conversely, sulfideoxidation genes were well represented in the collected MAGs (found in 28 % compared to the 7 %of MAGs that contained sulfate reduction genes) and they represented just less than 20 % of theoverall community metabolic potential (B.4). The MAGs that did contain sulfide oxidation genes,however, were those that represented lower abundance phyla (e.g. Omnitrophota). The densityof sulfide oxidation genes and the abundance of the phylum Omnitrophota follow the samedepth-dependent structure (Fig. 5.2a) of the sulfide concentrations in the chemocline (Fig. 5.1c),thus further supporting the link between these members of the microbial community and sulfideoxidation. Despite the overall lower abundances of the genes implicated in sulfate reduction andsulfide oxidation in Kabuno Bay, minimal pyrite export from the chemocline (6 mmol m – 2 yr – 1)compared to the depth integrated sulfate reduction rates (90 mmol m – 2 yr – 1) demonstrates strongsulfur cycling. Furthermore, the relatively low rates of pyrite export and low rates of sulfatereduction indicate that sulfide production plays a negligible role in Fe(III) reduction, thus directlylinking Fe(III) reduction to microbial respiration.Fe(III) reduction is the most abundant form of respiration in the Kabuno Bay chemocline asthe genes required, and the absolute abundance of the micoorganisms likely involved, outnumberthose associated with other forms of respiration such as sulfate reduction and, in particular,methanogenesis. Genes implicated in Fe(III) reduction were abundant throughout the watercolumn (up to 30 % throughout the chemocline; Fig. 5.3, B.4) and accompanied high rates of Fe(III)reduction (0.4 ± 0.04 µmol L – 1 h – 1) compared to those of sulfate reduction (0.0125 µmol L – 1 h – 1).These high rates of Fe(III) reduction are conferred by the presence of Fe(III) reduction genes inabundant members of the microbial community (e.g. Bacteroidia and Gammaproteobacteria; Fig.5.5, 5.6, 5.7b, B.7). Furthermore, the abundance of Fe(III) reduction genes and of phyla that containFe(III) reduction genes, such as Bacteriodetes, peak at 11 m (Fig. 5.2a) where concentrations ofFe(III) (oxyhydr)oxides are at their highest (Fig. 5.1b), corroborating the importance of thesemembers of the community in Fe(III) reduction. Conversely, the genes for methanogenesis (mcrA)were only found in one MAG (Methanoregula; Arch 01, Table 5.1, B.8) and peaked at 2 % ofthe overall metabolic potential of the microbial community (Fig. B.4). Thus, while both Fe(III)124reduction and methanogenesis appear to be supported by members of the microbial communityin Kabuno Bay, those likely capable of Fe(III) reduction far outnumber the methanogens, whichmay be due to the availability of organic carbon compounds for each metabolism.5.5.3 Microbial community network and tight carbon and iron couplingFe(III) reduction rates and the flux of Fe(III) (oxyhydr)oxides from the Kabuno Bay chemoclineindicate that there is a tight coupling between the carbon and iron biogeochemical cycles. Fe(III)reduction rates are sufficiently high that 80 % of the Fe(III) produced through phototrophic Fe(II)oxidation is converted back to Fe(II) in the euphotic portion of the water column. The theoreticalflux of Fe(III), calculated from the difference between the Fe(II) oxidation and Fe(III) reductionrates in the water column, indicates that 0.1 to 1.7 mol m – 2 yr – 1 Fe(III) will leave the chemocline,which encompasses the measured highly reactive Fe(III) mineral flux captured in sediment traps(0.4 mol m – 2 yr – 1). This appreciable Fe export from the euphotic zone is comparable to thosereported from other modern ferruginous lakes, namely Lake Matano (up to 0.1 mol m – 2 yr – 1)and Lake Towuti (up to 0.25 mol m – 2 yr – 1) [273]. In these environments, therefore, the interplaybetween the microbial mediated rates of iron oxidation and reduction dictates the amount ofhighly reactive Fe(III) exported from the chemocline. Thus, iron based microbial metabolisms,coupled to organic matter degradation, are likely crucial in maintaining the microbial communityunder ferruginous conditions.The partitioning of the metabolic pathways for complex carbon degradation and fermentationcoupled to Fe(III) reduction influence the capacity for coupled carbon and iron cycling. Oneof the most abundant classes of microorganisms in Kabuno Bay, the Bacteroidia (∼7 x 103 cellsmL – 1 at each depth), contained pathways for complex carbon degradation, fermentation, andFe(III) reduction. Specifically, those of the family Lentimicrobiaceae and the genus Lentimicrobium(Lenti 01, Lenti 02, Lenti 03; Fig. 5.5, 5.6, Table 5.1) had metabolic potential for complex carbondegradation, fermentation, and Fe(III) reduction (Fig. 5.5, 5.6). Notably, the collection of thesepathways within single taxa may enhance the capacity for respiration by minimizing extracellulartransport of intermediate organic species. Recent work with newly cultured representatives ofthe class Bacteroidia indeed suggest that members of this class grow through fermentation but125shuttle a portion of their terminal electrons onto iron (oxyhydr)oxides, effectively reducing ironunder laboratory conditions [280]. The pathways found in the Bacteroidia from Kabuno Bay (Fig.5.5, 5.6, B.7) are similar to those determined through laboratory growth experiments with similarBacteroidia strains [280], which lends further support to the claim that the Bacteroidia are involvedin both fermentation and Fe reduction in Kabuno Bay. The microbial community metabolisms inKabuno Bay are, thus, dominated by carbon cycling (degradation and fermentation) likely coupledto iron reduction, which has implications for the availability of organic carbon for other forms ofrespiration.Coupled C degradation, fermentation, and respiration pathways are also found in some sulfatereducers, suggesting that the lower sulfate reduction rates, compared to Fe(III) reduction rates,is due to alternative factors. Pathway coupling was found in most abundant class of sulfatereducers (Desulfobacterota; Fig. 5.6). The high rate of Fe(III) reduction, however, relative tosulfate reduction (∼40:1), despite the ∼600 µM sulfate available in the chemocline, suggests thatother facets of the microbial network outweigh the potential increase in sulfate reduction due topathway coupling. For example, the abundance of recovered MAGs delineated as putative Fe(III)reducers outnumber MAGs with the metabolic potential for sulfate reduction (∼4:1). Differentialgene expression of the metabolic potential within these different groups of organisms could alsoplay a role in the differential rates of Fe(III) and sulfate reduction. Further studies which includethe use of transcriptomics and proteomics would be necessary to confirm this hypothesis. Thus,while some sulfate reducers in Kabuno Bay have the metabolic potential to couple multiple Cdegradation and fermentation pathways with sulfate reduction, this coupling does not result inrates that are comparable to the Fe(III) reduction rates.The tight coupling between C degradation, fermentation, and respiration pathways in single,abundant microorganisms in both the Fe and S cycles restricts the leakage of carbon to alternativerespiration pathways, such as methanogenesis. The majority of the most abundant bacteria inthe Kabuno Bay chemocline (e.g. phylum Bacteroidetes) exhibit the metabolic potential to couplecarbon degradation and fermentation to either Fe(III) or sulfate reduction (Fig. 5.6). As such, theavailability of mid-sized carbon compounds or VFAs would depend on either their leakage fromthese microorganisms or their production by specific carbon degraders, such as the Ignavibacteria126(Fig. 5.6). Indeed, methanogenic Archaea that would likely rely on VFAs for growth represent asmall portion of the population (a small portion of the 2 % Archaea in Kabuno Bay), while theabundance of the key gene for methanogenesis (mcrA) is similarly low (Fig. 5.3, B.4). Thus, themetabolic potential for methanogenesis in Kabuno Bay is low, which is likely due to a lack ofcarbon leakage from either of the Fe or S cycles, both of which appear to be tightly coupled tocarbon degradation and fermentation pathways. Further research, particularly elucidating therates of microbial methane production are needed to confirm this hypothesis.5.5.4 Implications for coupled carbon and iron cycling in the Precambrian EonsThe identification of multiple modes of primary production with electron donors, and themetabolic potential for both oxygenic (H2O) and anoxygenic (Fe(II), H2S, H2) photosynthesis (Fig.5.3, 5.4a, 5.7a) can ultimately be applied to models of primary production for the ferruginousconditions that existed in the Archean and Proterozoic oceans. Prior studies have implicatedmultiple modes of photosynthesis in large-scale processes, such as the maintenance of a clementclimate during the Archean Eon [95]. These studies, however, could be further constrained by usingthe relationships between multiple modes of photosynthesis as seen in Kabuno Bay. The spatialand temporal coexistence and/or partitioning of these multiple modes of primary productionboth in the water column and along costal margins or in upwelling zones likely impacted thedistribution and circulation of key nutrients to other members of the microbial community whennutrients such as nitrogen or phosphate were scarce or limiting under ferruginous conditions[96, 115]. For example, phosphate is often scarce or limiting under ferruginous conditions [85]due to its absorbance to Fe(III) (oxyhydr)oxides. While phosphate transporters are ubiquitousacross the MAGs (Fig. B.7), including in the putative primary producers (Fig. 5.4c, B.7), deeperinvestigations into number of phosphate acquisition genes in different types of primary producersand how these translate to growth under the natural conditions of ferruginous Kabuno Baycould further constrain future model studies. Furthermore, the antiquity of these genes and therelationship between them could inform on the antiquity of processes vital to primary productionand microbial life under ferruginous conditions.The tight coupling between complex carbon degradation and iron reduction has implications127for the cycling of iron and carbon under the Precambrian ferruginous conditions. As microor-ganisms evolved and acquired complete C degradation and fermentation pathways coupled toFe(III) reduction pathways, the efficiency of microbial iron reduction could have increased. Theuse of a more extensive and varied set of carbon sources that the bacteria themselves produced,rather than relying on a separate population of carbon degraders and fermenters, could have hada number of implications for the early Earth ocean-atmosphere system. For example, couplingthe C degradation, fermentation, and respiration pathways in individual bacteria would havelimited the concentration of mid-sized carbon compounds and VFAs. Without these intermediatecarbon compounds, the rates of alternative respiration pathways that rely on these compounds,such as methanogenesis, may have decreased, ultimately lowering the fluxes of methane to theatmosphere. Thus, the coupling of these pathways in single bacteria may have influenced thedegree of climate warming by reducing the concentration of atmospheric methane. Furthermodeling efforts are required to ascertain the putative effects of the reduction in these methanefluxes. Another potential implication is that as the rates of iron reduction increased, they wouldapproach those of phototrophic Fe(II) oxidation, thus effectively cycling iron in the water columnprior to its deposition on the seafloor. This in turn would have precluded the deposition of ferriciron rich deposits, such as banded iron formations. We hypothesize that the evolution of completecarbon degradation and fermentation pathways coupled, in one organism, to iron (oxyhydr)oxidereduction could have begun to alter the magnitude and number of iron depositions, thus reconcil-ing the lower number of these deposits throughout the Proterozoic Eon – despite the ferruginousconditions that likely still existed for much of the Eon [5, 171]. The metabolic potential andprocess rates of the microbial community found in Kabuno Bay, thus lays the groundwork for theelucidation of the relationships under the naturally complex ferruginous conditions and can beused to inform models of nutrient cycling throughout the Precambrian Eons.5.6 Supplementary materials• Table B1: Description of each Fe speciation extraction step• Table B2: List of genes discussed in the main text and their primary function128• Table B3: % of metagenomic reads that aligned to each set of MAGs• Fig. B1: The location of Kabuno Bay with the sampling site• Fig. B2: The relative abundance of the reconstructed 16S rRNA gene for the 16S rRNA geneswith a greater than 1 % relative abundance for each depth• Fig. B3: Reads per kilobase mapped (RPKM) values for taxonomic marker genes related toclass Chlorobia and glycosyl hydrolases• Fig. B4: Gene abundances of the key genes for several pathways• Fig. B5: The depth integrated absolute abundance of each phyla compared to the number ofMAGs recovered from each of those groups• Fig. B6: Comparison of the Chlorobi 01 MAG to the genome of Chlorobium phaeoferrooxidansstrain KB01 at the pathway level• Fig. B7: Comparison of the genomes of several key MAGs at the pathway level• Fig. B8: Representation of the metabolic potential for two MAGs129Chapter 6ConclusionsThis dissertation creates new knowledge on photoferrotrophy in both laboratory conditions andin the natural environment through data collected on the physiology and metabolic capacityof pelagic photoferrotroph Chlorobium phaeoferrooxidans strain KB01, as well as process ratemeasurements and analyses of the microbial community in ferruginous Kabuno Bay. Thisnew knowledge was subsequently integrated into models to examine the antiquity of nutrientacquisition in the photoferrotrophic Chlorobia, the effect of competition for phosphorus betweenphotoferrotrophs and oxygenic photosynthetic bacteria on the early Earth ocean-atmospheresystem, and the role of photoferrotrophs as primary producers during the Archean. These models,developed in this thesis, provide insight on how photoferrotrophs could have sustained thebiosphere, deposited BIFs as a by-product of their growth, fueled microbial methanogenesis, and,therefore, helped to stabilize Earth’s climate under a dim early Sun.6.1 Extant photoferrotrophyThis dissertation presents new knowledge on extant photoferrotrophy through a detailed ex-amination of Chlorobium phaeoferrooxidans strain KB01, comparing strain KB01 to other knownphototrophs, and analyses of microbial community networks supported by photoferrotrophy. Priorto this work, knowledge of extant photoferrotrophs was restricted to benthic isolates, limiting ourability to define characteristic traits of photoferrotrophs and the role of photoferrotrophs in otherenvironmental settings. In Chapter 2, we describe the physiology and metabolic potential of strainKB01 and compare traits, such as its growth response to a range of light intensities, to other knownphototrophs. This work created a framework for the remainder of the dissertation by establishingthe key aspects of pelagic photoferrotrophy. In Chapter 3, we delve deeper into the growth130of strain KB01 by examining, and biochemically verifying, its metabolic potential for nutrientacquisition—specifically its sulfur and nitrogen requirements. Furthermore, we quantify strainKB01’s usage of phosphorus and determine that strain KB01 is capable of acquiring bio-essentialphosphorus under very low phosphate conditions [96]. Further knowledge on photoferrotrophy ispresented in Chapter 4 where we examine the cell surface chemistry and characteristics of strainKB01 and other photoferrotrophs. These data allow us to elucidate the mechanism by which strainKB01 avoids association with the Fe(III) (oxyhydr)oxides it produces as a by-product of its growth,ultimately causing a physical separation between the cell surface and mineral particles. Con-versely, the surface chemistry of strain KB01’s benthic relative, strain KoFox, results in cell-mineralassociation. To further the collective knowledge on photoferrotrophs in complex ferruginousenvironments, we determine the putative role of a dominant group of photoferrotrophs (relatedto strain KB01) as primary producers in a modern ferruginous lake (Kabuno Bay) in Chapter5. Through this work we establish that there is potential for multiple modes of photosynthesisin the Kabuno Bay chemocline, despite the dominant population of photoferrotrophs, and de-scribe the flow of organic carbon the microbial community network, following its production byphotoferrotrophs. Specifically, we examine the relationship between primary production and thepartitioning of organic carbon between terminal electron processes—namely Fe(III) reduction andmethane production.6.2 Photoferrotrophy during the Archean EonThis dissertation uses integrated models to connect extant photoferrotrophy to the role thatputative photoferrotrophs may have played during the Archean Eon. Prior to this work, photofer-rotrophy during the Archean Eon remained largely unconstrained and, in particular, hypotheseswere limited by experiments conducted with benthic isolates, which restricts the application ofphotoferrotrophy to the Archean ocean. In Chapter 2, we establish that strain KB01 is likelyanalogous to Precambrian photoferrotrophs through a detailed comparison of its key traits to otherknown phototrophs. The framework established in Chapter 2 is further developed in Chapter3 where we specifically examine the antiquity of nutrient acquisition in strain KB01 and the131other members of its phylogenetic class—Chlorobia. We determine that the ability to acquireboth sulfur and nitrogen under nutrient limiting conditions likely existed in the ancestors ofmodern Chlorobia and strain KB01. To link nutrient acquisition in extant environments to theArchean oceans we created a competition-based model between photoferrotrophs and oxygenicphototrophs, utilizing phosphorus and light availability, to determine how competition amongthese phototrophs could have played a role in delaying oxygenation of the Earth [96]. In Chapter4, we revise the Archean iron budget and delineate a broader range of potential iron fluxes forthe Archean oceans using a model constrained by the rock record. We subsequently couple thenewly revised iron fluxes with the physical cell-mineral separation of pelagic photoferrotrophsto establish the role of photoferrotrophs in BIF deposition during the Archean Eon. We furtherutilize this integrated modeling approach to establish that pelagic photoferrotrophs could notonly have deposited BIFs, but also fueled microbial methanogenesis, effectively supporting a fluxof methane to the atmosphere, which would have contributed to the maintenance of a clementclimate under the faint early Sun. In Chapter 5, we establish that a tight coupling exists betweenFe(II) oxidation and Fe(III) reduction in a complex modern ferruginous environment. Such acoupling limits the export of Fe(III) (oxyhydr)oxides from the chemocline and the flow of organiccarbon to microbial methanogenesis. Thus, the evolution of this coupling could have altered theearly Earth ocean-atmosphere system, potentially explaining the lack of BIFs for the majority ofthe Proterozoic, despite persistent ferruginous conditions.6.3 Looking aheadThe data and hypotheses generated in this dissertation provide an opportunity to continue todevelop and constrain models of the early Earth ocean-atmosphere system. The specific growthkinetics of Chlorobium phaeoferrooxidans strain KB01 (Chapter 2, 3) can be further utilized toexamine the relationship between nutrient acquisition and BIF deposition. For example, a one-dimensional flux balance model of an Archean ocean could be better constrained with the kineticequation found in Chapter 2 (Eq. 2.3). Such a model could link ocean driven nutrient fluxesto primary production with the potential to predict the conditions under which BIFs would132form. Furthermore, this type of model could be used to examine some of the remaining enigmassurrounding BIF deposition, such as their banded character [9, 21, 42, 109]. Ultimately, the resultsgleaned from this small-scale model could be applied to a global modeling system, such as thecoupled C-N-P-O2-S cycles model (CANOPS [96]), to interrogate the impact of nutrient limitedprimary production on the whole ocean-atmosphere system. Finally, other changes in the rockrecord throughout the Archean and Proterozoic Eons may be examined in the context of kineticstudies of key types of primary producing phototrophs and further evaluation of the interactionsamong them. This dissertation takes steps to increase the knowledge of photoferrotrophy andits implications in the Archean oceans. The next stage of investigation will require evaluation ofmore photoferrotrophs in laboratory cultures, increased kinetic studies on these photoferrotrophs,and the application of these data in informing global models. Integrated research of this typewill also provide more constraints on the role of photoferrotrophs during the Precambrian Eon.Additionally, given that isolates do not reflect the natural complexity of a microbial community in aferruginous environment, continued study of ferruginous environments, analogous to Precambrianoceans, is needed to better constrain the interactions between photoferrotrophs and the microbialcommunities they often support.6.4 ClosingThroughout Earth’s history, life has shaped the planet’s surface chemistry through its propagationand, therefore, transformation of numerous electron donors and acceptors. These interactionsbetween life and Earth’s surface have effectively modified the composition of oceans and the rocksdeposited within them, shifted redox balances, and, most notably, underpinned biogeochemicalcycles. To elucidate both the small- and large-scale implications of the microbial driven biogeo-chemical cycles throughout Earth’s history, in-depth knowledge of individual metabolisms as wellas the complexity of microbial community interactions, under conditions relevant to Earth’s past,are required. This dissertation adds to the available knowledge on phototrophic Fe(II) oxidationand utilizes this knowledge to develop a more informed modeling framework for phototrophicprimary production during the Archean Eon. In particular, this work delineates the likely role of133photoferrotrophs in the deposition of the world’s largest iron ore formations, their influence onthe Archean biosphere, and the resulting maintenance of a clement climate prior to proliferationof oxygenic phototrophic bacteria. The models generated in this dissertation utilize an integratedapproach. For example, micro-scale interactions between a cell surface and an Fe(III) particle,major events in the geologic record, and modern ocean upwelling currents are combined toproduce data on the whole ocean-atmosphere system during the Archean Eon. These integratedapproaches can be applied to any time in Earth’s history, but more importantly, they can beapplied to examining Earth’s future. As the drivers of biogeochemical cycles have largely shiftedfrom microbial to anthropogenic, it is imperative that we continue to predict and constrain thepast to better predict the future. A greater recognition of potential future outcomes provides uswith the opportunity to put integrated approaches into action, creating new bio/geoengineeringsolutions that could maintain a life-sustaining planet.134Bibliography[1] D. E. Canfield, M. T. Rosing, and C. Bjerrum. Early anaerobic metabolisms. Philosophical Transactionsof the Royal Society B-Biological Sciences, 361(1474):1819–1834, Oct 2006.[2] Heinrich D Holland. The oxygenation of the atmosphere and oceans. Philosophical Transactions of theRoyal Society of London B: Biological Sciences, 361(1470):903–915, 2006.[3] Paul G Falkowski, Tom Fenchel, and Edward F Delong. The microbial engines that drive earth’sbiogeochemical cycles. Science, 320(5879):1034–1039, 2008.[4] Cornelis Klein. Some precambrian banded iron-formations (bifs) from around the world: Theirage, geologic setting, mineralogy, metamorphism, geochemistry, and origins. American Mineralogist,90(10):1473–1499, 2005.[5] S. W. Poulton and D. E. Canfield. Ferruginous conditions: A dominant feature of the ocean throughearth’s history. Elements, 7(2):107–112, Apr 2011.[6] P. Cloud. Paleoecological significance of banded iron formation. Economic Geology, 68(7):1135–1143,1973.[7] RM Garrels, EA Perry, and FT Mackenzie. Genesis of precambrian iron-formations and the develop-ment of atmospheric oxygen. Economic Geology, 68(7):1173–1179, 1973.[8] Kurt O Konhauser, Tristan Hamade, Rob Raiswell, et al. Could bacteria have formed the precambrianbanded iron formations? Geology, 30(12):1079–1082, 2002.[9] A. Kappler and D. K. Newman. Formation of fe(iii)-minerals by fe(ii)-oxidizing photoautotrophicbacteria. Geochimica Et Cosmochimica Acta, 68(6):1217–1226, Mar 2004.[10] Jin Xiong and Carl E Bauer. Complex evolution of photosynthesis. Annual Review of Plant Biology,53(1):503–521, 2002.[11] William F Martin, Donald A Bryant, and J Thomas Beatty. A physiological perspective on the originand evolution of photosynthesis. FEMS microbiology reviews, 2017.[12] C Magnabosco, KR Moore, JM Wolfe, and GP Fournier. Dating phototrophic microbial lineages withreticulate gene histories. Geobiology, 16(2):179–189, 2018.[13] James Farquhar, Huiming Bao, and Mark Thiemens. Atmospheric influence of earth’s earliest sulfurcycle. Science, 289(5480):756–758, 2000.[14] AA Pavlov and JF Kasting. Mass-independent fractionation of sulfur isotopes in archean sediments:strong evidence for an anoxic archean atmosphere. Astrobiology, 2(1):27–41, 2002.[15] D. E. Canfield. The early history of atmospheric oxygen: Homage to robert a. garrels. In AnnualReview of Earth and Planetary Sciences, volume 33 of Annual Review of Earth and Planetary Sciences, pages1–36. Annual Reviews, Palo Alto, 2005.135[16] Sean A Crowe, Lasse N Døssing, Nicolas J Beukes, et al. Atmospheric oxygenation three billion yearsago. Nature, 501(7468):535–538, 2013.[17] Stefan V Lalonde and Kurt O Konhauser. Benthic perspective on earth’s oldest evidence for oxygenicphotosynthesis. Proceedings of the National Academy of Sciences, 112(4):995–1000, 2015.[18] Heinrich D Holland and Karl K Turekian. Treatise on geochemistry. Elsevier Science, 2004.[19] Noah Planavsky, Andrey Bekker, Olivier J Rouxel, et al. Rare earth element and yttrium compositionsof archean and paleoproterozoic fe formations revisited: new perspectives on the significance andmechanisms of deposition. Geochimica et Cosmochimica Acta, 74(22):6387–6405, 2010.[20] F. Widdel, S. Schnell, S. Heising, et al. Ferrous iron oxidation by anoxygenic phototrophic bacteria.Nature, 362(6423):834–836, Apr 1993.[21] A. Kappler, C. Pasquero, K. O. Konhauser, and D. K. Newman. Deposition of banded iron formationsby anoxygenic phototrophic fe(ii)-oxidizing bacteria. Geology, 33(11):865–868, Nov 2005.[22] Heinrich D Holland. The chemical evolution of the atmosphere and oceans. Princeton University Press,1984.[23] HEINRICH D Holland. The geologic history of seawater. Treatise on geochemistry, 6:583–625, 2003.[24] SW Poulton and R Raiswell. The low-temperature geochemical cycle of iron: from continental fluxesto marine sediment deposition. American Journal of Science, 302(9):774–805, 2002.[25] Simon W Poulton and Donald E Canfield. Development of a sequential extraction procedure for iron:implications for iron partitioning in continentally derived particulates. Chemical geology, 2005.[26] Louis A Derry and Stein B Jacobsen. The chemical evolution of precambrian seawater: evidence fromrees in banded iron formations. Geochimica et Cosmochimica Acta, 54(11):2965–2977, 1990.[27] Sean A Crowe, Guillaume Paris, Sergei Katsev, et al. Sulfate was a trace constituent of archeanseawater. Science, 346(6210):735–739, 2014.[28] Timothy W Lyons, Christopher T Reinhard, and Noah J Planavsky. The rise of oxygen in earth/’searly ocean and atmosphere. Nature, 506(7488):307–315, 2014.[29] David C Catling, Kevin J Zahnle, and Christopher P McKay. Biogenic methane, hydrogen escape,and the irreversible oxidation of early earth. Science, 293(5531):839–843, 2001.[30] Mojtaba Fakhraee, Sean A Crowe, and Sergei Katsev. Sedimentary sulfur isotopes and neoarcheanocean oxygenation. Science advances, 4(1):e1701835, 2018.[31] Ariel D Anbar, Yun Duan, Timothy W Lyons, et al. A whiff of oxygen before the great oxidationevent? Science, 317(5846):1903–1906, 2007.[32] Eva E Stu¨eken, Roger Buick, and Ariel D Anbar. Selenium isotopes support free o2 in the latestarchean. Geology, 43(3):259–262, 2015.[33] Robert Frei, Claudio Gaucher, Simon W Poulton, and Don E Canfield. Fluctuations in precambrianatmospheric oxygenation recorded by chromium isotopes. Nature, 461(7261):250–253, 2009.[34] Devon B Cole, Christopher T Reinhard, Xiangli Wang, et al. A shale-hosted cr isotope record of lowatmospheric oxygen during the proterozoic. Geology, 44(7):555–558, 2016.[35] Woodward W Fischer, James Hemp, and Jena E Johnson. Evolution of oxygenic photosynthesis.Annual Review of Earth and Planetary Sciences, 44:647–683, 2016.136[36] Lewis M Ward, Joseph L Kirschvink, and Woodward W Fischer. Timescales of oxygenation followingthe evolution of oxygenic photosynthesis. Origins of Life and Evolution of Biospheres, 46(1):51–65, 2016.[37] Donald E Canfield, Kirsten S Habicht, and BO Thamdrup. The archean sulfur cycle and the earlyhistory of atmospheric oxygen. Science, 288(5466):658–661, 2000.[38] AG Cairns-Smith. Precambrian solution photochemistry, inverse segregation, and banded ironformations. Nature, 276:807–808, 1978.[39] Paul S Braterman, A Graham Cairns-Smith, and Robert W Sloper. Photo-oxidation of hydrated fe(ii):significance for banded iron formations. Nature, 1983.[40] Kurt O Konhauser, Larry Amskold, Stefan V Lalonde, et al. Decoupling photochemical fe (ii) oxidationfrom shallow-water bif deposition. Earth and Planetary Science Letters, 258(1):87–100, 2007.[41] Madeline Vargas, Kazem Kashefi, Elizabeth L Blunt-Harris, and Derek R Lovley. Microbiologicalevidence for fe (iii) reduction on early earth. Nature, 395(6697):65–67, 1998.[42] KO Konhauser, DK Newman, and A Kappler. The potential significance of microbial fe (iii) reductionduring deposition of precambrian banded iron formations. Geobiology, 3(3):167–177, 2005.[43] David Emerson and Craig Moyer. Isolation and characterization of novel iron-oxidizing bacteria thatgrow at circumneutral ph. Appl. Environ. Microbiol., 63(12):4784–4792, 1997.[44] Katrina J Edwards, Daniel R Rogers, Carl O Wirsen, and TM McCollom. Isolation and char-acterization of novel psychrophilic, neutrophilic, fe-oxidizing, chemolithoautotrophic alpha- andgamma-proteobacteria from the deep sea. Appl. Environ. Microbiol., 69(5):2906–2913, 2003.[45] James J Anderson and Allan H Devol. Deep water renewal in saanich inlet, an intermittently anoxicbasin. Estuarine and Coastal Marine Science, 1(1):1–10, 1973.[46] Don E Canfield, Frank J Stewart, Bo Thamdrup, et al. A cryptic sulfur cycle in oxygen-minimum–zonewaters off the chilean coast. Science, 330(6009):1375–1378, 2010.[47] Sean A Crowe, CarriAyne Jones, Sergei Katsev, et al. Photoferrotrophs thrive in an archean oceananalogue. Proceedings of the National Academy of Sciences, 105(41):15938–15943, 2008.[48] Marc Lliro´s, Tamara Garcı´a–Armisen, Franc¸ois Darchambeau, et al. Pelagic photoferrotrophy andiron cycling in a modern ferruginous basin. Scientific reports, 5, 2015.[49] Katherine Barbeau. Photochemistry of organic iron (iii) complexing ligands in oceanic systems.Photochemistry and photobiology, 82(6):1505–1516, 2006.[50] Debra J Lonergan, Harry L Jenter, John D Coates, et al. Phylogenetic analysis of dissimilatory fe(iii)-reducing bacteria. Journal of bacteriology, 178(8):2402–2408, 1996.[51] Armin Ehrenreich and Friedrich Widdel. Anaerobic oxidation of ferrous iron by purple bacteria, anew type of phototrophic metabolism. Applied and environmental microbiology, 60(12):4517–4526, 1994.[52] Silke Heising and Bernhard Schink. Phototrophic oxidation of ferrous iron by a rhodomicrobiumvannielii strain. Microbiology, 144(8):2263–2269, 1998.[53] S. Heising, L. Richter, W. Ludwig, and B. Schink. Chlorobium ferrooxidans sp nov., a phototrophicgreen sulfur bacterium that oxidizes ferrous iron in coculture with a ”geospirillum” sp strain. Archivesof Microbiology, 172(2):116–124, Aug 1999.137[54] Kristina L Straub, Frederick A Rainey, and Friedrich Widdel. Rhodovulum iodosum sp. nov. andrhodovulum robiginosum sp. nov., two new marine phototrophic ferrous-iron-oxidizing purplebacteria. International Journal of Systematic and Evolutionary Microbiology, 49(2):729–735, 1999.[55] F. Hegler, N. R. Posth, J. Jiang, and A. Kappler. Physiology of phototrophic iron(ii)-oxidizing bacteria:implications for modern and ancient environments. FEMS Microbiology Ecology, 66(2):250–260, Nov2008.[56] Katja Laufer, Annika Niemeyer, Verena Nikeleit, et al. Physiological characterization of a halotolerantanoxygenic phototrophic fe (ii)-oxidizing green-sulfur bacterium isolated from a marine sediment.FEMS microbiology ecology, 93(5), 2017.[57] Casey Bryce, Nia Blackwell, Caroline Schmidt, et al. Microbial anaerobic fe (ii) oxidation–ecology,mechanisms and environmental implications. Environmental microbiology, 20(10):3462–3483, 2018.[58] S. A. Crowe, A. S. Hahn, C. Morgan-Lang, et al. Draft genome sequence of the pelagic photoferrotrophchlorobium phaeoferrooxidans. Genome Announc, 5(13), Mar 30 2017.[59] Yongqin Jiao and Dianne K Newman. The pio operon is essential for phototrophic fe (ii) oxidation inrhodopseudomonas palustris tie-1. Journal of bacteriology, 189(5):1765–1773, 2007.[60] Laura R Croal, Clark M Johnson, Brian L Beard, and Dianne K Newman. Iron isotope fractionationby fe (ii)-oxidizing photoautotrophic bacteria. Geochimica et cosmochimica acta, 68(6):1227–1242, 2004.[61] S Scha¨dler, C Burkhardt, F Hegler, et al. Formation of cell-iron-mineral aggregates by phototrophicand nitrate-reducing anaerobic fe (ii)-oxidizing bacteria. Geomicrobiology Journal, 26(2):93–103, 2009.[62] Nicole R Posth, Sonia Huelin, Kurt O Konhauser, and Andreas Kappler. Size, density and compositionof cell–mineral aggregates formed during anoxygenic phototrophic fe (ii) oxidation: Impact onmodern and ancient environments. Geochimica et Cosmochimica Acta, 74(12):3476–3493, 2010.[63] Tina Gauger, James M Byrne, Kurt O Konhauser, et al. Influence of organics and silica on fe (ii)oxidation rates and cell–mineral aggregate formation by the green-sulfur fe (ii)-oxidizing bacteriumchlorobium ferrooxidans kofox–implications for fe (ii) oxidation in ancient oceans. Earth and PlanetaryScience Letters, 443:81–89, 2016.[64] Lidia Romero-Viana, Brendan J Keely, Antonio Camacho, Eduardo Vicente, and M Rosa Miracle.Primary production in lake la cruz (spain) over the last four centuries: reconstruction based onsedimentary signal of photosynthetic pigments. Journal of Paleolimnology, 43(4):771–786, 2010.[65] Corinne Biderre-Petit, Delphine Boucher, Jan Kuever, et al. Identification of sulfur-cycle prokaryotes ina low-sulfate lake (lake pavin) using apra and 16s rrna gene markers. Microbial ecology, 61(2):313–327,2011.[66] SA Crowe, JA Maresca, C Jones, et al. Deep-water anoxygenic photosythesis in a ferruginouschemocline. Geobiology, 12(4):322–339, 2014.[67] Xavier A Walter, Antonio Picazo, Maria R Miracle, et al. Phototrophic fe (ii)-oxidation in thechemocline of a ferruginous meromictic lake. Frontiers in microbiology, 5, 2014.[68] SL Schiff, JM Tsuji, Lingling Wu, et al. Millions of boreal shield lakes can be used to probe archaeanocean biogeochemistry. Scientific reports, 7:46708, 2017.[69] SA Crowe, S Katsev, K Leslie, et al. The methane cycle in ferruginous lake matano. Geobiology,9(1):61–78, 2011.138[70] A. Zegeye, S. Bonneville, L. G. Benning, et al. Green rust formation controls nutrient availability in aferruginous water column. Geology, 40(7):599–602, Jul 2012.[71] LB Kuntz, TA Laakso, DP Schrag, and SA Crowe. Modeling the carbon cycle in lake matano.Geobiology, 13(5):454–461, 2015.[72] Ce´dric Morana, Fleur AE Roland, Sean A Crowe, et al. Chemoautotrophy and anoxygenic photosyn-thesis within the water column of a large meromictic tropical lake (lake kivu, east africa). Limnologyand Oceanography, 61(4):1424–1437, 2016.[73] Ce´line C Michiels, Franc¸ois Darchambeau, Fleur AE Roland, et al. Iron-dependent nitrogen cycling ina ferruginous lake and the nutrient status of proterozoic oceans. Nature GeoScience, 10(3):217–221,2017.[74] Nicolas J Beukes and Cornelis Klein. Geochemistry and sedimentology of a facies transition—frommicrobanded to granular iron-formation—in the early proterozoic transvaal supergroup, south africa.Precambrian Research, 47(1):99–139, 1990.[75] Alan J Kaufman, JM Hayes, and C Klein. Primary and diagenetic controls of isotopic compositions ofiron-formation carbonates. Geochimica et Cosmochimica Acta, 54(12):3461–3473, 1990.[76] A. F. Trendall. The hamersley basin. Developments in Precambrain Geology, 6:69–129, 1983.[77] DE Ayres. Genesis of iron-bearing minerals in banded iron formation mesobands in the dales gorgemember, hamersley group, western australia. Economic Geology, 67(8):1214–1233, 1972.[78] ME Baur, JM Hayes, SA Studley, and MR Walter. Millimeter-scale variations of stable isotopeabundances in carbonates from banded iron-formations in the hamersley group of western australia.Economic Geology, 80(2):270–282, 1985.[79] Birger Rasmussen, Bryan Krapezˇ, and Daniela B Meier. Replacement origin for hematite in 2.5 gabanded iron formation: Evidence for postdepositional oxidation of iron-bearing minerals. Bulletin,126(3-4):438–446, 2014.[80] Birger Rasmussen, Janet R Muhling, Alexandra Suvorova, and Bryan Krapezˇ. Dust to dust: Evidencefor the formation of “primary” hematite dust in banded iron formations via oxidation of iron silicatenanoparticles. Precambrian Research, 284:49–63, 2016.[81] Nicholas J Tosca, Stephen Guggenheim, and Peir K Pufahl. An authigenic origin for precambriangreenalite: Implications for iron formation and the chemistry of ancient seawater. Bulletin, 128(3-4):511–530, 2016.[82] P Kharecha, J Kasting, and J Siefert. A coupled atmosphere–ecosystem model of the early archeanearth. Geobiology, 3(2):53–76, 2005.[83] Ann E Isley and Dallas H Abbott. Plume-related mafic volcanism and the deposition of banded ironformation. Journal of Geophysical Research: Solid Earth, 104(B7):15461–15477, 1999.[84] Heinrich D Holland, B Lazar, and Mark McCaffrey. Evolution of the atmosphere and oceans. Nature,320(6057):27–33, 1986.[85] C Jones, Suolong Nomosatryo, Sean Andrew Crowe, Christian J Bjerrum, and Donald Eugene Canfield.Iron oxides, divalent cations, silica, and the early earth phosphorus crisis. Geology, 43(2):135–138,2015.[86] Birger Rasmussen, Daniela B Meier, Bryan Krapezˇ, and Janet R Muhling. Iron silicate microgranulesas precursor sediments to 2.5-billion-year-old banded iron formations. Geology, 41(4):435–438, 2013.139[87] Birger Rasmussen, Bryan Krapezˇ, Janet R Muhling, and Alexandra Suvorova. Precipitation of ironsilicate nanoparticles in early precambrian oceans marks earth’s first iron age. Geology, 43(4):303–306,2015.[88] Leslie J Robbins, Sean P Funk, Shannon L Flynn, et al. Hydrogeological constraints on the formationof palaeoproterozoic banded iron formations. Nature GeoScience, 12(7):558–563, 2019.[89] Cornelis Klein and Nicolas J Beukes. Geochemistry and sedimentology of a facies transition fromlimestone to iron-formation deposition in the early proterozoic transvaal supergroup, south africa.Economic Geology, 84(7):1733–1774, 1989.[90] David Antoine, Jean-Michel Andre´, and Andre´ Morel. Oceanic primary production: 2. estimationat global scale from satellite (coastal zone color scanner) chlorophyll. Global biogeochemical cycles,10(1):57–69, 1996.[91] Paul G Falkowski and John A Raven. Aquatic photosynthesis. Princeton University Press, 2013.[92] Robert E Blankenship. Origin and early evolution of photosynthesis. Photosynthesis research, 33(2):91–111, 1992.[93] Robert E Blankenship, Michael T Madigan, and Carl E Bauer. Anoxygenic photosynthetic bacteria,volume 2. Springer Science & Business Media, 2006.[94] Noah J Planavsky, Dan Asael, Axel Hofmann, et al. Evidence for oxygenic photosynthesis half abillion years before the great oxidation event. Nature GeoScience, 7(4):283–286, 2014.[95] Kazumi Ozaki, Eiichi Tajika, Peng K Hong, Yusuke Nakagawa, and Christopher T Reinhard. Effectsof primitive photosynthesis on earth’s early climate system. Nature GeoScience, page 1, 2017.[96] Kazumi Ozaki, Katharine J Thompson, Rachel L Simister, Sean A Crowe, and Christopher T Rein-hard. Anoxygenic photosynthesis and the delayed oxygenation of earth’s atmosphere. Naturecommunications, 10(1):1–10, 2019.[97] Patricia Sa´nchez-Baracaldo and Tanai Cardona. On the origin of oxygenic photosynthesis andcyanobacteria. New Phytologist, 2019.[98] Wolf-Dieter Schubert, Olaf Klukas, Wolfram Saenger, et al. A common ancestor for oxygenic andanoxygenic photosynthetic systems: a comparison based on the structural model of photosystem i.Journal of molecular biology, 280(2):297–314, 1998.[99] Katharine J Thompson, Paul A Kenward, Kohen W Bauer, et al. Photoferrotrophy, deposition ofbanded iron formations, and methane production in archean oceans. Science Advances, 5(11):eaav2869,2019.[100] David J Des Marais. When did photosynthesis emerge on earth? Science, 289(5485):1703–1705, 2000.[101] J. Xiong, W. M. Fischer, K. Inoue, M. Nakahara, and C. E. Bauer. Molecular evidence for the earlyevolution of photosynthesis. Science, 289(5485):1724–1730, Sep 2000.[102] John M Olson and Robert E Blankenship. Thinking about the evolution of photosynthesis. InDiscoveries in Photosynthesis, pages 1073–1086. Springer, 2005.[103] Antonio Camacho, Xavier A Walter, Antonio Picazo, and Jakob Zopfi. Photoferrotrophy: remains ofan ancient photosynthesis in modern environments. Frontiers in microbiology, 8:323, 2017.[104] Genming Luo, Shuhei Ono, Nicolas J Beukes, et al. Rapid oxygenation of earth’s atmosphere 2.33billion years ago. Science advances, 2(5):e1600134, 2016.140[105] Pascal Philippot, Janaı´na N A´vila, Bryan A Killingsworth, et al. Globally asynchronous sulphurisotope signals require re-definition of the great oxidation event. Nature communications, 9(1):2245,2018.[106] Stephanie L Olson, Lee R Kump, and James F Kasting. Quantifying the areal extent and dissolvedoxygen concentrations of archean oxygen oases. Chemical Geology, 362:35–43, 2013.[107] Nicole R Posth, Florian Hegler, Kurt O Konhauser, and Andreas Kappler. Alternating si and fedeposition caused by temperature fluctuations in precambrian oceans. Nature GeoScience, 1(10):703,2008.[108] LR Croal, Y Jiao, A Kappler, and DK Newman. Phototrophic fe (ii) oxidation in an atmosphere of h2:implications for archean banded iron formations. Geobiology, 7(1):21–24, 2009.[109] Nicole R Posth, Kurt O Konhauser, and Andreas Kappler. Microbiological processes in banded ironformation deposition. Sedimentology, 60(7):1733–1754, 2013.[110] Pawan K Dadheech, Gernot Glo¨ckner, Peter Casper, et al. Cyanobacterial diversity in the hot spring,pelagic and benthic habitats of a tropical soda lake. FEMS microbiology ecology, 85(2):389–401, 2013.[111] Gideon Mazor, Giora J Kidron, Ahuva Vonshak, and Aharon Abeliovich. The role of cyanobacterialexopolysaccharides in structuring desert microbial crusts. FEMS Microbiology Ecology, 21(2):121–130,1996.[112] Natascha Staats, Ben De Winder, Lucas Stal, and Luuc Mur. Isolation and characterization ofextracellular polysaccharides from the epipelic diatoms cylindrotheca closterium and naviculasalinarum. European Journal of Phycology, 34(2):161–169, 1999.[113] Juergen Schieber. Microbial mats on muddy substrates-examples of possible sedimentary featuresand underlying processes. Atlas of microbial mat features preserved within the siliciclastic rock record,2:117e133, 2007.[114] Martin Schmid, Michel Halbwachs, Bernhard Wehrli, and Alfred Wu¨est. Weak mixing in lakekivu: new insights indicate increasing risk of uncontrolled gas eruption. Geochemistry, Geophysics,Geosystems, 6(7), 2005.[115] Katharine J Thompson, Rachel L Simister, Aria S Hahn, Steven J Hallam, and Sean A Crowe. Nutrientacquisition and the metabolic potential of photoferrotrophic chlorobi. Frontiers in Microbiology, 8:1212,2017.[116] Friedrich Widdel and Friedhelm Bak. Gram-negative mesophilic sulfate-reducing bacteria. In Theprokaryotes, pages 3352–3378. Springer, 1992.[117] Elmar Pruesse, Christian Quast, Katrin Knittel, et al. Silva: a comprehensive online resource forquality checked and aligned ribosomal rna sequence data compatible with arb. Nucleic acids research,35(21):7188–7196, 2007.[118] Christian Quast, Elmar Pruesse, Pelin Yilmaz, et al. The silva ribosomal rna gene database project:improved data processing and web-based tools. Nucleic acids research, page gks1219, 2012.[119] Mark A Larkin, Gordon Blackshields, NP Brown, et al. Clustal w and clustal x version 2.0. Bioinfor-matics, 23(21):2947–2948, 2007.[120] Koichiro Tamura, Glen Stecher, Daniel Peterson, Alan Filipski, and Sudhir Kumar. Mega6: molecularevolutionary genetics analysis version 6.0. Molecular biology and evolution, 30(12):2725–2729, 2013.141[121] Sudhir Kumar, Glen Stecher, and Koichiro Tamura. Mega7: Molecular evolutionary genetics analysisversion 7.0 for bigger datasets. Molecular biology and evolution, page msw054, 2016.[122] Jared T Simpson, Kim Wong, Shaun D Jackman, et al. Abyss: a parallel assembler for short readsequence data. Genome research, 19(6):1117–1123, 2009.[123] Stefan Kurtz, Adam Phillippy, Arthur L Delcher, et al. Versatile and open software for comparinglarge genomes. Genome biology, 5(2):R12, 2004.[124] Jan P Meier-Kolthoff, Alexander F Auch, Hans-Peter Klenk, and Markus Go¨ker. Genome sequence-based species delimitation with confidence intervals and improved distance functions. BMC bioinfor-matics, 14(1):60, 2013.[125] Doug Hyatt, Gwo-Liang Chen, Philip F LoCascio, et al. Prodigal: prokaryotic gene recognition andtranslation initiation site identification. BMC bioinformatics, 11(1):119, 2010.[126] Takuya Aramaki, Romain Blanc-Mathieu, Hisashi Endo, et al. Kofamkoala: Kegg ortholog assignmentbased on profile hmm and adaptive score threshold. bioRxiv, page 602110, 2019.[127] Minoru Kanehisa, Michihiro Araki, Susumu Goto, et al. Kegg for linking genomes to life and theenvironment. Nucleic acids research, 36(suppl 1):D480–D484, 2008.[128] ED Graham, JF Heidelberg, and BJ Tully. Potential for primary productivity in a globally-distributedbacterial phototroph. The ISME journal, 12(7):1861, 2018.[129] Sara El-Gebali, Jaina Mistry, Alex Bateman, et al. The pfam protein families database in 2019. Nucleicacids research, 47(D1):D427–D432, 2018.[130] Vincent Lombard, Hemalatha Golaconda Ramulu, Elodie Drula, Pedro M Coutinho, and Bernard Hen-rissat. The carbohydrate-active enzymes database (cazy) in 2013. Nucleic acids research, 42(D1):D490–D495, 2013.[131] Tatiana A Tatusova and Thomas L Madden. Blast 2 sequences, a new tool for comparing protein andnucleotide sequences. FEMS microbiology letters, 174(2):247–250, 1999.[132] Sean R Eddy. A new generation of homology search tools based on probabilistic inference. In GenomeInformatics 2009: Genome Informatics Series Vol. 23, pages 205–211. World Scientific, 2009.[133] E Viollier, PW Inglett, K Hunter, AN Roychoudhury, and P Van Cappellen. The ferrozine methodrevisited: Fe (ii)/fe (iii) determination in natural waters. Applied geochemistry, 15(6):785–790, 2000.[134] Niels-Ulrik Frigaard, Kim L Larsen, and Raymond P Cox. Spectrochromatography of photosyntheticpigments as a fingerprinting technique for microbial phototrophs. FEMS microbiology ecology, 20(2):69–77, 1996.[135] DK Button. Nutrient uptake by microorganisms according to kinetic parameters from theory asrelated to cytoarchitecture. Microbiol Mol Biol R, 1998.[136] CL Woldringh, JS Binnerts, and A Mans. Variation in escherichia coli buoyant density measured inpercoll gradients. Journal of bacteriology, 148(1):58–63, 1981.[137] Carles M Borrego, Juan B Arellano, Carles A Abella, Tomas Gillbro, and Jesu´s Garcia-Gil. Themolar extinction coefficient of bacteriochlorophyll e and the pigment stoichiometry in chlorobiumphaeobacteroides. Photosynthesis Research, 60(2-3):257–264, 1999.[138] A. K. Manske, J. Glaeser, M. A. M. Kuypers, and J. Overmann. Physiology and phylogeny of greensulfur bacteria forming a monospecific phototrophic assemblage at a depth of 100 meters in the blacksea. Applied and Environmental Microbiology, 71(12):8049–8060, Dec 2005.142[139] Misha Kudryashev, Aikaterini Aktoudianaki, Dimitrios Dedoglou, Henning Stahlberg, and Geor-gios Tsiotis. The ultrastructure of chlorobaculum tepidum revealed by cryo-electron tomography.Biochimica et Biophysica Acta (BBA)-Bioenergetics, 1837(10):1635–1642, 2014.[140] CM Borrego, LJ Garcia-Gil, X Vila, et al. Distribution of bacteriochlorophyll homologs in naturalpopulations of brown-colored phototrophic sulfur bacteria. FEMS Microbiology Ecology, 24(4):301–309,Dec 1997.[141] Jo¨rg Overmann. The family chlorobiaceae. The Prokaryotes: Volume 7: Proteobacteria: Delta, EpsilonSubclass, pages 359–378, 2006.[142] Donald A Bryant, Zhenfeng Liu, Tao Li, et al. Comparative and functional genomics of anoxygenicgreen bacteria from the taxa chlorobi, chloroflexi, and acidobacteria. In Functional genomics andevolution of photosynthetic systems, pages 47–102. Springer, 2012.[143] Inger Andersson and Anders Backlund. Structure and function of rubisco. Plant Physiology andBiochemistry, 46(3):275–291, 2008.[144] Thomas E Hanson and F Robert Tabita. A ribulose-1, 5-bisphosphate carboxylase/oxygenase (rubisco)-like protein from chlorobium tepidum that is involved with sulfur metabolism and the response tooxidative stress. Proceedings of the National Academy of Sciences, 98(8):4397–4402, 2001.[145] Roman A Barco, David Emerson, Jason B Sylvan, et al. The proteomic profile of an obligateiron-oxidizing chemolithoautotroph reveals new insight into microbial iron oxidation. Applied andEnvironmental Microbiology, pages AEM. 01374–15, 2015.[146] William B Whitman, David C Coleman, and William J Wiebe. Prokaryotes: the unseen majority.Proceedings of the National Academy of Sciences, 95(12):6578–6583, 1998.[147] Christopher T Reinhard, Noah J Planavsky, Benjamin C Gill, et al. Evolution of the global phosphoruscycle. Nature, 2016.[148] Niels-Ulrik Frigaard, Aline Gomez Maqueo Chew, Hui Li, Julia A Maresca, and Donald A Bryant.Chlorobium tepidum: insights into the structure, physiology, and metabolism of a green sulfurbacterium derived from the complete genome sequence. Photosynthesis Research, 78(2):93–117, 2003.[149] Evelyn Marschall, Mareike Jogler, Uta Henßge, and Jo¨rg Overmann. Large-scale distribution andactivity patterns of an extremely low-light-adapted population of green sulfur bacteria in the blacksea. Environmental microbiology, 12(5):1348–1362, 2010.[150] Douglas Cork, Jeremy Mathers, Andrea Maka, and Anna Srnak. Control of oxidative sulfurmetabolism of chlorobium limicola forma thiosulfatophilum. Appl. Environ. Microbiol., 49(2):269–272,1985.[151] Matthew T Cottrell, Josephine Ras, and David L Kirchman. Bacteriochlorophyll and communitystructure of aerobic anoxygenic phototrophic bacteria in a particle-rich estuary. The ISME journal,4(7):945, 2010.[152] Mikhail V Zubkov, Bernhard M Fuchs, Heike Eilers, Peter H Burkill, and Rudolf Amann. Determina-tion of total protein content of bacterial cells by sypro staining and flow cytometry. Appl. Environ.Microbiol., 65(7):3251–3257, 1999.[153] Hans Van Gemerden. The sulfide affinity of phototrophic bacteria in relation to the location ofelemental sulfur. Archives of microbiology, 139(4):289–294, 1984.[154] Sean A Crowe, Raymond P Cox, CarriAyne Jones, et al. Decrypting the sulfur cycle in oceanic oxygenminimum zones. The ISME journal, page 1, 2018.143[155] J Jeffrey Morris. Black queen evolution: the role of leakiness in structuring microbial communities.Trends in Genetics, 31(8):475–482, 2015.[156] Elena Kazamia, Hjo¨rdis Czesnick, Thi Thanh Van Nguyen, et al. Mutualistic interactions betweenvitamin b12-dependent algae and heterotrophic bacteria exhibit regulation. Environmental microbiology,14(6):1466–1476, 2012.[157] Mak A Saito, Daniel M Sigman, and Franc¸ois MM Morel. The bioinorganic chemistry of the ancientocean: the co-evolution of cyanobacterial metal requirements and biogeochemical cycles at thearchean–proterozoic boundary? Inorganica Chimica Acta, 356:308–318, 2003.[158] I Halevy and A Bachan. The geologic history of seawater ph. Science, 355(6329):1069–1071, 2017.[159] ME Barley, AL Pickard, and PJ Sylvester. Emplacement of a large igneous province as a possiblecause of banded iron formation 2.45 billion years ago. Nature, 385(6611):55, 1997.[160] AF Trendall. The significance of banded iron formation (bif) in the precambrian stratigraphic record.Geoscientist, 10:4–7, 2000.[161] Ivan Valiela and Ivan Valiela. Marine ecological processes, volume 686. Springer, 1995.[162] Paul G Falkowski. Evolution of the nitrogen cycle and its influence on the biological sequestration ofco 2 in the ocean. Nature, 387(6630):272–275, 1997.[163] Christopher B Field, Michael J Behrenfeld, James T Randerson, and Paul Falkowski. Primaryproduction of the biosphere: integrating terrestrial and oceanic components. Science, 281(5374):237–240, 1998.[164] Robert W Howarth, Roxanne Marino, Judith Lane, and Jonathan J Cole. Nitrogen fixation infreshwater, estuarine, and marine ecosystems. 1. rates and importance. Limnology and Oceanography,33(4part2):669–687, 1988.[165] D Karl, R Letelier, L Tupas, et al. The role of nitrogen fixation in biogeochemical cycling in thesubtropical north pacific ocean. Nature, 388(6642):533–538, 1997.[166] Douglas G Capone and Edward J Carpenter. Nitrogen fixation in the marine environment. Science,217(4565):1140–1142, 1982.[167] Douglas G Capone, James A Burns, Joseph P Montoya, et al. Nitrogen fixation by trichodesmiumspp.: An important source of new nitrogen to the tropical and subtropical north atlantic ocean. GlobalBiogeochemical Cycles, 19(2), 2005.[168] Edward J Carpenter and Kristen Romans. Major role of the cyanobacterium trichodesmium innutrient cycling in the north atlantic ocean. Science, 254(5036):1356, 1991.[169] Andrew D Czaja, Clark M Johnson, Brian L Beard, et al. Biological fe oxidation controlled depositionof banded iron formation in the ca. 3770ma isua supracrustal belt (west greenland). Earth and PlanetaryScience Letters, 363:192–203, 2013.[170] Donald E Canfield, Simon W Poulton, Andrew H Knoll, et al. Ferruginous conditions dominatedlater neoproterozoic deep-water chemistry. Science, 321(5891):949–952, 2008.[171] Noah J Planavsky, Peter McGoldrick, Clinton T Scott, et al. Widespread iron-rich conditions in themid-proterozoic ocean. Nature, 477(7365):448–451, 2011.[172] James CG Walker. Was the archaean biosphere upside down? Nature, 329(6141):710–712, 1987.144[173] Hyman Hartman. The evolution of photosynthesis and microbial mats: A speculation on the bandediron formations. MBL LECT. BIOL. 1984., 1984.[174] CM Moore, MM Mills, KR Arrigo, et al. Processes and patterns of oceanic nutrient limitation. NatureGeoScience, 6(9):701–710, 2013.[175] Arpita Bose and Dianne K Newman. Regulation of the phototrophic iron oxidation (pio) genes inrhodopseudomonas palustris tie-1 is mediated by the global regulator, fixk. Molecular microbiology,79(1):63–75, 2011.[176] L Pereira, IH Saraiva, R Coelho, et al. Crystallization and preliminary crystallographic studies offoxe from rhodobacter ferrooxidans sw2, an feii oxidoreductase involved in photoferrotrophy. ActaCrystallographica Section F: Structural Biology and Crystallization Communications, 68(9):1106–1108, 2012.[177] George M Garrity, John G Holt, Jo¨rg Overmann, et al. Phylum bxi. chlorobi phy. nov. In Bergey’sManual R© of Systematic Bacteriology, pages 601–623. Springer, 2001.[178] Sumedha Sadekar, Jason Raymond, and Robert E Blankenship. Conservation of distantly relatedmembrane proteins: photosynthetic reaction centers share a common structural core. Molecular biologyand evolution, 23(11):2001–2007, 2006.[179] Soichirou Satoh, Mamoru Mimuro, and Ayumi Tanaka. Construction of a phylogenetic tree ofphotosynthetic prokaryotes based on average similarities of whole genome sequences. PloS one,8(7):e70290, 2013.[180] C. M. Borrego and L. J. Garciagil. Rearrangement of light-harvesting bacteriochlorophyll homologsas a response of green sulfur bacteria to low-light intensities. Photosynthesis Research, 45(1):21–30, Jul1995.[181] Lea H Gregersen, Kirsten S Habicht, Sandro Peduzzi, et al. Dominance of a clonal green sulfurbacterial population in a stratified lake. FEMS microbiology ecology, 70(1):30–41, 2009.[182] Toby Tyrrell. The relative influences of nitrogen and phosphorus on oceanic primary production.Nature, 400(6744):525–531, 1999.[183] Donald E Canfield, Alexander N Glazer, and Paul G Falkowski. The evolution and future of earth’snitrogen cycle. Science, 330(6001):192–196, 2010.[184] Ben Johnson and Colin Goldblatt. The nitrogen budget of earth. Earth-Science Reviews, 148:150–173,2015.[185] C Mark Moore, Matthew M Mills, Eric P Achterberg, et al. Large-scale distribution of atlantic nitrogenfixation controlled by iron availability. Nature GeoScience, 2(12):867–871, 2009.[186] Jacques Meyer, Bruce C Kelley, and Paulette M Vignais. Nitrogen fixation and hydrogen metabolismin photosynthetic bacteria. Biochimie, 60(3):245–260, 1978.[187] Jason Raymond, Janet L Siefert, Christopher R Staples, and Robert E Blankenship. The natural historyof nitrogen fixation. Molecular biology and evolution, 21(3):541–554, 2004.[188] Eric S Boyd, Amaya M Garcia Costas, Trinity L Hamilton, Florence Mus, and John W Peters. Evolutionof molybdenum nitrogenase during the transition from anaerobic to aerobic metabolism. Journal ofbacteriology, 197(9):1690–1699, 2015.[189] Madeline C Weiss, Filipa L Sousa, Natalia Mrnjavac, et al. The physiology and habitat of the lastuniversal common ancestor. Nature Microbiology, 1:16116, 2016.145[190] JJR Frausto Da Silva and Robert Joseph Paton Williams. The biological chemistry of the elements: theinorganic chemistry of life. Oxford University Press, 2001.[191] James CG Walker and Peter Brimblecombe. Iron and sulfur in the pre-biologic ocean. PrecambrianResearch, 28(3-4):205–222, 1985.[192] G Paris, JF Adkins, Alex L Sessions, SM Webb, and WW Fischer. Neoarchean carbonate–associatedsulfate records positive delta33s anomalies. Science, 346(6210):739–741, 2014.[193] Iadviga Zhelezinskaia, Alan J Kaufman, James Farquhar, and John Cliff. Large sulfur isotopefractionations associated with neoarchean microbial sulfate reduction. Science, 346(6210):742–744,2014.[194] Niels-Ulrik Frigaard and Donald A Bryant. Genomic insights into the sulfur metabolism of pho-totrophic green sulfur bacteria. In Sulfur metabolism in phototrophic organisms, pages 337–355. Springer,2008.[195] Russell L Cuhel, Craig D Taylor, and Holger W Jannasch. Assimilatory sulfur metabolism in marinemicroorganisms: sulfur metabolism, protein synthesis, and growth of pseudomonas halodurans andalteromonas luteo-violaceus during unperturbed batch growth. Archives of Microbiology, 130(1):8–13,1981.[196] Kishori M Konwar, Niels W Hanson, Maya P Bhatia, et al. Metapathways v2. 5: quantitative functional,taxonomic and usability improvements. Bioinformatics, 31(20):3345–3347, 2015.[197] Kishori M Konwar, Niels W Hanson, Antoine P Page´, and Steven J Hallam. Metapathways: a modularpipeline for constructing pathway/genome databases from environmental sequence information.BMC bioinformatics, 14(1):202, 2013.[198] Ron Caspi, Tomer Altman, Kate Dreher, et al. The metacyc database of metabolic pathways andenzymes and the biocyc collection of pathway/genome databases. Nucleic acids research, 40(D1):D742–D753, 2012.[199] Michael Kaufmann. The role of the cog database in comparative and functional genomics. CurrentBioinformatics, 1(3):291–300, 2006.[200] BL Cantarel, PM Coutinho, C Rancurel, et al. The carbohydrate-active enzymes database (cazy): anexpert resource for glycogenomics. Nucleic acids research, 37:233–238, 2009.[201] R Agarwala, T Barrett, J Beck, et al. Database resources of the national center for biotechnologyinformation. Nucleic Acids Research, 43:6–17, 2015.[202] Luis M Rubio and Paul W Ludden. Biosynthesis of the iron-molybdenum cofactor of nitrogenase.Annu. Rev. Microbiol., 62:93–111, 2008.[203] Rolf D Joerger, Paul E Bishop, and Harold J Evans. Bacterial alternative nitrogen fixation systems.CRC Critical reviews in microbiology, 16(1):1–14, 1988.[204] Eric S Boyd, Trinity L Hamilton, and John W Peters. An alternative path for the evolution of biologicalnitrogen fixation. Frontiers in microbiology, 2:205, 2011.[205] Basem Soboh, Eric S Boyd, Dehua Zhao, John W Peters, and Luis M Rubio. Substrate specificityand evolutionary implications of a nifdk enzyme carrying nifb-co at its active site. FEBS letters,584(8):1487–1492, 2010.[206] Zhenfeng Liu, Niels-Ulrik Frigaard, Kajetan Vogl, et al. Complete genome of ignavibacterium album,a metabolically versatile, flagellated, facultative anaerobe from the phylum chlorobi. Frontiers inmicrobiology, 3:185, 2012.146[207] Jennifer Hiras, Yu-Wei Wu, Stephanie A Eichorst, Blake A Simmons, and Steven W Singer. Refiningthe phylum chlorobi by resolving the phylogeny and metabolic potential of the representative of adeeply branching, uncultivated lineage. The ISME journal, 10(4):833–845, 2016.[208] Lawrence A David and Eric J Alm. Rapid evolutionary innovation during an archaean geneticexpansion. Nature, 469(7328):93–96, 2011.[209] Eric Boyd and John W Peters. New insights into the evolutionary history of biological nitrogenfixation. Frontiers in microbiology, 4:201, 2013.[210] Jo¨rg Overmann. Mahoney lake: a case study of the ecological significance of phototrophic sulfurbacteria. In Advances in microbial ecology, pages 251–288. Springer, 1997.[211] Mauro Tonolla, Sandro Peduzzi, Antonella Demarta, Raffaele Peduzzi, and Hahb Dittmar. Phototropicsulfur and sulfate-reducing bacteria in the chemocline of meromictic lake cadagno, switzerland.Journal of Limnology, 63(2):161–170, 2004.[212] Ryuji Kondo, Akie Nakagawa, Lisa Mochizuki, et al. Dominant bacterioplankton populations in themeromictic lake suigetsu as determined by denaturing gradient gel electrophoresis of 16s rrna genefragments. Limnology, 10(1):63–69, 2009.[213] KM Meyer, JL Macalady, JM Fulton, et al. Carotenoid biomarkers as an imperfect reflection of theanoxygenic phototrophic community in meromictic fayetteville green lake. Geobiology, 9(4):321–329,2011.[214] Kjell Magne Fagerbakke, Mikal Heldal, and Svein Norland. Content of carbon, nitrogen, oxygen,sulfur and phosphorus in native aquatic and cultured bacteria. Aquatic Microbial Ecology, 10(1):15–27,1996.[215] Chen-Tung Arthur Chen, Chi-Ming Lin, Being-Ta Huang, and Lei-Fong Chang. Stoichiometry ofcarbon, hydrogen, nitrogen, sulfur and oxygen in the particulate matter of the western north pacificmarginal seas. Marine Chemistry, 54(1-2):179–190, 1996.[216] Yanan Shen, Roger Buick, and Donald E Canfield. Isotopic evidence for microbial sulphate reductionin the early archaean era. Nature, 410(6824):77–81, 2001.[217] Robert P Lowell and Susan M Keller. High-temperature seafloor hydrothermal circulation overgeologic time and archean banded iron formations. Geophysical Research Letters, 30(7), 2003.[218] Donald E Canfield. A new model for proterozoic ocean chemistry. Nature, 396(6710):450, 1998.[219] Jeremy B Fein, Christopher J Daughney, Nathan Yee, and Thomas A Davis. A chemical equilibriummodel for metal adsorption onto bacterial surfaces. Geochimica et Cosmochimica Acta, 61(16):3319–3328,1997.[220] Dimitri A Sverjensky and Nita Sahai. Theoretical prediction of single-site surface-protonationequilibrium constants for oxides and silicates in water. Geochimica et Cosmochimica Acta, 60(20):3773–3797, 1996.[221] Kurt O Konhauser. Diversity of bacterial iron mineralization. Earth-Science Reviews, 43(3):91–121,1998.[222] Kurt O Konhauser, Stefan V Lalonde, Larry Amskold, and Heinrich D Holland. Was there really anarchean phosphate crisis? Science, 315(5816):1234–1234, 2007.[223] Vernon R Phoenix, Dave G Adams, and Kurt O Konhauser. Cyanobacterial viability during hy-drothermal biomineralisation. Chemical Geology, 169(3):329–338, 2000.147[224] Elizabeth D Swanner, Aleksandra M Mloszewska, Olaf A Cirpka, et al. Modulation of oxygenproduction in archaean oceans by episodes of fe (ii) toxicity. Nature GeoScience, 8(2):126–130, 2015.[225] G Saini and CS Chan. Near-neutral surface charge and hydrophilicity prevent mineral encrustationof fe-oxidizing micro-organisms. Geobiology, 11(2):191–200, 2013.[226] Oleg Stukalov, Anton Korenevsky, Terry J Beveridge, and John R Dutcher. Use of atomic forcemicroscopy and transmission electron microscopy for correlative studies of bacterial capsules. Appliedand environmental microbiology, 74(17):5457–5465, 2008.[227] Carel J Van Oss. Interfacial forces in aqueous media. CRC press, 2006.[228] Nathan Yee and Jeremy Fein. Cd adsorption onto bacterial surfaces: a universal adsorption edge?Geochimica et Cosmochimica Acta, 65(13):2037–2042, 2001.[229] Ann E Isley. Hydrothermal plumes and the delivery of iron to banded iron formation. The Journal ofGeology, 103(2):169–185, 1995.[230] Lee R Kump and William E Seyfried. Hydrothermal fe fluxes during the precambrian: effect of lowoceanic sulfate concentrations and low hydrostatic pressure on the composition of black smokers.Earth and Planetary Science Letters, 235(3):654–662, 2005.[231] Jun Korenaga. Archean geodynamics and the thermal evolution of earth. Archean geodynamics andenvironments, pages 7–32, 2006.[232] Norman H Sleep, K Zahnle, and PS Neuhoff. Initiation of clement surface conditions on the earliestearth. Proceedings of the National Academy of Sciences, 98(7):3666–3672, 2001.[233] John H Martin, George A Knauer, David M Karl, and William W Broenkow. Vertex: carbon cycling inthe northeast pacific. Deep Sea Research Part A. Oceanographic Research Papers, 34(2):267–285, 1987.[234] Allan H Devol and Hilairy E Hartnett. Role of the oxygen-deficient zone in transfer of organic carbonto the deep ocean. Limnology and Oceanography, 46(7):1684–1690, 2001.[235] Jillian F Banfield, Susan A Welch, Hengzhong Zhang, Tamara Thomsen Ebert, and R Lee Penn.Aggregation-based crystal growth and microstructure development in natural iron oxyhydroxidebiomineralization products. Science, 289(5480):751–754, 2000.[236] David Kadko. Upwelling and primary production during the us geotraces east pacific zonal transect.Global Biogeochemical Cycles, 31(2):218–232, 2017.[237] Christian J Bjerrum and Donald E Canfield. Ocean productivity before about 1.9 gyr ago limited byphosphorus adsorption onto iron oxides. Nature, 417(6885):159, 2002.[238] Alexander A Pavlov, Lisa L Brown, and James F Kasting. Uv shielding of nh3 and o2 by organichazes in the archean atmosphere. Journal of Geophysical Research: Planets, 106(E10):23267–23287, 2001.[239] Eugene L Madsen. Identifying microorganisms responsible for ecologically significant biogeochemicalprocesses. Nature Reviews Microbiology, 3(5):439–446, 2005.[240] Kurt O Konhauser. Introduction to geomicrobiology. John Wiley & Sons, 2009.[241] Jizhong Zhou, Kai Xue, Jianping Xie, et al. Microbial mediation of carbon-cycle feedbacks to climatewarming. Nature Climate Change, 2(2):106–110, 2012.[242] Thomas W Crowther, Stephen M Thomas, Daniel S Maynard, et al. Biotic interactions mediate soilmicrobial feedbacks to climate change. Proceedings of the National Academy of Sciences, 112(22):7033–7038, 2015.148[243] Mo´nica Torres-Beltra´n, Alyse K Hawley, David W Capelle, et al. Methanotrophic community dynamicsin a seasonally anoxic fjord: Saanich inlet, british columbia. Frontiers in Marine Science, 3:268, 2016.[244] J Schieber. Pyritic shales and microbial mats: Significant factors in the genesis of stratiform pb-zndeposits of the proterozoic? Mineralium Deposita, 25(1):7–14, 1990.[245] Celia Me´ndez-Garcı´a, Ana I Pela´ez, Victoria Mesa, et al. Microbial diversity and metabolic networksin acid mine drainage habitats. Frontiers in microbiology, 6:475, 2015.[246] Alyse K Hawley, Mo´nica Torres-Beltra´n, Elena Zaikova, et al. A compendium of multi-omic sequenceinformation from the saanich inlet water column. Scientific data, 4:170160, 2017.[247] Ben J Woodcroft, Caitlin M Singleton, Joel A Boyd, et al. Genome-centric view of carbon processingin thawing permafrost. Nature, 560(7716):49–54, 2018.[248] Elena Zaikova, David A Walsh, Claire P Stilwell, et al. Microbial community dynamics in a seasonallyanoxic fjord: Saanich inlet, british columbia. Environmental microbiology, 12(1):172–191, 2010.[249] Jody J Wright, Kishori M Konwar, and Steven J Hallam. Microbial ecology of expanding oxygenminimum zones. Nature Reviews Microbiology, 10(6):381–394, 2012.[250] Stilianos Louca, Alyse K Hawley, Sergei Katsev, et al. Integrating biogeochemistry with multiomicsequence information in a model oxygen minimum zone. Proceedings of the National Academy ofSciences, 113(40):E5925–E5933, 2016.[251] Steven J Robbins, Paul N Evans, Donovan H Parks, Suzanne D Golding, and Gene W Tyson. Genome-centric analysis of microbial populations enriched by hydraulic fracture fluid additives in a coal bedmethane production well. Frontiers in microbiology, 7:731, 2016.[252] Inka Vanwonterghem, Paul D Jensen, Korneel Rabaey, and Gene W Tyson. Genome-centric resolutionof microbial diversity, metabolism and interactions in anaerobic digestion. Environmental microbiology,18(9):3144–3158, 2016.[253] Kurt O Konhauser, Noah J Planavsky, Dalton S Hardisty, et al. Iron formations: A global record ofneoarchaean to palaeoproterozoic environmental history. Earth-Science Reviews, 172:140–177, 2017.[254] Mirna Daye, Vanja Klepac-Ceraj, Mihkel Pajusalu, et al. Light-driven anaerobic microbial oxidation ofmanganese. Nature, pages 1–4, 2019.[255] Klaus Grasshoff, Klaus Kremling, and Manfred Ehrhardt. Methods of seawater analysis. John Wiley &Sons, 2009.[256] Derek R Lovley and Elizabeth JP Phillips. Availability of ferric iron for microbial reduction in bottomsediments of the freshwater tidal potomac river. Applied and Environmental Microbiology, 52(4):751–757,1986.[257] Bo Thamdrup, Henrik Fossing, and Bo Barker Jørgensen. Manganese, iron and sulfur cycling in acoastal marine sediment, aarhus bay, denmark. Geochimica et Cosmochimica Acta, 1994.[258] B Cusack, T Stanton, and E Richelson. acid guanidinium thiocyanate-phenol-chloroform extraction.Anal biochem, 162:156–159, 1987.[259] Dinghua Li, Chi-Man Liu, Ruibang Luo, Kunihiko Sadakane, and Tak-Wah Lam. Megahit: anultra-fast single-node solution for large and complex metagenomics assembly via succinct de bruijngraph. Bioinformatics, 31(10):1674–1676, 2015.[260] Heng Li and Richard Durbin. Fast and accurate long-read alignment with burrows–wheeler transform.Bioinformatics, 26(5):589–595, 2010.149[261] Christopher S Miller, Brett J Baker, Brian C Thomas, Steven W Singer, and Jillian F Banfield. Emirge:reconstruction of full-length ribosomal genes from microbial community short read sequencing data.Genome biology, 12(5):R44, 2011.[262] Christopher S Miller. Assembling full-length rrna genes from short-read metagenomic sequencedatasets using emirge. In Methods in enzymology, volume 531, pages 333–352. Elsevier, 2013.[263] Benjamin J Callahan, Paul J McMurdie, Michael J Rosen, et al. Dada2: high-resolution sampleinference from illumina amplicon data. Nature methods, 13(7):581, 2016.[264] Matthew Kearse, Richard Moir, Amy Wilson, et al. Geneious basic: an integrated and extend-able desktop software platform for the organization and analysis of sequence data. Bioinformatics,28(12):1647–1649, 2012.[265] Jordan A Fish, Benli Chai, Qiong Wang, et al. Fungene: the functional gene pipeline and repository.Frontiers in microbiology, 4:291, 2013.[266] Yu-Wei Wu, Blake A Simmons, and Steven W Singer. Maxbin 2.0: an automated binning algorithm torecover genomes from multiple metagenomic datasets. Bioinformatics, 32(4):605–607, 2016.[267] Donovan H Parks, Michael Imelfort, Connor T Skennerton, Philip Hugenholtz, and Gene W Tyson.Checkm: assessing the quality of microbial genomes recovered from isolates, single cells, andmetagenomes. Genome research, 25(7):1043–1055, 2015.[268] Pierre-Alain Chaumeil, Aaron J Mussig, Philip Hugenholtz, and Donovan H Parks. Gtdb-tk: a toolkitto classify genomes with the genome taxonomy database. Bioinformatics, 2018.[269] Norman I Bishop. The influence of the herbicide, dcmu, on the oxygen-evolving system of photosyn-thesis. Biochimica et biophysica acta, 27(1):205, 1958.[270] Donald E Canfield. Reactive iron in marine sediments. Geochimica et Cosmochimica Acta, 53(3):619–632,1989.[271] Christelle Hyacinthe, Steeve Bonneville, and Philippe Van Cappellen. Reactive iron (iii) in sediments:chemical versus microbial extractions. Geochimica et cosmochimica acta, 70(16):4166–4180, 2006.[272] Kurt O Konhauser, Andreas Kappler, and Eric E Roden. Iron in microbial metabolisms. Elements,7(2):89–93, 2011.[273] Kohen Bauer, James Byrne, Paul Kenward, et al. Magnetite biomineralization in ferruginous watersand early earth evolution. 2020.[274] Donald A Bryant and ZF Liu. Green bacteria: insights into green bacterial evolution through genomicanalyses. Adv Bot Res, 66:99–150, 2013.[275] Johannes F Imhoff. Taxonomy, phylogeny, and general ecology of anoxygenic phototrophic bacteria.In Photosynthetic Prokaryotes, pages 53–92. Springer, 1992.[276] Y Cohen, BB Jørgensen, E t PADAN, and M Shilo. Sulphide-dependent anoxygenic photosynthesis inthe cyanobacterium oscillatoria limnetica. Nature, 257(5526):489–492, 1975.[277] Shimshon Belkin and Etana Padan. Hydrogen metabolism in the facultative anoxygenic cyanobacteria(blue-green algae) oscillatoria limnetica and aphanothece halophytica. Archives of Microbiology,116(1):109–111, 1978.[278] Judith M Klatt, Mohammad AA Al-Najjar, Pelin Yilmaz, et al. Anoxygenic photosynthesis controlsoxygenic photosynthesis in a cyanobacterium from a sulfidic spring. Appl. Environ. Microbiol.,81(6):2025–2031, 2015.150[279] RE Danczak, MD Johnston, Christopher Kenah, et al. Members of the candidate phyla radiation arefunctionally differentiated by carbon-and nitrogen-cycling capabilities. Microbiome, 5(1):112, 2017.[280] Verona Vandieken, Ian PG Marshall, Helge Niemann, Bert Engelen, and Heribert Cypionka. Labilibac-ulum manganireducens gen. nov., sp. nov. and labilibaculum filiforme sp. nov., novel bacteroidetesisolated from subsurface sediments of the baltic sea. Frontiers in microbiology, 8:2614, 2018.[281] Anton Korenevsky and Terry J Beveridge. The surface physicochemistry and adhesiveness ofshewanella are affected by their surface polysaccharides. Microbiology, 153(6):1872–1883, 2007.[282] D Suter, C Siffert, B Sulzberger, and W Stumm. Catalytic dissoluion of iron (iii)(hydr) oxides byoxalic acid in the presence of fe (ii). Naturwissenschaften, 75(11):571–573, 1988.[283] Eric E Roden and John M Zachara. Microbial reduction of crystalline iron (iii) oxides: Influence ofoxide surface area and potential for cell growth. Environmental Science & Technology, 30(5):1618–1628,1996.[284] Raul E Martinez, D Scott Smith, Karsten Pedersen, and F Grant Ferris. Surface chemical heterogeneityof bacteriogenic iron oxides from a subterranean environment. Environmental Science & technology,37(24):5671–5677, 2003.[285] D Scott Smith and F Grant Ferris. Proton binding by hydrous ferric oxide and aluminum oxidesurfaces interpreted using fully optimized continuous p k a spectra. Environmental Science & technology,35(23):4637–4642, 2001.[286] Raul E Martinez, D Scott Smith, Ezra Kulczycki, and F Grant Ferris. Determination of intrinsicbacterial surface acidity constants using a donnan shell model and a continuous pk a distributionmethod. Journal of Colloid and Interface Science, 253(1):130–139, 2002.[287] JP Cowen. Morphological study of marine bacterial capsules: implications for marine aggregates.Marine Biology, 114(1):85–95, 1992.[288] CJ Van Oss. Hydrophobicity of biosurfaces—origin, quantitative determination and interactionenergies. Colloids and Surfaces B: Biointerfaces, 5(3-4):91–110, 1995.[289] Derick G Brown and Peter R Jaffe´. Effects of nonionic surfactants on the cell surface hydrophobicityand apparent hamaker constant of a sphingomonas sp. Environmental science & technology, 40(1):195–201, 2006.[290] Willem Norde and J Lyklema. Protein adsorption and bacterial adhesion to solid surfaces: acolloid-chemical approach. Colloids and surfaces, 38(1):1–13, 1989.[291] Joseph N Ryan and Philip M Gschwend. Effects of ionic strength and flow rate on colloid release:Relating kinetics to intersurface potential energy. Journal of colloid and interface Science, 164(1):21–34,1994.[292] Scott M McLennan and SR Taylor. Continental freeboard, sedimentation rates and growth ofcontinental crust. Nature, 306(5939):169–172, 1983.[293] DR Lowe. Major events in the geological development of the precambrian earth. The ProterozoicBiosphere: A Multidisciplinary Study, pages 67–75, 1992.[294] Kent C Condie. Chemical composition and evolution of the upper continental crust: contrastingresults from surface samples and shales. Chemical geology, 104(1-4):1–37, 1993.[295] Keith D Stolzenbach, Kathleen A Newman, and Charles S Wong. Aggregation of fine particles at thesediment-water interface. Journal of Geophysical Research: Oceans, 1992.151[296] William S Reeburgh. Oceanic methane biogeochemistry. Chemical reviews, 107(2):486–513, 2007.[297] Paul R Craddock and Nicolas Dauphas. Iron and carbon isotope evidence for microbial ironrespiration throughout the archean. Earth and Planetary Science Letters, 303(1-2):121–132, 2011.[298] Donald E Canfield. Sulfate reduction and oxic respiration in marine sediments: implications fororganic carbon preservation in euxinic environments. Deep Sea Research Part A. Oceanographic ResearchPapers, 36(1):121–138, 1989.152Appendix AChapter 4: supplemental material• Materials and Methods• Section 1: Cell surface features and acid-base chemistry• Section 2: Cell-iron surface interaction and extended DVLO modeling• Section 3: Iron concentration and supply• Section 4: Physical separation of ferric iron oxyhydroxides and cellular biomass in an ocean setting• Section 5: Box model of Archean marine carbon and iron cycles• Section 6: Organic carbon burial and diagenesis• Table A1: Media concentrations• Table A2: Cell surface characteristics• Table A3: Modern and Archean Fe fluxes• Table A4: Physical separation model results• Table A5: Data compilations for Fig. 4.1 and Figure S1• Fig. A1: BIF redox state• Fig. A2: Growth curve for Chlorobium phaeoferrooxidans strain KB01• Fig. A3: Additional SEM and TEM images of strains KB01 and KoFox• Fig. A4: Surface charge of strains KB01 and KoFox• Fig. A5: Additional cell surface characteristics• Fig. A6: Fe particle aggregation and physical separation model results• Fig. A7: Modelled weight % organic carbon• Fig. A8: Iron and carbon box model sensitivity resultsA.1 Supplementary materials and methodsA.1.1 Experimental and growth mediaThe basal media was prepared after Hegler et. al., 2008 [55] and allocated into serum bottles (100 mLmedia, 160 mL total volume), with 0.3 g L – 1 NH4Cl, 0.5 g L– 1 MgSO4·7H2O, 0.1g L – 1 CaCl2·2H2O. Afterautoclaving, 22 mmol L – 1 bicarbonate, trace elements, mixed vitamin solution, selenate-tungstate andvitamin B12 were added and the pH was adjusted to 6.8-6.9 under an N2:CO2 atmosphere (80:20). KH2PO4,Fe(II), and silica were added in a range of concentrations depending on the specific experiment (see TableA.1).All experiments were initially conducted using the standard media with 10 mM Fe(II) and 4.4 mMKH2PO4. In subsequent experiments, the media composition was adjusted to test the impact of a rangeof concentrations of Fe(II), PO43 – , and Si (Table A.1) on the percentage of cells that remain suspended as153opposed to depositing. This same range of Fe(II), PO43 – , and Si concentrations was further used to test thesurface charge of the Fe(III) oxyhydroxides precipitated under these conditions (Fig. 4.2, Fig. A.5a). Therange of Fe(II), PO43 – , and Si concentrations was used to create a range of P:Fe(III) and Si:Fe(III) ratios (Fig.4.2b). Specifically, Si concentrations were chosen to reflect the likely Si concentrations of the Precambrianoceans based on Jones et. al., 2015 [85].A.1.2 Fe(II) oxidation and cell growthTo track the growth kinetics of Chlorobium phaeoferrooxidans strain KB01 and Chlorobium ferrooxidans strainKoFox, sub-samples were taken from the serum bottles at the time of inoculation and every day thereafter.The first of these sub-samples were analyzed for Fe(II) and Fe(III) concentrations using spectrophotometricanalyses. Specifically, Fe(II) and Fe(III) concentrations were measured using the ferrozine method andsamples were measured directly as well as after being fixed in 1 N HCl – after Voillier et. al., 2000 [133].Fe(II) and Fe(III) concentrations were further used to identify when all of the Fe(II) had been oxidized toFe(III) oxyhydroxides and therefore the appropriate time to determine the percentage of cells that remainedsuspended. Additional sub-samples were taken from these same serum bottles to measure the pigmentconcentrations. Pigments were measured spectrophotometrically after 24 hour extractions of 1 mL ofpelleted cells in acetone:methanol (7:2 v/v) [134]. Pigment concentrations were then used as a proxy forcell abundance for both strains. To further confirm the growth of the strains, sub-samples were taken fromthe serum bottles and cells were fixed in gluteraldehyde (final concentration of 0.1 %). After the cells werefixed they were subsequently stained with SYBR green (0.25 % final concentration) and directly counted ina 96 well plate using a Miltenyi Biotec MACSQuant, with a flow rate of medium. Pigment concentrationswere also used to determine the fraction of cells that remained suspended. An example of a growth curvefor strain KB01 is shown in Fig. A.2 where the decrease in Fe(II) concentrations and an increase in cellcounts confirm the growth of the strain.A.1.3 Determination of cellular association to Fe(III)C. phaeoferrooxidans strain KB01, C. ferrooxidans strain KoFox, and Synechococcus sp. were grown to late logphase in standard growth media (Fig. A.2). The bottles were gently shaken by inversion to mix cells andFe(III) oxyhydroxides and allowed to settle for 24 hours. Mixing was done to resuspend cells that had settledout of the water column as a result of Stokes settling without any association to Fe(III) oxyhydroxides,and to allow maximum potential exposure of cell surfaces to Fe(III) oxyhydroxide particles. Sub-samplesfor Fe(II)/Fe(III) and pigments were taken from the upper portion of the serum bottle, avoiding settledFe(III) oxyhydroxides. A second set of sub-samples was taken from each bottle after the bottle had beenwell mixed. Fe(II)/Fe(III) and pigment concentrations were measured as described above. The fraction ofcells associated with Fe(III) oxyhydroxides was calculated by dividing pigment concentrations, a proxy forcell density (6.3 x 10 – 10 pigment/cell/mL for C. phaeoferrooxidans and 5.8 x 10 – 10 pigment/cell/mL for C.ferrooxidans), from the water column by the fully mixed suspensions. These measurements were conductedin 5 replicates and across a range of conditions (Table A.1) Fig. 4.2, Fig. A.5a. Association between bothorganisms and Fe(III) oxyhydroxide particles was further assessed through electron microscopy (describedabove). Images of strain KB01 are shown in the main text (Fig. 4.3a, b), while images of KoFox and moreTEM images of KB01 are shown below (Fig. A.3).A.1.4 Zeta potentialTo assess the surface charge of Fe(III) oxyhydroxides and cells, we determined the zeta potential of bothusing a Particle Metrix: ZetaView c©. To prepare the Fe(III) oxyhydroxides for zeta potential measurements,they were first concentrated through centrifugation at 14 g for 10 seconds. The supernatant was thendiscarded and the Fe(III) oxyhydroxides were sonicated for 1 minute to loosen any cells from the Fe(III)oxyhydroxide particles. The Fe(III) oxyhydroxides were subsequently rinsed once in sterile MQ water andconcentrated again via centrifugation. This process was repeated three times to reduce the numbers of cells154associated with the Fe(III) oxyhydroxides to low (<100 cells mL – 1, counted through flow cytometry asdescribed above) numbers. We confirmed that such low cell numbers had no impact on the surface chargeof the Fe(III) oxyhydroxide particles by determining the zeta potential of Fe(III) oxyhydroxides producedthrough abiotic Fe(II) oxidation in sterile 1.0 mM Si growth media. The zeta potential of these Fe(III)oxyhydroxides was the same as those produced in our experiments (data not shown). Fe(III) oxyhydroxidesthat we precipitated abiotically in sterile MQ water and in 0.7 M NaCl exhibited a positive surface charge,as generally observed [220]. The surface charges of biotically precipitated Fe(III) oxyhydroxides under therange of media compositions tested (described above) are shown in Fig. 4.2.To determine the surface charge of the photoferrotrophic strains C. phaeoferrooxidans strain KB01 andC. ferrooxidans strain KoFox, both strains were grown to late log phase in 400 µM Fe(II), 1.0 mM Si, 3 µMPO –4 media. The bottles were shaken and allowed to settle for 24 hours until all the Fe(III) oxyhydroxideshad reached the bottom serum bottle. Cells were then collected from water column and suspensions ofFe(III) oxyhydroxides and cells after mixing. The samples from the water column were centrifuged (10000g) for 7 minutes and washed three times (7 minutes centrifugation at 10000 g between washes) in eitherFe free growth media or 0.1 N NaCl. Each sample was then diluted 1/10 with filter-sterilized dH2O (tolower the conductivity of the sample to <2000 µS cm – 1) and measured in triplicate using a Particle Metrix:ZetaView c©. The samples of mixed cell-Fe(III) oxyhydroxide suspensions were first treated with pH 7dithionite to reduce the Fe(III) to Fe(II). Fe(II) and Fe(III) concentrations were measured (as describedabove) before and after the dithionite treatments and the subsequent rinses to confirm that any residualFe had been effectively removed (<2.5 µM) (data not shown). Following dithionite treatment, the mixedsuspensions were measured in the same manner as the water column samples. Zeta potentials for eachstrain are summarized in Fig. A.4, while the zeta potentials for cells and Fe(III) oxyhydroxides measured ingrowth media and used in DVLO modeling are in Table A.2.A.1.5 Surface contact anglesTo calculate the interfacial forces between the C. phaeoferrooxidans strain KB01, C. ferrooxidans strain KoFox,and the Fe(III) oxyhydroxides formed as by-product of their growth, we measured static contact anglesfor strain KB01, strain KoFox, and Fe(III) oxyhydroxides precipitated abiotically from silica-rich media.These measurements were conducted following Korenevsky and Beveridge, 2007 [281] and Saini and Chan,2013 [225] and the contact angle data are summarized in Table A.2. To conduct these measurements, cellsand Fe(III) oxyhydroxides were collected onto a 0.22 µm polycarbonate filter until there was a thick lawnof cells or Fe(III) oxyhydroxides coating the entire filter. The filters were subsequently dried for 45-60minutes to remove excess liquid prior to measurement. Contact angle measurements were conducted usinga contact angle goniometer with three liquids that have known surface tension properties: water, glycerol,and diiodomethane (Table A.2). 1 µL of each liquid was placed on the sample, 30 images were taken overthe course of one and a half minutes, and contact angles were measured and averaged from these 30 images.This process was repeated three times for each of the reference liquids.A.1.6 Cell surface titrationsTo determine the acid-base chemistry and interrogate cell-surface functional groups, both strains C.phaeoferrooxidans strain KB01 and C. ferrooxidans strain KoFox were grown in 1.0 mM silica-rich mediaas described above. The cells were then removed from the media and pelleted through centrifugation.To remove residual Fe(III) oxyhydroxides the pelleted cells were treated for 10 minutes with 10 mL ofoxalate/oxalic acid (pH 3) and 1 mL of 100 mM Fe(II) for every 1 mL of cells [256, 282, 283]. To removeresidual Fe(II) the cells were rinsed with anoxic iron-free growth media [55]. Finally, the cells wereresuspended in 200 mL of that same anoxic iron-free media. Cell suspensions were acid/base titratedfollowing the protocol detailed in Martinez et. al. 2003 [284]. Cells were centrifuged four times for 8minutes at 6200 x g. In between these centrifugations the cells were rinsed three times with degassedMQ water and once with sterile, degassed 0.1 M KNO3. The final pellet was resuspended in 0.1 M KNO3.Sub-samp