Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Integrating geochemical and microbiological information for better modeling of the N-cycle – past and… Michiels, Céline Chantal 2019

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


24-ubc_2019_september_michiels_celine.pdf [ 24.72MB ]
JSON: 24-1.0378547.json
JSON-LD: 24-1.0378547-ld.json
RDF/XML (Pretty): 24-1.0378547-rdf.xml
RDF/JSON: 24-1.0378547-rdf.json
Turtle: 24-1.0378547-turtle.txt
N-Triples: 24-1.0378547-rdf-ntriples.txt
Original Record: 24-1.0378547-source.json
Full Text

Full Text

Integrating geochemical andmicrobiological information for bettermodeling of the N-cycle – past andpresentbyCe´line Chantal MichielsMSc, Universite´ Libre de Bruxelles, 2012BSc, Universite´ Libre de Bruxelles, 2010A 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)April 2019c© Ce´line Chantal Michiels 2019The following individuals certify that they have read, and recommend to the Faculty of Graduateand Postdoctoral Studies for acceptance, the dissertation entitled:Integrating geochemical and microbiological information for better modeling of the N-cycle – pastand presentSubmitted by Ce´line C. Michiels in partial fulfillment of the requirements for the degree of Doctorof Philosophy in Microbiology and ImmunologyExamining Committee:Prof. Sean A. Crowe, Microbiology and Immunology & Earth, Ocean and Atmospheric SciencesSupervisorProf. Steven J. Hallam, Microbiology and ImmunologySupervisory Committee MemberProf. Mary O’Connor, ZoologyUniversity ExaminerProf. Susan Baldwin, Chemical and Biological EngineeringUniversity ExaminerProf. Gregory Dake, ChemistryChairAdditional Supervisory Committee Members:Prof. Philippe Tortell, Earth, Ocean and Atmospheric SciencesSupervisory Committee MemberProf. Roger Franc¸ois, Earth, Ocean and Atmospheric SciencesSupervisory Committee MemberiiAbstractCycling of N occurs through a multitude of microbial reactions used by microorganisms to harnessenergy and generate growth. These microbial reactions are the main controls on the availabilityof fixed-N and can often limit primary production in marine ecosystems. The microorganismsinvolved in the N-cycle are diverse and the metabolic pathways are further distributed acrossmany taxa, rendering the modeling of the N-cycle complex. Indeed, models of N-cycling fall shortof making robust and explicit predictions, in part due to a lack of ecophysiological informationdescribing the relevant processes at a molecular scale. Direct ecophysiological information isobtained from process rate measurements, yet these generally lack coupled information onmicrobial community composition limiting their extensibility across multiple environments. Thisdissertation creates a new framework for the modeling of the N-cycle by measuring the rates andpathways of N-cycling in anoxic pelagic environments. This new and quantitative knowledgeis incorporated into models of N-cycling to improve reconstructions of past and future N-cycle.I describe the rates and pathways of Fe-dependent NO–3 reduction in a ferruginous pelagicenvironment, analogous to the Proterozoic oceans. I then describe the nutrients status and theimplications of NO–3 reduction through DNRA and denitrification for biological productionthrough a flux-balance model for ancient oceans. I also study the environmental factors thatinfluence the partitioning of N-loss between anammox and denitrification in an anoxic fjord(Saanich Inlet). A flux-balance model was built to describe the competition between anammoxand denitrification based on the rates of N2 production as well as changes in microbial communitycomposition and ecophysiological parameters. We show that recycling of N through DNRA, ratherthan N-loss, dominates annual NO–3 reduction in Saanich Inlet, challenging current assumptionsthat DNRA does not need to be considered as an important pathway of N-cycling in the ocean.Overall, the work presented here offers a new and integrated approach that combines geochemicalinformation such as nutrient profiles and process rate measurements, microbiological informationsuch as microbial community composition, structure and functions analysis, and applies it toquantitative models that can be used to further test hypotheses about the N-cycle.iiiLay SummaryAvailability of oxygen is an organizing principle for life in marine ecosystems. As oxygen declines,microbial metabolisms prevail over higher trophic activities. Low oxygen conditions are foundin large zones of the modern oceans where anaerobic microbial activities play an important rolein biogeochemical cycling of essential elements like nitrogen, which affects nutrient availability,primary production, and CO2 sequestration. These low oxygen conditions also existed on theEarly Earth, and microbial activities would have been primordial regulators of biogeochemicalcycling of essential nutrients, which likely impacted biological productivity and climate. My thesiscreates new knowledge on how nitrogen is used under low-oxygen conditions by specific groupsof microorganisms, for past and present marine systems. I then apply this new information tomodeling approaches that inform on biogeochemical cycles in the Earth system, for the ancientand modern moceans, which can lead to better predictions for future climate.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 partlyor wholly published, in press, or in review. Copyright licenses were obtained and are listed below.• Chapter 1: Ce´line C. Michiels wrote the main text with editorial support from Sean A.Crowe.• Chapter 2: Ce´line C. Michiels wrote the main text with Sean A. Crowe. Sean A. Crowe andFranois Darchambeau designed the research. Sean A. Crowe and Franois Darchambeaucollected samples. Sean A. Crowe, Franois Darchambeau and Fleur Roland performedlaboratory work, Ce´line C. Michiels and Sean A. Crowe analyzed and interpreted the data,as well as developped the model found in the published paper. Editorial support wasreceived from the entire list of authors. The reference for the published paper can be foundas follows:C. C. Michiels, F. Darchambeau, F. A. E. Roland, et al. Iron-dependent nitrogen cyclingin a ferruginous lake and the nutrient status of proterozoic oceans. Nature Geoscience,10(3):217U176, Mar 2017.• Chapter 3: Ce´line C. Michiels wrote the main text with help from Sean A. Crowe. K.E.Giesbrecht and D. E. Varela supplied the Chlorophyll a data. J. A. Huggins, R. L. Simisterand J. S. Spence helped in the analysis of the isotope pairing technique and taxonomic data.S. J. Hallam and D. E. Varela reviewed the mauscript and provided editorial support.This chapter was submitted to the journal Frontiers in a special edition titled ”Facing Marinedeoxygenation”. The manuscript was accepted in January 2019.The citation can be found as: C. C. Michiels, J. A. Huggins, K. E. Giesbrecht et al. Rates andpathways of N2 production in a persistently anoxic fjord: Saanich Inlet, British Columbia.Frontiers in Marine Science, 6:27, 2019.• Chapter 4: Ce´line C. Michiels wrote the main text with editorial support from Sean A.Crowe. Ce´line C. Michiels and Sean A. Crowe interpreted the data. J. A. Huggins helpedin the analysis of the isotope pairing data. C. Morgan-Lang and R. L. Simister helped invthe processing of the metagenomic data. S. J. Hallam provided help in the analysis of themetagenomic data and editorial support.• Chapter 5: Ce´line C. Michiels wrote the main text.Throughout this dissertation the word ‘we’ refers to Ce´line C. Michiels unless otherwise stated.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 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiiAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xivDedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 The emergence of the N-cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41.2 The modern N-cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81.3 Environmental distribution of anammox, denitrification and DNRA . . . . . . . . . 111.4 Controls on anammox, denitrification and DNRA . . . . . . . . . . . . . . . . . . . . 181.5 Distributed metabolisms and the N-cycle . . . . . . . . . . . . . . . . . . . . . . . . . 261.6 Problem statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301.7 Dissertation overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312 Iron-dependent nitrogen cycling in a ferruginous lake and the nutrient status of Pro-terozoic oceans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352.3 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353 Rates and pathways of N2 production in sulphidic Saanich Inlet . . . . . . . . . . . . . 423.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51vii3.2.1 Study site and sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513.2.2 Nutrient and process rate measurements . . . . . . . . . . . . . . . . . . . . . 523.2.3 Microbial community profiling . . . . . . . . . . . . . . . . . . . . . . . . . . 543.2.4 Flux balance modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 553.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 553.3.1 General water column physical, chemical, and biological properties . . . . . 553.3.2 Rates of denitrification, anammox, and dark carbon fixation . . . . . . . . . 583.3.3 Response of denitrification and anammox to amendments . . . . . . . . . . 583.3.4 Depth-Integrated rates of N-loss . . . . . . . . . . . . . . . . . . . . . . . . . 613.3.5 Microbial community composition . . . . . . . . . . . . . . . . . . . . . . . . 633.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 673.4.1 Partitioning of N-loss in SI, and the seasonality of anammox and denitrifica-tion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 673.4.2 Kinetics of denitrification and anammox . . . . . . . . . . . . . . . . . . . . . 713.4.3 Vertical partitioning of the microbial communities in SI . . . . . . . . . . . . 723.4.4 Model of NO–2 competition between anammox and complete denitrification 773.4.5 SI as a model ecosystem for coastal OMZs . . . . . . . . . . . . . . . . . . . . 804 Combining microbiological and geochemical information to constrain energy flow throughthe marine N-cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 844.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 854.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 894.2.1 Study site and sample collection . . . . . . . . . . . . . . . . . . . . . . . . . . 894.2.2 Nutrient analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 894.2.3 Process rate measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 904.2.4 DNA extraction, qPCR and absolute cell abundance . . . . . . . . . . . . . . 904.2.5 Metagenome sequencing and assembly . . . . . . . . . . . . . . . . . . . . . . 914.2.6 Taxonomy of of the microbial community recovered through metagenomicanalyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 914.2.7 Quantification of functional genes . . . . . . . . . . . . . . . . . . . . . . . . . 924.3 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 934.3.1 Dynamics in rates and pathways of N-cycling . . . . . . . . . . . . . . . . . . 934.3.2 Dynamics of the microbial community in response to physical perturbations 964.3.3 Power supply and ecophysiology of anaerobic N-metabolisms . . . . . . . . 1054.4 Implications and extensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1075 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1095.1 Dynamics in rates and pathways of anaerobic N-cycling . . . . . . . . . . . . . . . . 1095.2 Integrated approach for better modeling of biogeochemical cycling . . . . . . . . . 110viii5.3 Looking ahead . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1115.4 Closing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115AppendicesA Chapter 1: supplemental material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132A.1 Isotope pairing technique protocol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132A.1.1 Sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132A.1.2 Start of the incubation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132A.1.3 Taking time points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132A.1.4 Analysis of samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132A.2 Summary of pelagic and benthic rates of denitrification, anammox and DNRA . . 133B Chapter 2: supplemental material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137B.1 Fe-dependent NO–3 reduction – thermodynamic considerations . . . . . . . . . . . . 137B.2 Denitrification and DNRA rates summary in Kabuno Bay . . . . . . . . . . . . . . . 137B.3 Dark carbon fixation in Kabuno Bay . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137B.4 Box-model of C, N, S and Fe cycling for a hypothetical Proterozoic upwelling system 137B.5 Global N-fixation and N-loss in the Archean and Proterozoic . . . . . . . . . . . . . 141C Chapter 3: supplemental material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148C.1 NH+4 sediment fluxes in Saanich . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148C.2 Microbial communities in SI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148C.3 Model of NO–2 competition between anammox and complete denitrification . . . . 153C.3.1 General remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155C.3.2 Stability of the model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155C.3.3 Stagnation phenotype: partitioning of N2 production through anammox andcomplete denitrification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157C.3.4 Matlab code for model NO–2 competition . . . . . . . . . . . . . . . . . . . . 161D Chapter 4: supplemental material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165D.1 Geochemical profiles in Saanich Inlet (SI) . . . . . . . . . . . . . . . . . . . . . . . . . 165D.2 Potential and scaled rates of anaerobic N-metabolisms . . . . . . . . . . . . . . . . . 167D.3 Taxonomy and functional gene abundances . . . . . . . . . . . . . . . . . . . . . . . 169D.4 Energy availability and power supply . . . . . . . . . . . . . . . . . . . . . . . . . . . 173D.5 Methods supplement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174ixList of Tables1.1 N-budgets for the Earth system (marine and terrestrial) based on 1) Gruber andGalloway (2008) [1] and 2) Canfield et al. (2010) [2] and references therein. Inputs arecharacterized by positive numbers whereas outputs from the systems are negative. *indicates that this flux was not mentioned but could have been merged with anotherflux without mention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121.2 Thermodynamic calculations for denitrification and DNRA with different organicand inorganic ED. We varied the N-content of the organic molecules. The ∆G◦ foreach reaction was calculated based on the second law of thermodynamics. . . . . . 211.3 Main genes and enzymes involved in the N-cycle, based on Kuypers et al. (2018)[3] and references therein. The subunits for the genes are not specified here. . . . . 273.1 Addition of labeled N-species and electron donors to incubations in 2015. . . . . . . 523.2 Michaelis-Menten parameters, Km (µM) and Vmax (nmol L– 1 hr– 1) for NO–3 depen-dency of denitrification at 165m in August 2015. Note that anammox kinetics arenot following Michaelis-Menten model in this case . . . . . . . . . . . . . . . . . . . 613.3 Parameters used in model for competition of NO–2 between anammox and completedenitrification (Fig. 3.12) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82A.1 Summary of benthic rates of denitrification, anammox and DNRA . . . . . . . . . . 134A.2 Summary of benthic rates of denitrification, anammox and DNRA, cont’d . . . . . . 135A.3 Summary of pelagic rates of denitrification, anammox and DNRA . . . . . . . . . . 136B.1 Free Gibbs Energy yield under standard conditions (4G◦) and for Kabuno Bayconcentrations (4G). Values for 4G◦ can be found in [4] . . . . . . . . . . . . . . . . 142B.2 Chemical species concentrations (in µM) representative for the chemocline inKabuno Bay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142B.3 Summary of DNRA and denitrification rates for KBs water column. Rates werecalculated over 48 hours unless stated otherwise next to the calculated rates. . . . . 143B.4 Rates and ratio considered for calculations . . . . . . . . . . . . . . . . . . . . . . . . 143B.5 Description of the different parameters used in the current model . . . . . . . . . . 144xC.1 Summary of samples and the number of sequences and OTUs observed in eachsample, as well as bacterial small subunit ribosomal RNA (SSU or 16S rRNA) geneabundance obtained through qPCR. The chao diversity index was also calculatedfor each sample based on OTUs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149C.2 Stoichiometric coefficients for the metabolites considered in the model and theirrespective reactions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153C.3 Kinetic parameters used in the stability analysis of the model. . . . . . . . . . . . . . 156C.4 Kinetic parameters for complete denitrifiers. . . . . . . . . . . . . . . . . . . . . . . . 157D.1 Partitioning of N-loss and N-retention through NOx reduction in moles m– 2 d– 1. . 169D.2 List of genes and their acronyms used in this paper for the metagenomic analysis . 172D.3 Examples of free energy yields calculated for 2 months anoxic water column in SI(in kJ moles N– 1). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173D.4 Sampling dates and type of 15N-labeled incubations. . . . . . . . . . . . . . . . . . . 174D.5 Accession numbers for NCBI raw reads of samples. . . . . . . . . . . . . . . . . . . . 175xiList of Figures1.1 Overview of the microbial reactions comprising the N-cycle. . . . . . . . . . . . . . . 21.2 Schematics of distributed metabolisms within the N-cycle over time. . . . . . . . . 61.3 15N-labeling incubations workflow. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161.4 Environmental distribution of anammox, denitrification and DNRA. . . . . . . . . . 171.5 ∆Greaction as a function of the logarithm of the activities . . . . . . . . . . . . . . . . 252.1 Vertical distribution of selected physical and chemical properties of Kabuno Bay forFebruary 2012. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362.2 Rates and pathways in Kabuno Bay for February 2012. . . . . . . . . . . . . . . . . . 372.3 Model outputs describing coupled C, N, S and Fe cycling in an idealized Proterozoicupwelling system. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403.1 Saanich Inlet (SI), a model ecosystem for the study of microbial metabolisms in OMZs 493.2 Geochemical profiles for SI, 2015 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 573.3 Process rate measurements for SI, 2015. . . . . . . . . . . . . . . . . . . . . . . . . . . 593.4 NO–3 dependency in SI. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 603.5 HS– dependency in SI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 623.6 Depth Integrated N-loss. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 633.7 Microbial communities in SI from 16S rRNA gene sequencing. . . . . . . . . . . . . 653.8 Clustering of the microbial community composition of SI in 2015. . . . . . . . . . . . 663.9 Relative abundance of Planctmycetes, SUP05, Marinimicrobia and Arcobacter. . . . . 673.10 Anammox rates vs. in situ NO–3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 723.11 Marinimicrobia OTUs in SI, 2015 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 783.12 Model of NO–2 competition between anammox and complete denitrification . . . . 814.1 Rates and pathways of anaerobic N-metabolisms. . . . . . . . . . . . . . . . . . . . . 954.2 Taxonomic composition of the microbial community. . . . . . . . . . . . . . . . . . . 974.3 Depth-integrated taxonomic composition of microbial communities. . . . . . . . . . 984.4 Functional gene abundances of anaerobic N-metabolisms. . . . . . . . . . . . . . . . 1024.5 Re-networking of anaerobic N-metabolisms linked to power supply. . . . . . . . . . 103xiiB.1 Box-model for C, N, S and Fe cycling in hypothetical Precambrian upwelling systemadapted from [5] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145B.2 Fe-pyrite to highly reactive Fe ratio for 50 and 100% DNRA . . . . . . . . . . . . . . 145B.3 Role of Fe(II) concentrations in dictating the FePY/FeHR ratio across a suite ofdifferent model conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146B.4 Run of the model with 20µM Fe(II) in the intermediate box (I) but no NO–3 . . . . . . 147C.1 Chao1 diversity index from amplicon sequencing . . . . . . . . . . . . . . . . . . . . 150C.2 Relative abundance of 15 most abundant taxa for the surface waters of SI in 2015(10m). These taxa were the most abundant ones found in average throughout thesamples. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151C.3 Relative abundance of 15 most abundant taxa for the deeper waters of SI in 2015(100, 120, 135, 150, 200m). These taxa were the most abundant ones found in averagethroughout the samples. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152C.4 Lower limit of the stability of the model for the set conditions found in table C.3. . 156C.5 Simulation of the model for a stagnation phenotype that shows anammox dominat-ing N2 production. See table C.3 for kinetic parameters used here. . . . . . . . . . . 158C.6 Decrease of km,NO2 (see table C.4) for complete denitrification shows rates ofdenitrification dominating over anammox rates after 100 days. . . . . . . . . . . . . 159C.7 Increase of Vmax,DEN (See table C.4) for complete denitrification shows rates ofdenitrification dominating over anammox rates after 100 days. . . . . . . . . . . . . 160C.8 Increase of YDEN (see table C.4) for complete denitrification shows rates of denitrifi-cation dominating over anammox rates after 100 days. . . . . . . . . . . . . . . . . . 161D.1 Nutrient concentrations in SI for the years 2015-2016 at station S3. . . . . . . . . . . 166D.2 Potential and scaled rates of denitrification, anammox and DNRA. . . . . . . . . . . 168D.3 Taxonomic composition of microbial communities at the OTU level in SI. . . . . . . 170D.4 RPKM counts for functional genes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171xiiiAcknowledgementsThis thesis has profoundly changed my life by uprooting me from my birth country. AlthoughI have missed my Belgian family and friends dearly, I can with certainty say that I now have aCanadian family and have gathered around me some dear friends in Vancouver. This thesis alsoallowed me to visit some amazing places in BC, the middle of the Pacific Ocean and Indonesia. Ihave met some incredible people along this journey and would like to thank them all for theirkindness. However, the list is long, and I will try to keep it more concise in a few dedicated thankyous. First and foremost, I would like to thank Sean A. Crowe for welcoming me in his brand-newlab in 2013 and expanding my vision on how to ”do science”. Your incredibly vast knowledgehas been a never-ending source of new ideas and I am very grateful to have been able to thriveunder your supervision. Thank you to my committee, Steven J. Hallam, Philippe Tortell andRoger Franc¸ois for your support during our meetings. Thank you to my External Examiner, Prof.Gregory Dick from U. Michigan, who reviewed my thesis carefully and thoughtfully. Thank youto my lab mates in the Crowe lab, past and current members – Kate Thompson, Ashley Davidson,Julia Huggins, Jenifer Spence, Rachel Simister, Niko Finke, Arne Sturm, Kohen Bauer, AndrewHefford – for your help and guidance in the lab, your knowledge, the discussion at bread andbeer meeting (!!), and numerous coffee dates! A special thank you to Annie Cheng and BelleCheng who have helped me measure my million samples without ever complaining. To all the”Saanichers” – Jade Shiller, Monica Torres-Beltran, Melanie Sorensen, Andreas Muller, Chris Payneand Lora Pakhomova, and the crew from the Strickland, thank you! And finally, a special ”thankyou” to a few Hallam lab members – CarriAyne Jones, Aria Hahn, Connor Morgan-Lee, ColleenKellogg, Alyse Hawley. When I sought out your expertise, you were always there and willing toshare it with me and morexivDedicationA` mes parents et Camille, pour leur soutien inconditionnel malgre´ les 7825km de distance quinous se´parent.To Kev for being there no matter what.xvChapter 1IntroductionNitrogen (N) is required as an elemental constituent for all life on Earth, as it is one of the mostabundant elements in nucleic acids and proteins. For every 100 atoms of carbon, cells typicallyrequire between 2 to 20 atoms of nitrogen depending on the specific organism [6]. The majority ofEarth’s accessible nitrogen resides in the atmosphere, with nitrogen comprising 78% by mass inits inert form N2 gas. This N2 gas is primarily made available to life through the bioenergeticallyexpensive microbial reaction of N-fixation, which requires an enormous activation energy (16ATPs or ∼544 kJ mol– 1 [7]) to break the triple bound found in the N2 molecule. Rates of N-fixationcan limit primary production both in terrestrial and marine ecosystems, and this limitation ofproductivity is a first-order control on biological CO2 sequestration globally [8, 9]. BioavailableN is thus fundamental to sustaining life on Earth, and has the potential to further influence theglobal carbon pump and climate.The modern biogeochemical cycling of N, its biological availability, and operation underdifferent redox states depends almost entirely on a suite of oxidoreductive metabolic reactionsconducted by a complex network of microorganisms (Fig. 1.1). N2 is incorporated from theatmosphere into the biosphere through the fixation of gaseous N2 into organic amines (R-NH3),which upon biomass degradation are liberated to the surrounding environment as NH+4 viaammonification (Fig. 1.1). NH+4 is then sequentially oxidized to NO–2 and NO–3 through theprocess of nitrification, requiring molecular oxygen to operate (Fig. 1.1). These oxidized N-species are important to eukaryotic primary producers and are usually assimilated as a sourceof fixed-N (Fig. 1.1). NO–3 is alternatively either reduced back to the atmosphere as N2 throughthe chemotrophic reactions of denitrification and anaerobic ammonium oxidation (anammox) orreduced back to NH+4 through the chemotrophic reaction of dissimilatory nitrate reduction toammonium (DNRA) (Fig. 1.1). When assimilated into biomass, N can be released through organic1matter remineralization and ammonification. However, a small portion of this organic matterescapes the remineralization process and ends up buried in sediments [10]. During subduction,some of the buried N (bound to the organic matter) returns to the atmosphere but a significantpart leaves the biosphere for the geosphere and is sequestered in the solid Earth [11].NH4+ N2 N2O NO NO2- NO3-Org Noxicanoxic8 593412nrfnirnifanfvnfnos nor nirnarnap, nashzshao/hzoamonxrhaoN"fixationAssimilation/RemineralizationNitrification/commamoxDNRN5" Dissimilatory5Nitrate5Reduction5to5NitritePDNO5– Partial5denitrification5to5Nitrous5OxideComplete5denitrificationAnammox – Anaerobic5ammonium5oxidationDNRA5" Dissimilatory5Nitrite5Reduction5to5Ammonium156892347 676710102-III 0 +I +II +III +V oxidationreductionFigure 1.1: Overview of the microbial reactions constituting the N-cycle. Reactions are defined in the greybox. Orange arrows correspond to an oxidation reaction of the N-species whereas blue arrows show areduction of the N-species. The grey arrows encompass assimilation of the N-species by an organism andremineralization of organic N-species. Reactions with arrows above the dotted line are generally consideredto be done under oxic conditions whereas arrows drawn below the dotted line are usually considered doneunder anoxic conditions. Chemical N-species follow the redox state highlighted in the grey box. In italicare the main genes found to execute the corresponding enzymatic reaction. This figure was made with thehelp of Julia Huggins and is inspired from Kuypers et al. (2018) [3].The main pools of N in the Earth system are distributed between the atmosphere and the solidEarth: the crust and the mantle [12]. Recycling of N between these different pools occurs both2rapidly through biological processes (Fig. 1.1 and description above) with a turnover time foratmospheric N2 of ∼1000 years [2], and slowly through tectonics of the crust and mantle overtimescales of 1 Ga [13]. Biological reactions govern the processes of N fixation and remineralization,which, in combination, determine the overall rate at which biomass is buried in marine sedimentsand subducted into the mantle. Thus, the rapid turnover of N through biological activity couldhave caused long-term changes in atmospheric N abundance. Such dynamics in atmosphericN abundance through time can further influence climate and impose a negative feedback onbiological productivity. For example, lower rates of N-fixation result in higher N2 pressure inthe Earth’s atmosphere, which increases the greenhouse effect of existing gases by broadeningtheir absorption lines [14]. A lowering of N2 pressure through high rates of N-fixation andburial of organic matter will in turn induce a negative feedback on N-fixation, which wouldbecome limited by lower N2 partial pressure [15]. Despite best efforts to date, however, likelychanges in atmospheric N2 over Earth’s history, and in the future, remained unconstrained dueto high uncertainties in the sizes and exchange rates between the different reservoirs and theircorresponding mechanisms of regulation [16, 17].Cycling of N through biological activity is mainly conducted by microorganisms through themany metabolic reactions transforming N, between its oxidized and reduced states under oxicand anoxic conditions (Fig. 1.1). The metabolic potential for these reactions are widely distributedacross the 3 domains and across mutliple phyla within these domains [3] (see example in Fig.1.2c). Furthermore, microorganisms have established specific niches and have developed complexinteractions between functional metabolic groups in order to carry out these transformations [18].The complexity of the microbial metabolic network responsible for N-cycling has, as a conse-quence, rendered biogeochemical modeling of this element challenging. Biogeochemical modelingapproaches often use thermodynamic calculations in combination with susbtrate availability, aswell as ecophysiological parameters for the relevant microorganisms. Ecophysiological parametersremain elusive, however, due to the fact that the majority of relevant microorganisms remainuncultivated and that there is a lack of mechanistic links between process rate measurements andthe underlying microbial community in environmental systems (see section 1.4). Modeling effortsof N-cycling, to date, are, therefore, hampered by the lack of information needed to constrain3these models.The emergence of humans as geobiological agents is causing changes to the Earth system atunprecedented rates, and specifically to the N-cycle [2, 19]. An acute example is the doubling ofglobal rates of N-fixation through the use of the Haber-Bosch process to produce fertilizers sincethe early 20th century [2]. Such rapid perturbations to fluxes of matter and energy strongly deviatefrom the pre-industrial dynamics in biogeochemical cycles that were established over billions ofyears; this will almost certainly alter rates of pathways in the N-cycle and have unconstrainedfeedbacks on climate. Such changes in rates and pathways will be largely determined by changesin the fluxes of N through the different metabolic pathways. It is currently difficult, however,to predict these responses and feedbacks either qualitatively or quantitatively, since we lacksufficient information on these metabolic pathways to construct meaningful models that willultimately allow robust forecasts. In particular, we need new ecophysiological information on therelevant microorganisms involved in key metabolic pathways in the N-cycle. The extent of currentknowledge and gaps will be discussed in the following sections.This chapter introduces the evolution of the N-cycle over time, the details of different metabolicpathways comprising the modern N-cycle, the distribution of these pathways amongst a broaddiversity of microorganisms, and the controls that constrain the rates of these processes. It alsoaims to highlight the existing gaps in current knowledge of the N-cycle and the uncertaintiesassociated with global budgets and N-cycling modeling. The following chapters of the thesisare then outlined in the last section. Overall, my thesis creates new knowledge on the N-cyclethrough direct measurements of rates and pathways of N-cycling in anoxic pelagic environments.This new and quantitative knowledge can be incorporated into models of N-cycling to improvereconstructions of the past and make predictions about the future.1.1 The emergence of the N-cycleThe distribution of N-species in modern pools of N is the product of billions of years of feedbackbetween biological N-cycling and geophysical processes (tectonics). Dynamics in the N-cycleand the distribution of N between Earth’s major N inventories is archived in the rock record,4which reveals exchange between atmospheric N and the solid Earth mediated through biologicalN-fixation, transfer of fixed-N into marine sediments, and subduction of these sediments intothe mantle [17]. Life emerged on Earth approximately 4.2 Ga ago [20] and N-fixation likelyemerged as a key metabolic pathway as early as 3.8 Ga [21], in a plausible response to scarcityof bioavailable N in the early biosphere. It was not until the advent of oxygenic photosynthesis,however, and the introduction of molecular oxygen (O2) into Earth’s surface environments, thatthe full-suite of oxidative and reductive pathways in the biological N-cycle likely emerged (Figs.1.1 and 1.2) [16, 22].Before biological N-fixation evolved ∼3.8Ga ago, bioavailable N came only from extraterrestrialinput and slow photochemical and hydrothermal activities [23–25]. It is assumed that bioavailableN was extremely limited, which likely restricted life on the early Earth [2]. Initial input of Non Earth occurred through the bombardment of the proto-Earth with solid ammonia that thenoutgassed to the atmosphere under extreme heat conditions [2]. Heat shock in the atmosphere,associated with lightning activity and meteorite impacts, turned N2 into NO, which, throughphotochemical and aqueous reactions, could have been oxidized to NO–2 and NO–3 [23, 25, 26].These processes were extremely slow and yielded up to 2 · 10 8 mol N yr– 1 [27], producing onlya fraction of todays marine biological fixation (10 · 10 12 mol N yr– 1 or 10 Tmol N yr– 1 [1]). Inaddition to atmospheric reactions, hydrothermal activity could have theoretically transformedN2/NO–3 /NO–2 into NH+4 . However, there is no evidence for such reactions today and theiroperation in the past thus seems unlikely [28, 29]. A need for alternative pathways to fix N wouldhave emerged when biological activity exhausted geological sources.In the absence of atmospheric oxygen, the early biological N-cycle was only composed ofbiological N-fixation, NH+4 assimilation and release of the NH+4 through biomass degradation, alsocalled ammonification (Fig. 1.2a), followed by transfer of this fixed-N to the sediments. With theadvent of biological N-fixation during the PaleoArchean (4.0 – 3.2Ga), NH+4 would likely haveemerged as the principal N-species, accumulating in the oceans under anoxic and ferruginous(Fe2+–rich) marine conditions. N-fixation is known to have a relatively small fractionation effect,leaving the N-pool fixed in the biomass with a small negative fractionation relative to the source N2[30], thus agreeing with the N isotopic fractionation observed in the geological record, which shows5a small excursion of δ15N between 0 and -4h. In addition to the rock record, the phylogeneticdiversity of well-conserved nitrogenase, the enzyme necessary for N-fixation, shows that themetabolic potential for N-fixation is widespread across two domains of life, implying a strongneed for this reaction to occur [31] and suggesting that it evolved early in the evolution of life andwas subsequently spread through vertical inheritance and horizontal gene transfer [32]. Altogether,the evidence suggests that biological N-fixation evolved early in Earth’s history, not long after thecapacity to assimilate NH+4 into biomass.NH4+N2N2ONONO2-NO3-Org N Org N?NH4+N2N2ONONO2-NO3-???????OMNH4+N2N2O?NO3-NO2-N2O?NH4+?NO3-H2SO2NH4+S3PlanctomycetesSUP05 uncultured bacteriumNitrospora defluvii/Nitrospina gracilisSAR11 cladeThaumarcheotarg Na. b. c.Today4Ga 2.5Ga4.5Ga? ???Figure 1.2: Schematics of distributed metabolisms within the N-cycle over time. Colored circles representmicrobial species, when known, performing a reaction (function) shown by the arrows. Several speciescan perform the same reaction, and the same species can perform multiple reactions depending on themetabolic potential contained in their genomes and their expressed metabolism. For the ancient N-cycle,however (a and b), it is, to date, impossible to associate species with pathways (a) represents the earlybiosphere with constrained chemical N-species present. (b) highlights the changes to a modern N-cycle dueto the oxygenation of the atmosphere and thus the presence of oxidized N-species such as NO–3 and NO–22.5 Ga ago (c) is a specific example of the modern N-cycle with the distribution of the N-cycle reactionsthrough the microbial communities in Saanich Inlet (BC, Canada). (c) was adapted from Hawley et al.(2014) [33].Oxygenic photosynthesis likely evolved around 3.0 Ga, during the Meso-Neo Archean (3.2 –2.5Ga) [34–37], and the local accumulation of O2 in regions known as ”O2 oases” [38] triggered the6onset of the oxidative part of the biological N-cycle (Fig. 1.2b). This allowed for the developmentof nitrification and, for the first time, the accumulation of oxidized N-species (NO–3 /NO–2 ) in theoceans (Figs. 1.1 and 1.2b). Denitrification and/or anammox likely evolved shortly thereafter asNO–3 became increasingly available. The advent of biological N2 production has been implicatedin a bottleneck for biological productivity, because the NO–3 produced through nitrification wouldhave been effectively returned to the atmosphere through denitrification/anammox, with thecorresponding re-imposition of N-limitation on biological productivity [39–41]. It was onlywhen atmospheric O2 increased appreciably, likely in the late Proterozoic (1.6 – 0.6 Ga), that themarine NO–3 pool started to stabilize and become more widely available and similar to modernconcentrations [41]. The rock record supports the evolution to a modern N-cycle, showing positiveexcursions of δ15N throughout the late Archean and Proterozoic, indicating an early onset ofnitrification and denitrification processes associated with a small pool of NO–3 in the oceans[22, 40–43]. Indeed, denitrification is associated with isotopic fractionation that leads to a heavierNO–3 pool, which is subsequently recorded in sediment biomass containing assimilated NO–3 .This signal can be amplified if the pool of NO–3 is small in the water column, leading to increasedpositive δ15N signal. However, the δ15N of the Meso- and Neoproterozoic (1.6 – 0.6 Ga) is close tothat of modern marine sediments (+5h) implying a stabilization of the NO–3 pool in the ocean,which indicates the widespread establishment of the modern N-cycle due to the stabilization ofO2 in the atmosphere and the ocean [41, 43].Most information about the ancient N-cycle comes from the rock record [21, 22, 40–43] andfrom associated modeling using information from the rock record [39, 42, 44, 45]. Few modelsintegrate microbial and molecular information in their design in part due to a lack of essentialquantitative ecophysiological information on the relevant microbial processes. One way to inferwhat the microbial activities would have looked like during the Archean and the Proterozoic is tostudy modern environments that harbor geochemical conditions analogous to the ancient oceans(e.g. ferruginous conditions), and host microbial communities that manifest metabolic potentialrelevant to the past. The oceans in the Archean and Proterozoic eons were mainly ferruginous[46] with transient occurrences of euxinic (sulphide-rich) conditions in coastal and closed basinareas in the late Archean and through the Proterozoic [47–49]. Finding ferruginous conditions7under the modern oxidized atmosphere has proven challenging, however, because ferrous iron(Fe2+) oxidizes to ferric iron (Fe3+) when oxygen is present, thus restricting the occurrence of suchconditions on today’s Earth. Nevertheless, a few modern analogues to the ferruginous Archeanand Proterozoic oceans exist on Earth today and information from these analogues can be usedto inform our view of the past [Kabuno Bay in Lake Kivu (RDC) [50], Lake La Cruz (Spain) [51],Lake Pavin (France) [52], Lake Lugano (Switzerland) [53] and Lake Matano (Indonesia) [54]]. Forexample, studies in these environments reveal the ecological role of photoferrotrophy (anoxygenicphotosynthesis with Fe2+) in illuminated Fe-rich environments. By extension, studies in thesesame environments may inform qualitative and quantitative models of N-cycling under theseconditions.The Neo-Archean and the Proterozoic oceans transitioned periodically from ferruginous toeuxinic conditions in coastal areas and closed basins, as a result of increases in organic carbonavailability in the water column and/or with increased seawater sulphate concentrations [41, 49].These euxinic conditions appear to have been transient and mainly restricted to coastal shelveswhere biological activity would have been high [47–49, 55]. This could have influenced ratesof N-recycling through denitrification and anammox as it has been shown that microorganismsperforming denitrification can use sulphide as an electron donor and the presence of sulphidemight be an inhibitor of anammox [56]. However, we are lacking quantifiable microbial informationto assess how the transition from ferruginous to euxinic conditions would have affected the N-cycle.1.2 The modern N-cycleThe modern N-cycle is comprised of multiple biological and geochemical processes and thetransmission of N between the different pools at the Earth’s surface depends almost entirelyon the activity of microorganisms. Throughout the oceans and in soils, microorganisms calleddiazotrophs fix N2 from the atmosphere and incorporate it into their biomass [57]. When thesemicroorganisms die and decompose, fixed-N contained in their biomass is released into thesurrounding environment as NH+4 through the process of ammonification (Fig. 1.1). Under oxic8conditions, this NH+4 can be oxidized sequentially to NO–2 and then NO–3 , through the microbialreactions of nitrification. Under the low oxygen conditions that can develop when respirationrates exceed oxygen supply from photosynthetic production and/or advective and diffusivetransport from the atmosphere, specific microorganisms use NO–3 as a terminal electron acceptorin respiration instead of oxygen, through canonical denitrification, leading to the production ofN2 and closing the N-cycle by returning N to the atmosphere (Fig. 1.1).In addition to complete denitrification, several other microbial reactions also utilize NO–3 orNO–2 as an electron acceptor (Fig. 1.1): dissimilatory NO–3 reduction to NO–2 (DNRN), partialdenitrification to nitrous oxide (PDNO), and anaerobic ammonium oxidation (anammox) [3]. Theseprocesses, like complete denitrification, usually operate under low oxygen or anoxic conditions.Microorganisms that carry out partial denitrification (DNRN and PDNO) are often grouped withcomplete denitrifies and collectively are often referred to simply as ’denitrifiers’. Thus, many’denitrifiers’ lack the metabolic potential to perform all the steps of complete denitrification, andeven when they do, environmental conditions may favour incomplete denitrification. Denitrifiersoxidize inorganic (HS– [2], Fe(II) [58, 59]) and/or organic electron donors (e.g. formate, lactate,and other C-containing compounds [60]) depending on their metabolic potential and the substratesavailable. Anammox bacteria, on the other hand, reduce NO–2 while oxidizing NH+4 autotrophically[61]. This metabolism is confined to the bacterial phylum Planctomycetes and is phylogeneticallyrestricted in comparison to the very broad diversity of the denitrifiers which spans bacterial andarchaeal phyla [3, 62]. Together, complete denitrification and anammox contribute to the loss ofbioavailable nitrogen back to the atmosphere, often referred to as N-loss [63].The microbial reaction of DNRA provides a shunt in the N-cycle (Fig. 1.1) and precludes Nrecycling back to the atmosphere. Indeed, as it reduces NO–3 and NO–2 to NH+4 , DNRA retainsfixed-N, possibly enhancing the transfer of N2 from the atmosphere to marine sediments. AsDNRA consumes NO–2 , it also competes with PDNO, complete denitrification, and anammox forthe same substrates in the environment. This process has, however, rarely been measured, andit could be cryptically active where/when anammox is also present, as anammox consumes theNH+4 produced rather than allowing its accumulation [64]. This renders DNRA invisible to manytechniques used to determine rates and pathways of microbial N transformations [64]. Therefore,9the extent to which DNRA contributes to global N-cycling is poorly defined, so far [3].Overall, the relative fluxes of N between anammox, DNRN, PDNO, complete denitrificationand DNRA likely influence global climate and biogeochemical processes. These processes affect therates of N transfer between different reservoirs, the availability of N to primary producers, and theinteractions of N with other biogeochemical cycles in soils, sediments, and pelagic environments.The differential movement of N through this metabolic network has the potential to tip the balancebetween export of fixed-N to marine sediments versus N recycling back to the atmosphere throughN2 production. Naturally, this balance plays an important role in global biological productivityand climate over multiple time-scales. Moreover, leakage of intermediates such as greenhouseactive N2O gas can have feedbacks on climate. The complex metabolic pathways that underpinthe N-cycle have thus influenced Earth’s climate over geological time-scales by regulating fluxesof N from the atmosphere, to the crust, and the mantle. Today, climate change is mainly drivenby human activity and this will have specific feedbacks on primary production and microbialactivities. However, the extent to which primary production and microbial activities will respondto these changes is poorly constrained.Current anthropogenic activities have lead to the increased loading of agricultural soils with Nthrough the industrial Haber-Bosch process, at a current rate of 9.7 Tmoles yr– 1 [2], the equivalentof current estimates for marine N-fixation. The increase in soil NO–3 , in turn, leaches N to coastalwaters and causes eutrophication by increasing biological productivity in coastal areas. Indeed,an increase in organic matter promotes aerobic respiration that consumes oxygen, and, withoutsufficient ventilation, can lead to the development of anoxia. Oxygen minimum zones (OMZs)form under similar conditions, with upwelling of nutrient-rich deep waters promoting primaryproduction in surface water and aerobic respiration in underlying waters. Anoxia in coastal areasand OMZs supports heavy N-loss [60, 65–69] in its anoxic cores [0.1% of the oceanic volume, [63]],and to the development of transient plumes of HS– [70, 71]. As coastal anoxia and marine OMZsare currently expanding, due to global warming leading to poor ventilation of these zones andincreased nutrient loading with increased primary production [72], it is likely that N-loss willbe enhanced in the oceans. Thus, a shift in the balance between N-fixation and N-loss couldoccur, but our current knowledge of the N-cycle is insufficient to enable consistent and robust10predictions of future N-cycling. It is therefore important to improve quantitative models that willenable us to make predictions for the future biogeochemical cycling of essential elements for lifeon Earth.This will be discussed in the next three sections of this introduction.1.3 Environmental distribution of anammox, denitrification andDNRAIt is currently unknown whether the global N-cycle is balanced, and this is due to uncertainties inthe estimates of N-budgets and rates of N-species transformations. The abundance of fixed-Nin the environment is controlled by the balance between sources and sinks, which are primarilybiological N-fixation and N-loss through N2 production and NH+4 burial. Some analyses suggestglobal N-budgets (Table 1.1), are currently balanced ([1] and references therein) while othersimply that anthropogenic activities have increased atmospheric N-fixation to such a degree thatit outpaces N-loss back to the atmosphere ([2] and references therein). Analyses that indicatebalance, however, are based primarily on estimates that carry large uncertainties of up to 20 to50%, or more [1]. Global input and output fluxes for the N-cycle are mostly based on extrapolationfrom budgets built at smaller scales based on process rate measurements that are limited bothspatially and temporally. Refinements and expansions of these measurements would likely leadto more robust scaling and could promote consensus on global N-budgets; such concensus isessential to the predictions of future N-cycling and climate models.Robust measurements of N transformation rates can be measured by amending soils, sediments,and waters with 15N labeled N-species and tracking their movement through different N-pools[73, 74]. 15N is not naturally abundant [0.4% [75]] and thus mass spectrometry provides asensitive way to detect the accumulation of small amounts of excess 15N in natural N pools thatresults from amendments of 15N labeled reactants or substrates. For example, it is possible tomeasure N2-fixation by exposing environmental microbial communities to 15N2 and followingthe incorporation of labeled 15N into biomass (Fig. 1.3a). In the same way, it is also possibleto discriminate N2 production between anammox and denitrification by separately providingmicrobial communities with 15NO–3 /15NO–2 or15NH+4 and following the production of15N2 (Fig.11Table 1.1: N-budgets for the Earth system (marine and terrestrial) based on 1) Gruber and Galloway (2008)[1] and 2) Canfield et al. (2010) [2] and references therein. Inputs are characterized by positive numberswhereas outputs from the systems are negative. * indicates that this flux was not mentioned but could havebeen merged with another flux without mentionSystem Input/Output Flux (Tg yr-1) Gruber and Galloway (2008) Flux (Tg yr-1) Canfield et al. (2010) Terrestrial N-fixation 145 110.6  Atmospheric deposition 40 25.2  Anthropogenic activity 205 182  N-loss -317 -99.4  Riverine export to ocean -80 -68.6   ∆= -7 ∆=149.8 Marine N-fixation 140 140  Atmospheric deposition 50 ?*  Riverine export to ocean 80 68.6  N-loss -244 -238  N-burial -25 ?*   ∆= 1 ∆= -29.4   ∆∆ = -6 ∆∆ = 120.4 !1.3a). Nitrification and DNRA can be quantified in these same experiments by tracking theaccumulation of 15NO–3 /15NO–2 or15NH+4 , respectively (Fig. 1.3a). The15N/14N compositionof the tracked products (usually 15N2) can be measured by using gas source isotope ratio massspectrometry (IRMS). If the resulting products are dissolved species (15NH+4 or15NO–3 /15NO–2through DNRA or nitrification, respectively), these products are first chemically reduced oroxidized to 15N2 before being measured by Istotope Ratio Mass Spectrometry (IRMS [76]). 15Nlabeling incubations and the subsequent measurement of 15N excess by IRMS has proven to besensitive [74], with detection limits depending on the variation in sensitivity between instruments.Measurements based on this 15N-labeling technique form the backbone of my thesis and aschematic figure detailing the handling of the samples can be found in Fig. 1.3b. A fully detailedprotocol can be found in Appendix A.Other approaches to measuring rates of N-cycling include analyses of N2:Ar ratios to determineproduction/consumption of N2 relative to atmospheric levels [77], measuring variability in thenatural abundance of N isotopes [78], and application of inhibitors such as acetylene which blocks12the final step in denitrification [79]. These methods are less widely used today as they are usuallyless sensitive and allow for a less detailed insight into processes and are generally blind to specificmetabolic pathways. In addition to these methods, another way to look at N-cycling is to compareconcentrations of NO–3 and PO2 –4 , i.e. N*. Concentrations of NO–3 and PO2 –4 in the ocean follow atrend directed by the Redfield ratio (16N:1P, [80]), corresponding to the average composition ofmarine photosynthetic community and its following remineralization. Thus, concentrations ofNO–3 and PO2 –4 are commonly compared by calculating N* according to Eq. 1.1.N∗ = [NO−3 ]− 16 ∗ [PO2−4 ] + 2.9µM (1.1)The inference of N-loss based on N* relationship between the two nutrients is constant for theentire ocean, and a negative deviation from it indicates a NO–3 deficit due to more N-loss thanN-fixation, signifying active denitrification or anammox. The advantage of N* is that it allowsinvestigation of the effect of N-fixation and N-loss on nutrient levels in the ocean without havingto use direct rate measurements, which can be costly and time-consuming [81]. However, similarto the other methods mentioned in this paragraph, N* calculations cannot discriminate specificprocesses and are semi-quantitative estimates of N-loss. Overall, application of these tools andtechniques over the last century has yielded remarkable insights into how N-cycling operates in adiverse suite of environments. For example, the anammox process was discovered in a wastewaterreactor in the early 1990’s [82] and, shortly after, found to operate all over the world in sediments[83] and pelagic environments [84, 85]. Rates and pathways of N-loss and N-processes operatingunder low oxygen conditions, namely anammox, denitrification and DNRA, were compiled inFig. 1.4, and Tables A.1, A.2 and A.3 to provide a summary of available information (spatialdistribution and magnitude of rates) on rates and pathways of microbial N transformations. Wespecifically inventoried benthic and pelagic rates found for both lacustrine and marine systems.Anammox, denitrification, and DNRA do, however, operate in terrestrial soils as well, and moreinformation about this can be found in the following references and references therein [86–89].N-cycling is intense in most marine and lake sediments, contributing to more than 50%of global marine N-loss [9]. N-cycling processes in sediments are vertically distributed, with13nitrification occurring in the oxic sediments, often causing a subsurface accumulation of NO–3and NO–2 . Denitrification, anammox and DNRA occur below this where O2 is depleted andNO–3 accumulates. This vertical distribution of N metabolism is intrinsically linked to otherbiogeochemical cycles such as oxygen, carbon, sulfur, and iron-cycles. Indeed, respiration insediments generally proceeds using a suite of terminal electron acceptors, in order of progressivelydecreasing free energy yields (O2>NO–3 >Fe3+>SO2 –4 ). This respiration is fueled by organic matterdeposited from the overlying water column, acting as an electron donor for respiration. Theprogressive depletion of electron acceptors and donors below the sediment-water interface leadsto a vertical cascade in redox couples, which leads to stratified microbial communities. When O2is exhausted, NO–3 is generally used as the next most favourable electron acceptor for anaerobicmicrobial respiration. This is where heterotrophic denitrification and DNRA usually occur, as wellas anammox supplied with NH+4 from the remineralization of organic matter and NO–2 from NO–3reduction. Rates of denitrification, anammox and DNRA have been reported mostly on coastalshelves, in riverine estuaries, in lakes, and, on a few accounts, on the continental slope (Tables A.1and A.2). Rates vary over several orders of magnitude, with rates of DNRA varying between 0.024[90] to 24 · 10 7 µmoles m– 2 d– 1 [91], rates of anammox between 1.2 [92] and 5 · 10 3 µmoles m– 2d– 1 [93], and rates of denitrification between 10 [94] and 2.4 · 10 7 µmoles m– 2 d– 1 [91](Tables A.1and A.2). The magnitude of the rates generally appears to increase with decreasing latitude, withthe highest rates of DNRA and denitrification reported in tropical estuaries and marine sediments[91, 93, 95–97]. However, an overwhelming majority of the measurements have taken place innorthern latitude temperate regions and more measurements in tropical sediments are needed toaccurately assess the contribution of these systems to global N-loss.Pelagic environments support 30 to 50% of the global marine N-loss [9]. In these environments,anammox and denitrification have been reported in coastal and open ocean OMZs, in closedbasins (e.g. fjords), and in inland waters such as stratified lakes (Fig. 1.4ab and Table A.3). Usually,denitrification and anammox operate near oxic-anoxic boundary layers, where O2 is low and NO–3is available. When detected, volumetric rates of denitrification vary between 0.05 nM d– 1 in theETSP [98] and 1700 nM d– 1 in Wintergreen Lake (USA) [99]. Rates of anammox vary between0.12 nM d– 1 in the Arabian Sea [69] to 480 nM d– 1 in the Golfo Dulce (Costa Rica) [100]. Out14of 27 studies (Table A.3), only 11 studies have attempted to measure DNRA. Rates of DNRArange between 0.48 nM d– 1 in the Eastern Tropical South Pacific (ETSP) [60] and 151 nM d– 1 in asulphidic hydrothermal vent [101] . Most of these studies represent single measurements at singlestations or along transects, likely missing spatial and temporal variability that can be found insuch environments. A specific example would be a drastic change in nutrients with the occurrenceof a transient plume of HS– [102], or increased N2-production through meso-scale eddy events inthe Peruvian upwelling system [103]. These events are extremely transient and could thereforeelude sampling. Thus, without sampling coverage of such events, we might be overlooking thetrue variations in rates and pathways of N-cycling in marine anoxic waters, preventing accuratedescriptions of marine N-budgets, and overlook important controls on these pathways, such asnutrient availability.15N-FIXANDENDNRAANNITR15N2 15NO3- 15NH4+15N-org 29N2 30N2 15NH4+29N2 15NO3-** ***  If DNRA is active, Anammox will also produce 30N2** If nitrification is active, Anammox will also produce 30N2a.b. 1He2He headspaceSampleAddition of 15N-labelT0T1T2T331mL gas sample2mL liquid sample4Nutrient analysis: NO3-, NO2-, NH4+ 15NH4+ 28,29,30N2 GC-IRMSTime15 N productionT0 no labelFigure 1.3: 15N-labeling incubations workflow. (a) shows which active processes of the N-cycle can bemeasured based on specific addition of 15N-labels. In blue shows the addition of the label. In the serumbottles, the reactions (N-FIX=N-fixation, AN=anammox, DEN=Denitrification, NITR=nitrification) thatcan be detected by adding the specified labels and in the white outlined box, the products coming fromthe transformation of the labeled N-species with the processes that will be measured. (b) is a workflowdiagram of the incubation experiment. (1) The serum bottle is filled with anoxic water, overflowed 3x andthen closed with blue butyl stoppers to limit O2 contamination. (2) A headspace is added to the serumbottle to further limit O2 contamination, the 15N-label is added to the sample and the bottle is then shakenfor gas species to equilibrate. (3) The samples are incubated in the dark and several time points are takento follow the course of 15N2 production. (4) The production of labeled N-species can be measured byGC-IRMS and the concentration of nutrients can be measured by spectrophotometry.16−50050−100 0 100 200longlat0.50.5PWAYANAMMOXDENITRIFICATIONDNRARATE_LN0.−50050−100 0 100 200longlat0.80.8RATE_LN0481216PWAYANAMMOXDENITRIFICATIONDNRA−50050−100 0 100 200longlat0.80.8RATE_LN0481216PWAYANAMMOXDENITRIFICATIONDNRAa.b.−50050−100 0 100 200longlat0.50.5PWAYANAMMOXDENITRIFICATIONDNRARATE_LN0. (nM d-1)−50050−100 0 100 200longlat0.80.8RATE_LN0481216PWAYANAMMOXDENITRIFICATIONDNRARate (µmol m-2 d-1)1023*1031 5107Figure 1.4: Environmental distribution of anammox, denitrification and DNRA. (a) Rates of anammox, denitrifi-cation and DNRA found in marine and freshwater sediments (in µmol m– 2 d– 1). (b) Pelagic lacustrineand marine rates of DNRA, denitrification, and anammox (in nM d– 1). Rate magnitude is described by theblack circle on the right of the figure. Anammox is in green, denitrification in blue and DNRA in orange. Adot in the circle indicates that the process was looked for but not detected. No dot or no circle indicatesthat there was no experiment done to measure this process.171.4 Controls on anammox, denitrification and DNRAMost biogeochemical models, when attempting to reproduce the specific rates and pathways ofN-cycling, will need constraints on the controls for the different processes described in thesemodels. Metabolic processes are usually controlled primarily by whether or not a reaction isthermodynamically favourable, and can lead to the harnessing of energy and growth for themicroorganisms conducting the reaction. For example, it is usually considered that microorganismswill consume first the available electron acceptors that are the most energetic, then consume thenext most energetic available acceptors, if their metabolic potential allows it, when the first one isdrawn down to inaccessible concentrations [68, 104]. Thus free energy yield calculations for themetabolic reactions will be a first-order determinant on which pathways can occur under specificconditions of the system in a model. Secondly, the metabolic pathways conducted by specificenzymes are limited by how fast these enzymes can process the substrates, depending on theconcentrations of these substrates. This is defined by the kinetic features of the enzymes, which canlimit the rates of reactions. These parameters can be measured either for lab cultured organismsor for environmental microbial communities, which is non-taxon specific. However, becausethe kinetic parameters usually vary between the type of enzyme and are also taxon-specific,environmental kinetic parameters are not extensible to other environments, without previousknowledge of the microorganisms involved in the pathways and their individual kinetic traits(or ecophysiological parameter). Beyond thermodynamic and kinetic information, other factorscan also control the rates and pathways of N-cycling, such as inhibitors and physical factors (e.g.temperature). These factors further confound attempts at modeling the dynamics in rates andpathways under changing system conditions. Below is summarized the state of the knowledgefor the controls on anammox, denitrification and DNRA, and how they are currently used inmodeling approaches.Organisms catalyze redox reactions to harness energy from electron transport that allowsthem to perform anabolic metabolism for growth and reproduction. The yield obtained fromthese biochemical reactions is constrained by thermodynamic properties, which dictate how muchenergy a given reaction yields under specific conditions. This energy can be quantified as the18Gibbs free energy or ∆G◦ of a reaction and is calculated based on the free energy of formation forthe reactants and products involved in the reaction (Eq. 1.2). These are based on standard-statereference conditions, which are rarely found in natural environments. These standard stateGibbs free energies of reaction can be translated from the standard-state reference to any set ofenvironmental conditions by correcting for the activities of the individual products and reactantsin the reaction (Q=quotient of products and reagents activities), as well as temperature andpressure (Eq. 1.3).∆G◦ = ∆G◦fprod − ∆G◦freac (1.2)∆G = ∆G◦ + RT ∗ ln(QprodQreac) (1.3)Using Gibbs free energies it is possible to assess whether a particular biochemical reaction isfavorable under a given set of environmental conditions. Negative Gibbs free energies signifyexergonic reaction yields, which release energy and can power microbial metabolism. Reactionswith positive Gibbs free energies require energy input to occur, whereas Gibbs free energies of 0signify equilibrium. Because changes in product and reactant concentrations affect the value of theGibbs free energy of reaction, it follows then that specific metabolisms yield different quantitiesof energy based on the concentrations of the substrates present. Reaction free energy yield isthus a first order determinant on microbial niche partitioning. This is a particularly relevantconsideration for the N-cycle where multiple pathways competing for the same substrates havevery different geochemical outcomes. Denitrification, for example, produces N2 and leads to areturn of N to the atmosphere. Whereas DNRA generates NH+4 that remains bioavailable or canbe sequestered in sediments and ultimately subducted to the mantle. In this case, concentrationsof electron donors and acceptors influence which reaction, DNRA or denitrification, is the mostenergetically favorable. We calculated the free energy yield of the reactions with different organicand inorganic electron donors (ED), as both DNRA and denitrification can be chemolithotrophicor heterotrophic [56, 65, 105–107]. We estimated the free energy yield of the half-reaction for19natural organic compounds as follows:∆Goxidation = 60.3− 28.5 ∗ NOSC [108] (1.4)Where NOSC is the nominal oxidation state of carbon, calculated as:NOSC = −(e−a) + 4 [108] (1.5)for the generic half reaction:CaHbNcOdPe + (3a + 4e− d)H2O⇒ aHCO−3 + cNH+4 + ePO2−4 + (5a + b− 4c− 2d + 7e)H+ + (4a + b− 3c− 2d + 5e)e− [108](1.6)The results show that DNRA is more competitive with an organic ED than denitrification permole of NO–3 reduced, no matter the C/N ratio content of the organic ED studied (Table 1.2 andFig. 1.5a). Denitrification, however, was more competitive with HS– as an inorganic ED thanDNRA (Fig. 1.5b). In comparison, H2 as an inorganic ED was more favorable for DNRA when theratio between H2 and NO–3 is high. Fe(II) was favorable for denitrification only at very high ratiosas well, and was endergonic for DNRA. Thus, thermodynamics are an effective way to determinewhether a reaction will occur or not based on the conditions present in the ecosystem and providea first-order approach to predicting the outcomes of potential competition between reactions usingthe same substrate. However, other factors, such as enzyme kinetics and inhibition, growth yield,and viral infection are also important considerations.Although thermodynamics are useful in determining whether a reaction will occur and howmuch energy can be harvested out of it, reaction rates also play a role in predicting the outcome ofcompetition. The kinetics of enzymatic reactions can often be described using a Michaelis-Mentenmodel which describes rates as a function of a limiting-substrate concentrations, a half-saturationconstant (Km), which is the concentration of a substrate at half of the maximum rates, and the20Table 1.2: Thermodynamic calculations for denitrification and DNRA with different organic and inorganicED. We varied the N-content of the organic molecules. The ∆G◦ for each reaction was calculated based onthe second law of thermodynamics. Reactions ∆G˚ (kJ moles N-1) DNRA/Redfield C106H263N16O110P + 53 NO3- + 53 H2O + 14 H+ Þ 106 HCO3- + 69 NH4+ + HPO42- -559.01 DNRA/Redfield -50% N C106H239N8O110P + 53 NO3- + 53 H2O + 6 H+ Þ 106 HCO3- + 61 NH4+ + HPO42-  DNRA/Redfield -75% N C106H227N4O110P + 53 NO3- + 53 H2O + 14 H+ Þ 106 HCO3- + 57 NH4+ + HPO42- +2 H+  DNRA/Redfield +50% N C106H287N24O110P + 53 NO3- + 53 H2O + 22 H+Þ 106 HCO3- + 77 NH4+ + HPO42-  Denitr/Redfield C106H263N16O110P + 84.8 NO3- Þ 106 HCO3- + 16 NH4+ + 42.4 N2(g) + HPO42- + 42.4 H2O +7.2 H+  -524.87 Denitr/Redfield -50% N C106H239N8O110P + 84.8 NO3- Þ 106 HCO3- + 8 NH4+ + 42.4 N2(g) + HPO42- + 42.4 H2O +15.2 H+  Denitr/Redfield -75% N C106H227N4O110P + 84.8 NO3- Þ 106 HCO3- + 4 NH4+ + 42.4 N2(g) + HPO42- + 42.4 H2O +19.2 H+  Denitr/Redfield +50% N C106H287N24O110P + 84.8 NO3- + 0.8 H+ Þ 106 HCO3- + 24 NH4+ + 42.4 N2(g) + HPO42- + 42.4 H2O  DNRA/ HS- NO3- + HS- + H2O + H+ Þ SO42- + NH4+ -487.54 Denitr/ HS- 8 NO3- + 5 HS- + 3 H+  Þ 5 SO42- + 4 N2 (g) + 4 H2O -480.20 DNRA/ H2 NO3- + 4 H2 + 2 H+ Þ NH4+ + 3 H2O -679.61 Denitr/ H2 2 NO3- + 5 H2 + 2 H+ Þ N2 (g) + 6 H2O -600.24 DNRA/ Fe2+ NO3- + 8 Fe2+ + 21 H2O Þ NH4+ + 8 Fe(OH)3 + 14 H+ 401.99 Denitr/ Fe2+ NO3- + 5 Fe2+ + 12 H2O Þ ½ N2 (g)  + 5 Fe(OH)3 + 9 H+ 75.76  maximum rate of reaction (Vmax) when the enzyme is substrate-saturated (Eq. 1.7).Rreaction =Vmax ∗ [S][S] +Km(1.7)These kinetic parameters can either be measured in pure culture or in the environment. Thelatter involves an added layer of complexity, as interactions with the environment and othermembers of in situ microbial communities are likely to influence the rates measured. However,as most microorganisms have not been cultured to date, environmental kinetics are likely togive us the most environmentally relevant information. Unfortunately, only a handful of studies,summarized here for marine settings, have measured these parameters. Half-saturation constants(Km) are usually reported as apparent substrate dependency constants for environmental studies.Denitrification, anammox and DNRA compete for NO–3 and NO–2 as substrates. Denitrificationwas shown to have an apparent NO–3 dependency (or km) of 2.9 µM in an anoxic fjord of the BalticSea [56]. Contrary to this, another study in the Baltic Sea reported no effect of NO–3 additionbetween 1 to 10µM, suggesting high NO–3 affinity and enzymatic saturation above 1µM [106].NO–2 dependency (or km) for anammox was reported to be below 3µM [109], and as low as 0.1µMin marine sediments [110]. This finding was supported by a comparison to rates of anammox and21the corresponding ambient NO–2 concentrations from OMZs, showing no correlation between thetwo [68]. Thus, it appears that anammox bacteria are likely not limited by NO–2 concentrations inthe environment and may have a higher affinity for NO–2 than denitrifiers. It is, however, harderto conclude something about NO–3 and NO–2 dependency for denitrification, as it has not beentested extensively. Moreover, only one study has tested the NO–3 dependency of DNRA and thisstudy found no effect of NO–3 concentration on rates of DNRA [106].Electron donor (ED) availability appears important in regulating rates of denitrification andDNRA as well, with increasing electron donor concentrations often correlating to high rates[105, 106, 111]. Both denitrification and DNRA have been reported to be either organotrophic(organic ED) or lithotrophic (inorganic ED) processes [56, 65, 105–107]. Addition of different EDcan thus help to determine how the N-cycle is coupled with other cycles such as the C-, S- andFe-cycles. Only denitrification has been shown to depend on organic matter in marine waters[105, 106] and marine sediments [90, 112]. The reactive DOC dependency constant reported fordenitrification in marine waters was 0.08 µM [106], and rates of denitrification increased in marinesediments with shallowing depth as well as with increased organic matter loading [90, 112].Additionally, denitrification has been shown to depend on sulphide in sulphidic environmentssuch as the Mariager fjord in the Baltic sea [56], and other stations in the Baltic Sea chemocline[106], with km for HS– varying between 1.7 and 3.5 µM in the Baltic Sea [106]. A linear dependencywas, however, observed in Mariager fjord, suggesting that enzyme saturation was not reachedunder the concentrations studied (0-50µM HS– - [56]). DNRA also responded to HS– amendmentsin the Baltic Sea and sulphidic sediments [106, 113], with a km between 6.8 and 8.6 µM [106].Finally, DNRA appears to be coupled with Fe(II) oxidation in estuarine sediments with a kmof 33.8µM [111]. Denitrification, although it was simultaneously detected with DNRA, did notrespond to increased Fe(II) concentrations.NH+4 is the ED used in the anammox process, and can also be a limiting substrate foranammox in marine pelagic environments, as NH+4 concentrations are very low (<5µM - [68]).A single measurement of enrichment cultures of Ca. Scalindua sp. suggests a km for NH+4 of3µM for anammox bacteria [109]. A collection of rates of anammox compared to ambient NH+4concentrations in OMZs reveals a positive correlation between the two, implying that the supply22rate or concentration of NH+4 in seawater can regulate rates of anammox in OMZs [68]. A positivecorrelation can also be found between rates of anammox and organic matter concentrations[105]. Indeed, when organic matter is remineralized through heterotrophic processes, such asdenitrification under low oxygen conditions, NH+4 is released and available to be used by anammox.Hence, it has been argued that organic matter stoichiometry controls N-loss in open ocean OMZs,constraining the amount of NH+4 released during respiration and its supply rate to anammox[105]. In addition to the stoichiometry of organic matter, it has been argued that organic matterconcentrations could regulate the partitioning of N-loss between anammox and denitrification inmarine sediments [83, 90]. However, organic matter quality and quantity cannot always explainthe partitioning between denitrification and anammox as other processes can also be active, suchas sulphate reduction, organotrophic or sulphide-dependent DNRA, and sulphide-dependentdenitrification making the deciphering of the different interactions complex. Therefore, otherfactors, such as inhibitors, could be at play in regulating rates of anammox, denitrification andDNRA, in addition to substrate availability.Oxygen also regulates and sometimes inhibits the occurrence of anammox, denitrification, andDNRA at different levels. It has been shown that anammox proceeds at O2 concentrations up to13.5µM [114], whereas denitrification proceeds at concentrations up to 20µM [115]. It has alsobeen reported that rates of DNRA in sediments increased with increasing concentrations of O2 inoverlying estuarine waters [96]. Similarly, nitrification was shown to have a very high affinity foroxygen, rendering the operation of nitrification possible under very low oxygen concentrations(km = 0.3 to 0.8 µM) [116]. This is generally unexpected and could indicate that many canonicallyanaerobic metabolisms operate under mildly oxygenated conditions, whereas some canonicallyaerobic metabolisms proceed at vanishingly low oxygen blurring the lines between geochemicalconditions and the corresponding energy driven metabolic cascade.Other factors also likely contributed to the regulation of anammox, denitrification and DNRAand their absolute and relative rates. For example, HS– appears to inhibit anammox withthresholds as low as 1.5µM HS– [114]. Temperature was tested as another factor influencing ratesof anammox and denitrification with the highest rates between 15 and 35◦C for both processes[110]. Salinity was also shown to influence certain processes such as DNRA and denitrification in23a oligohaline estuary, with higher salinity corresponding to high rates of DNRA and low rates ofdenitrification [117] Both temperature and salinity are likely to influence the physiology of themicroorganisms responsible and thus should also be considered as important regulating factors ofN-cycling.Overall, we need to expand the current knowledge about how substrate availability and thepresence of potential inhibitors control DNRA, anammox and denitrification across environments.Only a few studies, described above, have explored the kinetics of anammox, DNRA and denitrifi-cation, and a consensus has not always been found for these processes (i.e. HS– for denitrificationor NO–3 for DNRA), leaving kinetics for these processes poorly constrained. Further, it is essentialto link the environmental kinetic information with the individual microbial taxa associated withN-cycling in order to construct ecophysiologically-constrained biogeochemical models associatedwith specific community compositions, such as gene-centric modeling approaches [118, 119].Indeed, if these microbial taxa are found more broadly, this ecophysiological information can beextensible to other environments and used to constrain modeling efforts globally.24a.b.-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6log( aC/ aN)-560-550-540-530-520-510∆Greaction (kJ/mol N)150% N Denitrification100% N Denitrification50% N Denitrification25% N Denitrification150% N DNRA100% N DNRA50% N DNRA25% N DNRA-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6Log (aED/aTEA)-650-600-550-500-450-400-350∆Greaction (kJ/mol N) HS- into SO42- DNRAHS- into SO42- DenitrH2 DNRAH2 DenitrFigure 1.5: ∆Greaction in function of the logarithm of the activity (a) of the ED divided by the activity of the electronacceptor (NO–3 ) a) With organic ED and varied N-content b) With inorganic electron donors, HS– and H2251.5 Distributed metabolisms and the N-cycleThe N-cycle is composed of a set of reactions performed (Fig. 1.1) by a wide array of microorgan-isms (Fig. 1.2). The study of the genes and enzymes involved in the N-cycle is complex, as severalenzymes are sometimes capable of catalyzing the same reaction and the diversity of the microor-ganisms involved is broad. Most studies investigating microbial communties in the environmentoften offer an overview of the entire community, without deciphering which taxa are potentiallyinvolved in specific pathways (for example: [120]). Investigating which taxa are involded in themetabolic pathways of N-cycling, however, would facilitate the determination of ecophysiologicalinformation, and thus improve the specificity of models for N-cycling. Furthermore, knowingwhich taxa are involved will make this information generally extensible to other environmentswhere pathways of N-cycling are supported by similar key-players. Such studies have started toemerge, however, thanks to the advances in high-throughput sequencing and computing power.For example, Hawley et al. (2014)[33] describe the key-players involved in N-cycling for ananoxic fjord (Saanich Inlet, BC, Canada) based on meta’omic data (Fig. 1.2c) –metagenomic,metatranscriptomics and proteomics; this conceptual model highlighted the involvment of SUP05as a main player in partial denitrification, Ca. Scalindua (from the Planctomycetes phylum) asthe key-player for the anammox pathway, and finally the archaea Thaumarchaeota and bacteriaNitrospira sp.were implicated in nitrification. The information generated for the inlet is generallyextensible to OMZs as the same taxa can be found in OMZs as well (e.g. [18, 121]), and thus, ifrates and pathways of N-cycling are measured in the inlet, they can also be compared to ratesmeasured in other low oxygen zones with the same supporting microbial populations. Thus,Saanich Inlet can be considered as a model ecosystem for the metabolic activities found in OMZs.In this section, we highlight more generally the main genes and enzymes reported in the literaturefor N-fixation, nitrification, denitrification, DNRA and anammox.N-fixation is carried out by the enzyme nitrogenase. Three versions of nitrogenase exist,requiring different metal-cofactors such as Molybdenum (Mo), Iron (Fe) and Vanadium (Va) [122].There are thus 3 different genes of interest: nif, anf and vnf, requiring one of the three differentco-factors, respectively (Fig. 1.1 and table 1.3). In the modern ocean and in terrestrial settings,26Table 1.3: Main genes and enzymes involved in the N-cycle, based on Kuypers et al. (2018) [3] andreferences therein. The subunits for the genes are not specified here.Gene abbr. Enzyme Processes associated Reaction nif Mo-Nitrogenase N-fixation N2 + 8e– + 8H+ + 6ATP ⇒ 2NH3 +H2 +16ADP + 16Pi  anf Va-Nitrogenase N-fixation See above vnf Fe-Nitrogenase N-fixation See above amo Ammonia monooxygenase Nitrification NH4+ + O2 + 2e– + H+ ⇒ NH2OH + H2O  hao Hydroxylamine oxidoreductase Nitrification  NH2OH ⇒ NO + 3e– + 3H+  nxr Nitrite oxidoreductase Nitrification NO2– + H2O ⇒NO3– + 2e– + 2H+  nar/nap Cytoplasmic/ Periplasmic Nitrate reductase DNRN NO3– + 2e– + 2H+ ⇒ NO2– + H2O  nir* *Cu-Nitrite reductase Denitrification/PDNO/Anammox? NO2– + e– + 2H+ ⇒ NO + H2O  nir** **Assimilatory nitrite reductase Fermentative DNRA NO2–  + 6e– + 8H+ ⇒ NH4+ + 2 H2O  nor Nitric oxide reductase Denitrification/PDNO 2 NO + 2e– + 2H+ ⇒N2 O + H2O  nos Nitrous oxide reductase Complete denitrification N2O + 2e– + 2H+ ⇒N2 + H2O  nrf NADH-dependent nitrite reductase DNRA NO2–  + 6e– + 8H+ ⇒ NH4+ + 2 H2O  hzs Hydrazine synthase Anammox NO + NH4+ + 3e– + 2H+ ⇒ N2H4 + H2O  hzo/hdh Hydrazine oxidoreductase or hydrazine dehydrogenase Anammox N2H4 ⇒ N2 + 4e– + 4H+  otr Octaheme tetrathionate reductase DNRA? NO2– + 6e– + 8H+ ⇒ NH4+ + 2 H2O  onr Octaheme nitrite reductase DNRA? NO2– + 6e– + 8H+ ⇒ NH4+ + 2 H2O    Fe and Mo are often scarce, respectively, and is therefore a source of limitation for N-fixation[123]. Further, nitrogenases operate under anoxic conditions [124]. Therefore, microorganismsevolved mechanisms to protect the enzyme from oxygen with, for example, spatial separationof oxygenic photosynthesis and N-fixation in a nitrogenase-containing heterocyst or temporalseparation of both processes [124]. Nitrogenases likely evolved during the early proliferation oflife on Earth when N became limiting, and this is supported by the fact that it contains oxygen-sensitive co-factors that would have been widely available in the anoxic oceans of the Precambrianand thus became distributed across prokaryotes [30, 31] . Indeed, both archaea and bacteriapossess the ability to fix N. The genes for N-fixation have been found in both photosynthetic andnon-photosynthetic organisms, such as Trichodesmium spp. [124], UCYN-A in the oceans [125]and members of the Planctomycetes and Proteobacteria phyla [126], as well as members of theRhizobiales order in terrestrial settings [127]. In particular, the nif gene was found in 189 differenttaxa [128]. These microorganisms sometimes live in symbiosis with eukaryotes, providing the Nnecessary for their growth.Nitrification has historically been divided in to a two-step reaction (Fig. 1.1), with phyloge-27netically separated microorganisms performing the distinct steps of the process. The first stepinvolves the oxidation of NH+4 to hydroxylamine with ammonium oxidase enzyme (AMO), areaction that is endergonic (Table 1.3) [129]. The energy is then conserved through the oxidationof hydroxylamine to NO or directly to NO–2 using the octaheme hydroxylamine oxidoreductase(HAO) (Table 1.3). This step is conducted by either so-called ammonium-oxidizing bacteria (AOB),such as Beta- and Gammaproteobacteria, Nitrosomas, or Nitrospira, or by ammonium oxidizingarchaea (AOA) such as Thaumarcheaota [130], although the gene for the archaeal hao remainselusive, to date [131]. A recent discovery, however, showed that a Nitrospira sp. possess all thegenes necessary for the whole reaction and is able to perform the entire nitrification process [132].This complete nitrification was dubbed Comammox for complete ammonia oxidation [132]. Thisdiscovery confirms previous hypotheses based on the complete energetic yield (∆G◦ = -349kJ(mol NH+4 )– 1) which is greater than performing the 2 steps separately (∆G◦ = -275kJ (mol NH+4 )– 1and ∆G◦ = -74kJ (mol NO–2 )– 1). The second step of nitrification, NO–2 oxidation of NO–3 , isconducted using the enzyme nitrite oxidoreductase (NXR) (Table 1.3) [133]. This enzyme can befound across bacterial phyla, with members in Alpha-, Beta-, Gammaproteobacteria, Chloroflexi,Nitrospinae and Nitrospirae, in anoxygenic photosynthetic organisms such as Thioploca sp. KS1and in anammox bacteria (Kuypers et al. (2018) [3] and references therein).Similar to nitrification, denitrification is a multi-step reaction and it is usually distributed acrossmultiple phylogenetic groups, although certain microorganisms possess the suite of enzymesneeded to perform the entire process. NO–3 reduction to NO–2 is performed under low oxygenconditions where NO–3 is available, with either a periplasmic or membrane-bound enzyme (NAPor NAR Fig. 1.1 and Table 1.3) [134]. Many organisms perform only this step, such as themembers of the SAR11 clade, microorganisms that comprise up to half of the total microbial cellsfound in oxic marine waters and seem also to be ecologically relevant in OMZs [135]. For othermicroorganisms, like Parococcus denitrificans and Beggiatoa sp., this first step is followed by furtherreactions including NO–2 reduction to NO, N2O, or N2, or to NH+4 [134, 136]. NO–2 reduction toNO is conducted via nitrite reductases (NIR) that can either be heme-containing or Cu-containingenzymes [137]. The genes coding for these enzymes (nirK and nirS) are usually used as markergenes for canonical denitrifiers. However, they are present in many other organisms, and are28widespread in bacteria and archaea [138]. Nitric oxide reductase, an enzyme that catalyzesthe reduction of NO to N2O, refers to a suite of enzymes, from flavoproteins to haem copper-oxidases (NOR), that are distributed throughout the tree of life [3]. The enzymes are usedeither for detoxification of NO or for respiration and have special environmental relevance, asthey are responsible for the production of the greenhouse gas N2O. The final step for completedenitrification involves two versions of nitrous oxide reductases (NOS), one typical and anothercalled atypical found in soil bacteria [139]. Generally, the genes coding for the enzymes were foundin diverse bacterial phyla such as members of the Proteobacteria, Bacteroidetes and Chlorobi, aswell as archaeal phyla such as Crenarcheota and Halobacteria [3]. As denitrification is a multistepreaction, with each step distributed across multiple diverse taxa, the interactions between thesedifferent taxa can control the balance between sources and sinks of N2O in the environment.Dissimilatory nitrite reduction to ammonium (DNRA) is a fermentative or respiratory pathwaythat uses either a cytoplasmic nitrite reductase (NIR) [140] or a periplasmic cytochrome c nitritereductase (NRF) [141], respectively. The latter enzyme is the most studied and the most used as amarker gene for DNRA [141]. It has also been hypothesized that DNRA might also be conductedby octaheme nitrite reductase (ONR) or the octaheme tetrathionate reductase (OTR). Indeed, thesetwo enzymes (ONR and OTR) have been shown to be closely related to the cytochrome c nitritereductase as it has a similar active sites [142]. Hydroxylamine appears to be produced as anintermediate of the reaction when catalyzed by the cytochrome c enzyme, however, it does notaccumulate [143]. This process can be carried out by most bacterial lineages ( e.g. Bacteroidetes,Firmicutes, Proteobacteria, Planctomycetes - [144]), some archaea, diatoms and fungi, making thisreaction widespread, phylogenetically [3].The anammox reaction is biochemically challenging to conduct as it produces hydrazine (N2H4-rocket fuel) as an intermediate [61]. This intermediate molecule needs to be contained as it can behighly reactive and toxic. Anammox bacteria have thus evolved a specialized intracytoplasmiccompartment called the anammoxosome in order to enclose hydrazine [61, 145]. Due to this highlyspecific function, anammox appears to be confined to bacteria from 5 generas of the phylumPlanctomycetes [61, 62]. Three enzymes are involved in the multiple step reaction. An unknownnitrite reductase (NIR) transforms NO–2 to NO, similar to the enzyme in denitrification [146].29Then, a hydrazine synthase (HZS) combines NO and NH+4 into hydrazine [147]. Hydrazine issubsequently transformed to N2 using hydrazine dehydrogenase (HDH or HZO) [148]. Thus,anammox is a very specialized process with the use of hydrazine as an intermediate, and theconfinement of this process to one phylum only is highly uncommon in the N-cycle.Metabolic reactions of the N-cycle are widely distributed across the 3 domains and acrossmultiple phyla in each domain. Mainly, microorganisms involved in the N-cycle seem to havedeveloped specific niches and have crafted complex interactions between functional metabolicgroups in order to recycle N. Despite the complexity of the N-cycle apparent from existingknowledge, the true complexity may be much greater given that most of this knowledge comesfrom lab cultures, whereas the vast majority of microbial diversity remains uncultivated [3]. It isthus likely that we are missing much of the metabolic and taxonomic diversity connected to theN-cycle, and there is appreciable scope for the discovery of novel taxa, and perhaps genes andenzymes that catalyze N-species transformations.1.6 Problem statementModels of N-cycling fall short of making robust and explicit predictions of future N-cycling orreconstructions of the past. This is, in part, due to lack of constraints on the factors that regulatethe partitioning between denitrification, anammox and DNRA, as well as a lack of informationon the ecophysiology describing the relevant microorganisms. Indeed, with a small fraction ofmicroorganisms cultured to date, model parameters are mostly set with information from labcultures with limited extensibility to the environment. Direct ecophysiological information comesthrough process rate measurements, yet these generally lack coupled information on microbialcommunity composition, thereby limiting their extensibility across multiple environments. Fur-thermore, information from key environments, like those with ferruginous conditions similarto the Precambrian oceans, is almost entirely lacking. Quantitative information on dynamicsof rates and pathways of N-cycling that is accompanied by relevant information on microbialcommunity dynamics and ecophysiologies is thus needed across diverse environments to improvereconstructions of the N-cycle in the past and make better predictions of the N-cycle in the future.301.7 Dissertation overviewThe overall goal of my thesis is to generate new information on the rates and pathways of N-cycling under low-oxygen conditions that can be used to improve models of once and futureN-cycling. More specifically I aim to determine:i the rates and pathways of pelagic N-cycling under ferruginous conditions extensible to thePrecambrian oceans.I also aim to:ii quantitavely describe rates and pathways of N-cycling in modern anoxic marine environmentsand:iii investigate dynamics in the microbial community structure and their metabolic potential thatare relevant to N-cyclingThis information will be incorporated into:iv quantitative models that will enable reconstructions of past N-cycling and reproduction ofrates and pathways of modern N-cycling.These aims are achieved in the following chapters:Chapter 2: Iron-dependent nitrogen cycling in a ferruginous lake and the nutrient statusof Proterozoic oceansThis chapter elucidates (i) the rates and pathways of Fe-dependent NO–3 reduction in aferruginous pelagic environment. It then takes the in-situ process rates measurements and (iv)integrates these results in a box-model for the Proterozoic oceans to study how these processesimpact cycling of N and biological productivity during the Proterozoic Eon.Chapter 3: Rates and pathways of N2 production in sulphidic Saanich Inlet31Chapter 3 presents a detailed investigation of (ii) the environmental factors that influence thepartitioning of N-loss between anammox and denitrification for an anoxic and sulphidic fjord,Saanich Inlet (BC), with a year-long time-series of process rate measurements. (iv) A kineticmodel was also built to study the competition between anammox and complete denitrification forNO–2 based on the rates obtained in the study. Finally, (iii) the vertical and temporal changes inmicrobial community composition were shown to confirm the conceptual model of distributedmetabolism in SI previously built in Hawley et al. 2014.Chapter 4: Combining microbiological and geochemical information to constrain energyflow through the marine N-cycleIn chapter 4, I further address the partitioning of N-cycling through anammox, denitrifi-cation and DNRA in an anoxic fjord (Saanich Inlet, BC), and link the dynamics in anaerobicN-metabolisms to renewal in the inlet. We show that higher energy fluxes are coupled withhigher rates of DNRA and with changes to the microbial community structure and metabolicpotential. I thus combined here (ii) process rate measurements and (iii) metagenomic analysisof the microbial community composition and structure, as well as metabolic potential, in orderto study the changes in substrate supply rates in SI and the associated changes in the microbialcommunities and their metabolic activities.Chapter 5: ConclusionsThis chapter addresses the current and future challenges to the study of the N-cycle and thedistribution of the metabolisms involved, as well as to the integration of the newly producedknowledge into informative and quantitative models for the past and future N-cycle.32Chapter 2Iron-dependent nitrogen cycling in aferruginous lake and the nutrient statusof Proterozoic oceansNitrogen limitation during the Proterozoic has been inferred from the great expanse of oceananoxia under low-O2 atmospheres, which could have promoted NO–3 reduction to N2 and fixedN loss from the ocean. The deep oceans were Fe rich (ferruginous) during much of this time,yet the dynamics of N cycling under such conditions remain entirely conceptual, as analogueenvironments are rare today. Here we use incubation experiments to show that a modernferruginous basin, Kabuno Bay in East Africa, supports high rates of NO–3 reduction. Although60% of this NO–3 is reduced to N2 through canonical denitrification, a large fraction (40%) isreduced to NH+4 , leading to N retention rather than loss. We also find that NO–3 reduction isFe dependent, demonstrating that such reactions occur in natural ferruginous water columns.Numerical modelling of ferruginous upwelling systems, informed by our results from KabunoBay, demonstrates that NO–3 reduction to NH+4 could have enhanced biological production,fuelling sulfate reduction and the development of mid-water euxinia overlying ferruginous deepoceans. This reduction to NH+4 could also have partly offset a negative feedback on biologicalproduction that accompanies oxygenation of the surface ocean. Our results indicate that N lossin ferruginous upwelling systems may not have kept pace with global N fixation at marinephosphorous concentrations (0.04–0.13 µM) indicated by the rock record. We therefore suggestthat global marine biological production under ferruginous ocean conditions in the Proterozoiceon may thus have been P not N limited.332.1 IntroductionAs an element essential to life, nitrogen (N) often limits biological production [19]. N is madeavailable to life through microbial fixation of atmospheric N2. This N is liberated as NH+4 fromdecaying biomass, and oxidized to NO–3 in the presence of oxygen. N is returned to the atmospherethrough NO–3 reduction to N2 under low O2 conditions. Two microbial processes are responsiblefor N2 production; denitrification, which reduces NO–3 through a series of intermediates to N2,and anammox, which forms N2 by directly coupling NO2 with NH+4 . Organisms responsible fordenitrification and anammox proliferate in O2 minimum zones (OMZs) of the modern oceans,which support 20-40% of global fixed N loss to the atmosphere [63].Under the well-oxygenated modern atmosphere, OMZs (O2<20µM) comprise 7% by volumeof the global ocean [149], and their anoxic cores, which sustain most fixed N loss, occupy only0.1% [63]. During the Proterozoic eon, however, atmospheric O2 levels were lower than todayand vast regions of the ocean were anoxic [45]. Loss of fixed N is predicted under ocean anoxiaand such expansive anoxia could have led to extreme N limitation [39]. N isotope distributionsfrom Palaeoproterozoic upwelling systems, however, imply relatively little fixed N loss [42]. Thissuggests either modest rates of denitrification or N retention, possibly through reduction of NO–3to NH+4 [42]. Notably, anoxia and a supply of NO–3 will not support fixed N loss without electrondonors to drive denitrification, or NH+4 to support anammox. In the modern ocean, denitrificationis fuelled through both organic electron donors and H2S [18, 60]. Organic electron donors mayhave been scarce under the generally low productivity of Proterozoic oceans [39], and H2S wouldhave been scarce except during episodic euxinic periods that punctuate the Proterozoic geologicrecord [42, 47–49, 55]. Ferruginous conditions were much more prevalent than euxinia, dominatingocean chemistry throughout the Proterozoic [150]. Ferrous Fe (Fe(II)) is known to support NO–3reduction in laboratory experiments [58, 59] but the environmental operation, significance andpathways of Fe-dependent NO–3 reduction remain untested in natural ferruginous water columns.342.2 MethodsPhysico-chemical parameters as well as 15N-labelled incubations were performed during a sam-pling expedition to Kabuno Bay (Lake Kivu, East Africa – 1.58◦to 1.70◦S, 29.01◦to 29.09◦E) inFebruary 2012. In situ vertical conductivity - temperature - depth (CTD) profiles were collectedvia two multi-parameter probes (Hydrolab DS5, OTT Hydromet; and Sea&Sun CTD90, Sea andSun Technology). NO–2 and NH+4 concentrations were measured spectrophotometrically [151].Additionally, NO–3 concentrations were determined by subtracting NO–2 from the NOx (NO–3 andNO–2 ) measurements (via chemiluminescence [152]). Fe speciation was measured according toViollier et al. 2000 [153]. H2S and SO2 –4 concentrations were determined using the Cline method[151] and ion chromatography (Dionex), respectively. 15N-labelled incubations were performed induplicate in 12ml Exetainers, allowing water to be incubated under anoxic conditions. Microbialactivity was arrested with ZnCl2 at several time points for each experiment. The 15N – N2 and15N – NH+4 produced was quantified with isotopic ratio mass spectrometry. Rates of DNRA,denitrification and anammox were determined according to Thamdrup et al. 2006 [154]. Rateswere calculated on the basis of linear regressions with the least-squares method over the mostlinear data intervals (24 or 48h). The structure of the box model set-up here is the same as thatdeveloped in Canfield et al. 2006 [5] and adapted for the Proterozoic eon in Boyle et al. 2013 [44].Details on model parameters can be found in Appendix B.2.3 Results and discussionKabuno Bay (KB) is a ferruginous sub-basin of Lake Kivu, which straddles the border of Rwandaand the Democratic Republic of Congo, East Africa [50]. Saline springs feed KB causing permanentstratification, anoxia below 10m (Fig. 2.1 a and b), and Fe(II)-rich deep waters (500 µM – Fig. 2.1d).Such ferruginous conditions are analogous to those that prevailed through much of the Proterozoiceon [50]. A strong gradient of NH+4 between 10 and 11.5m depth (Fig. 2.1c) indicates high ratesof NH+4 oxidation to NO–3 and NO2 within this depth interval. Since KB’s oxic surface watersare devoid of NO–3 and NO–2 (concentrations<1µM), NO–3 and NO2 produced through NH+4oxidation are advected to the main basin, assimilated, or rapidly reduced.35a. b. c. d.0.0 0.5 1.0 1.5NOx (µM)051015Depth (m)0 50 100 150 200 250NH4+ (µM)NOx (µM)NH4+ (µM)0 100 200 300 400 500 600Fe(II)aq (µM)0.0 40.0 80.0Fe (II) and (III) particulate (µM)Fe(II)aq (µM)Fe(III) part (µM)Fe(II) part (µM)0 50 100 150 200 250DO (µM)051015Depth (m)22 23 24 25Temperature (°C)DO (µM)Temp (°C)1,000 3,000 5,000Specific Conductivity (µS/cm)0510156 7 8pH SpCond (µS/cm)pH0 1 2 3 4 5 6 7 8H2S (µM)250 300 350 400 450 500 550SO42- (µM)H2S (µM)SO42- (µM)0 5 10 15 20Turbidity (NTUs)0510150. 0.5 1.0 1.5NOx (µM)0510150 50 100 150 200 250NH4+ (µM)NOx (µM)NH4+ (µM)0 100 200 300 400 500 600Fe(II)aq (µM)0510150 40 80Fe (II) and (III) particulate (µM)Fe(II)aq (µM)Fe(II) part (µM)Fe(III) part (µM)0 50 100 150 200 250DO (µM)051015Depth (m)22 23 24 25Temperature (°C)DO (µM)Temp (°C)1,000 3,000 5,000Specific Conductivity (µS/cm)0510156 7 8pH SpCond (µS/cm)pH0 1 2 3 4 5 6 7 8H2S (µM)250 300 350 400 450 500 550SO42- (µM)H2S (µM)SO42- (µM)0 5 10 15 20Turbidity (NTUs)051015Figure 2.1: Vertical distribution of selected physical and chemical properties of Kabuno Bay for February 2012. a)Dissolved O2 (DO) concentration (µM) and temperature (◦C). b) pH and specific conductivity (SpCond, µScm– 1). c) NH+4 and NOx concentration (µM). d) Fe(II)aq, Fe(II)part and Fe(III)part concentrations (µM)We determined rates and pathways of microbial N transfomations in KB using incubationswith 15N-labelled NO–3 . Both denitrification and dissimilatory nitrate reduction to ammonium(DNRA) occur between 11 and 11.5 m, but anammox was below our limit of detection (6nmolN l– 1 d– 1). Rates of denitrification and DNRA were up to 80 ± 10 and 50 ± 10nmol N l– 1 d– 1,respectively (Fig. 2.2a,b and Fig. B.3), exceeding those typically found in marine OMZs [60, 67] butsimilar to coastal marine anoxic basins such as the Baltic Sea [56]. While 60% of NO–3 reduced islost from the KB through denitrification, 40% is retained as NH+4 through DNRA. Substantial NO–3recycling to NH+4 has also been periodically observed in the Peruvian and Omani OMZs [65, 67],but such a high fraction appears to be unusual for modern pelagic marine environments [60, 98].Our observations imply that some biogeochemical feature of KB favours DNRA compared withother environments studied to date. Fe(II), which is unusually abundant in KB, indeed promotesDNRA in estuarine sediments [155], and may also do so in KB.To test for such Fe dependency, we amended a subset of our 15N incubations with 40 µM Fe(II).We found that Fe(II) addition considerably enhanced both denitrification and DNRA to 230 ± 40360 10 20 30 40 50Time (hr)05010015015NH4+ production (nM) 11m11m, Fe11.5m11.5m, Fe0 10 20 30 40 50Time (hr)05010015020025030035015N 2 production (nM) 11m11m, Fe11.5m11.5m, Fea.b.9.510.010.511.011.5Depth (m)0 50 100 150 200 250Denitrification (nmol N L-1 d-1)No Fe addedFe added9.510.010.511.011.5Depth (m)0 50 100 150 200 250DNRA (nmol N L-1 d-1)Figure 2.2: Rates and pathways in Kabuno Bay for February 2012 a&b Denitrification (a) and DNRA (b) inKabuno Bay water column. Samples were collected in February 2012, with (in orange) or without (in blue)addition of the electron donor Fe(II) to the incubations. Insets show time-course evolution of 15N-labelledmetabolic products. Rates were calculated on the basis of linear regressions over the linear data intervals(24 or 48h). Table B.3 contains the detailed rates and associated errors. The error on the rate is the standarderror of the slope for the linear regression.and 70 ± 20nmol N l– 1 d– 1, respectively (Fig. 2.2a,b and Table B.3), suggesting a role for Fe(II)in NO–3 reduction. Our results support measurements from estuarine sediments, which invokemicrobial mediation [155], but the nearly equivalent stimulation between both NO–3 reductionto NH+4 and denitrification provides no evidence that Fe(II) favours DNRA and instead mayindicate that Fe(II) enhances the reduction of an intermediate (for example, NO–2 ) common toboth reactions.Thermodynamic considerations reveal that reduction of NO–3 , and a number ofintermediate N species, by Fe(II) is energetically favourable in KB (see Appendix B) yieldingsufficient free energy for microbial growth. While the precise pathway remains unresolved, Fe(II)clearly plays a role in NO–3 reduction in KB.To assess the biogeochemical role of Fe-dependent NO–3 reduction in KB, we compared rates37of NO–3 reduction with other key processes [50]. While Fe(II) supports NO–3 reduction, thecorresponding Fe(II) oxidation rates of 1,700nmol Fe l– 1 d– 1 (based on stoichiometry) are only aminor fraction (1%) of the observed phototrophic Fe(II) oxidation in the KB chemocline [50]. Bycomparison, NO–3 reduction rates are an order of magnitude lower than SO2 –4 reduction rates,which are up to 410nmol S l– 1 d– 1 [50]. We also compared rates of NO–3 reduction with darkcarbon fixation, and on the basis of growth yields for chemoautotrophic NO–3 reduction (seeAppendix B), this comparison suggests that NO–3 -driven chemoautotrophy could support up to2% of the total dark carbon fixation in KB’s water column [50]. The overall contribution of NO–3reduction to biogeochemical cycling, therefore, is largely to regulate recycling and loss of fixed Nfrom KB, and here, the partitioning of NO–3 reduction between DNRA and denitrification is key.We have shown that NO–3 reduction both to N2 and NH+4 takes place at relatively high ratesunder ferruginous conditions, and further, that this NO–3 reduction is partly coupled to theoxidation of Fe(II). By extension, the ferruginous oceans of the Proterozoic eon could also havesupported large-scale NO–3 reduction, possibly through both denitrification and DNRA, and withFe(II) as the electron donor [39, 156]. To quantitatively link our observations in KB to possiblebiogeochemical cycling under ancient marine ferruginous conditions, we set up a box modelfor N cycling in ocean upwelling systems [44, 156]. Our model describes mass balances for C,N, S, O and Fe species and their biogeochemical reactions (see Fig. 2.3a and full description inAppendix B). NH+4 and Fe were supplied through upwelling, and these, along with nutrientrecycling, ultimately controlled primary production in the overlying surface waters. Productionin the surface waters is sustained exclusively through upwelled N with no productivity by localN fixation. Settling of organic matter generated through primary production drives respirationand nutrient recycling in intermediate waters. Chemotrophic processes such as nitrification,and Fe-dependent denitrification and DNRA were included (Fig. 2.3a). Our model is basedon previous studies[44, 156], but we considered a ferruginous system where N cycling wasdriven first by Fe-dependent NO–3 reduction, with the remaining NO–3 reduced by organic matteroriginating through primary production (Fig. 2.3a). We also included DNRA, in accordance withour results from KB, to evaluate its impact on coupled C, N, Fe and S cycling under ferruginousconditions. Without explicit constraints on the fraction of NO–3 reduced to NH+4 versus N2 in38ferruginous oceans, we varied its contribution from 0 to 40%, with the balance occurring throughdenitrification.Deep-ocean Fe(II) and NH+4 concentrations throughout the Proterozoic eon are uncertain.If global N-fixation is limited by phosphorous supply according to the Redfield ratio (16N:1P;[157]), we can set deep-ocean NH+4 concentrations in our model at 16 times the phosphorousconcentration (0.04–0.13 µM) of Proterozoic seawater [158] (see Appendix B), which yields up to2 µM NH+4 . To validate this assumption, we excluded DNRA and ran our model with differentratios of N/P in deep waters. When deep ocean NH+4 is greater than 13 µM, sulfidic conditionsdevelop without DNRA or N fixation under all reasonable upwelling rates [156] (Fig. 2.3e). Thus,if deep-ocean NH+4 concentrations were more than 13 µM, sulfidic conditions would have beenwidespread during the Proterozoic eon. Such widespread euxinia is not supported by the geologicrecord [150], indicating that NH+4 concentrations were generally less than 13 µM in the deep ocean.We thus chose 2 µM NH+4 as the benchmark for our modelling, but also explored a concentrationrange from 0.6 through to 13 µM. In line with considerations for both siderite solubility [46]and nutrient dynamics that permit marine oxygenic photosynthesis [158] we chose 42 µM as ourbenchmark Fe(II) concentration (see Appendix B). We also considered a broader range of Fe(II)concentrations (see Appendix B), which may be possible if siderite formation was kineticallyinhibited [159].Our model shows that in ferruginous upwelling systems the balance between DNRA and deni-trification strongly influences coupled C, N, S and Fe cycling with enhanced primary productionwhen DNRA is an appreciable NO–3 reduction pathway. Indeed, when 40% of NO–3 reductionis channelled through DNRA, primary production rates increase by up to 170% (Fig. 2.3b,c). Anotable effect of this enhanced primary production is greatly increased H2S production. Thisoccurs even in the absence of N fixation as DNRA provides the nitrogen to stimulate additionalorganic matter production, which ultimately fuels microbial sulfate reduction (see Appendix B).At 40% DNRA, strong upwelling leads to sulfate reduction and pyrite deposition at rates sufficientto yield sediment iron speciation (Fepy/FeHR) that indicates possible water column euxinia. SuchFepy/FeHR values (>0.7) exist in Proterozoic sedimentary rocks [42, 47–49, 55], which could thusrecord a contribution of DNRA to NO–3 reduction at this time.39 b. d. c.Primary ProductionEuphotic zoneOrganic MatterUpwelling zoneO2 respiration + NitrificationNO3- ReductionSO42- ReductionNO3-NH4+NH4+Bottom watersDNRAFe2+Bottom watersH2SN2Denitrificationy1-y a. e.1 3 5 7 9 11 13 15NH4+ concentration in deep ocean (µM) HRUpwelling 1 cm hr-1Upwelling 2 cm hr-1Upwelling 3 cm hr-1Possible euxinia0.5 1.0 1.5 2.0 2.5 3.0Upwelling rate (cm hr-1) Fe(II) (µM)05101520253035Export production (nmol C cm-2 hr-1)NH4+Fe(II)NH4+ UM_1Fe_1Export productionEP_10.5 1.0 1.5 2.0 2.5 3.0Upwelling rate (cm hr-1) Fe(II) (µM)05101520253035Export production (nmol C cm-2 hr-1)0.5 1.5 2.5Upwelling rate (cm hr-1) HR0.5 1.5 2.5Upwelling rate (cm hr-1) HR0 2 4 6 8 10 12%PAL0102030405060708090100110120130140150Export production (103 nmol C yr-1 cm-2)0100200300400% of increased EP through 40% DNRAEP with 0% DNRAEP with 40% DNRA% of increased EP through 40% DNRAFigure 2.3: Model outputs describing coupled C, N, S and Fe cycling in an idealized Proterozoic upwelling system.a) Model structure illustrating reactions included and their reactants and products. b&c) Model runs with0% (b) and 40% NO–3 reduction (c) through DNRA. The solid lines represent model runs with the surfacewater oxygen concentrations of 3.8% PAL, whereas the dashed lines represent runs at 6.2% PAL (blue,export production; orange, NH+4 concentrations in the upwelling zone; black, Fe(II) concentrations in theupwelling zone; the insets show the Fe pyrite to highly reactive Fe ratio (Fepy/FeHR) where the grey linedelineates plausible euxinic conditions [150]). d, Yearly export production (EP) for 0% and 40% DNRA for arange of surface waters oxygen concentrations (from 0 to 12% PAL) at an upwelling rate of 2 cm h– 1. At 0%PAL, nitrification is not present in our model; however, NO–3 is still supplied from the intermediate watersthrough advection and diffusion, therefore feeding NO–3 reduction through DNRA and denitrification (seeAppendix B). e, Fepy/FeHR ratios for a range of deep-ocean NH+4 concentrations at 0% DNRA for threedifferent upwelling rates at 3.8% PAL (the grey dashed line delineates plausible euxinic conditions[150]).See Appendix B for details.40Our model also reveals a negative feedback between primary production and surface oceanoxygen. This negative feedback develops when N loss increases in response to enhanced NO–3supply due to stimulation of nitrification by O2. An increase from 3.8% to 6.2% PAL (presentatmospheric level) O2, values thought possible for the Mesoproterozoic [160] (values as low as0.1% PAL have been proposed [161]), reduces primary production by up to 20% when NO–3reduction occurs exclusively through denitrification. This effect is muted by DNRA (Fig. 2.3b andc), which can play an increasingly important role in supplying N for primary production with theprogressive oxygenation of the surface ocean (Fig. 2.3d).It is widely assumed that the Proterozoic oceans were N limited due to massive N loss [39].While the euphotic waters directly overlying upwelling systems can be locally N–limited dueto N–loss from below, the global expression of N limitation ultimately depends on the balancebetween the geographic expansiveness of upwelling systems, and ocean-wide N fixation. Weextrapolated N loss from our model to an area equivalent to upwelling regions in the modernocean (0.36 10 12 m2 – [162]) yielding a modelled global N-loss from Proterozoic oceans of upto 1.6 Tg N yr– 1 (see Appendix B). By comparison, Proterozoic phosphorous concentrations[158] could have supported 4.8 Tg of N fixation per year based on an equivalent ratio of N fixedto deep ocean phosphorous as in the modern ocean [157] (see Appendix B). It has also beenproposed that molybdenum (Mo) limited N fixation due to its scavenging from seawater as sulfideminerals [163]. Mo limitation seems unlikely, as Mo scavenging from seawater generally requiresstrong euxinia (see Appendix B), which as we show here, would not likely have developed inProterozoic upwelling systems. To balance global N fixation with N loss in the Proterozoic eon,an upwelling area three times that of the modern ocean would have been needed. This suggeststhat N limitation in the Proterozoic was unlikely and that productivity would, instead, have beenlimited by phosphorous. Our modelling results are well supported by the Palaeoproterozoic rockrecord [42], which implies upwelling systems with euxinic conditions (possibly supported byDNRA) that induce little fixed N loss (also possibly the result of DNRA). The operation of theseprocesses throughout the Proterozoic eon can be further tested through an expansion of the Nisotope record, and through simulations in global biogeochemical models informed by our data.41Chapter 3Rates and pathways of N2 production insulphidic Saanich InletMarine oxygen minimum zones (OMZs) support 30-50% of global fixed-nitrogen (N) loss butcomprise only 7% of total ocean volume. This N-loss is driven by canonical denitrification andanaerobic ammonium oxidation (anammox), and the distribution and activity of these two pro-cesses vary greatly in space and time. Factors that regulate N-loss processes are complex, includingorganic matter availability, oxygen concentrations, and NO–2 and NH+4 concentrations. While bothdenitrification and anammox produce N2, the overall geochemical outcome of these processes aredifferent, as incomplete denitrification, for example, produces N2O, which is a potent greenhousegas. Information on rates of anammox and denitrification and more detailed ecophysiologicalknowledge of the microorganisms catalyzing these processes are needed to develop more robustmodels of N-loss in OMZs. To this end, we conducted monthly incubations with 15N-labeled Nunder anoxic conditions and during a deep-water renewal cycle in Saanich Inlet, British Columbia,a persistently anoxic fjord. Both denitrification and anammox operated throughout the low oxygenwater column with depth integrated rates of anammox and denitrification ranging from 0.15±0.03to 3.4±0.3 and 0.02±0.006 to 14±2 mmol N2 m– 2 d– 1, respectively. Most N2 production in SaanichInlet was driven by denitrification, with high rates developing in response to enhanced substratesupply from deep water renewal. Dynamics in rates of denitrification were linked to shifts inmicrobial community composition. Notably, periods of intense denitrification were accompaniedby blooms in an Arcobacter population against a background community dominated by SUP05and Marinimicrobia. Rates of N2 production through denitrification and anammox, and theirdynamics, were then explored through flux-balance modeling with higher rates of denitrification42linked to the physiology of substrate uptake. Overall, both denitrification and anammox operatedthroughout the year, contributing to an annual N-loss of 2 · 10 – 3 Tg N2 yr– 1, 37% of which weattribute to anammox and 63% to complete denitrification. Extrapolating these rates from SaanichInlet to all similar coastal inlets in BC (2478 km2), we estimate that these inlets contribute 0.1% toglobal pelagic N-loss.3.1 IntroductionNitrogen (N) is an essential element to life and it is used as a building block for proteins andnucleic acids in all terrestrial and marine organisms. The bioavailability of N, therefore, can limitprimary production in both terrestrial and aquatic compartments of the biosphere [8, 32]. Thelargest pool of N at the Earths surface is N2 in the atmosphere and this N2 is made available tolife mostly through energetically expensive microbial N-fixation [2]. The abundance of fixed-N inthe oceans is governed by the balance between this N-fixation into biomass, biomass depositionand ultimate burial in marine sediments, and the return of fixed-N to the atmosphere through asuite of redox reactions that ultimately lead to anaerobic N2 production [9]. The processes thatcomprise the N-cycle are spatially decoupled with most N-fixation occurring in the euphoticsurface ocean [164], the oxidative components distributed throughout much of the ocean, andanaerobic N2 production partitioned between the low oxygen waters (30-50%) that typicallydevelop at intermediate water depths and in eutrophic coastal regions, as well as in bottomsediments (50-70%) [165]. The availability of N to marine life, therefore, depends on the relativerates of N-fixation versus N-loss, and N-loss is expected to scale with the extent and intensityof low oxygen marine waters, which are currently expanding with unconstrained feedbacks onmarine N inventories [72, 166, 167].Under low oxygen conditions (<20 µM O2 concentration), NO–3 is used as an electron acceptorin anaerobic microbial energy transduction leading, in part, to N2 production and closure ofthe N- cycle. Such low oxygen conditions commonly develop in the open ocean at intermediatewater depths, in restricted basins, and in eutrophic coastal regions, when respiratory oxygenconsumption outpaces physical mixing and oxygenic photosynthesis. Low oxygen marine waters43are commonly referred to as Oxygen Minimum Zones (OMZs), and are pervasive features of themodern oceans comprising 7% of their total volume (with O2 < 20µM) [149]. The anoxic coresof OMZs, which contain oxygen concentrations below the limit of detection of oxygen sensorsgenerally used in oceanographic research (<5µM, but as low as 1nM), constitute only 0.1% of theoceans total volume [63, 121]. Despite their relatively low volumes, OMZs play an outsized role inN biogeochemistry sustaining up to 50% of global marine fixed N-loss with an annual N-sink of150 Tg of N [165].N2 production and thus N-loss in OMZs is driven by two entirely different microbialmetabolisms: canonical denitrification and anaerobic ammonium oxidation (anammox). Indenitrification a suite of either inorganic [sulphide (HS– ), ferrous iron (Fe(II))] or organic electrondonors is used to reduce NO–3 through a series of intermediates; NO–2 , NO, and N2O to ultimatelyproduce N2 (see Eqs. 3.1 and 3.2).Organotrotrophic denitrification [168]:84.8NO−3 + (CH2O)106(NH3)16H3PO4⇒ 42.4N2 + 106HCO−3 + 42.4H2O+HPO2−4 + 16NH+4 + 7.2H+(3.1)Chemotrophic denitrification: 8NO−3 + 5HS− + 3H+ ⇒ 4N2 + 5SO2−4 + 4H2O (3.2)Anammox directly couples NO–2 reduction to the oxidation of NH+4 through hydroxylamineand hydrazine intermediates to produce N2 (Eq. 3.3).Anammox: NO−2 +NH+4 ⇒ N2 + 2H2O (3.3)Both pathways are fuelled, in part, by relatively oxidized N species and yield N2 as theirultimate metabolic products, and they thus occupy overlapping niches. Denitrification andanammox, however, diverge in both their ecophysiology and their biogeochemical outcomesincluding possible leakages of intermediate N species and their overall influence on the carbon (C)cycling [169]. Denitrification, for example, can either consume or produce CO2 depending on the44electron donor used, as denitrifiers can be heterotrophic or autotrophic. Anammox, on the otherhand, is considered exclusively autotrophic and only consumes CO2. Denitrification, furthermore,yields N2O as an intermediate, a potent greenhouse gas, that may accumulate during partialdenitrification and play a role in global climate forcing [170]. The differing ecophysiologies of theorganisms conducting denitrification and anammox are thus expected to interact with one anotherin different ways across a spectrum of anaerobic conditions. These differences confound attemptsto model N-cycle dynamics and its interactions with other cycles, without explicit descriptions forboth anammox and denitrification and their regulation.Process rate measurements are beginning to define the relationships between anammox, den-trification and the N-cycle. In OMZs globally, anammox appears to support most N2 production[18, 60, 65–67, 84, 98, 154], while denitrification may dominate ephemerally [69, 105, 171, 172]. Inopen ocean OMZs, the relative contributions of anammox and denitrification to N2 productionare theoretically constrained by the stoichiometry of settling organic matter and the NH+4 supplyfrom remineralization of organic matter to anammox [63, 105, 173–175]. This constraint developswhen anammox is limited by NH+4 supplied through ammonification of organic N during het-erotrophic NO–3 respiration (organotrophic denitrification). In this case, N2 production shouldoccur 71% through denitrification and 29% through anammox based on Redfieldian organic matterstoichiometry of 106C:16N (Eq. 3.4) [100, 173, 175].Coupling of organotrophic denitrification and anammox in open ocean OMZ:94.4NO−3 + (CH2O)106(NH3)16H3PO4⇒ 55.2N2 + 106HCO−3 + 71.2H2O+HPO2−4 + 13.6H+(3.4)Here, denitrification produces 16 moles of NH+4 and 16 moles of NO–2 that fuels anammox,producing 16 moles out of a total 55.2 moles N2, hence 29% of the total N2 production (Eqs. 3.5453.6).Denitrification products fuel anammox entirely through organic matter degradation:94.4NO−3 + (CH2O)106(NH3)16H3PO4⇒ 39.2N2 + 106HCO−3 + 39.2H2O+HPO2−4 + 16NH+4 + 16NO−2 + 13.6H+(3.5)Anammox consumes NH+4 and NO–2 from denitrification:16NH+4 + 16NO−2 ⇒ 16N2(3.6)This expected ratio between anammox and denitrification, however, is rarely observed in theocean and deviations from the ratio can be at least partially explained through variability inorganic matter composition and departures from the Redfield ratio [105]. While this stoichiometricvariability appears to account for differences observed in the role of anammox in N2 production inopen ocean OMZs, it remains unclear to what extent organic matter stoichiometry can explain theapparently outsized role of anammox in global N2 production, more generally. Denitrification iscommonly undetected in many OMZs, and this then raises the question as to what supplies NH+4to anammox when denitrification appears absent. One possibility is microbial NO–3 reduction toNH+4 (DNRA), which has been detected in the Peruvian OMZ and above the Omani shelf, andcould be partially responsible for directly supplying NH+4 to anammox [65, 67]. While DNRAcould provide NH+4 for anammox in some cases, it is unlikely the universal source as the rates ofDNRA measured are generally insufficient to fully support the NH+4 requirements of concurrentanammox [67]. Other possible sources of NH+4 include remineralization of organic matter throughNO–3 reduction to NO–2 , microaerobic respiration, and sulphate reduction [18], and in certainshallow settings, benthic release of NH+4 [60, 67]. While we are gaining a clearer picture of thecontrols on rates of anammox and denitrification in OMZs, there remain no universal rules thatallow quantitative prediction of the partitioning between these two pathways.Beyond observations from canonical OMZs, anammox and denitrification have been reportedfrom other anoxic environments, including marine sediments, anoxic fjords and lakes, and46wastewater treatment facilities. In marine sediments for example, it has been shown that therelative contribution of anammox to N2 production increases dramatically with both distancefrom the coast and water depth [83, 90] with anammox comprising up to 80% of the total N-lossat 700m depth [176]. This trend may be attributed to decreased organic matter content in deepersediments [90]. Indeed, availability of organic matter, rather than its reactivity or quality, appearsto regulate the relative importance of denitrification and anammox in estuarine sediments [112].The relative contribution of anammox to sediment N2 production also appears to increase whenNO–3 concentrations are persistently high in overlying waters [112, 177]. Notably, in sedimentsunderlying low oxygen marine waters, nearly all N2 production was supported by anammox [64].In these sediments, NH+4 was supplied to anammox through DNRA. This implies then that therelative importance of anammox to sediment N2 production may in part depend on the activityof DNRA. HS– may also play a role in regulating anammox and denitrification. While HS– isa common electron donor and thus a suitable substrate for denitrification, it has been shownto inhibit anammox at micromolar levels, possibly through toxicity [56]. This is consistent withthe distribution of anammox, which appears to operate above the sulphidic zone in the BlackSea [114]. Likewise, anammox contributes up to 30% of the N2 production in lacustrine watercolumns, but the highest rates of anammox occur in nearly HS– free waters [107]. In contrast,anammox appears entirely excluded from very iron-rich lake waters and sediments [111, 178],and ferruginous estuarine sediments [97, 155]. Taken together, the emerging picture suggests thatthe regulation of the relative importance of anammox and denitrification to total N2 production isconvoluted and development of predictive knowledge will require comprehensive and detailedstudies across the broad range of systems where these processes are known to operate.We have conducted a time-series study of the rates of denitrification and anammox and theirrelative contribution to N2 production in Saanich Inlet (SI). SI is a persistently anoxic fjord thatprovides a tractable ecosystem in which to study anaerobic microbial metabolisms relevant andextensible to low oxygen environments globally (Fig. 3.1 a and b). The choice to use the word”persistent” for SI is recent, however, as only partial renewals have been recorded between 2014and 2019. Part of the water column therefore remains anoxic throughout the year. Biogeochemicalresearch has been conducted in SI since 1965 [179] and has culminated with instrumented real-time47monitoring and a more than 10 year continuous time-series experiment [180–183], making it oneof the best studied anoxic fjords on Earth. The inlet is situated on the southern tip of VancouverIsland (Fig. 3.1a) and is up to 228m deep with a 75m deep sill at its entrance that restrictshydrological connection to the Strait of Georgia and the mixing of waters in its deep basin. Similarto OMZs, aerobic respiration in SI water column outpaces O2 supply through physical watermixing and photosynthesis in the surface waters, rendering low oxygen conditions for most of theyear below 100m for most of the year (Fig. 3.1b). In contrast to most open ocean OMZs, however,sulphidic conditions develop in the bottom waters of SI as a result of either sulphate reductionin the water column [184] and/or sulphide efflux from the underlying sediments [185]. Mostyears, SI stagnant deep waters transition from sulphidic to oxic at the end of the summer (lateAugust - early September) in response to upwelling off the coast of Vancouver Island that forcesdense well-oxygenated waters into the Strait of Georgia and over the sill into the inlet [184], inconnection to weak tidal currents [186]. The inlet thus exists in two main states during the year ifrenewal occurs: a state of stagnation referring to low oxygen concentrations in the deep-watersand a state of renewal when oxygenated waters penetrate the inlet and mix with low oxygendeep-waters. These physical-chemical characteristics combine to support microbial communitieswith anaerobic metabolisms that couple the C, N and S-cycles and are broadly analogous to thosewe expect to find in other low oxygen and anoxic marine waters globally [104].48NO3-H2SO2NH4+S3a. b. c.DenitrificationOMNH4+N2N2ONO3-NO2-N2O?NH4+?NitrificationAnammoxAmmonificationSUP05PlanctomycetesMarinimicrobiaNitrospira/NitrospinaThaumarchaeotaSAR11Figure 3.1: Saanich Inlet (SI), a model ecosystem for the study of microbial metabolisms in OMZs a) Sampling ofStation S3 in Saanich Inlet, on Vancouver Island, British Columbia (Canada). Sampling of S3 happenedonce a month within a historic time series data collection. b) Typical redox gradients found in SI. Thesegradients move vertically depending on the season. c) Active microbial metabolisms of the N-cycle presentalong the redox gradients of SI.N-cycling and its interactions with the other cycles in SI have been previously interrogatedusing a variety of geochemical and microbiological analyses. Geochemical data indirectly implythat SI supports relatively high rates of both pelagic and benthic N-loss that vary seasonally,with the highest rates in the winter (December to February, 8.1 mmol m– 2 d– 1) and lowest in thesummer (May to August, 1.7 mmol m– 2 d– 1) [186]. Multi-omic analyses revealed that microbialcommunities in SI harbour the metabolic potential to catalyze many components of the N-cycleand to link it to cycling of C and S [33, 182, 187]. These metabolic pathway reconstructions haveled to a conceptual model describing the microbial interactions that underpin N-cycling in SI, and49low oxygen waters more broadly (Fig. 3.1c). Specifically, this model (reproduced in Fig. 3.1c from[33]) proposes that Thaumarchaeota are responsible for the first step of nitrification (NH+4 to NO–2 )and that two different species of bacteria, Nitrospina gracilis and Nitrospira defluvii, oxidize NO–2 toNO–3 . Along with nitrification, the SAR11 are the most abundant aerobic heterotrophs, and theyare thought to degrade settling organic matter and release NH+4 to the oxic water column (Fig.3.1c). Lower in the water column, the model suggests that Planctomycetes produce N2 through theanammox process, while bacteria from the SUP05 clade (Gammaproteobacteria) were implicatedin reducing NO–3 to N2O (Fig. 3.1c). The final step of denitrification remained more elusive butanalyses of Single Cell Amplified Genomes (SAGs) reveal metabolic potential for N2O reductionto N2 in the Marinimicrobia ZA3312c-A and SHBH1141 (previously known as Marine Group-A)[187]. Notably, the taxonomic affiliations and genomic make-up of the key organisms that driveN-cycling in SI are closely related to those found across OMZs and other anoxic environmentsglobally [104]. For example, the Gammaproteobacteria SUP05 – with a single cultivated member,Ca. T. autotrophicus strain EF1 [188] – appears to be a ubiquitous member of OMZ microbialcommunities with the metabolic potential for partial denitrification [33, 120, 182], along withbacteria from the group Marinimicrobia that reduce N2O to N2 [187] and are some of the mostwidely distributed and abundant taxa in marine OMZs.The conceptual metabolic model for coupled C, N, and S cycling in OMZs was furtherexpanded into a quantitative gene-centric model that integrates metabolic potential derived frommulti-omic information with geochemical data to predict process rates [118]. Modeled rates werevalidated through direct measurements, but these rates were an order of magnitude lower thanthe rates needed to support previous geochemical data [186]. These observations highlight adiscontinuity between current conceptual and quantitative models of the N-cycle in SI and a needfor data that more fully capture and integrate the dynamics of N-cycling across multiple seasons.Here, we used isotope labeling experiments to directly quantify rates and pathways of anaero-bic N cycling in SI over an entire year. These measurements allowed us to calculate annual N-lossfrom Saanich inlet, determine the specific microbial pathways that are responsible, and to assessthe biogeochemical controls on the rates and pathways of N-loss in the inlet. Overall, our datareveal that fixed N-loss from SI has strong seasonality and that periods of intense N-loss during50the summer are driven primarily through sulphide-dependent denitrification, which is likelyfuelled by benthic sulphide supply and new input of NO–3 from a partial renewal of the watercolumn. Anammox also contributed to N-loss at relatively constant rates throughout the year.3.2 Methods3.2.1 Study site and samplingSaanich Inlet (SI) is a marine fjord located on the West Coast of Vancouver Island, British Columbia,Canada (Fig. 3.1). We conducted a monthly time series experiment between January and December2015 (Table 3.1) and sampled at station S3 (Fig. 3.1a 48 ◦ 35.5 N and 123◦ 30.3 W, 227m deep).A standard profile of 16 depths was sampled every month with 12L GO-FLO bottles attachedin-series to a steel cable (10, 20, 40, 60, 75, 85, 90, 97, 100, 110, 120, 135, 150, 165, 185 and 200m).Depths were set using a metered winch cable with a precision of plus or minus 0.5m and theaccuracy of the depth reached was checked with the CTD depth profile. CTD profiles [pressure(SBE 29), conductivity (SBE 4C), temperature (SBE 3F), and oxygen (SBE 43)] were obtained withthe SBE25 Sealogger CTD (SBE). Oxygen concentrations measured with the SBE 43 sensor werecalibrated monthly against Winkler titrations [151] and its limit of detection is <1µM. The CTD,attached at the end of the winch cable, and the bottles were lowered to their final depths and leftthere to equilibrate with surrounding water for at least a minute before closing.Samples for nutrient concentration measurements were immediately filtered and put on icefor later analysis. Samples for sulfide analyses were fixed in 0.5% Zinc Acetate final concentrationwithout prior filtration and frozen at -20◦C for later analysis. 250mL serum bottles destined forincubations were overfilled 3 times with water from 7 depths (90m, 100m, 120m, 135m, 150m, 165mand 200m). The overfilling of the bottle as well as capping with blue halobutyl stoppers (Bellco,UK) minimized oxygen contamination (De Brabandere et al., 2012). Samples for chlorophyll adetermination were collected in carboys from four depths corresponding to 100%, 50%, 15% and1% of the surface incident irradiance as measured by the PAR sensor on the CTD. Carboys werekept cool and dark until further subsampling back in the lab. 500 mL subsamples from eachcarboy were filtered for phytoplankton biomass (chl a). Filters were kept frozen at -20◦C until51Table 3.1: Addition of labeled N-species and electron donors to incubations in 2015.Exact date of sampling Type of 15N labeled-incubationJanuary 2015 14 January 2015 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM)February 2015 11 February 2015 /March 2015 11 March 2015 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM)April 2015 8 April 2015 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM)May 2015 13 May 2015 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM)June 2015 3 June 2015 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM), 15NO3-(10µM) & HS- (1, 5, 10, 15, 20µM)July 2015 8 July 2015 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM)August 2015 12 August 2015 15NO3- (2.5, 5, 10, 15, 25µM), 15NH4+ & 14NO3-(10µM&10µM)September 2015 9 September 2015 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM)October 2015 22 October 2015 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM)November 2015 18 November 2015 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM)December 2015 9 December 2015 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM)analysis.3.2.2 Nutrient and process rate measurementsSamples for NO–2 , NH+4 and HS– determinations were thawed immediately prior to analysis andmeasured with spectrophotomectric assays: the Griess assay, the indophenol blue method, andthe Cline assay, respectively [151]. NOx (NO–3 and NO–2 ) was measured by chemiluminescencefollowing reduction to NO with vanadium [189], and we subtracted NO–2 from the total NOxconcentrations to obtain NO–3 concentrations (Antek instruments 745 and 1050, Houston TX).Chlorophyll a samples collected on filters (0.7m nominal porosity) were extracted for 24 hourswith 90% acetone at -20C, and the extracted chlorophyll a measured in a Turner Designs 10AUfluorometer, using an acidification method and corrected for phaeopigment interference [190].DIN deficit (DINdef) was calculated according to Bourbonnais et al. (2013) [191] and correctedfor the release and dissolution of iron and manganese oxyhydroxide-bound PO3 –4 under anoxicconditions.Dark Carbon fixation rates were measured by overfilling 60mL serum bottles 3 times tominimize O2 contamination and ammending 14C – HCO3 to the incubation bottles following theJGOFS protocol [192].The protocol used for measuring rates of denitrification and anammox was modified from52[110]. In an attempt to minimize bottle effects arising from the use of small sample volumes, weincubated the water in 250mL serum bottles closed with blue butyl rubber stoppers. At the startof the incubation, we inserted a 20mL oxygen scrubbed helium headspace into the bottle and thenadded the 15N labeled N-species and electron donors to the bottles according to table 3.1. Gasentering the serum bottle was passed through an oxygen scrubber (Cu-CuO, Glasgertebau Ochs -Germany) to limit O2 introduction to incubations of anoxic water to below our detection limitmeasured with flow-through cell oxygen sensor (<0.2µmol L– 1, Pyroscience). For incubationsof oxygen contaminated water, adding a 20 mL headspace decreases the amount of oxygen inthe seawater by about 30 times due to preferential partitioning of O2 into the headspace gas. Incontrast, given the distribution of sulphide between aqueous and gaseous species in seawater, lessthan 2% of the total sulphide in our incubation vessels resides in the headspace. Samples weretaken approximately every 6, 12, 24 and 48 hours during the incubations to both allow maximumsensitivity and capture intervals with constant rates. In between time points, incubations werekept in the dark at 15◦C. To determine the time-course of 15N labeled-N2 production, gas sampleswere taken with a 1 mL gas-tight syringe (Hamilton) previously flushed with He and then withthe headspace gas. Gas samples were stored in 3mL exetainers previously filled with milliQwater. Liquid samples were taken to follow the production or consumption of NO–2 , NO–3 , orNH+4 . Liquid samples were taken with a plastic 5 mL syringe previously flushed with He, filteredand then stored at -20◦C for later analysis. The 15N content of N2 was determined in gas samplescollected during the incubations on an Isotope-Ratio Mass Spectrometer (Delta V with continuousflow inlet, thermoscientific). Concentrations of N2 were calibrated with standards by injectingdifferent amounts of gas from N2 flushed Exetainer vials at 1 atmosphere. The excess 14N15Nand 15N15N in the gas samples was calculated as described by [193]. Then, rates were calculatedthrough least squares fitting of the slope of 15N accumulation versus time for the linear region of15N excess ingrowth (i.e. constant rates), correcting for the 15N labelling percentages of the initialsubstrate pool and accounting for the initial pool of substrate present. Rates were determinedto be significant if the slope of the linear regression was considered different from 0 (p¡0.05).Denitrification rates were determined from the accumulation of 30N2 in the bottle headspace fromthe 15NO–3 additions, and anammox rates were calculated from the accumulation of29N2 from the5315NH+4 +14NO–3 additions according to [154] with modifications and compared to the accumulationof 29N2 from the 15NO–3 additions. The detection limit on these rates were calculated as themedian of the standard error on the slope used to calculate all significant rates [98, 106] and wasdetermined to be 0.04 nM h– 1 and 0.4 nM h– 1 for anammox and denitrification, respectively. Toproduce integrated rates of denitrification and anammox for each month, we first scaled potentialrates (Rpot) to in situ rates or corrected rates (Rcor) by using Michaelis-Menten half saturationconstants Km and in situ substrate concentrations for each process, respectively (Eq. 3.7).Rcor =[S][S] +Km∗ Rpot (3.7)For denitrification, we used Km determined through the addition of different 15NO–3 concentra-tions in the incubations (Fig. 3.4a and table 3.1). For anammox, we used a Km from the literature[109]. Then, we integrated the corrected rates over the sampling depth intervals to attain areaspecific process rates.3.2.3 Microbial community profilingSix different depths (10, 100, 120, 135, 150, and 200m) were sampled for microbial communityprofiling and water from these depths was returned to the lab for same-day filtering. 10L of waterwas filtered onto Sterivex 0.22µm (Millipore) filters with a 2.7µm glass fiber pre-filter. Filteredbiomass was soaked in lysis buffer then frozen immediately in liquid nitrogen. Filters werestored at -80C until further analysis. DNA was extracted according to [194]. Extracted DNAwas quantified using the picogreen assay (Invitrogen) and checked for amplification of the smallsubunit ribosomal RNA (SSU or 16S rRNA) gene using universal primers targeting the V4-V5region of the bacterial and archaeal 16S rRNA gene (515F-Y and 926R) [195]. DNA was sent to theJoint Genome Institute (California, USA) for 16S rRNA amplicon sequencing on the Illumina MiSeqplatform ( sequenced, amplicons were quality filtered using the JGI ”itaggerReadQC” pipeline(source: and Quality filtered reads were run through USE-54ARCH [196] and QIIME [197]. First, we identified chimeras using UCHIME [196]. We then pickedOTUs de novo with the sumaclust method at 97% OTU threshold [198]. We filtered singletonsfrom the OTU table and then assigned taxonomy to a representative set of sequences with rdpclassifier using the QIIME release Silva database V128 [199]. Chao1 diversity index was calculatedwith R. Clustering of the samples was also performed in R based on a dissimilarity matrix withEuclidean method ( also quantified total bacterial and archaeal 16S rRNA genes present in our samples viaqPCR by targeting the region V1-V3 regions of the bacterial and archaeal 16S rRNA genes withthe primers 27F/20F (5’-AGAGTTTGATCCTGGCTCAG, 5’-TTCCGGTTGATCCYGCCRG) andDW519R (5’-GNTTTACCGCGGCKGCTG) [183]. Standards used for total bacteria and totalarchaea quantification were obtained from SSU rRNA gene clone libraries as described in [183].qPCR program was as followed: (1) 95◦C for 3 minutes, (2) 95◦C for 20 seconds, (3) 55◦C for 30seconds, (4) plate read, repeat (2) to (4) 44 times, obtain melting curve by incrementing 0.5◦C from55◦C to 95◦C every second. qPCR reactions were performed in low-profile PCR 96 well-plates(BioRad) in a 20µl reaction volume on a CFX Connect Real-Time thermocycler (BioRad). Resultscan be found in Appendix C.3.2.4 Flux balance modelingFlux balance modeling was conducted to describe rates of anammox, NO–3 reduction to NO–2and complete denitrification (NO–2 to N2) based on cell abundance, input fluxes of substrates,and kinetic descriptions of these processes. The script for the simulation was written in Matlab(version R2015b) and can be found in the Appendix C. More details can be found in the discussionsection that follows as well as in Appendix C.3.3 Results3.3.1 General water column physical, chemical, and biological propertiesA salinity profile (Fig. 3.2a), shows relatively uniform bottom waters with monthly variabilityin the surface waters. Figure 3.2b shows the relatively homogeneous temperature in the SI55water column with a warming in the surface waters during the summer (June to September2015) and the extension of this warming to deeper water depths in the following months. Thechlorophyll a data (Fig. 3.2c) shows peaks of fluorescence in the surface waters in March, May,and September 2015 with the highest peak, at 43.82 µg L– 1 chlorophyll a in March just below thesurface. O2 concentration profiles (Fig. 3.2d) were also determined with the CTD probe, revealingO2 depletion at depth to less than the sensor limit of detection (<1µM) for all of 2015. The upperboundary of the oxycline (depths where there is a sharp gradient in oxygen concentration) isgenerally around 80m and oxygen penetrates at least to 120m, though penetration can be as deepas 150m, as seen in July and September. Low O2 concentrations and anoxia thus characterize thedeeper waters of SI (>120m depth) throughout 2015. NO–3 concentrations (Fig. 3.2e) are highin surface waters (up to 32µM) and generally decline with increasing depth within the oxyclineand often remain detectable in deeper low-oxygen waters. A peak in NO–2 concentrations (Fig.3.2f) can be detected sporadically in both the surface waters and/or around 120m-135m depthwhere it can reach concentrations as high as 2.5µM. Surface waters are largely devoid of anyNH+4 (Fig. 3.2g), which tends to accumulate below 140m in anoxic waters and reaches the highestmeasured concentrations (up to 32µM) by 200m. Sulphide (HS– ) was only present in bottomwaters, reaching concentrations up to 41µM in February 2015, and was generally detected at 135mand below (Fig. 3.2h). In figure 3.2i, we show DIN deficit [191] calculated for the year 2015 withvalues varying from 0 in the surface waters to 60 in the bottom waters for February 2015. Overall,values reflected a DIN deficit in the anoxic waters and increased with depth (Fig. 3.2i).56Jan15 Mar15 Apr15 Jun15 Jul15 Sep15 Oct15 Dec150510152025303540Depth (m)05101520253035404550Chlorophyll a (g L-1)µFigure 3.2: Geochemical profiles for SI, 2015 a) Salinity (g kg-1); b) Temperature (C); c) Chlorophyll a µg L– 1) d) Oxygen profiles (µM); e) NO–3concentrations (µM); f) NO–2 concentrations (µM); g) NH+4 concentrations (µM); h) HS– concentrations (µM) i) calculated DIN deficit values(see methods in main text) for Saanich Inlet during the year 2015 at station S3. a, b and d values have been obtained from the CTD profilesmonthly. Intermediate values have been interpolated in matlab using the gridfit function (specifically the nearest neighbor). c, d-g values wereobtained from discrete sampled depths as indicated by black dots on graphs and interpolated interpolated in matlab using the gridfit function(specifically the nearest neighbor). Note that Fig. 3.2c only goes down to 100m depth as the values obtained for the chlorophyll a profile weresampled mostly above 50m.573.3.2 Rates of denitrification, anammox, and dark carbon fixationBoth anammox and denitrification were active throughout the year in the low oxygen waterswhere we conducted 15N-labeled incubations. Rates of denitrification, corrected for in situsubstrate concentrations, varied between 0.28±0.03 and 140±14 nM hr– 1 (Fig. 3.3b) based on theaccumulation of 30N2 in 15NO–3 amended incubations (Table 3.1). Similarly, rates of anammoxvaried throughout the year, between 0.07±0.01 and 13.2±0.4 nM hr– 1 (Fig. 3.3c) based on theaccumulation of 29N2 in 15NH+4 +14NO–3 amended incubations. We also compared rates ofanammox obtained through the accumulation of 29N2 with the addition of 15NO–3 and found thatthey were of the same order of magnitude as the rates obtained from 15NH+4 +14NO–3 incubations(Fig. 3.3a). Overall, rates of denitrification, when detected, were equal to or higher than ratesof anammox, although anammox dominated N2 production in 55% of the measurements made.However, the fact that rates of denitrification were generally higher, when detected, led to a higherannual proportion of N2 production through denitrification (see section 2.3.4: depth-integratedrates of N-loss). Dark carbon fixation rates were measured for most of the water column andranged between 0.24 to 400 nmoles C L– 1 hr– 1 (Fig. 3.3d).3.3.3 Response of denitrification and anammox to amendmentsBetween 1 and 20µM 15NO–3 was amended to seawater collected from 165 m depth in August 2015.This depth contained 1 µM NO–3 in situ and was therefore at the lower end of NO–3 concentrationsfound within Saanich inlet’s anoxic waters. Hence, NO–3 concentrations may be expected to limitdenitrification, NO–3 reduction to NO–2 , and anammox at this depth. Rates of dentirification, basedon the accumulation of 30N2, increased with increasing NO–3 concentrations up to 20µM (Fig. 3.4a),and the relationship between rates and NO–3 concentration could be modeled with a Michaelis-Menten formulation. Our data could be described with a maximum rate of denitrification (Vmax)and a half-saturation constant, Km, for NO–3 of 112 nmol L– 1 hr– 1 and 5µM (Fig. 3.4d and f, table3.2), respectively. The rate of denitrification found at 20µM NO–3 , however, is lower than for 15 µMand did not follow predictions from the Michaelis-Menten model in Figure 3.4e and f. Anammoxwas not detected. We also determined changes in the concentrations of NOx and NH+4 when we58Jan15 Mar15 Apr15 Jun15 Jul15 Sep15 Oct15 Dec15100120140160180200Depth (m)0.5151015Anammox rates nM hr-1Jan15 Mar15 Apr15 Jun15 Jul15 Sep15 Oct15 Dec15100120140160180200Depth (m)0.5151015Anammox rates nM hr-1Jan15 Mar15 Apr15 Jun15 Jul15 Sep15 Oct15 Dec15100120140160180200Depth (m)0.5151050100150Dentirification(nMhr-1)Jan15 Mar15 Apr15 Jun15 Jul15 Sep15 Oct15 Dec15100120140160180200Depth (m)0.5110100200400Dark Carbon Fixation nM Chr-1b.c.a.d.15NO3- 15NO3-15NH4++14NO3- 14HCO3-Figure 3.3: Process rate measurements for SI, 2015 Potential rates of a) anammox (nM hr– 1) and b) denitrifica-tion (nM hr– 1) calculated from the incubated samples with 15NO–3 for the year 2015 at station S3. c) showspotential rates of anammox of samples incubated with 15NH+4 +14NO–3 . In d), graph shows rates of darkcarbon fixation from incubation with H14CO3- in nM hr– 1. The scale bar is in log scale.added different 15NO–3 concentrations (Fig. 3.4a and b). We observed that NO–2 accumulates withconcentrations reaching a maximum of 9µM when 20 µM 15NO–3 was added. NH+4 concentrations,on the other hand, remain relatively constant between 8 and 12µM. Rates of NO–3 reduction varied,between 0 and 430 nM hr– 1, and rates of NO–2 accumulation varied between 0 and 286 nM hr– 1(Fig. 3.4e). The latter rates combined with the rates of denitrification are enough to explain therates of NO–3 reduction and thus no accumulation of other intermediates such as N2O is requiredor expected.We also amended seawater collected from 120m depth in June 2015 with HS– ranging from 1to 10µM, in addition to 10µM 15NO–3 to examine the influence of HS– on rates of denitrification590 10 20 30 40 50Time (hr)03691215NO2- and NO 3-  (µM)1 µM5 µM10 µM15 µM20 µM1 µM_15µM_110µM_115µM_120µM_10 10 20 30 40 50Time (hr)024681012NH4+ (µM)0 10 20 30 40 50Time (hr)05010015020025030029N 2 (nM)1_295_2910_2915_2920_290 10 20 30 40 50Time (hr)01,0002,0003,0004,0005,00030N 2 (nM)0 5 10 15 20 25 30NO3- (µM)050100150200Denitrification (nmol L-1 hr-1)Modeled rateModeled highModeled lowMeasured rate1 5 10 15 20NO3- concentrations (µM)0100200300400500NO3- reduction and NO2- accumulation (nM hr-1) NO3- reductionNO2- accumulationa bfedcFigure 3.4: NO–3 dependency in SI. a) NOx accumulation/consumption over time with the addition ofdifferent NO–3 concentrations NO–3 concentrations in dashed lines, NO–2 concentrations in solid linesb) NH+4 accumulation/consumption over time with the addition of different NO–3 concentrations. c)Production of 29N2 in the incubations (nM) d) Production of 30N2 in the incubations (nmol), e) rates ofNO–3 reduction and NO–2 accumulation (nM hr– 1) f) Michaelis Menten curve and measured denitrificationrates for different NO–3 concentrations in August 2015, see table 3.2 for details on the Michaelis Mentenparameters used in f)60Table 3.2: Michaelis-Menten parameters, Km (µM) and Vmax (nmol L– 1 hr– 1) for NO–3 dependency ofdenitrification at 165m in August 2015. Note that anammox kinetics are not following Michaelis-Mentenmodel in this caseDenitrificationKm (µM) 5 ± 0.5Vmax (nmol L-1 hr-1) 112 ± 10and anammox. This depth was chosen because it does not contain any detectable sulphide in situ.Instead, it immediately overlies the sulphidic deep waters and thus likely receives a flux of HS–from below that fails to accumulate to detectable concentrations at 120m depth and signifies sulfideoxidation. Results show an increase in denitrification rates with increasing HS– concentrations(Fig. 3.5d and f) above an apparent threshold of 2.5 µM HS– . These experiments reveal a seeminglylinear trend, but scarcity in data precludes the delineation of a definitive relationship (Fig. 3.5eand f). Anammox occurs (Fig. 3.5c and e) with 1 and 2.5µM HS– amendments but was notdetected with 5 and 10µM HS– amendments. NOx concentrations were constant over time inthese experiments except for the highest HS– concentrations (Fig. 3.5a) and NH+4 concentrationsdecreased over time (Fig. 3.5b).3.3.4 Depth-Integrated rates of N-lossDepth-integrated rates of N2 production varied over the year, with a greater contribution fromdenitrification (63%) than anammox (37%) (Fig. 3.6). Rates of denitrification ranged between0.02±0.006 to 14±2 mmol m– 2 d– 1 (Fig. 3.6), with the highest rates following renewal in Julyand August. Anammox rates, on the other hand, were comparatively constant throughout theyear, and varied between 0.15±0.03 and 3.4±0.3 mmol m– 2 d– 1 (Fig. 3.6). Anammox dominatedN2 production in January, April, May, June, October and November (>50% of N2 production).Nevertheless, results show that denitrification overall dominates the yearly N2 production in thewater column (Fig. 3.6).610 5 10 15 20 25Time (hr)04812NO2- and NO 3- (µM)1 µM2.5 µM5 µM10 µM1 µM_12.5 µM_15 µM_110 µM_10 5 10 15 20 25Time (hr)01234NH4+ (µM)1 µM2.5 µM5 µM10 µM0 10 20 30 40 50Time (hr)020040060080029N 2 (nM)1_292.5_295_2910_290 10 20 30 40 50Time (hr)02004006008001,0001,20030N 2 (nM)1_302.5_305_3010_300 2 4 6 8 10H2S (µM)024Anammox (nmol L-1  hr-1)Group: 10 2 4 6 8 10H2S (µM)01020304050Denitrification (nmol L-1  hr-1)Figure 3.5: HS– dependency in SI a) NOx accumulation/consumption over time with the addition ofdifferent HS– concentrations NO–3 concentrations in dashed lines, NO–2 concentrations in solid lines b)NH+4 accumulation/consumption over time with the addition of different HS– concentrations. c) Productionof 29N2 in the incubations (nM) d) Production of 30N2 in the incubations (nM) e) measured anammoxrates for different HS– concentrations in June 2015 f) measured denitrification rates for different HS–concentrations in June 2015621 2 3 4 5 6 7 8 9 10 11 12Month of 20150.0000.0050.0100.015N-loss (mol N m-2 d-1)020406080100% anammox DenitrificationAnammox% anammoxFigure 3.6: Depth Integrated N-loss. Depth Integrated N-loss rates over the year through denitrification inorange and through anammox in blue. The black dotted line represents the percentage of N-loss occurringthrough anammox (in %). The stacked bar represents the averaged N-loss over the entire year.3.3.5 Microbial community compositionFrom a total of 6,889,880 sequences quality filtered by the JGI, 0.3% of the reads were discardedbecause they were too short or too long and 2% of the sequences were identified as chimerasand discarded. The final read count per sample can be found in table C.1. After clusteringat the 97% identify threshold, 28,947 OTUs were resolved across 72 samples. The estimatedcommunity diversity (chao1) was low and variable in the surface waters and comparably higherand more stable at deeper depths. These results are summarized in figure C.1 and table C.1 of theAppendix C.The microbial community in SI is vertically stratified with strong shifts in community com-positions apparent between the surface waters (10m) and the deeper waters (100m and below)(Figs. 3.7 and 3.8). In particular, there is a shift between high relative abundances of Alphapro-teobacteria and Bacteroidetes (together, 42 to 85.3%) in the surface to a higher relative abundanceof Gammaproteobacteria (23.4 to 68.5%) in the deeper waters (between 100 and 200m) (Fig.633.7). In the surface waters, Alphaproteobacteria were mainly comprised of the SAR11 clade andBacteroidetes of the Flavobacteriales. The cyanobacterial population present early in the yeardecreases to <1% during the spring bloom (April, May, June), along with a sharp increase inFlavobacteriales for these 3 months (Fig. 3.7 and 3.8 ). In the deeper waters, the overwhelmingmajority of the Gammaproteobacteria are associated with two OTUs belonging to the SUP05cluster (Oceanospiralles clade) (Figs. 3.7 and 3.9, and Appendix C). This trend was constantthroughout the year. Another Gammaproteobacterial group Ectothiorhospiraceae (purple sulfurbacteria) were present throughout the year as well in the deeper waters (100 to 200m), withone OTU present between 1 and 30% (Fig. C.3). Thaumarcheotal (Marine Group 1) relativeabundance was generally low in the surface waters and increased up to 28% at 100m where NO–3concentrations generally peak (Figs. 3.7, C.2 and C.3). In the deeper waters (100 to 200m), theMarinimicrobia clade totalled a few percent throughout the year and increased to up to 12% inNovember at 135m. Epsilonbacteria were mostly comprised of an OTU from the genus Arcobacter,which reached up to 30% at 200m in July 2015 during deep water renewal and remained presentat relatively high abundances until September (Figs. 3.7, 3.9 and C.3). Several OTUs from thegenera Ca. Scalindua (Planctomycetes) were present throughout the water column with a total upto 5.7% at 100m in December 2015 (Figs. 3.7 and 3.9). Members of the Woesearcheota phylumwere most abundant at depths from 100 to 200m (0.3% to 12.2%). These results indicate a strongvertical stratification of the water column microbial community and relative consistency in thisstratified community throughout the year, with notable exceptions (Fig. 3.8). Surface waters, forexample, exhibited considerable dynamics in microbial communities during the spring blooms(April to June), and deeper waters shifted composition following renewal in July (Fig. 3.8).6425 50 75 10000 25 50 75100JAN15_10mFEB15_10mMAR15_10mAPR15_10mMAY15_10mJUN15_10mJUL15_10mAUG15_10mSEP15_10mOCT15_10mNOV15_10mDEC15_10mE1JAN15_100mFEB15_100mMAR15_100mAPR15_100mMAY15_100mJUN15_100mJUL15_100mAUG15_100mSEP15_100mOCT15_100mNOV15_100mDEC15_100mE2JAN15_120mFEB15_120mMAR15_120mAPR15_120mMAY15_120mJUN15_120mJUL15_120mAUG15_120mSEP15_120mOCT15_120mNOV15_120mDEC15_120mE3JAN15_135mFEB15_135mMAR15_135mAPR15_135mMAY15_135mJUN15_135mJUL15_135mAUG15_135mSEP15_135mOCT15_135mNOV15_135mDEC15_135mE4JAN15_150mFEB15_150mMAR15_150mAPR15_150mMAY15_150mJUN15_150mJUL15_150mAUG15_150mSEP15_150mOCT15_150mNOV15_150mDEC15_150mE5JAN15_200mFEB15_200mMAR15_200mAPR15_200mMAY15_200mJUN15_200mJUL15_200mAUG15_200mSEP15_200mOCT15_200mNOV15_200mDEC15_200mSampleSequence_proportion0255075100JAN15_10mFEB15_10mMAR15_10mAPR15_10mMAY15_10mJUN15_10mJUL15_10mAG15_10mSEP15_10mOCT15_10mNOV15_10mDEC15_10mE1JAN15_100mFEB15_100mMAR15_100mAPR15_100mMAY15_100mJUN15_100mJUL15_100mAG15_100mSEP15_100mOCT15_100mNOV15_100mDEC15_100mE2JAN15_120mFEB15_120mMAR15_120mAPR15_120mMAY15_120mJUN15_120mJUL15_120mAG15_120mSEP15_120mOCT15_120mNOV15_120mDEC15_120mE3JAN15_135mFEB15_135mMAR15_135mAPR15_135mMAY15_135mJUN15_135mJUL15_135mAG15_135mSEP15_135mOCT15_135mNOV15_135mDEC15_135mE4JAN15_150mFEB15_150mMAR15_150mAPR15_150mMAY15_150mJUN15_150mJUL15_150mAG15_150mSEP15_150mOCT15_150mNOV15_150mDEC15_150mE5JAN15_200mFEB15_200mMAR15_200mAPR15_200mMAY15_200mJUN15_200mJUL15_200mAG15_200mSEP15_200mOCT15_200mNOV15_200mDEC15_200mSampleSequence_proportionTaxonomyActinobacteriaAlphaproteobacteriaBacteroidetesBetaproteobacteriaChloroflexiCyanobacteriaDeltaproteobacteriaEpsilonproteobacteriaEuryarchaeotaFirmicutesGammaproteobacteriaGemmatimonadetesLentisphaeraeMarinimicrobia (SAR406 clade)NitrospinaeParcubacteriaPAUC34fPlanctomycetesSBR1093SpirochaetaeThaumarchaeotaTM6 (Dependentiae)unclassified_allUnclassified_proteobacteriaVerrucomicrobiaWoesearchaeota (DHVEG−6)10m100m120m135m150m200mRelative abundance (%)JDFMAM JJASONJDFMM JJASONAJDFMM JJASONAJDFMM JJASONAJDFMM JJASONAJDFMM JJASONA0255075100JAN15_10mFEB15_10mMAR15_10mAPR15_10mMAY15_10mJUN15_10mJUL15_10mAG15_10mSEP15_10mOCT15_10mNOV15_10mDEC15_10mE1JAN15_100mFEB15_100mMAR15_100mAPR15_100mMAY15_100mJUN15_100mJUL15_100mAG15_100mSEP15_100mOCT15_100mNOV15_100mDEC15_100mE2JAN15_120mFEB15_120mMAR15_120mAPR15_120mMAY15_120mJUN15_120mJUL15_120mAG15_120mSEP15_120mOCT15_120mNOV15_120mDEC15_120mE3JAN15_135mFEB15_135mMAR15_135mAPR15_135mMAY15_135mJUN15_135mJUL15_135mAG15_135mSEP15_135mOCT15_135mNOV15_135mDEC15_135mE4JAN15_150mFEB15_150mMAR15_150mAPR15_150mMAY15_150mJUN15_150mJUL15_150mAG15_150mSEP15_150mOCT15_150mNOV15_150mDEC15_150mE5JAN15_200mFEB15_200mMAR15_200mAPR15_200mMAY15_200mJUN15_200mJUL15_200mAG15_200mSEP15_200mOCT15_200mNOV15_200mDEC15_200mSampleSequence_proportionTaxonomyActinobacteriaAlphaproteobacteriaBacteroidetesBetaproteobacteriaChloroflexiCyanobacteriaDeltaproteobacteriaEpsilonproteobacteriaEuryarchaeotaFirmicutesGammaproteobacteriaGemmatimonadetesLentisphaeraeMarinimicrobia (SAR406 clade)NitrospinaeParcubacteriaPAUC34fPlanctomycetesSBR1093SpirochaetaeThaumarchaeotaTM6 (Dependentiae)unclassified_allUnclassified_proteobacteriaVerrucomicrobiaWoesearchaeota (DHVEG−6)Figure 3.7: Microbial communities in SI from 16S rRNA gene sequencing. Microbial communities composition of SI in 2015 for 6 depths (10, 100, 120,135, 150 and 200m) at the phylum level in relative abundance.65APR15.10mMAY15.10mJUN15.10mDEC15.100mOCT15.100mAUG15.100mFEB15.100mMAR15.100mAPR15.100mMAY15.100mJAN15.100mAUG15.120mJUL15.100mJUN15.100mSEP15.100mNOV15.100mDEC15.10mNOV15.10mOCT15.10mJAN15.10mFEB15.10mSEP15.10mMAR15.10mJUL15.10mAUG15.10mFEB15.120mDEC15.150mSEP15.120mAUG15.135mJUL15.120m0 10000 20000 30000 40000hclust (*, "average")dist(raw_cluster, method = "euclidean")HeightAPR15.10MAY15.10JUN15.10DEC15.100OCT15.100AUG15.100FEB15.100MAR15.100APR15.100MAY15.100JAN15.100AUG15.120JUL15.100JUN15.100SEP15.100NOV15.100DEC15.10NOV15.10OCT15.10JAN15.10FEB15.10SEP15.10MAR15.10JUL15.10AUG15.10FEB15.120DEC15.150mSEP15.120mAUG15.135mJUL15.120mNOV15.150mMAY15.120mMAY15.200mAPR15.200mMAR15.200mFEB15.200mDEC15.120mSEP15.135mMAR15.150mMAY15.135mJUN15.200mJAN15.200mJAN15.120mJAN15.150mAPR15.135mFEB15.135mJAN15.135mOCT15.120mAPR15.150mMAR15.120mDEC15.135mNOV15.135mJUN15.120mMAY15.150mJUN15.150mAPR15.120mOCT15.135mJUL15.135mSEP15.150mFEB15.150mOCT15.150mMAR15.135mJUN15.135mNOV15.120mSEP15.200mNOV15.200mAUG15.150mJUL15.150mDEC15.200mOCT15.200mJUL15.200mAUG15.200mhclust (*, "average")dist(raw_cluster, method = "euclidean")APR15.1MAY15.1J N .DEC15.10OCT15.10AUG15.10FEB15.10MAR15.10AP 15.10MAY15.10J N15.10AUG . 2J L ..SEP .NOV . 0DEC15.1NOV15.1OCT15.1JAN15.1FEB15.1S P15.1MAR15.1JUL15.1A G .FEB15.12D C15.15S P15.120AUG15.135J L15.120NOV15.150MAY15.12015.200APR15.200MA 15.200FEB15.200D C15.120S P15.135MAR15.150Y15.135JUN15.200JA 15.200J 15.120J 15.150APR15.135FEB15.135JAN15.135OCT15.120APR15.150MA 15.120DEC15.135NOV15.135JUN15.120MAY15.150JUN15.150APR15.120OCT15.135JUL15.135SEP15.150F B15.150OCT15.150MAR15.135JUN15.135NOV15.120SEP15.200NOV15.200AUG15.150J L15.150DEC15.200OCT15.200JUL15.200A G15.200hclust (*, "average")dist(raw_cluster, method = "euclidean")15.10JU 15.10J 2J L 0N15.1JJ L15.125235J L 20522012355035J 20012535J 205235J 205J 235J L 5035J 202015J L 20J Lhclust (*, "average")dist(raw_cluster, method = "euclidean")Surface spring bloomsRenewal ecotypeTypical surface  communityChemocline communityApr - 10mMay - 10mJun  - 10mAug  - 10mJul  - 10mMar  - 10mSep  - 10mFeb  - 10mJan  - 10mOct  - 10mNov  - 10mDec  - 10mNov  - 100mSep  - 100mJun  - 100mJul  - 100mAug  - 120mJan  - 100mMay  - 100mApr  - 100mMar  - 100mFeb  - 100mAug  - 100mOct  - 100mDec  - 100mFeb  - 120mDec  - 150mSep  - 120mAug  - 135mJul  - 120mNov  - 150mMay  - 120mMay  - 200mApr  - 200mMar  - 200mFeb  - 200mDec  - 120mSep  - 135mMar  - 150mMay  - 135mJun  - 200mJan  - 200mJan  - 120mJan  - 150mApr  - 135mFeb  - 135mJan  - 135mOct  - 120mApr  - 150mMar  - 120mDec  - 135mNov  - 135mJun  - 120mMay  - 150mJun  - 150mApr  - 120mOct  - 135mJul  - 135mSep  - 1 0mFeb  - 150mOct  - 150mMar  - 135mJun  - 135mov  - 120mSep  - 200mNov  - 200mAug  - 200mJul  - 200mDec  - 200mOct  - 200mJul  - 200mAug  - 200mStagnation ecotypeFigure 3.8: Clustering of the microbial community composition of SI in 2015. Clustering of the microbial community composition of SI in 2015 for 6depths (10, 100, 120, 135, 150 and 200m) and 12 months. Dissimilarities between samples is shown by the height of the fusion of the dendrogram:the higher the fusion, the more dissimilar samples are between each other. Clustering of samples was performed in R with the Euclideanmethod.66Arcobacter (Campylobacter)Marinimicrobia (SAR406 clade)Oceanospiralles (SUP05)PlanctomycetesJAN15_100mFEB15_100mMAR15_100mAPR15_100mMAY15_100mJUN15_100mJUL15_100mAUG15_100mSEP15_100mOCT15_100mNOV15_100mDEC15_100m NASample2Group Abundance0102030400.7Arcobacter (Campylobacter)Marinimicrobia (SAR406 clade)Oceanospiralles (SUP05)PlanctomycetesJAN15_120mFEB15_120mMAR15_120mAPR15_120mMAY15_120mJUN15_120mJUL15_120mAUG15_120mSEP15_120mOCT15_120mNOV15_120mDEC15_120m NASample2Group Abundance0102030400.7Arcobacter (Campylobacter)Marinimicrobia (SAR406 clade)Oceanospiralles (SUP05)PlanctomycetesJAN15_135mFEB15_135mMAR15_135mAPR15_135mMAY15_135mJUN15_135mJUL15_135mAUG15_135mSEP15_135mOCT15_135mNOV15_135mDEC15_135m NASample2Group Abundance0102030400.7Arcobacter (Campylobacter)Marinimicrobia (SAR406 clade)Oceanospiralles (SUP05)PlanctomycetesJAN15_150mFEB15_150mMAR15_150mAPR15_150mMAY15_150mJUN15_150mJUL15_150mAUG15_150mSEP15_150mOCT15_150mNOV15_150mDEC15_150m NASample2Group Abundance0102030400.7Arcobacter (Campylobacter)Marinimicrobia (SAR406 clade)Oceanospiralles (SUP05)PlanctomycetesJAN15_200mFEB15_200mMAR15_200mAPR15_200mMAY15_200mJUN15_200mJUL15_200mAUG15_200mSEP15_200mOCT15_200mNOV15_200mDEC15_200m NAGroup Abundance0102030100m120m135m150m200mJ F M A M J J A S O N DArcobacter (Campylobacter)Marinimicrobia (SAR406 clade)Oceanospiralles (SUP05)PlanctomycetesJAN15_100mFEB15_100mMAR15_100mAPR15_100mMAY15_100mJUN15_100mJUL15_100mAUG15_100mSEP15_100mOCT15_100mNOV15_100mDEC15_100m NASample2Group Abundance0102030400.7Relativ  Abundance (%)Marinimicrobia (SAR406 clade)Oceanospiralles (SUP05)PlanctomycetesGroupArcobacter (Campylobacter)Marinimicrobia (SAR406 clade)Oceanospiralles (SUP05)PlanctomycetesAbundance010203040200m150m135m120m100m10mJANFEBMARAPRMAYJUN JULAUGSEPOCTNOVDECDepth2qPCR1e+071e+081e+091e+100.70.7205135120m100m10mJANFEBMARAPRMAYJUN JULAUGSEPOCTNOVDECSample2Depth2qPCR1e+071e+081e+091e+100.70.7200150135m120m100m10mJANFEBMARAPRMAYJUN JULAUGSEPOCTNOVDECSample2Depth2qPCR1e+071e+081e+091e+100.70.7205135120m100m10mJANFEBMARAPRMAYJUN JULAUGSEPOCTNOVDECSample2Depth2qPCR1e+071e+081e+091e+100.70.7200m150m135m120m100m10mJANFEBMARAPRMAYJUN JULAUGSEPOCTNOVDECSample2Depth2qPCR1e+071e+081e+091e+100.70.7200m150m135m120m100m10mJANFEBMARAPRMAYJUN JULAUGSEPOCTNOVDECSample2Depth2qPCR1e+071e+081e+091e+100.70.716S counts (16S L-1)Total 16S -Total 16S -Total 16S -Total 16S -Total 16S -+071e+081e+09+10Figure 3.9: Relative abundance of Planctmycetes, SUP05, Marinimicrobia and Arcobacter. Comparison of therelative abundance of Planctomycetes, SUP05 cluster Marinimicrobia, and newly highlighted Arcobacterbacteria. In addition to the relative abundance of these clades, we added total 16S counts (16S L– 1) for eachof these samples.3.4 Discussion3.4.1 Partitioning of N-loss in SI, and the seasonality of anammox anddenitrification15N-labeled incubations indicate that both anammox and denitrification operated simultaneouslythroughout the year in the anoxic water column of SI. Although anammox dominated (responsiblefor >50% N-loss) in 55% of the measurements in which N2 production was detected (Fig. 3.3a,b and c), depth-integrated rates of denitrification show that it accounts for up to 63% of thetotal N-loss from SI (Fig. 3.6). Overall, depth-integrated rates of denitrification and anammoxranged between 0.02 to 14.4 mmol m– 2 d– 1 and 0.15 and 3.36 mmol m– 2 d– 1, respectively (Fig.3.6). These integrated rate measurements agree well with rates previously reported based solely67on geochemical measurements [186], which imply N-loss of between 1.7 to 8.1 mmol m– 2 d– 1.Annual N-loss for the inlet was calculated from these measurements by taking the average of thedepth-integrated rates, and multiplying these by the surface area of the anoxic basin of SI (33km2). Annual N-loss totalled 0.002 Tg N yr– 1 in 2015. Given that 50% of the N-loss previouslyreported is from benthic N2 production, the rates that we measure here, that only capture pelagicN-loss, appear appreciably higher and may thus suggest inter-annual variability in N2 productionrates. Rates of denitrification and anammox were previously reported for 2 months during peakstagnation in SI in 2010 [118] and while these are at the lower end of the range of rates measuredhere, they generally agree with the rates we detect during peak stagnation.The monthly variability in rates of N-loss from Saanich inlet are driven through dynamics inrates of both anammox and denitrification. Thus, knowledge on the regulation of both anammoxand denitrification is key to knowing how N-budgets in Saanich inlet, and by extension, otheranoxic fjords, vary. Partitioning of N-loss between these pathways for the entire year reveals63% denitrification and 37% anammox. This ratio is close to the theoretical ratio calculatedfor the partitioning of N-loss in the open ocean through anammox and denitrification (29 to71% ratio anammox/denitrification ratio). This ratio applies when substrate (NH+4 ) supply ratesfor anammox are constrained by the stoichiometry of settling organic matter [63, 105, 173–175].Excursions beyond this ratio might indicate additional sources of NH+4 , such as sulphate reduction,and/or an input of NH+4 from the underlying sediments. Excursion below this ratio more likelysignals competition for nitrite or chemoautotrophic denitrification, which would not liberate NH+4 .In July and December of 2015, N-loss was close to the theoretical ratio (27 and 31%, respectively),which is consistent with the canonical scenario in which heterotrophic denitrification suppliesanammox with NH+4 . The ratio deviates from this throughout much of the rest of the year, withgenerally higher proportions of anammox (40% and beyond), implying that an additional supplyof NH+4 , beyond that supplied through heterotrophic denitrification, is needed.Rates of denitrification and anammox are expected to respond to the rates of supply of theprincipal substrates: NO–3 , organic matter or HS– for denitrification, and NH+4 and NO–2 foranammox; as well as possible inhibitors like HS– for anammox and O2 for both anammox anddenitrification. In July 2015, O2 and NO–3 both penetrate to 150m (Fig. 3.2d and e) signalling the68intrusion of oxygenated NO–3 rich waters to intermediate depths in SI, and although devoid of O2,deep waters (185 and 200m) contain detectable NO–3 . These observations indicate intermediate-and deep-water renewal in July. Although our monthly nutrient profiles do not record a strongdeep-water renewal that would have oxygenated the deep waters, we do observe changes innutrient concentrations, which coincide with higher rates of denitrification in July and August(Fig. 3.6). Thus, the regulation of denitrification in SI appears linked to renewal, and is furtherenhanced by the accumulation of HS– in August.A dramatic increase in deep water NH+4 concentrations is reflected by relatively high ratesof anammox recorded in September, October and November (Fig. 3.6), along with lower con-centrations of HS– , implying that increased deep-water NH+4 leads to higher rates of anammox.The NH+4 in the deep waters could originate from the remineralization of sinking organic mattersupplied through primary production in the surface waters. Chlorophyll a peaks in the surfacewaters of SI, a proxy for the abundance of photosynthetic organisms, vary over the year, andincreases during the spring/early summer (Fig. 3.2c). Organic matter from primary productionis exported to the deep waters and sediments as particles and fecal pellets. Given that bothparticles and fecal pellets would sediment to the deep waters in less than a week [200], weexpect deep water NH+4 concentrations to respond to blooms in the surface waters within 15days. As we sampled approximately every 4 weeks, it is possible that we lacked the temporalresolution to capture intense degradation activity following a bloom. However, as blooms are acommon occurrence during the summer months [33, 201, 202], the increase in the deep-water NH+4concentrations in September (Fig. 3.2g) likely originates from a corresponding increase in surfacewater primary production. Therefore, the combination of relatively high productivity in surfacewater and the ensuing high NH+4 concentrations in the deep waters likely support relatively highrates of anammox towards the end of the summer.Though ultimately sourced from primary production, the detailed biogeochemical pathwaysthrough which NH+4 is made available to anammox can vary. These pathways include: ammoni-fication due to organotrophic denitrification; DNRA; ammonification associated with sulphatereduction; or benthic NH+4 efflux. We thus consider these possible sources and their relative fluxesin relation to rates of anammox. Rates of denitrification measured in SI could have supplied6925% of the NH+4 needed to support co-occuring anammox on average, assuming Redfieldian OMstoichiometry and 100% organotrophic denitrification (in August 2015, Fig. 3.6). This is unlikely asHS– clearly influences rates of denitrification, and is consistent with previously reported genomicinformation [33, 182], which implies chemoautotrophic denitrification in SI. Another source ofNH+4 could be organotrophic NO–3 reduction to NO–2 .We calculate that the highest potential rates of NO–3 reduction to NO–2 recorded in August2015 (Fig. 3.4e, 400nM N hr– 1 for 20µM NO–3 addition), are sufficiently high such that all of theNH+4 needed to support the highest rates of anammox found could come from this reaction (13.7nM hr– 1). However, in situ NO–3 concentrations are generally not as high as the concentrations inthese amended incubations, and thus rates of NO–3 reduction to NO–2 might not supply all theNH+4 . Remineralization of organic matter through partial or complete denitrification is thus likelyonly partly responsible for the NH+4 supply to anammox. Similarly, SO2 –4 reduction could alsoproduce NH+4 through remineralization of organic matter in the water column. In SI, however,sulfate reduction remains unmeasured through direct process rate experiments but functionalmarkers for canonical sulfate reduction were found in the metaproteomes generated to date forSUP05 [33]. DNRA can also supply NH+4 to anammox, as it does in the Peruvian OMZ [67]. ADNRA catalyzing-like protein, hydroxylamine-oxidoreductase, was recovered in metaproteomesand appears to be associated with the dentrifier SUP05 [33]. DNRA, however, has not beendetected in SI to date, though modeling predicts appreciable DNRA for September 2009, andDNRA, if operating in SI, could thus contribute to dynamics in anammox activity [118]. Given thatthe pelagic pathways for NH+4 delivery to annamox appear insufficient to support the measuredrates, we consider the possibility that NH+4 efflux from the bottom sediments also contributesNH+4 to anammox. Indeed, high rates of organic matter remineralization through SO2 –4 reductioncharacterize SI sediments [185]. Some of the NH+4 liberated in the process would diffuse fromthe sediment and could advect upwards to fuel anammox in the overlying water. Based on ourcalculations (see Appendix C), NH+4 fluxes from the sediment in SI could fuel 88 to 100% of theNH+4 required to support anammox. We thus expect a combination of these multiple NH+4 sourcesfuels anammox and contributions to its variability throughout the year.703.4.2 Kinetics of denitrification and anammoxExternal forcing by substrate supply rates places overall constraints on material fluxes and thusmicrobial community metabolism, but microbial community structure and function also dependon the specific ecophysiologies of relevant organisms, such as an organisms ability to take up andmetabolize a given substrate. We showed that denitrification in SI appears to depend on NO–3concentrations (Fig. 3.4c and table 3.2), and the Km for NO–3 obtained at 165m in August 2015was 5 µM and in the same range as earlier reports from both environmental measurements andcultured denitrifiers (1.7 to 10µM) [56]. These prior kinetic constants come from pure cultures oforganotrophic denitrifiers [203, 204], sediment microbial communities [110, 205], and an anoxicsulfidic fjord [56]. When NO–3 concentrations exceeded 15µM, however, the rates of completedenitrification decreased (Fig. 3.4f). This is in line with the observation that denitrifiers tend tofavor the first step of denitrification, NO–3 reduction to NO–2 , over complete NO–3 reduction to N2,when NO–3 concentrations are high. Similar observations were made previously in Mariager fjord[56] where NO–3 reduction to NO–2 took over when NO–3 concentrations exceeded 5µM.Rates of denitrification and anammox in SI are sensitive to the HS– concentrations present.When seawater from 120m depth (June 2015) was amended with HS– , rates of denitrificationincreased with respect to HS– concentrations (Fig. 3.5) for HS– concentrations higher than 2.5µM.The rates then seemed to exhibit a linear response, possibly because the enzyme saturationfor sulphide oxidation is much higher [56]. This, observation is similar to reports of a lineardependency of denitrification on HS– concentrations, with no sign of saturation, up to 40µMHS– in Mariager Fjord, Denmark [56]. Measurements of NO–3 and NO–2 indicate low or no NO–2accumulation during these incubations and, given the low rates of denitrification for 2.5µM HS– ,this implies a shunting of the NO–2 produced to anammox. Indeed, anammox occurs with HS–amendment (Fig. 3.5d). Anammox occurrence at low HS– concentrations was observed previouslyin a sulphidic alpine lake [107], in contrast with most previous marine observations, which foundthat anammox was inhibited by HS– at concentrations as low as 1.6µM [100, 114]. The stimulationof anammox with low HS– concentrations in SI may reflect the production of NO–2 throughpartial denitrification (when the relevant substrates such as HS– are abundantly available) and71the bypass of complete denitrification due to a higher affinity of anammox bacteria for NO–2 [110].This is supported by the fact that NO–2 did not accumulate during these incubations (Fig. 3.5a),implying that sulfide dependent partial denitrification (NO–3 to NO–2 ) underpins nitrite leakage toanammox in SI.As rates of anammox appear to be sensitive to higher fluxes of NH+4 in the water column(see above), we plotted the rates of anammox obtained for the whole year vs. the in situ NH+4concentrations (Fig. 3.10). However, the lack of a coherent positive relationship between rates ofanammox and NH+4 concentrations generally implies insensitivity of anammox to NH+4 concentra-tions higher than 2µM. These results could indicate that the Km for NH+4 of anammox bacteria islower than the in situ NH+4 concentrations. Alternatively, this could also indicate that anammoxbacteria could obtain the NH+4 needed through tight coupling between anammox and DNRA [64]or through ammonification in particle-associated processes [206], which would not be specificallyrecorded in the ambient NH+4 concentrations in SI.0 5 10 15 20 25 30NH4+ (µM)02468101214Anammox (nmol L-1  hr-1)15NO315NH4+14NO3 incubationFigure 3.10: Comparison of anammox rates vs in situ NO–3 concentrations (µM)3.4.3 Vertical partitioning of the microbial communities in SIThe strongest difference in microbial community composition was between the surface waters at10m depth and deep waters below 100m depth (Fig. 3.8), while temporal variations were most72notable in the surface waters (Fig. 3.8). Variation in community composition between 120 and200m depth was comparatively small as were dynamics in deep water community compositionthroughout the year, with the exception of OTUs assigned to the Epsilonproteobacterium, Arcobac-ter (Fig. 3.7). Such vertical stratification in microbial community composition is typical for aquaticecosystems including OMZs [104, 120, 207], and has been previously observed in SI [33, 183]. Indeed, niche partitioning along redox gradients is generally expected [104]. A conceptualmodel previously developed [33] describes microbial community structure and function in SIand provides a benchmark framework through which to view temporal and vertical dynamicsin microbial community composition (Figs. 3.1c, 3.7 and 3.8 , 3.9, C.2 and C.3). The key taxathat comprise this model, including SUP05, Marinimicrobia, Thaumarcheota, SAR11, and Plancto-mycetes were prevalent community members throughout SI in 2015. Thamarcheaota and SAR11were the most abundant at 100m and SUP05 increasing in abundance with depth. Planctomyceteswere low in the surface water and increased to a few percent in the deeper waters, similar toMarinimicrobia. In addition to these taxa, our community profiles reveal dynamics in relativelyabundant Bacteroidetes, which increase in the surface waters during the spring bloom (from 20%to 65% in April, May and June) and Arcobacter that appears to bloom in the deep waters (from<1% up to 30% at 200m) in association with renewal in July and subsequently decreases in relativeabundance in the following month.A closer analysis of microbial community dynamics in the surface waters reveals that of the 15most abundant OTUs, there were high relative abundances of 3 OTUs of the Flavobacteriaceaeand 1 OTU of the Rhodobacterales family in April, May and June, and correspondingly lowabundances of an OTU belonging to the SAR11 clade (Fig. C.3). This particular microbialcommunity composition appears contemporaneous with photosynthetic blooms. Flavobacteriaceaeand Rhodobacterales are generally considered participants in biomass degradation [208] and theirrelatively high abundance in the spring may thus be a response to relatively strong photosyntheticactivity (Fig. 3.2c). While photosynthetic blooms are evident from pigment distributions (Fig.3.2c), we did not observe correspondingly high relative abundances of photosynthetic bacterialtaxa (cyanobacteria) at this time. This likely indicates that cyanobacteria play a limited role inthis bloom, which instead is the response of diatom growth, as previously reported [200]. Diatom73blooms in April, May, and June thus appear to stimulate a number of microbial taxa linked toorganic matter degradation in the surface waters, while cyanobacterial contributions to microbialcommunity composition are marginalized at this time.One of the most abundant OTUs present throughout the water column was assigned to theSUP05 cluster (C.3), which varied between a few percent in the surface water to a maximumof 48% in July and September at 150 and 200m, respectively (Figs. 3.9 and C.3). Based on itsmetabolic potential to couple sulphide oxidation to NO–2 reduction to N2O and its relatively highabundance, SUP05 has been implicated as a key-player in coupled C, N and S cycling and N-lossfrom SI [33, 182], and more broadly throughout low oxygen marine waters globally [209–211].Consistent with this, we find that N2 production through denitrification was active throughoutthe year when SUP05 was a ubiquitously abundant community member (Figs. 3.7, 3.9 and C.3).Likewise, water in collected from 120m in June had a microbial community composition with 28%SUP05, and rates of denitrification in this water increased in response to HS– addition, indirectlylinking SUP05 to sulphide dependent denitrification. However, N2O did not accumulate in ourincubations, and we thus suspect that other taxa also play a role in denitrification, by reducingN2O to N2.Some Marinimicrobia clades indeed possess the nosZ gene and have the metabolic potential toperform this last step in dentrification [187]. Like SUP05, Marinimicrobia were relatively abundantin the deep waters where they comprised 4 different OTUs that together comprise up to 12% ofthe total microbial community at 135m in November 2015 (Figs. 3.7, 3.9, 3.11a and C.3). These4 OTUs were phylogenetically compared to previously identified Marinimicrobia genome binsand SAGs (Fig. 3.11b, [187]) and were found to be affiliated to 4 different clades: 3 SI clones(SHBH1141, SHBH319 and SHAN400) as well as an Arctic clone (Arctic96B-7). Interestingly,only SHBH1141 appear to carry the nosZ gene [187], making it the most likely microorganismin SI to reduce N2O to N2 coupled to HS– oxidation [33]. The SHBH1141 clade increased withdepth, with the overall highest relative abundance at 150m (Fig. 3.11a). However, SHBH1141srelative abundance decreased in July, in association with the renewal. Comparatively, SHAN400clade stays relatively constant between 100 and 200m and both Arctic96B-7 and SHBH391 havehigher relative abundance at 120m than at 100m and remain constant down to 200m. From SAGs,74SHAN400 and Arctic96B-7 were shown to carry NO–3 reduction to NO coupled to HS– oxidation,thus participating to partial denitrification, whereas no genes involved in the N-cycle was foundfor SHBH391 [187].Anammox was also operative throughout the entire year in 2015, and accordingly, we foundmembers of the Planctomycetes phylum present at up to 5% in the water column (Figs. 3.7,3.8 and3.9). Indeed, Planctomycetes is the only phylum known to contain bacteria that perform anammox.The metabolic potential for anammox is restricted in the Planctomycetes to the order Brocadiales[62]. SI hosts mainly Ca. Scalindua, a well-known marine anammox bacterium which comprisedup to 2.6% of the community at 135m in May (Fig. 3.9). Altogether, microbial community profilingreveals that the key taxa that comprise previous conceptual models for coupled microbial C,N, and S cycling in SI are present and relatively abundant at depths between 100 and 200mthroughout the year. At the community level, these taxa collectively underpin N cycling and lossfrom Saanich inlet, which we demonstrate through contemporaneous process rate measurements.In addition to the taxa discussed above, it appears that an OTU assigned to the Epsilonpro-teobacteria Arcobacter increases dramatically in relative abundance in the deep waters, notablyat 200m where it goes from <1% in June to 30% in July, becoming one of the 15 most abundantOTUs in SI, and then drops to 20% in August (Fig. 3.9a and Fig. 3.11a). This increase in relativeabundance appears to be a response to deep water renewal and is strongly correlated with theenhanced rates of denitrification found in July and August, relative to the rest of the year, as wellas the highest rates of dark carbon fixation (Fig. 3.3d). Indeed, a number of Arcobacter isolatesare known to perform complete denitrification (NO–3 to N2) [212], as well as sulphide oxidation[213]. The high relative abundances of Arcobacter in July and the fact that Marinimicrobia OTUsdecreased at the same time posits an important role for Arcobacter in the SI N-cycle (Fig. 3.9).Considering our observations of microbial community and biogeochemical dynamics acrossthe year, we suggest that the inlet exists in two principle biogeochemical states: throughoutmuch of the year, the inlet is relatively stagnant, anaerobic N2 production is distributed betweenanammox and denitrification and we suggest that Ca. Scalindua, SUP05, and Marinimicrobiaare the key taxa responsible; during the summer renewal the input of NO–3 to deep sulphide-rich waters stimulates the growth of Arcobacter, which drives most N2 production through75denitrification marginalizing anammox. These two states thus define microbial communityphenotypes representing background or stagnation periods and renewal periods, respectively.Shifts between stagnation and renewal phenotypes imply that the relevant community memberspossess differing ecophysiology. In particular, the bloom in Arcobacter in response to renewalimplies that these organisms have higher maximum cell specific growth rates and/or lower biomassyield than the combination of SUP05 and Marinimicrobia. Without any existing informationon biomass yield, we thus estimated cell-specific rates of N2 production through denitrificationduring stagnation and renewal periods. We expect for the combined Marinimicrobia/SUP05population to have lower cell specific rates in comparison to the Arcobacter population, whichproduces N2 at higher rates for similar cell abundance. For the stagnation phenotype we used anaverage cell abundance of 1.64 · 10 9 cells L– 1 for the combined abundance of Marinimicrobia andSUP05, which we estimated by combining qPCR of the bacterial 16S rRNA gene as a proxy for totalcommunity size with the relative Marinimicrobia/SUP05 abundance from our amplicon sequencedata. Marinimicrobia, in association with SUP05, are likely responsible for the production of N2throughout most of the year, and we used the lowest and highest rates of denitrification in thestagnant period (0.01 – 38.45 nM hr– 1) to come up with a range of cell specific denitrification ratesfor Marinimicrobia/SUP05 between 0.0001 – 0.6 fmol N2 cell– 1 d– 1.To compare against the renewal phenotype, we estimated Arcobacter cell abundance (basedon total bacterial 16S rRNA gene copies for July at 200m, 3.05 · 10 9 cells L– 1 combined with therelative abundance of Arcobacter from our amplicon sequence data) and with the correspondingrates of denitrification obtained cell-specific rates of 1.08 fmol N2 cell– 1 d– 1 for Arcobacter. Thecell-specific rates for Marinimicrobia are therefore lower than the cell-specific rate calculated forArcobacter. Thus, it is likely that SUP05/Marinimicrobia population has a higher growth yield toArcobacter, shown by similar cell abundance but lower cell-specific rate for the former.The low relative abundance of Planctomycetes associated with lower N2 production ratesindicate that the anammox bacteria present in SI have a high cell-specific growth rate with alow growth yield. Again, based on the 16S abundance obtained from qPCR analysis appliedto the average relative abundance of anammox bacteria, cell-specific rates for anammox varybetween 0.02 and 6.72 fmoles NH+4 cell– 1 d– 1, using an average cell counts for anammox for the76year (2.3 · 10 7 anammox cells L– 1) and the highest and lowest rates measured in SI in 2015. Ourmeasured rates encompass the cell specific rates obtained from the Namibian OMZ [4.5 fmol NH+4cell– 1 d– 1, [66]], the Black Sea [3-4. fmol NH+4 cell– 1 d– 1, [114] ], and diverse bioreactors [2-20fmol NH+4 cell– 1 d– 1, [61]]. The lower end of our measured rates might be explained by smallfractions of active versus total anammox bacteria present, which would increase the cell-specificrates calculated here. This highlights that, even though anammox bacteria are generally present inat lower relative abundances than denitrifiers (SUP05, Marinimicrobia and/or Arcobacter), theyplay a similar role in N-species transformations and N2 production as well as overall energytransduction in low-oxygen and anoxic marine waters.3.4.4 Model of NO–2 competition between anammox and complete denitrificationBased on the results described above, we built a flux balance model to study the competitionfor NO–2 between anammox and complete denitrification, testing if we could reproduce therates corresponding to the two community phenotypes proposed (high or low N2 production).Lower rates of denitrification are attributed to a stagnation phenotype, whereas higher ratesof denitrification correspond to a renewal phenotype (Fig. 3.12). The rates of anammox, NO–3reduction to NO–2 , and complete denitrification (NO–2 to N2), are described through Michaelis-Menten equations, depending on both substrates (electron donors and electron acceptors), theirrespective kinetic parameters (km and Vmax) for each of these substrates, cell abundance andbiomass yield (Y) (see Appendix C for a complete description of the model). Both NO–3 reductionto NO–2 , and complete denitrification are sulphide-dependent. Nutrient concentrations of interest(NO–3 , NO–2 , NH+4 and HS– ) are calculated based on the rates of anammox, NO–3 reductionto NO–2 , and complete denitrification, as well as fixed input fluxes of the substrates throughpossible advection and diffusion (see Appendix C). These fluxes, however, are fixed throughoutthe simulation and do not reflect the highly dynamic nature of the nutrient fluxes found inSI, specifically through a renewal event. This model has thus been built to represent the twophenotypes introduced in the previous section in a steady-state scenario.The stagnation phenotype represents the background state and characterizes the inlet through-out most of the year, with limited input of NO–3 and higher fluxes of NH+4 and HS– coming from77100m120m135m150m200mJ F M A M J J A S O N DRelative Abundance (%)SHBH1141 SHBH391 Arctic96B-7 SHAN400SHBH1141 SHBH391 Arctic96B-7 SHAN400SHBH1141 SHBH391 Arctic96B-7 SHAN400SHBH1141 SHBH391 Arctic96B-7 SHAN400SHBH1141 SHBH391 Arctic96B-7 SHAN400denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●●●●●●●●denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●●●●●●●denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH39denovo12; SHBH 141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade2●●●●●●●denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●●●●●●denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●●●●●●denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●●●●●●denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●●; SHAN400denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade2●●●●●●●denovo15; SHAN400denovo13;Arctic96B−7denovo11; S BH391denovo12; SHBH1141denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH 141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●●●●●●denovo15; SHAN400denovo13;Arctic96B−7denovo11; S BH391denovo12; SHBH1141denovo15; S AN 00denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●●●●●●denovo15; SHAN400denovo13;Arctic96B−7denovo11; S BH391denovo12; SHBH1141denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●●●●●●denovo15; SHAN400denovo13;Arctic96B−7denovo11; S BH391denovo12; SHBH1141denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●●; SHAN400denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade2●●●●●●●denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●●●●●5; AN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141denovo15; SHAN400denovo13;Arctic96 −7denovo11; SHBH39denovo12; SHBH 14JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●●●●●5; AN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●●●●●5; AN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●●; AN400denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade2●●●●●●●denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●●●●●●denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●●●●●●denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141denovo15; SHAN4denovo13;Arctic9 −denovo11; SHBH3denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●●●●●●denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●●; SHAN400denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade2●●●●denovo15; SHAN400denovo13;Arctic96B−71; BH3912; SHBH1141denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●● 2●5 AN400denovo13;Arctic96B−71; BH3912; SHBH1141denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHB 1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●● 2●5 AN400denovo13;Arctic96B−71; BH3912; SHBH1141denovo15; SHAN400denovo13;Arctic96B−7denovo11; SHBH391denovo12; SHBH1141JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●● 2●5 AN400denovo13;Arctic96B−71; BH3912; SHBH1141denovo15; SHAN400denovo13;Arctic96 −denovo11; SHBH39denovo12; SHBH 14JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DECMonth2Clade20.90.9Abundance●● 25 AN400a.Denitrovibrio acetiphiphilus (AF146526)Caldithrix abyssi (A 430587)SI 200m clone SHBH1141 (GQ350776)SI 200m clone SHBH435 (GQ350809)SI Marinimicrobia OTU 2015 (denovo12)Atlantic Ocean clone ZA3648c (AF382142)HOT clone HF770D10 (DQ300775)P26 2000m clone P262000D03 (HQ674365)P4 1300m clone P41300E03 (HQ673390)HOT fosmid HF0010_18O13 (GU474850)Atlantic Ocean clone ZA3312c (AF382116)Arctic Ocean clone Arctic96B-7 (AF355047)HOT fosmid HF0500_01L02 (GU474916)SI Marinimicrobia OTU 2015 (denovo15)P26 2000m clone P262000N21 (HQ674572)HOT fosmid HF4000_22B16 (GU474892)SI 200m clone SHBH391 (GQ350786)SI 200m clone SHBH680 (GQ350930)SI 100m clone SHAG399 (GQ348837)SI 120m clone SHAN400 (GQ349121)SI Marinimicrobia OTU 2015 (denovo11)SI Marinimicrobia OTU 2015 (denovo13)Arctic95A-2 (AF355046)Monterey Bay 750m fosmid EBAC750-03B02 (AY458631)Arctic Ocean 400 m clone CB1343b.27 (GQ337204)Arabian Sea clone A714018 (AY907803)SAR406 (U34043)90502050100b.Figure 3.11: Marinimicrobia OTUs in SI, 2015 OTUs from the Marinimicrobia phylum in SI. (a) Relativeabundance of the 4 most abundant OTUs of the Marinimicrobia phylum in SI plotted by depths and months(b) maximum-likelihood phylogenetic tree of the small subunit rRNA for sequences of the Marinimicrobiaclade (ML implemented with RaxML, 100 bootstraps). OTUs identified in this study are highlighted ina colored box. Previously identified SAGs or genome bins are in bold. The bar represents 5% estimatedsequence divergence. Bootstrap value at the branch node indicate the robustness of the branching of thetree78the sediments underlying the anoxic water. To mimic this situation, we thus chose lower inputflux of NO–3 in comparison to NH+4 and HS– (Fig. 3.12 and table 3.3). Kinetic parameters (Km andVmax) for anammox bacteria were taken from the literature [109]. Kinetic parameters measuredfor denitrification in SI were determined during the renewal phenotype (August 2015, 165m) andtherefore, the stagnation pehnotype kinetic parameters remained unconstrained. Thus, kineticparameters for the SUP05/Marinimicrobia consortium (or stagnation phenotype) were fit to yieldrates of the same order of magnitude for denitrification as those measured in SI during stagnation(Fig. 3.12, table 3.3 and Appendix C). However, cell-specific growth rates (Vmax) and growthyield (Y) were estimated based on rates of denitrification and cell abundance (see above). Withnutrient fluxes appropriate for the stagnation period and reasonable physiological parametersfor the relevant organisms, we find that rates of both denitrification and anammox are similar,and fall within the range observed in SI outside of a renewal period (Fig. 3.12a, Fig. 3.3). Thedominance of one pathway over another could be inverted by changing the relative Km values fordenitrification and anammox, as these appear to be similar to one another in the stagnation period(see Appendix C). Modeled abundances of anammox bacteria were similar to those observed inSI (2.3*10 7 anammox cells L– 1) as well as the modeled abundances of complete denitrifiers thanobserved in SI (cell abundance of Marinimicrobia/SUP05=10 9 cells L– 1). Therefore, we couldreproduce with a simple flux balance model rates of the same order of magnitude for anammoxand complete denitrification measured during the year 2015 during peak stagnation in SI.Deep water renewal in SI introduces oxygenated water to the deep basin, where NO–3 isproduced through rapid nitrification, and the NO–3 produced is in turn lost through anaerobicNO–3 reduction and N2 production following renewal. The renewal event is thus far away fromsteady-state, and we therefore tried to reproduce the high rates of denitrification in 60 days of thesimulation, corresponding to the approximate duration of the event based on the geochemicalprofiles and rate measurements. The kinetic parameters for anammox parameters were identicalto simulations of the stagnant phenotype, however, we changed the kinetic parameters fordenitrification to represent Arcobacter in the renewal phenotype. As NO–3 dependency wasmeasured in August 2015, corresponding to higher rates of N2 production, we chose to use theMichaelis-Menten constants modeled from this data to describe complete denitrification (NO–3 to79N2) (tables 3.2 and 3.3). We also chose a higher cell-specific growth rate than in the ’background’phenotype (table 3.3), as it appears that the Arcobacter population grows rapidly from <1% to30% of relative abundance within a months time (table 3.3). The simulation reveals that underthe conditions described here (Fig. 3.12b and table 3.3), rates of denitrification reach 10 – 6 N2 Md– 1, which corresponds to the highest rates measured in July 2015 at 200m (Fig. 3.3). Rates ofanammox remain similar to the rates that can be found after renewal in SI (Figs. 3.3 and 3.11b).The abundance of complete denitrifiers reaches 10 9 cells L– 1 after only 15 days of simulation,which is again similar to abundances observed in SI during renewal. However, to fully mimic theinput fluxes in SI, the model would need to have dynamic fluxes that can be changed over time. Inaddition to dynamic fluxes, competition between two different population of complete denitrifiersshould be implemented to fully study the transition between one phenotype to the other.3.4.5 SI as a model ecosystem for coastal OMZsWe have extrapolated the annual N-loss calculated for SI (33 km2) to all similar coastal inlets inBC (2478 km2) in order to estimate the possible importance of BC coastal fjords to N-loss from theNorth Eastern Sub-Arctic Pacific Ocean. We estimate that these inlets could contribute up to 0.12Tg N yr-1, which constitutes 0.1 % to global pelagic N-loss [165] if they are all anoxic and similarto SI. On an area-specific basis this is extremely high in comparison to the ETSP, for example,which has a surface area of 1.2 · 10 6 km2 and supports up to 10 Tg N yr– 1 [60]. This highlightsthat coastal OMZs are hotspots for N-loss and could also, in the near future, be subject to changesdue to increased anthropogenic influence.The low oxygen conditions in SI support pelagic anaerobic microbial metabolisms includingdenitrification and anammox that co-occur and underpin high rates of N-loss from the watercolumn. We showed that denitrification is the most important contributor to N2 productionand its rates and the organisms responsible vary seasonally. Rates of anammox, in contrast, arerelatively constant throughout the year, contributing 37% of the N-loss from SI. Anammox isoften reported as the primary pathway of N-loss from OMZs [18, 60, 65–67, 84, 98, 154], and ourtime-series observations from SI may be more broadly extensible to low oxygen marine watersglobally. In SI, rates and pathways of N-loss and the responsible microbial taxa are dynamic800 100 200 300days10-1010-910-810-710-610-510-4Nutrients (moles L-1)NH4+NO3-HS-NO2-0 100 200 300days1071081091010Cell abundance (cells L-1)ANNO3RDEN0 100 200 300days10-1010-910-810-7Reactionrate(molesL-1 d-1)ANNO3RDEN0 20 40 60days10-910-810-710-610-510-410-3Nutrients(molesL-1)NH4+NO3-HS-NO2-0 20 40 60days10610710810910101011Cell abundance (cells L-1)ANNO3RDEN0 20 40 60days10-1110-1010-910-810-710-610-5Reactionrate(molesL-1 d-1)ANNO3RDENOMNH4+N2N2ONO3-NO2-N2O?NH4+?NO3-H2SO2NH4+S3a. b. c.NH4+N2N2ONONO2-NO3-Org N Org NNH4+N2N2ONONO2-NO3-PlanctomycetesSUP05 uncultured bacteriumNitrospira defluvii/Nitrospina gracilisSAR11 cladeThaumarcheotaOrg Na. b. c.2.5 Ga4 Ga4.5 Ga TodayMarinimicrobiaOMNH4+N2N2ONO3-NO2-N2O?NH4+?NO3-H2SO2NH4+S3a. b. c.NH4+N2N2ONONO2-NO3-Org N Org NNH4+N2N2ONONO2-NO3-PlanctomycetesSUP05 uncultured bacteriumNitrospira defluvii/Nitrospina gracilisSAR11 cladeThaumarcheotaOrg Na. b. c.2.5 Ga4 Ga4.5 Ga TodayMarinimicrobiaArcobacter sp. ?a. b. c. d.e. f. g. f.Figure 3.12: Model of NO–2 competition between anammox and complete denitrification model of NO–2 competitionbetween anammox and complete denitrification for the 2 phenotypes found in SI, stagnation phenotypeand renewal phenotype. a) represents the phenotype found in SI during stagnant periods of time. (b, cand d) shows a simulation of the model that tested to see if the phenotype of the phenotype could bereproduced through modeling. Parameters used to model this simulation can be found in table 3.3. c)represents the second phenotype found in SI during renewal event, leading to higher N2 production (d, eand f) shows that the model reproduced higher rates of N2 production. Parameters used in this simulationcan be found in table 3.3.81Table 3.3: Parameters used in model for competition of NO–2 between anammox and complete denitrifica-tion (Fig. 3.12)Ecotype Parameters Value & units Reference ‘Stagnation’ RNO3 RNO2 RNH4 RHS Km_AN for NH4+ Km_AN for NO2- 5x10-9 moles L-1 d-1 5x10-9 moles L-1 d-1 5x10-8 moles L-1 d-1 5x10-8 moles L-1 d-1 3 µM 0.45 µM State of stagnation in SI “ “ “ Awata et al. 2013 Awata et al. 2013   Km_DEN & NO3R for HS- Km_DEN & NO3R for NO2- or NO3- 10 µM 1 µM and 5 µM Jensen et al. 2009 and this paper Fit the rates in SI   Vmax (AN, NO3R) Vmax (DEN) YAN YNO3R YDEN 2x10-14 moles N2 cell-1 d-1 2x10-15 moles N2 cell-1 d-1 5x1013 cell (moles ED)-1 5x1014 cell (moles ED)-1 5x1015 cell (moles ED)-1 Strous et al. 1999 Fit the rates in SI Louca et al. 2016 Fit the rates in SI “ ‘Renewal’ RNO3 RNO2 RNH4 RHS Km_AN for NH4+ Km_AN for NO2- 5x10-6 moles L-1 d-1 5x10-8 moles L-1 d-1 5x10-7 moles L-1 d-1 5x10-8  moles L-1 d-1 3 µM 0.45 µM After renewal in SI “ “ “ Awata et al. 2013 Awata et al. 2013   Km_DEN & NO3R for HS- Km_DEN & NO3R for NO2- or NO3- 10 µM 5 µM Jensen et al. 2009 and this paper This paper, August 2015 (165m)   Vmax (AN, NO3R) Vmax (DEN) YAN YNO3R YDEN 2x10-14 moles N2 cell-1 d-1 2x10-13 moles N2 cell-1 d-1 5x1013 cell (moles ED)-1 5x1014 cell (moles ED)-1 1.5x1015 cell (moles ED)-1 Strous et al. 1999 Fit the rates in SI Louca et al. 2016 Fit the rates in SI “  82responding to substrate fluxes driven by physical forcings. Analogous dynamics in upwellingand horizontal transport or large-scale eddies in open ocean OMZs may also lead to strongmicrobial responses with corresponding biogeochemical outcomes [103, 214]. While sulphidicconditions that characterize SI are rare in modern open ocean OMZs, they could become prevalentwith progressive ocean deoxygenation [121, 210]. Information on microbial responses to systemdynamics and on the ecophysiology the underpins coupled C, N, and S cycling in Saanich Inlet andother experimentally tractable coastal ecosystems is key for predicting broader global responsesto ocean deoxygenation and the expansion of marine anoxia.83Chapter 4Combining microbiological andgeochemical information to constrainenergy flow through the marine N-cycleModern oceans contain large volumes of anoxic water that are currently expanding due toanthropogenic activities. Importantly, high rates of anaerobic N-metabolisms characterize theseanoxic waters, resulting in intense cycling of N through microbial metabolisms. This can eitherlead to N-loss or N-retention, depending on the partitioning of N-reduction across denitrification,anammox, and dissimilatory NO–3 reduction to NH+4 (DNRA). The outcome therefore influencesmarine N-budgets and thus can impact biological production, the marine C-cycle, and climate.While substrate supply rates are a first order control on the rates of N-reduction, the controls onpartitioning across the different pathways remain uncertain and this confounds efforts to predictthe response of the marine N-cycle to deoxygenation. Here we show that DNRA dominatesN-reduction on an annual basis in Saanich Inlet, a persistently anoxic fjord that serves as ananalogue for anaerobic marine microbial metabolisms. While anammox and denitrificationplay an important role throughout most of the year, high rates of DNRA develop followingintroduction of new oxidants and substrates to the anoxic deep-waters during renewal events.These events provided enhanced energy fluxes, or power supply, that fueled higher rates ofN-reduction and altered the microbial community structure and metabolic potential. Notably,changes in microbial community abundance, structure and metabolic potential did not scale withcorresponding metabolic rates, and this undermines attempts to model biogeochemical cyclingwith gene-centric modeling theory, which inherently relies on such scaling. The observation that84DNRA emerges transiently, to dominate N-reduction, in response to physical perturbations thatenhance power supply, suggests that N-recycling should be considered in models that aim topredict biogeochemical responses to ocean deoxygenation.4.1 IntroductionNitrogen (N) is an essential nutrient for life and thus often limits primary production in theoceans. The marine N-cycle, therefore, is tightly coupled to biological CO2 sequestration, creatingfeedbacks between N biogeochemistry and climate [9, 215] – dynamics in ocean N-inventoriescan thus have large-scale effects on the Earth system [157]. Ocean N-inventories are primarily setby the balance between fixation of atmospheric N2, largely by photosynthetic microorganismsinhabiting the sunlit surface ocean [9], NO–3 -supply via terrestrial runoff, and removal by N2producing organisms in anoxic regions of the oceans and coastal sediments [9, 216]. N-scarcityin the oceans develops when biological N-fixation and terrestrial nitrate runoff is outpaced byN2 production, which can happen when marine anoxic water masses expand [16, 217]. N-lossunder anoxic conditions, however, can be short-circuited by dissimilatory nitrate reduction toammonium (DNRA), or counteracted by enhanced N-fixation within or beyond the euphotic zone.Dynamics between these interacting processes ultimately control ocean N-inventories impactingprimary production and climate [42, 44, 178].Connections between ocean N-inventories, marine anoxia, N-fixation, anaerobic N-metabolisms,primary production, and climate are evident throughout Earth’s history, with notable examplesduring the Paleoproterozoic [2.5 – 1.6 Ga ago, [16, 39, 44, 178, 217, 218]] and during PhanerozoicOceanic Anoxic events (OAEs)[219]. Indeed, N-scarcity may have characterized much of thePaleoproterozoic due to a combination of appreciable nitrification, supported by oxygen in thesurface ocean, and volumetrically expansive masses of underlying anoxic ocean waters supportingdenitrification and possibly anammox [16, 39, 217, 218]. Such N-scarcity could have limitedglobal biological productivity, influencing atmospheric chemistry and climate [16, 217, 218]. OtherN-metabolisms, such as N-fixation or DNRA however, likely counteracted N-loss to maintainN availability in the oceans during specific intervals [39, 44, 178], and can lead to changes in85ocean chemistry, from ferruginous to euxinic conditions. Maintenance of N availability, in light ofstrong N-sinks, would have been essential to supporting biological production with correspondinginfluences on ocean chemistry, atmospheric oxygen concentrations [178], and the strength of themarine C-sink [8, 39]. In the Phanerozoic eon, by contrast, abundantly available N may have drivenintervals with elevated primary production, which in turn contributed to ocean deoxygenation,and the onset of Oceanic Anoxic Events, along with biological crises [219–221]. The modern oceansare also losing oxygen, which is the combined result of anthropogenic climate warming andocean nutrient loading [72]. As anoxic water masses expand, so too may the strength of marineN2 production with corresponding implications for ocean N inventories, primary production,and climate. If deoxygenation favored DNRA and N-retention, however, it could instead leadto positive feedbacks on primary production and ocean deoxygenation. In addition to drivingmarine deoxygenation, both climate warming and nutrient loading thus influence the distributionof N in the oceans. In the absence of predictive models that accurately diagnose the responseof the N-cycle to ocean deoxygenation, however, predictions of future N inventories and thecorresponding feedbacks on biological production and climate remain unconstrained.Anoxic and low oxygen marine waters are characterized by high rates of anaerobic N-metabolisms, and the partitioning of these metabolisms between anammox, denitrification andDNRA dictates N-loss versus N-retention, and influences oceanic N-inventories. N-loss, throughN2 production, is the result of NO–3 and NO–2 reduction through the microbial metabolismsdenitrification and anammox. NO–3 and NO–2 reduction can also be channeled through DNRA,which short-circuits N2 production and retains fixed N in the ocean. Denitrification and anammoxreactions and corresponding N2 production, are widespread throughout marine oxygen minimumzones (OMZs) and in coastal shelf sediments. Rates of these microbial metabolisms have beenextensively determined [18, 56, 60, 65–67, 84, 85, 98, 100, 154, 171, 222], and together with mea-surements of N-fixation as well as models of the distribution of nutrients (for example: [223]),form the basis for current ocean N-budgets [1, 2, 63]. Despite the wealth of information on ratesof denitrification and anammox, the controls on partitioning between these metabolisms, remainuncertain. Measurements of DNRA, by contrast, are sparse, although it has been transientlydetected in OMZs [60, 65, 67, 224] and it can be an important pathway in N-cycling in estuarine86sediments [225]. While all three of these NO–3 /NO2- reduction pathways can be simultaneouslystudied using 15N-labeling experiments, there remains insufficient ecophysiological informationon the relevant organisms with which to build predictive models that would inform the responseof the marine N-cycle to ocean deoxygenation. Acquiring this ecophysiological information isbecoming possible using new meta’omic information coupled with process rate measurements,and their collective integration into gene-centric modeling frameworks [118, 119]. Even withthese approaches, determining the ecophysiologies of the relevant microorganisms, however,is confounded by the diversity and complexity of marine microbial communities, and theircorresponding interactions with the surrounding environment.Substrate availability is a key constraint of rates of microbial metabolism and this likelyplays a role in governing the relative importance of the different anaerobic N-metabolisms[60, 67, 105]. Conventional approaches often consider the interaction between microorganismsand their substrates in terms of free energy availability [105, 108, 226], and enzyme-kinetics [227].Classical studies, for example, predict the cascade of terminal electron acceptor use in redoxstratified environments based on the successive use of the electron acceptor yielding the mostfree energy upon reaction with a given electron donor [104, 216]. However, the rate at whichthis energy can be supplied – rather than its potential availability – also places constraints onmicrobial growth [228, 229]. The rate of energy supply, in essence the power supply [228, 229],can be calculated as the product of free energy yields and corresponding reaction rates [228, 229].Notably, since power supply depends on the rate at which a reaction substrate is supplied, it alsodepends on physical transport as well as ambient substrate concentrations. Power supply ratescan thus be inferred by combining geochemical information to estimate free energy yields withmeasurements of metabolic reaction rates [228, 229] that through mass balance necessarily reflectrates of substrate supply. Organic matter degradation rates, for example, have been combinedwith free energy yields for specific metabolic reactions that occur in marine sediments [228] toinfer the power supply in the marine subsurface. Power supply, however, has rarely been exploredmore broadly as a large-scale regulator of biogeochemical processes and microbial ecology in theoceans.We have used a time-series experiment in Saanich Inlet (SI) to determine the response of the87N-cycle and its underlying microbial catalysts to dynamics in power supply induced by oceancurrents. SI is an anoxic fjord on the east coast of Vancouver Island, Canada, that undergoespartial renewals of its water column due to a combination of weak tidal action and upwelling ofdense water into the Strait of Georgia [179]. SI represents an experimentally tractable ecosystemin which to study biogeochemical processes and microbial ecology more broadly extensible to lowoxygen marine waters in the global ocean. Multi-omic sequencing approaches have been used todevelop qualitative models that describe the microbial N metabolisms that underpin the N-cyclein SI [33]. These qualitative models have been extended to a quantitative reaction-transport andgene-centric modeling approach that couples multi-omic sequence information with geochemistryto link gene abundances to key microbial metabolic pathways and their rates under steady-stateconditions [118]. More recent studies, however, imply year-long dynamics in SI with strongvariation in the rates and pathways of N2 production associated with renewal events ([186, 230]and chapter 3) and these cannot be addressed through the existing gene-centric framework. Priorstudies in SI, furthermore, have overlooked the potential role of DNRA, and while its operationwas not evident through gene-centric modeling [118], proteins related to cytochromes that confermetabolic potential for DNRA have been detected in SI, implying a possible role for DNRA [33]. Wemeasured rates and pathways of NOx (NO–3 and NO–2 ) reduction through DNRA, denitrificationand anammox monthly in the water column of SI across two consecutive years, through stagnationperiods and renewal events. We combined geochemical profiles, thermodynamic calculations,process rate measurements, and analyses of the metabolic potential of microbial communitiesusing metagenomics to determine how N-metabolisms respond to dynamics in substrate supplyregimes. We show that DNRA becomes the dominant NO–3 reduction pathway and mode of energytransduction following renewal events. These renewal events were also followed by increasein power supply, linking power supply to the outcome of networked microbial N-metabolisms.Our results, therefore, suggest that dynamics in power supply influence microbial communitymetabolisms with consequences for N-budgets and elemental cycles more broadly.884.2 Methods4.2.1 Study site and sample collectionSampling took place in Saanich Inlet (BC, Canada) at Station 3 (48◦ 35.5 N and 123◦ 30.3 W,227m deep). The study was spanned from February 2015 to January 2017, with the exceptionof January 2016. Exact dates of sampling can be found in Appendix D (table C.4). For eachmonth, a CTD profile was taken for the following parameters with the corresponding sensors inparentheses: pressure (SBE 29), conductivity (SBE 4C), temperature (SBE 3F), and oxygen (SBE 43).Samples for chemical analyses were obtained from 16 depths (10, 20, 40, 60, 75, 85, 90, 97, 100,110, 120, 135, 150, 165, 185 and 200m) as described in chapter 3 [230], for NO–3 , NO–2 , NH+4 andHS– . Nutrient samples for NO–3 , NO–2 and NH+4 were filtered through a 0.2µm filter and the HS–samples were fixed in 0.5% ZnAc final concentration. Both sets of samples were then frozen untilanalysis. Water samples for DNA filtration were taken from 6 depths (10, 100, 120, 135, 150 and200m) with a volume of 10L and were filtered back in the lab the night after the sampling. Thefiltration apparatus consists of a pre-filter (glass fiber filter 2µm pore size) and a 0.2µm sterivexfilter (Millipore Sigma). Both filters were then filled with lysis buffer and frozen immediately inliquid nitrogen and kept at -80◦C. Samples for process rate measurements were collected from 7depths (90, 100, 120, 135, 150, 165 and 200m) in 250mL serum glass bottles, overflowed 3 times,and closed without any air bubble with blue butyl rubber stoppers (Bellco glass). The bottles wereput on ice before processing in the lab the night of the sampling.4.2.2 Nutrient analysesSamples for NO–2 , NH+4 and HS– were thawed before analysis and measured according tothe following spectrophotometric assays: the Griess assay, the indophenol blue method andthe Cline assay, respectively [151]. NOx (NO–2 and NO–3 ) concentrations were measured bychemiluminescence following reduction with vanadium [189], and NO–3 concentrations wereobtained by subtracting NO–2 from the total NOx concentrations.894.2.3 Process rate measurements15N-labeling incubations were performed according to chapter 3 [230]. Rates of anammox werecalculated from the net accumulation of 29N2 in the headspace of the samples, with addition of15NH+4 +14NO–3 or with addition of15NO–3 [83]. Rates of denitrification were calculated fromthe net accumulation of 30N2 in the headspace of the samples, with addition of 15NO–3 . Rates ofDNRA were calculated based on the net accumulation of 15NH+4 in the liquid samples from15NOxlabeled incubation, after transformation of 15NH+4 to15N2 [231]. The excess of 14N15N and 15N15Nin the samples were measured by Isotope-Ratio Mass Spectrometry (Delta V, thermoscientific).Concentrations of 29N2 and 30N2 were determined according to chapter 3 [230] and Thamdrup etal. (2006)[154]. Rates of the processes were calculated based on least squares fitting of the slopeof 15N accumulation, correcting for the initial 15N2 present and accounting for the initial pool ofunlabelled substrates present (chapter 3 and [230]). A detailed table of the incubations can befound in Appendix D (Table D.4).4.2.4 DNA extraction, qPCR and absolute cell abundanceDNA was extracted from sterivex filters following Wright et al. (2009) [194] for 4 months in2016 (April, August, September and October). Once extracted, DNA was checked for qualityand sent for high-throughput metagenomic sequencing at the Joint Genome Institute (WalnutCreek, California). We also quantified total bacterial and archaeal 16S rRNA genes (or SSUrRNA gene) via qPCR by targeting the bacterial and archaeal 16S rRNA genes with the primers27F/20F (5’-AGAGTTTGATCCTGGCTCAG, 5’-TTCCGGTTGATCCYGCCRG) and DW519R (5’-GNTTTACCGCGGCKGCTG) [183], respectively. Standards used for total bacteria and totalarchaea quantification were obtained from a SSU rRNA gene clone libraries as described inZaikova et al. (2010) [183]. qPCR program was as follows: (1) 95◦C for 3 minutes, (2) 95◦C for20 seconds, (3) 55◦C for 30 seconds, (4) read, repeat (2) to (4) 44 times, obtain melting curveby incrementing 0.5◦C from 55◦C to 95◦C every second. qPCR reactions were performed inlow-profile PCR 96 well-plates (BioRad) in 20µl volume reaction on a CFX Connect Real-Timethermocycler (BioRad). 16S rRNA gene abundance (16S rRNA gene L– 1) was multiplied by the90average number of 16S rRNA gene copies found in bacterial genomes (4.2) [232] and was used asa proxy for absolute cell abundance (cell L– 1).4.2.5 Metagenome sequencing and assemblyTwenty samples (4 months and 5 depths) were used to generate metagenomic datasets at the DOE-JGI (Walnut Creek, California) following the protocols for library production and sequencing onthe Illumina HiSeq platform and the accession numbers can be found in Table D.5. The sequenceswere first trimmed and filtered using Trimmomatic (Bolger et al., 2014). Adapters supplied inthe Illumina TruSeq3 adapter sequence file were removed by Trimmomatic’s ILLUMINACLIPcommand. Next, the first three and last three nucleotides were removed from each read ifbelow a quality threshold, and a sliding window of four nucleotides was checked based onaverage Phred score (Qscore). Nucleotides within these windows were removed until the averageQscore across the window was >15. Finally, the sequence reads with <36 bp were removed,along with their mate-pair reads. Paired-end sequencing reads were assembled into contigsusing Megahit [233], with default settings. Open Reading Frames (ORFs) were predicted usingProdigal v2.0 (, based on a minimum nucleotidelength of 60 as implemented in MetaPathways 2.5 [234] and Metagenomic data sets are accessible through the JGI IMG/M portal( under the study name ’Mapping the globalMethanome’ (project ID 503042) and raw reads at the NCBI Sequence Read Archive numbers forall the samples can be found in the Appendix D.4.2.6 Taxonomy of of the microbial community recovered through metagenomicanalyses16S rRNA gene sequences were retrieved from metagenomic datasets (trimmed reads) usingEMIRGE [235]. We then used the retrieved 16S rRNA gene sequences to reconstruct the taxonomiccomposition of the microbial community. The sequences were aligned and classified using thelatest SILVA database at a 90% identity cut-off.914.2.7 Quantification of functional genesCurated reference sequences for genes involved in nitrogen cycling were downloaded fromFunGene ( except for hzs, hao and hzo [236]. The other genes weredownloaded from GenBank and manually curated based on sequence length, annotations, andphylogeny [237]. A complete list of the genes studied in this paper can be found in Appendix D(Table D.3). Chosen functional genes were quantified using TreeSAPP, a phylogenetic-basedprotein profiling software (available at TreeSAPPwas used to both build the reference trees and map putative protein sequences to these treesfor functional classification and quantification. Specifically, the script create treesapp ref data.pywas used to cluster the reference protein sequences with UCLUST at some percent similaritysuch that trees contained between 150 and 600 sequences [238]; this range balanced highlyaccurate taxonomic annotations and reasonable compute time to determine the optimal sequenceplacements in the phylogenetic trees. Following clustering, MAFFT version 7.294b was usedwith the –maxiterate 1000 and –localpair settings to generate a multiple sequence alignmentand trimAl version 1.4.rev15 was used to remove non-conserved positions [239, 240]. RAxMLversion 8.2.0 was used to build the reference trees with the’–autoMRE’ to decide when to quitbootstrapping before 1000 replicates have been performed and PROTGAMMAAUTO to select theoptimal protein model [241][242].TreeSAPP’s was then used to map the query sequences (protein sequences fromcalled genes in contigs) onto these reference trees using the following procedure: Proteins werepredicted using Prodigal version 2.50 from the assembled metagenomic contigs [243]. Open-reading frame (ORF) protein sequences were aligned to HMMs using hmmsearch and the alignedregions were extracted [244]. hmmalign was used to include the new query sequences in thereference multiple alignment and then trimAl removed the unconserved positions from thealignment file [239]. RAxML was used to classify the query sequences using thorough sequenceinsertions [245]. Quality-controlled reads provided by the JGI were aligned to the nucleotide ORFsusing BWA MEM [246]. An executable that calculates reads per kilobase per million mappablereads (RPKM) from a SAM file was used to generate these normalized abundance values for each92predicted protein.4.3 Results and discussion4.3.1 Dynamics in rates and pathways of N-cyclingDuring most of the year, the water column in SI is stagnant, with O2 and NO–3 abundant inthe surface waters and scarce at depth. Below ∼100-110m, water in SI is usually devoid of O2and NO–3 . In the bottom waters, NH+4 and HS– accumulate (Fig. D.1), likely diffusing out ofthe underlying sediments [185]. Geochemical dynamics within the water column, however, areinduced transiently by renewal events in intermediate and deep-waters (100 to 200m). Theserenewal events result from the input of dense water from outside of the inlet that spills over the sillat the inlets entrance and settles to depths of equal density (isopycnals) [184]. This new mass ofwater introduces O2 and NO–3 that can accumulate to detectable concentrations depending on theflux of renewal water, its mixing with deep water, and the rate, abiotic or biologically catalyzed,at which O2 and NO–3 react with reducing agents like HS– and NH+4 . Monthly profiles of O2and NO–3 concentrations reveal 6 renewal events between 2015 and 2016 (Fig. 4.1 - grey shading,and Fig. D.1). These renewal events were marked by increases in concentrations of O2 and/orNO–3 in intermediate or deep-waters relative to previous months. The strongest renewal over thisperiod developed in September 2016 where high O2 and NO–3 concentrations were recorded indeep-waters (50µM and 12µM, respectively). During some events, O2 remained undetectable indeep and intermediate waters and yet NO–3 accumulated, for example in July of 2015, revealingrenewal, nonetheless (Fig. D.1). Renewal events thus often occurred at the end of the summer,such as in 2015. Several renewal events, however, were recorded throughout 2016 and the inletwent through several cycles of renewal/stagnation (at least 6 detected) over the 2 years studiedhere.To test the responses of anaerobic N-metabolisms to renewal-induced physical-chemical dy-namics, we measured rates and pathways of microbial N-metabolisms, specifically anammox,denitrification, and DNRA. Rates of anammox, denitrification and DNRA, integrated over thedepth of the anoxic water column (Fig. 4.1b and c), change in response to renewal, revealing93dynamics in the partitioning of NOx (NO–3 /NO–2 ) reduction between these processes with impli-cations for N-loss and retention. Overall, total NOx reduction (through DNRA, denitrificationand anammox) varies between 7 · 10 – 4 and 1.4 moles N m– 2 d– 1, spanning several orders ofmagnitude over the two-year period. The highest rates of NOx reduction followed renewal events.Five of the six renewal events were, indeed, followed by elevated rates of NOx reduction andalso corresponded to very high rates of DNRA (Fig. 4.1a and b – August and November 2015,February, May, October 2016). The single other renewal event, by contrast, was dominated by highrates of denitrification (Fig. 4.1a and b – August 2016). NOx reduction rates following renewalwere several orders of magnitude higher than those measured in stagnation periods (e.g. Juneand December 2015, Fig. 4.1b), and developed in association with a large advective supply of newoxidants, mainly O2 (Fig. 4.1a), which likely fueled nitrification. Nitrification, in turn, suppliedoxidized N-species (NOx) needed to support higher rates of anaerobic N-metabolisms. Rates ofnitrification were previously shown to increase after renewal [247] and the rates measured in thelow oxygen waters would have been sufficient to support the highest volumetric rates of anaerobicN-metabolisms measured in this study.The distribution of NOx reduction across the 3 pathways varies strongly over time. Shifts inanaerobic N-metabolisms were shown to occur, specifically between a regime of fixed N-retentionthrough DNRA following renewal (e.g. August 2015 and October 2016 in Fig. 4.1b) and one ofN-loss driven mainly through denitrification and to a lesser extent, anammox (e.g. March and July2016 in Fig. 4.1b and c). N-retention through DNRA totaled 58% of the total NOx reduction whenintegrated over the two years studied, making DNRA the dominant pathway for NOx reduction inSI. Few previous studies have observed DNRA in such a prominent role in N-cycling with notableexceptions in estuarine and coastal margin sediments [91, 225]. Appreciable DNRA has alsobeen detected in the Peruvian OMZ [67], but there, DNRA represented a minor fraction (<12%)of the total NOx reduction. Our results thus reveal strong dynamics in the rates and pathwaysof anaerobic N-metabolisms in SI, with a notable rise of DNRA to prominence in response torenewal events. These dynamics in anaerobic N-metabolisms are the likely phenotypic expressionof changes in the underlying microbial community structure and function, which we explorebelow.9453020010000. supply (kJ m-2 d-1)0100200Depth (m)250100501050O 2 (µM)DNRA & DEN (mol N m-2 d-1)03 10-35 10-37 10-31 10-2AN (mol N m-2 d-1)Power supplyDenitrificationDNRAAnammoxJ     F     M    A     M    J     J     A     S     O    N    D J     F     M    A     M    J     J     A     S     O    N    D2015 2016Figure 4.1: Rates and pathways of anaerobic N-metabolisms. (a) Oxygen concentrations (µM) and (b) Depth-integrated rates of denitrification (blue) and DNRA (pink) in moles N m– 2 d– 1, as well as power supply(kJ m– 2 d– 1) plotted as histograms (c) depth integrated rates of anammox (purple) in moles N m– 2 d– 1.Standard error on the depth-integrated rates were compiled in table D.1 of Appendix D, and were notdepicted in this figure as they were smaller than the square depicting the data point.954.3.2 Dynamics of the microbial community in response to physical perturbationsWe find that microbial community composition and structure changes in response to renewal.Environmental disturbances have previously been linked both to increased [248] and decreased[249] microbial community diversity, as well as to changes in activity. Here we observe that thediversity of the microbial community decreased following renewal events. Changes in diversitywere evident in both the number of observed species as well as in diversity indices (Figs. 4.2and 4.3e). We note that the decrease in diversity was also accompanied by an increase in cellabundance (see methods for description of cell abundance). Such an increase in cell abundanceand a decrease in diversity, in less than a month, suggests that only a few specific taxa have grownfollowing enhanced substrate supply rates – mainly O2 and NOx – and the decrease in diversityis the likely result of a few blooming taxa. The taxon Arcobacter, of the Epsilonproteobacteriawhich are commonly known as blooming organisms [250] for example, increased in absoluteabundance after renewal (Figs. 4.2 and 4.3 and see methods for absolute cell abundance), andcould be partially responsible for the decrease in diversity. Taken together these observationssuggest that renewal induces growth of blooming organisms and this occurs in association withincreased rates of N-metabolisms such as denitrification or DNRA. This may thus imply thatmicrobial growth is stimulated by enhanced N-supply and linked to specific N-metabolisms.96150mAprilAugustSeptemberOctober135mAprilAugustSeptemberOctober120mAprilAugustSeptemberOctober100mAprilAugustSeptemberOctober200mAprilAugustSeptemberOctoberγ-proteobacteriaLentisphaeraeMarinimicrobia (SAR406 clade)NanoarchaeotaNitrospinaePatescibacteriaPlantomycetesThaumarchaeotaUnclassifiedVerrucomicrobia16S rRNA gene counts Actinobacteriaα-proteobacteriaArchaea (<1%)Archaea (unclassified)Bacteria(<1%)Bacteria(unclassified)BacteroidetesChloroflexiδ-proteobacteriaε-proteobacteriaEuryarchaeotaTaxaOTUs observedα-diversity (Shannon index)Abundance (106 cell L-1)00.11010010005000OTUs observedα-diversity (Shannon)50100250200234Figure 4.2: Taxonomic composition of the microbial community. The taxonomic composition of the microbial communities is plotted at the phylumlevel, and was obtained from 16S rRNA gene extraction via EMIRGE from metagenomic samples for 4 months and 5 depths (100, 120, 135, 150and 200m).97a.b.c.d.Arcobacter sp. Ectothiorhodospiracaea unculturedSUP05 cladeNitrospina sp.Ca. NitrosopumilusCa. ScalinduaMarinimicrobia cladeDomainBacteria ArchaeaPhylumProteobacteriaε-ProteobacteriaNitrospinaePlanctomycetesOthersClassMarinimicrobiaγ-Proteobacteriaα-Proteobacteriaδ-ProteobacteriaThaumarchaeotaEuryarchaeotaOrderThiomicrospiralesEctothiorhodospiralesOther γ-proteobacteriaSAR11 cladeOther α-proteobacteriaOther ε-proteobacteriae.OTUs observedCell counts (cell m-2)Shannon index500 -250 -04e13 -2e13 -002 -4 -April Aug Sept OctFigure 4.3: Depth-integrated taxonomic composition of microbial communities. The taxonomic composition of the microbial communities is plottedat the OTU level (97% similarity), and was obtained from 16S rRNA gene extraction via EMIRGE from metagenomic samples for 4 months(a=April, b=August, c=September, d=October) and 5 depths (100, 120, 135, 150 and 200m). The absolute abundance (see methods) for eachsequence was integrated over the 5 depths mentioned here. Each concentric circle represents a taxonomic level. e) shows OTUs observed duringthe 4 months (April, August, September and October) cell abundances, and the alpha-diversity index – Shannon.98Emerging gene-centric modeling approaches imply that rates of biogeochemical reactionsscale with the abundance of genes that code for enzymes involved in that pathway and thus,based on these approaches we might expect a direct link between dynamics in organismal andgene abundances, substrate supplies, and metabolic rates [118, 119]. In gene-centric modelingapproaches, biogeochemical reaction rates are tied to the growth rates and abundances of theorganisms that host relevant genes and pathways. While it is often assumed that rates of genetranscription better reflect microbial activity than gene or organism abundances [251], multi-omicmodeling implied that, in anaerobic marine environmental systems, the time-scales of microbialgrowth and biogeochemical reaction rates were similar and thus best linked through gene andorganism abundances [118]. It follows then that the dynamics we observe in rates and pathwaysof anaerobic N-metabolisms in SI should be tied to changes in the abundances of genes andtheir host organisms involved in these pathways. Renewal events indeed induce both microbialcommunity growth (Figs. 4.2 and 4.3) and cause enhanced rates of anaerobic N-metabolisms(Fig. 4.1), implying a connection between growth and biogeochemical reaction rates in SI. It maybe possible, therefore, to connect the abundance of taxa involved in N-cycling to the reactionrates, in line with multi-omic modeling theory [118, 119]. Previous studies identified key-taxa thatsupport N-cycling in SI [33, 187], and provide a benchmark against which to evaluate microbialcommunity dynamics in response to perturbation and test the idea that biogeochemical reactionrates scale with the abundance of relevant genes and organisms.SUP05 is a prominent member of the microbial community throughout the period of obser-vation and its abundance does not vary strongly despite dynamics in the rates and pathwaysof anaerobic N-metabolisms (Fig.4.4). SUP05 has been implicated in partial denitrification, inessence NO–3 reduction to N2O, in SI and other low oxygen marine waters, globally [18, 33, 121].According to gene-centric models, we might thus have expected the abundances of SUP05 to scalewith rates of denitrification. While rates of denitrification vary over nearly 3 orders of magnitude(Fig. 4.1b), the abundance of SUP05 changes less than 10’s of percent over the same time period(Figs. 4.2 and 4.3). The relationship between SUP05 and the rates of denitrification may be con-founded by several factors. Notably, SUP05 in SI may operate in partnership with Marinimicrobiato achieve complete reduction from N2O to N2 [187] and thus partial denitrification may obscure99the relationship between rates of denitrification and SUP05 abundances. We note, however, thatN2O did not accumulate to appreciable concentrations during our experiments implying thatdenitrification, when operative, was complete. Marinimicrobia abundances, furthermore, arenegatively correlated with rates of denitrification (Fig. 4.3). These observations, therefore, suggestthat there is a decoupling between the rates of denitrification and the abundances of SUP05 andMarinimicrobia, and barring strong transcriptional regulation, this instead may suggest a likelyrole for other taxa in denitrification or that growth of SUP05 and Marinimicrobia is sustainedthrough alternative metabolisms, such as aerobic respiration and/or sulfur oxidation [187, 188].Other taxa implicated in denitrification, for example, include Arcobacter (chapter 3 and [230]),which is highest in abundance in August when rates of denitrification are highest (Figs. 4.2 and4.3), but otherwise Arcobacter abundance does not scale with rates of denitrification over the4 months of observation. We also note relatively high abundances of Ectothiorhodospiraceae(Gammaproteobacteria, Fig. 4.3), which has previously been linked to HS– oxidation and den-itrification in other environments [252] and may thus also contribute to denitrification in SI.Collectively, these observations suggest that some of the most abundant and conspicuous taxa inSI are decoupled from rates of the pathways that putatively support their growth based on prioranalyses. Other factors thus likely contribute both to controlling organism abundance and rates ofdenitrification.The abundances of the genes involved in denitrification, like the corresponding organisms, didnot scale with the rates of denitrification. This suggests a decoupling of gene abundances fromrates of biogeochemical reactions and is inconsistent with gene-centric modeling theory. Indeed,we find that the denitrification gene pool (nirS, norB and C, and nosZ) more than tripled betweenAugust and September (Figs. 4.4 and 4.5), achieving the highest gene abundances a month afterthe highest rates of denitrification. This increase in the gene abundances also matched an increasein overall cell abundance for the same month, indicating microbial growth when renewal occurred(Figs. 4.2 and 4.3). In contrast, the rates of denitrification detected in September, after renewal(Figs. 4.1 and 4.5), were an order of magnitude lower than the highest rates detected in August.Therefore, the growth of the denitrifying population, based on the gene abundances, did notscale with the rates of denitrification. This may further imply that other pathways are used to100generate growth of the microbial populations containing denitrification genes, as suggested above.Alternatively, a switch between the activity of multiple populations of denitrifying bacteria withdistinctive physiologies causing variable cell-specific rates and biomass yields could lead to eitherhigh gene abundance at low denitrification rates or low gene abundance and high denitrificationrates (Fig. 4.5). The discrepancy between the dynamics in the rates and the abundances of thegenes underpinning denitrification therefore suggests that gene abundance is not always a goodpredictor of biogeochemical activity.The abundances of anammox bacteria and their functional genes both remained relativelyconstant and did not follow the dynamics in the rates of anammox. Anammox is restricted tothe phylum Planctomycetes, and Ca. Scalindua was previously implicated in anammox in SI [33]and more globally in OMZs [66, 120, 121, 253]. In the 4 months studied here, Ca. Scalindua onlycomprised a few percent of the microbial community and varied less than an order of magnitudein absolute abundance (Figs. 4.2 and 4.3) despite an order of magnitude change in the rates ofanammox (Fig. 4.5). Similarly, the anammox gene pool (hzs and hzo – hydrazine synthase andhydrazine oxidoreductase) remained constant, within the same order of magnitude (Figs. 4.4and 4.5). Thus, like denitrification, rates of anammox did not scale with respective gene and cellabundances. The variations in the rates of anammox, however, are relatively small compared tothose of denitrification, with no appreciable changes in the abundance of anammox bacteria (Fig.4.5). The factors responsible for the inconsistency between the rates of anammox and their geneand cell abundances are likely different from denitrification. While taxa involved in denitrificationcan grow facultatively through other metabolisms, Ca. Scalindua is only known to grow throughthe anammox pathway [62] and therefore, the decoupling between abundances and rates is notthe likely result of growth of Ca. Scalindua through alternative metabolisms. Also, while thecapacity for denitrification is distributed across many taxa and thus conducted through differentphysiologies, the anammox metabolism is restricted to Planctomycetes [254] and it is unlikely tobe conducted by other taxa with different physiologies. The combined results for denitrificationand anammox thus show that the relationships between metabolic rates and gene and organismabundances appear more complex than currently accounted for in gene-centric modeling and thatthese relationships further diverge across pathways and organisms.101150mAprilAugustSeptemberOctober135mAprilAugustSeptemberOctober120mAprilAugustSeptemberOctober100mAprilAugustSeptemberOctober200mAprilAugustSeptemberOctobernapA nirSnarG norB norC nosZ nirA nrfA hzo hzsGene abundance (109 gene L-1)0150100200 nrfA/hzo/hzs other genes010010002000amoA haoFigure 4.4: Functional gene abundances of anaerobic N-metabolisms. Gene abundances (gene L– 1) for the follow-ing pathways and associated genes: nitrification (pmo/amoA=ammonia monooxygenase, hao=hydroxylamineoxidoreductase), anammox (hzo=hydrazine dehydrogenase, hzs=hydrazine synthase), NO–3 reduction(napA=periplasmic dissimilatory nitrate reductase, narG=membrane-bound dissimilatory reductase), den-itrification (nirS=nitrite reductase, norBC=nitric oxide reductase, nosZ=nitrous oxide reductase), DNRA(nrfA=dissimilatory periplasmic cytochrome c nitrite reductase, nirA=assimilatory nitrite reductase). Ofnote, the different scales for hzo, hzs and nrfA than for the other genes. 102April 2016August 2016September 2016October 2016APRILHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRF0 100% total NO3- red90100110120130140150160170180190200Depth (m)% DEN% DNRA% AN7.8 10-30 100% total NO3- red90100110120130140150160170180190200Depth (m)% DEN% DNRA% AN0 100% total NO3- red90100110120130140150160170180190200Depth (m)% DEN% DNRA% AN10-4NH4+ NO3-NO2-N22.71 10-2APRILHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFAPRILHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFAPRILHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFAPRILHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFAPRILHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFAPRILHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRF101510161017Gen  abundance (gene m-2)Process rate (% of NO3- reduction)DNRADenitrificationAnammoxGenenapA/narGnirSnorB/CnosZhzo/hzsnrfA/nirAkJ m-2 d-1a.b.c.d.APRILHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRF0 100% total NO3- red90100110120130140150160170180190200Depth (m)% DEN% DNRA% AN3.30 100% total NO3- red90100110120130140150160170180190200Depth (m)% DEN% DNRA% AN0 100% total NO3- red90100110120130140150160170180190200Depth (m)% DEN% DNRA% AN0.07NH4+ NO3-NO2-N211.01APRILHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFAPRILHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFAPRILHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFAPRILHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFAPRILHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFAPRILHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRF0.670.9431.960 100% total NO3- red90100110120130140150160170180190200Depth (m)% DEN_Aug% DNRA_Aug% AN_AugNH4+ NO3-NO2-N2AUGUSTHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFAUGUSTHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFAUGUSTHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFAUGUSTHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPINORNOSNRFAUGUSTHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFAUGUSTHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFAPRILHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFNH4+ NO3-NO2-N2OCTOBERHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZNAPNIRNORNOSNRFOCTOBERHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFOCTOBERHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFOCTOBERHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFOCTOBERHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFOCTOBERHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRF0.685592.760 100% total NO3- red90100110120130140150160170180190200Depth (m)% DEN_Oct% DNRA_Oct% AN_OctAPRILHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRF8.645.39NH4+ NO3-NO2-N21.71SEPTEMBERHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFSEPTEMBERHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFSEPTEMBERHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFSEPTEMBERHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFSEPTEMBERHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRFSEPTEMBERHZO NAP NIR NOR NOS NRFGene2Month2Abundance●●1.0e+151.0e+161.0e+171.5e+170.90.9Gene2●●●●●●HZONAPNIRNORNOSNRF0 100% total NO3- red90100110120130140150160170180190200Depth (m)% DEN_Sep% DNRA_Sep% AN_SepFigure 4.5: Re-networking of anaerobic N-metabolisms linked to power supply. Depth-integrated rates of power supply associated with DNRA,denitrification and anammox (black arrows, in kJ m– 2 d– 1) with depth-integrated gene abundances (in genes m– 2) associated with these rates(in bubble). The insets on the left of each conceptual model shows where most of the NOx reduction occurs in the water column and throughwhich process (blue for denitrification, pink for DNRA, purple for anammox).103In addition to denitrification and anammox, both of which are implicated conceptually andquantitatively in N-cycling in SI, we also found that DNRA contributed appreciably to the N-cyclefollowing renewal, and like denitrification and anammox, genes involved in DNRA did not scalewith rates of DNRA. While the taxa that conduct DNRA in SI remain unknown, SUP05 has beententatively linked to DNRA through the discovery of nrfA homologues in a SUP05 affiliatedopen reading frame. The abundances of the genes involved in DNRA (nrfA and nirA, [3]) tripledbetween August and September, following renewal (Fig. 4.4), but remained constant betweenSeptember and October despite two orders of magnitude increase in the rates of DNRA (Figs. 4.1and 4.5). The disconnect between DNRA rates and gene abundances, like denitrification, probablyreflect both the distribution of DNRA across multiple organisms as well as growth of nrfA/nirAgenes through metabolisms apart from DNRA. Connecting rates of DNRA to gene abundancesis further confounded by both uncertainty in the taxa that conduct DNRA and their underlyinggenes. Indeed, DNRA may be achieved through enzymatic pathways that do not contain nrfAand may instead be catalyzed by other enzymes encoded by other genes [3, 224, 255]. Again,it is difficult to reconcile rates of DNRA with corresponding gene abundances as required ingene-centric modeling frameworks.Nitrifying bacteria and Archaea as well as the nitrification gene pool respond to renewalin likely connection to elevated nitrification rates. Key organisms implicated in nitrificationincluding, Nitrospina sp. as well as Thaumarchaeota [33], along with the genes encoding enzymesthat catalyze key steps in nitrification (amoA and hao – ammonia oxidase and hydroxylamineoxidoreductase), appear to increase following renewal. Specifically, we find an increase in theabundance of organisms belonging to these clades at the depths where renewal was observed(135m in August and 200m in September/October - Fig. 4.4). The genes amoA and hao alsoexhibited similar increases in abundance at the same depths at the same time. Although wedid not measure rates of nitrification, it was previously detected in the low-oxygen waters of SIand shown to increase in rate following renewal in 2008 [247]. It is therefore likely that ratesof nitrification increased in response to the new input of oxygenated waters in SI following theSeptember 2016 renewal as well (Fig. 4.1a) and so did the nitrifying population according tothe abundance of key nitrifying organisms and the abundance of nitrifying genes. However, we104cannot confirm whether the rates and nitrifying microbial population growth scale as we arelacking rates of nitrification for the period studied.4.3.3 Power supply and ecophysiology of anaerobic N-metabolismsWe often think about microbial communities in redox stratified environments in terms of energyavailable or free energy yield (∆G). However, microbial communities can be limited by therate at which this energy is made available, or in other words, by the power supply [228, 229].We determined the power supply in the water column of SI for the anaerobic N-metabolismsdenitrification, anammox, and DNRA. This was accomplished by multiplying reaction rates bythe corresponding free energy yield at that depth and these were integrated over the depth of theanoxic water column (see Appendix D for more details on the power supply calculations). We alsouse reaction rates as a proxy for the substrate supply rate. This works because substrates did notaccumulate and thus reaction rates place maximum values on the supply rate. Likewise, reactionrates place minimum values on substrate supply rates because mass balance precludes reactionrates that exceed substrate supply rates. We show that the power supply spans several orders ofmagnitude and is higher following renewal (Figs. 4.1a, b, and 4.5). Power supply in SI variedbetween 0.3 and 560 kJ m– 2 d– 1 and was the highest in October 2016 (Fig. 4.1b). The dynamics inpower supply were mainly driven by changes in substrate supply rates rather than the changesin free energy yield that accompany changes in substrate and metabolite concentrations, whichremain within an order of magnitude despite relatively small changes in substrate concentrations(Table D.3). Dynamics in power supply thus appear to be the result of changes in the input ofoxidants to anoxic waters during renewal, which fuels nitrification and in turn supports higherrates of anaerobic N-metabolisms.Changes in power supply are tightly coupled to changes in N-metabolisms. This has theeffect of changing the overall outcome of reductive N-metabolisms, switching between N-lossand N-retention in connection to the substrate supply rate. Following renewal, the increase inpower supply was reflected mainly as enhanced rates of DNRA, which causes N-retention inthe inlet instead of N-loss through denitrification and anammox (Table D.1). A shift between N2production and N-retention through denitrification/anammox or DNRA, respectively, therefore105has important consequences for nutrient budgets in SI. The connection between this shift inpathways and dynamics in power supply implies that changes in power supply may play animportant role in controlling the biogeochemistry of systems above and beyond simple regulationof the rates at which reactions take place. In this case, it appears to cause a re-networking of theN-cycle, by diverting the rate at which the energy is processed from one pathway to another, herefrom anammox and denitrification, to DNRA.Power is used by organisms to maintain biomass, respond to environmental stressors, and ifsufficient, fuel growth [228, 229]. Here we partitioned energy flow between different metabolicpathways in the N-cycle and compared these to functional gene abundances, as a proxy formicrobial abundances (Fig. 4.5). Gene-centric modeling theory is based on scaling betweenreaction rates and functional gene abundances. In SI, however, changes in functional geneabundances associated with renewal did not scale to corresponding changes in rates. Sincedynamics in power supply in SI are mostly driven by substrate supply rates, functional geneabundances also do not scale with power supply (Figs. 4.4 and 4.5). Therefore, because poweris at least partly decoupled from microbial growth based on the results presented here, it mustbe dissipated through other mechanisms. These can include but are not limited to: extracellularsecretions; energy spilling reactions (i.e. heat loss); defense against chemical stresses; cell motility;and proofreading and synthesis of macromolecules, such as RNA and proteins [229]. The powerused in these processes could be referred to as maintenance energy and seems to vary betweentaxa and with growth conditions [256–259]. Thus, if the maintenance energy is low, there isgenerally more energy available to fuel growth that could support higher biomass yield. Here,anammox bacteria appear to have higher maintenance energy requirements, as we did not observesubstantial changes in the relevant gene abundances despite changes in the rates of anammox, andthis further implies a low biomass yield and this is consistent with previous observations fromwastewater treatment plants [62]. Conversely, gene abundances for denitrification and DNRAchanged substantially after renewal, although these did not scale with the corresponding increasein rates of these reactions. This therefore suggests lower maintenance energy requirementsfor the organisms involved in denitrification and DNRA, as well as relatively higher biomassyields. The lack of consistent relationship between gene abundances, relevant organisms and106rates imply that maintenance energy requirements and biomass yield are variable and depend ongrowth conditions. Beyond the complexities apparently associated with the relationship betweengrowth and reaction rate, differentials in taxon-specific death rates would further confound theserelationships. These observations demonstrate a need for more information on biomass yield,maintenance energy requirements and taxon-specific death rates. Such information should begenerated for key-relevant taxa and appropriate model organisms that would enable predictivebiogeochemical modeling that includes information on microbial ecology under dynamic (non-steady state) conditions.4.4 Implications and extensionsDNRA was the dominant pathway of NOx reduction in SI when summed over the year, withseasonal shifts between denitrification and anammox, and DNRA. This is notable as such apredominance of DNRA has rarely been observed in marine settings – and to date, only in somecoastal and estuarine sediments [64, 93, 97, 155, 225]. Nevertheless, DNRA has previously beendetected in pelagic settings, where it appeared to play a relatively minor role, such as in thePeruvian OMZ and the Baltic Sea [67, 106]. In SI, our time-series observations allowed us tocapture large-scale dynamics in anaerobic N-metabolisms, and these observations reveal temporalshifts in denitrification and anammox, and DNRA that, when integrated annually, have importantimplications for N budgets in SI. The extent to which such large temporal variations in ratesand pathways occur more broadly in coastal and open ocean low oxygen marine waters remainsunknown, but should be evaluated in future studies.Physical dynamics in the ocean influence power supply and, as we show here, can be accom-panied by changes in the rates and pathways of microbial metabolism, with consequences forbiogeochemical cycles and ocean chemistry. Models predict that current trends in climate will leadto increased frequencies of meso-scale eddies as well as increased upwelling, both of which havethe potential to enhance the supply of oxidants and oxidized N-substrates to low oxygen marinewaters [214, 260–263]. This enhanced supply of oxidants and N-substrates can lead to enhancedpower supply, and, given that enhanced power supply appears linked to a shift in anaerobic107N-metabolisms to DNRA, we suggest that predicted changes in ocean circulation may influenceN-inventories by promoting N-retention at the expense of N-loss. Anecdotally, detection of DNRAin the Peruvian upwelling system [67], which is currently characterized by frequent meso-scaleeddies, supports this idea. If DNRA were to become the dominant anaerobic N-metabolism inthe future oceans, it would dramatically influence global N-budgets and likely support a positivefeedback on ocean deoxygenation.Given the connection between power supply, physical dynamics and ocean circulation, andpathways of anaerobic N-metabolisms, there may be evidence in the geologic record that supportsthe positive feedbacks we propose here and their influence on the Earth system. For example, theoceans transiently became anoxic during Oceanic Anoxic Events, in the Cretaceous period [219],with links to changes in ocean circulation and enhanced primary productivity [219, 264, 265].N-retention through DNRA, in response to enhanced power supply triggered by changes in oceancirculation, provides a means to initiate a positive feedback on ocean deoxygenation that coulddrive the expansive ocean anoxia during OAEs. At even larger scales, the oceans have changedfrom ferruginous (iron-rich) to euxinic (sulphide-rich) conditions during intervals in the lateArchean and throughout the Proterozoic eons [42, 47–49, 55]. N-retention through DNRA versusloss through denitrification provides a means of sustaining euxinia under the widespread oceananoxia that characterized the Precambrian eons and indeed, the N-isotope record implies basinscale DNRA during some intervals [42]. The reasons for widespread DNRA in the Precambrianoceans are necessarily uncertain at this time but by analogy, we suggest that variations in powersupply linked to physical or chemical dynamics in the oceans could be responsible for changesbetween DNRA dominated and denitrification dominated anaerobic N-metabolisms. Changesin ocean circulation in response to deglaciation events in the Late Proterozoic eon have indeedbeen linked to the development of euxinia [266]. Taken together, we suggest that power supplydynamics can cause changes to the N-cycle that impact ocean nutrient status and can have large-scale effects on ocean chemistry, biological production, and the Earth system. Analogous changesin ocean circulation and power supply could mirror these events of deoxygenation in the future,and this deoxygenation could possibly lead to the development of widespread euxinia in theoceans with potential to induce biological crises of similar scale to the Cretaceous OAEs.108Chapter 5ConclusionsThis dissertation provides new knowledge about the dynamics in rates and pathways of bothancient and modern marine N-cycling, through the combination of information from geochemicalprofiles, process rate measurements, and analysis of microbial community composition, structureand function. This new knowledge was then integrated into models to study the effect of theserates on the nutrient status of the ancient oceans, the competition between metabolic pathways ofinterest leading to N2 production, and the bioenergetics of anammox, denitrification and DNRA.This chapter synthesizes the dissertation’s findings and concludes with a discussion on the futurechallenges related to the N-cycle and corresponding modeling efforts.5.1 Dynamics in rates and pathways of anaerobic N-cyclingThis dissertation describes rates and pathways of anaerobic N-metabolisms across different pelagicenvironments, which expands current knowledge on the factors that control the partitioningof these pathways. Prior to this work, the drivers of the partitioning were underexplored andremained unconstrained. In Chapter 2, we show the presence of Fe-dependent NO–3 reduction –through DNRA and denitrification – under pelagic ferruginous conditions, which can serve as ananalogue environment extensible to the Proterozoic oceans. The fact that these pathways are activeunder these analogous conditions to the ancient ocean can inform on ancient marine N-cyclingand gives us further insight, when combined with the geologic record, on how anaerobic N-metabolisms would have been active. In Chapter 3 we study the variations in rates and pathwaysof anaerobic N-metabolisms with fine-scale temporal and spatial resolution in an anoxic fjord. Thisreveals rate variations of previously unappreciated magnitude, in addition to temporal changesin the partitioning of the pathways that responds to physical perturbations in the inlet. Most109studies are currently limited, both temporally and spatially, therefore overlooking these changingenvironmental conditions and corresponding dynamics in rates and pathways of N-cycling. InChapter 4, we also show that DNRA is a dominant pathway of NOx reduction, challenging currentassumptions that DNRA constitutes a minor process under OMZ-like conditions. Rates of DNRAare regulated nearly exclusively by substrate supply rates, which changes the paradigm of howwe think about controls on rates and partitioning of these pathways. Rather than measuring thestate of physico-chemical conditions at a singular time-point (presence or absence of substrates,inhibitors, e.g.) – which is the currently accepted approach to studying these pathways – we showit can instead be the rate of change in these conditions that regulates the rates and pathways ofanaerobic N-metabolisms.5.2 Integrated approach for better modeling of biogeochemicalcyclingNew knowledge on the connection between rates and pathways of anaerobic N-cycling andcorresponding environmental conditions presented in this thesis, promotes more thorough N-cycle models. In Chapter 2, the presence of DNRA under ferruginous conditions was tested ina reaction-transport model for a coastal upwelling set for the Proterozoic oceans. This allowedus to study the potential impact of DNRA on the nutrient status of the ancient ocean, as well asits influence on biological production, and ocean and atmospheric chemistry. Ultimately, thesechanges can be found in the rock record, but the study of a modern analogue can give moreinsight on the microbial mechanisms that could have occurred under similar conditions in theancient oceans. In Chapter 3, we reproduce the variations in rates and pathways in a simpleflux-balance model mimicking the competition between anammox and denitrification, while usingkinetic parameters measured in situ, or inferred when the parameters were not available. Eventhough the model was straightforward in its structure, it generated similar rates of N-cyclingand similar growth rates of relevant cell populations to those observed in Saanich Inlet. Thus, bycombining geochemical (substrate concentrations) and microbiological (process rate measurementsand microbial community composition) information, we are able to broadly reconstruct the rates110and pathways observed in the water column. In Chapter 4, we show that DNRA is an importantN-cycle pathway in Saanich Inlet and should be considered both in conceptual and gene-centricmodeling. We further show that, under dynamic (non-steady state) conditions, rates of anaerobicN-metabolisms and corresponding gene abundances, used as a proxy for microbial populationsresponsible for the pathways, did not scale uniformly, contradicting the current framework ofgene-centric modeling.5.3 Looking aheadThe extent to which temporal and spatial variations in rates and pathways of N-metabolisms occurin marine systems needs to be further studied globally. Indeed, it is unclear whether N-budgetsare currently balanced (see Chapter 1), due to uncertainties in estimates of the N-budgets and ratesof N-transformations [2]. Most studies that measure rates and pathways of N-transformationsin OMZs, for example, are limited to a singular time-point, and remain relatively localizedspatially (e.g. [18, 65, 67, 98, 100]). It is therefore possible that changing environmental conditionsinfluencing changes in rates and pathways are not captured by current spatial and temporalresolution of most studies. In Chapters 3 and 4, we show that rates of denitrification and DNRAvary over several orders of magnitude on an annual basis in a coastal anoxic fjord, Saanich Inlet.These variations were linked to environmental perturbations of the inlet, which were coupledwith higher substrate supply rates and thus higher energy brought to the system. Environmentalperturbations can also occur in the coastal and open oceans. These can be large and meso-scaleeddies [260], for example, or increased upwelling currents due to increased winds that are aconsequence of climate change [263]. These changes have rarely been studied in combinationwith process rate measurements [103, 214] and lack altogether microbial community analysis.Thus, these observations call for a broadening of the research to rates and pathways of anaerobicN-cycling in OMZs under dynamic conditions. This would in turn allow for more accurateestimates of marine N-budgets, and further, better inform biogeochemical modeling efforts.Biogeochemical models of N-cycling should consider DNRA as an important pathway of NOxreduction for the reconstruction of rates and pathways of past, present and future N-cycling. In111current approaches, however, DNRA is often overlooked [5, 39, 44], and considered insignificantto global N-budgets. While DNRA is not a primary source of fixed-N, the activity of DNRA canimpact biological productivity by mitigating N-loss, through anammox and denitrification, bydiverting the products of NOx reduction to NH+4 rather than to N2, with the retained N beingupwelled to surface waters and feeding primary production. In Chapters 2 and 4, we indeeddetermine that DNRA plays an important role in NOx reduction, more than previously thought,both in ferruginous conditions that are analogous to the Proterozoic oceans, as well as in a moderncoastal anoxic fjord. Furthermore, in Chapter 2, a higher partitioning of NOx reduction throughDNRA was shown through modeling to lead to higher rates of primary production in the surfacewaters of the Proterozoic oceans, with corresponding impact on ocean and atmosphere chemistry.To confirm the activity of DNRA in the ancient oceans, it would be essential to test whether theN-isotope fractionation found in the rock record can be reproduced when adding DNRA to thepathways considered in models such as the one in Boyle et al. (2013) [44]. N-isotope fractionationthrough DNRA remains, however, poorly constrained so far [64], unlike denitrification andanammox, and further research is needed, both in natural and laboratory cultured settings toobserve the fractionation associated with DNRA. Results in Chapter 4 also shows that rates ofDNRA vary greatly in range, responding to physical perturbations in the system, and if found atlarger scale in the modern oceans, could have similar effect on biological productivity and oceanchemistry. This switch between denitrification and DNRA in response to dynamic conditionsshould also be considered in current modeling efforts, in light of deoxygenation of the modernoceans [72] and changes in ocean circulation due to climate change [262, 263] . Together, theseresults call for the consideration of DNRA in models for the past, and reconstruction of presentand future conditions.Key-players involved in anaerobic N-cycling in Saanich Inlet remain elusive, and their corre-sponding ecophysiological information unknown, and this information is key to refining currentmodels. These key-players found in Saanich Inlet are likely extensible to other parts of the oceans,and could greatly impact N-cycling under specific environmental conditions. The ecophysiologicalinformation associated with the key-players is difficult to constrain, however, due to the complexityof the microbial community involved in these pathways. In Chapter 3, we identified two differing112denitrifying populations – SUP05 and Marinimicrobia versus Arcobacter – and these populationswere linked to changes in rates of N2 production, with either low or high rates of N2-production,respectively. Although SUP05 and Marinimicrobia were previously recognized as key-playersin Saanich Inlet [33, 187], the involvement of Arcobacter needs to be further confirmed throughmeta’omic analysis that can be used to test its potential and expressed metabolic activities. InChapter 4, we also detected genes involved in DNRA, which was previously omitted in theconceptual metabolic model developed for Saanich Inlet [33]. The taxonomic identification ofkey-players involved in DNRA remains unknown so far and thus, missing taxa have yet to beuncovered for Saanich Inlet. This information will allow us to improve existing conceptual modelsof N-cycling, and likely add relevant key-players involved in NOx reduction. Another notablefinding from this thesis is that the gene-centric modeling approach, which states that changesin rates scale with changes in the abundance of the genes that code for the enzyme conductingthe metabolic pathway [118, 119], is not applicable under dynamic (non-steady state) conditions.This therefore asks for further details on ecophysiological parameters such as biomass yields ofrelevant taxa changing with different growth conditions, as well as taxon-specific growth anddeath rates. To do so, new information could be produced through cultivation of relevant taxa inthe lab. The culture of these taxa might not be feasible, as is the case for many environmental taxa(e.g. anammox bacteria), but other techniques could be used to measure taxon-specific growth anddeath rates in the environment, for example through stable isotope probing techniques [267]. Thefact that we cannot model rates and pathways of anaerobic N-cycling under dynamic conditionsunderscores the lack of knowledge on the metabolic and growth parameters of relevant taxa.5.4 ClosingOver the span of ∼4 billion years, the Earth’s surface redox state has been drastically alteredthrough coupling of geologic and microbial metabolic processes. The interactions between theseprocesses are extremely complex and require both in-depth knowledge of the micro-scale aswell as the global system in order to model the multiple feedback loops that constitute theseinteractions. This dissertation developed a more informed modeling framework for ancient113and contemporaneous N-cycle predictions that incorporates geochemical and microbiologicalinformation. By using an integrated approach in a specific environment to develop new knowledge,resulting models can be made extensible to other environments. As the Earth is entering theAnthropocene, a new geologic epoch where humankind is the main driver in changing theredox state of the Earth’s surface, unprecedented alterations to the environment and climate areoccurring. These alterations will be disruptors of current biogeochemical cycling and predictionsof the feedbacks from microbial processes that will arise from it are largely unconstrained.Collaboration between fields, from climate modeling to environmental microbiology, will thereforebe necessary in order to tackle future changes. By being able to predict what the future will looklike given the current path that we are headed, this will allow us to plan how humankind needsto bio/geoengineer changes in order to keep a life-sustainable planet.114Bibliography[1] N. Gruber and J. N. Galloway. An earth-system perspective of the global nitrogen cycle. Nature,451:293, 01/16/online 2008.[2] D. E. Canfield, A. N. Glazer, and P. G. Falkowski. The evolution and future of earth’s nitrogen cycle.Science, 330(6001):192–6, Oct 8 2010. Canfield, Donald E Glazer, Alexander N Falkowski, Paul G engResearch Support, Non-U.S. Gov’t Research Support, U.S. Gov’t, Non-P.H.S. Review New York, N.Y.2010/10/12 06:00 Science. 2010 Oct 8;330(6001):192-6. doi: 10.1126/science.1186120.[3] M. M. M. Kuypers, H. K. Marchant, and B. Kartal. The microbial nitrogen-cycling network. NatureReviews Microbiology, 16:263, 02/05/online 2018.[4] D. E. Canfield, E. Kristensen, and B. Thamdrup. Aquatic geomicrobiology. Adv Mar Biol, 48:1–599,2005. Canfield, Donald E Kristensen, Erik Thamdrup, Bo eng Review Adv Mar Biol. 2005;48:1-599.doi: 10.1016/S0065-2881(05)48017-7.[5] D. E. Canfield. Models of oxic respiration, denitrification and sulfate reduction in zones of coastalupwelling. Geochimica Et Cosmochimica Acta, 70(23):5753–5765, Dec 1 2006.[6] Sterner, Robert, and Elser. Ecological Stoichiometry: The Biology of Elements From Molecules to TheBiosphere. 2002.[7] H. Bothe, O. Schmitz, M. G. Yates, and W. E. Newton. Nitrogen fixation and hydrogen metabolism incyanobacteria. Microbiology and Molecular Biology Reviews : MMBR, 74(4):529–551, 2010.[8] P. G. Falkowski. Evolution of the nitrogen cycle and its influence on the biological sequestration ofco2 in the ocean. Nature, 387(6630):272–275, May 15 1997.[9] N. Gruber. The dynamics of the marine nitrogen cycle and its influence on atmospheric co2 variations.In M. Follows and T. Oguz, editors, The Ocean Carbon Cycle and Climate, pages 97–148. KluwerAcademic, Dordrecht, 2004.[10] P. A. Schroeder and A. A. McLain. Illite-smectites and the influence of burial diagenesis on thegeochemical cycling of nitrogen. Clay Minerals, 33(4):539–546, // 1998.[11] V. Busigny, P. Cartigny, and P. Philippot. Nitrogen isotopes in ophiolitic metagabbros: A re-evaluationof modern nitrogen fluxes in subduction zones and implication for the early earth atmosphere.Geochimica et Cosmochimica Acta, 75(23):7502–7521, 2011.[12] B. Johnson and C. Goldblatt. The nitrogen budget of earth. Earth-Science Reviews, 148:150–173,2015/09/01/ 2015.[13] R. A. Berner. Geological nitrogen cycle and atmospheric n2 over phanerozoic time. Geology, 34(5):413–415, 2006.[14] C. Goldblatt, M. W. Claire, T. M. Lenton, et al. Nitrogen-enhanced greenhouse warming on earlyearth. Nature Geoscience, 2:891, 11/15/online 2009.115[15] J. M. Klingler, R. L. Mancinelli, and M. R. White. Biological nitrogen fixation under primordialmartian partial pressures of dinitrogen. Advances in Space Research, 9(6):173–176, 1989/01/01/ 1989.[16] A. L. Zerkle, S. W. Poulton, R. J. Newton, et al. Onset of the aerobic nitrogen cycle during the greatoxidation event. Nature, 542:465, 02/06/online 2017.[17] B. W. Johnson and C. Goldblatt. A secular increase in continental crust nitrogen during the precam-brian. Geochemical Perspectives Letters, 4:24–28, 2017.[18] D. E. Canfield, F. J. Stewart, B. Thamdrup, et al. A cryptic sulfur cycle in oxygen-minimum-zonewaters off the chilean coast. Science, 330(6009):1375–1378, Dec 3 2010.[19] P. M. Vitousek. Human domination of earth’s ecosystems (vol 277, pg 494, 1997). Science, 278(5335):21–21, Oct 3 1997.[20] E. A. Bell, P. Boehnke, T. M. Harrison, and W. L. Mao. Potentially biogenic carbon preserved in a 4.1billion-year-old zircon. Proceedings of the National Academy of Sciences, 112(47):14518–14521, 2015.[21] D. L. Pinti and K. Hashizume. 15n-depleted nitrogen in early archean kerogens: clues on ancientmarine chemosynthetic-based ecosystems?: A comment to beaumont, v., robert, f., 1999. precambrianres. 96, 6282. Precambrian Research, 105(1):85–88, 2001.[22] V. Beaumont and F. Robert. Nitrogen isotope ratios of kerogens in precambrian cherts: a record ofthe evolution of atmosphere chemistry? Precambrian Research, 96(1):63–82, 1999/06/15/ 1999.[23] W. L. Chameides and J. C. G. Walker. Rates of fixation by lightning of carbon and nitrogen in possibleprimitive atmospheres. Origins of life, 11(4):291–302, 1981/12/01 1981.[24] J. F. Kasting. Stability of ammonia in the primitive terrestrial atmosphere. Journal of GeophysicalResearch-Oceans and Atmospheres, 87(Nc4):3091–3098, 1982.[25] J. F. Kasting. Bolide impacts and the oxidation state of carbon in the earth’s early atmosphere. Originsof life and evolution of the biosphere, 20(3):199–231, 1990/05/01 1990.[26] J.F. Kasting and J. C. G. Walker. Limits on oxygen concentration in the prebiological atmosphere andthe rate of abiotic fixation of nitrogen. Journal of Geophysical Research: Oceans, 86(C2):1147–1158, 1981.[27] D. P. Summers and B. Khare. Nitrogen fixation on early mars and other terrestrial planets: Experi-mental demonstration of abiotic fixation reactions to nitrite and nitrate. Astrobiology, 7(2):333–341,2007/04/01 2007.[28] J. A. Brandes, R. M. Hazen, and H. S. Yoder. Inorganic nitrogen reduction and stability undersimulated hydrothermal conditions. Astrobiology, 8(6):1113–1126, 2008/12/01 2008.[29] A. Smirnov, D. Hausner, R. Laffers, D. R. Strongin, and M. A. A. Schoonen. Abiotic ammoniumformation in the presence of ni-fe metals and alloys and its implications for the hadean nitrogencycle. Geochemical Transactions, 9:5–5, 05/19 10/02/received 05/19/accepted 2008.[30] A. L. Zerkle, C. K. Junium, D. E. Canfield, and C. H. House. Production of 15ndepleted biomass duringcyanobacterial n2fixation at high fe concentrations. Journal of Geophysical Research: Biogeosciences,113(G3), 2008.[31] J. B. Glass, F. Wolfe-Simon, and A. D. Anbar. Coevolution of metal availability and nitrogenassimilation in cyanobacteria and algae. Geobiology, 7(2):100–23, Mar 2009.[32] P. G. Falkowski, T. Fenchel, and E. F. Delong. The microbial engines that drive earth’s biogeochemicalcycles. Science, 320(5879):1034–1039, May 23 2008.116[33] A. K. Hawley, H. M. Brewer, A. D. Norbeck, L. Pasa-Tolic, and S. J. Hallam. Metaproteomics revealsdifferential modes of metabolic coupling among ubiquitous oxygen minimum zone microbes. ProcNatl Acad Sci U S A, 111(31):11395–400, Aug 05 2014.[34] R. Buick. When did oxygenic photosynthesis evolve? Philosophical Transactions of the Royal Society B:Biological Sciences, 363(1504):2731–2743, 05/09 2008.[35] S. A. Crowe, L. N. Dssing, N. J. Beukes, et al. Atmospheric oxygenation three billion years ago. Nature,501:535, 09/25/online 2013.[36] M. T. Rosing and R. Frei. U-rich archaean sea-floor sediments from greenland indications of ¿3700ma oxygenic photosynthesis. Earth and Planetary Science Letters, 217(3):237–244, 2004/01/15/ 2004.[37] C. Magnabosco, K. R. Moore, J. M. Wolfe, and G. P. Fournier. Dating phototrophic microbial lineageswith reticulate gene histories. Geobiology, 16(2):179–189, Mar 2018.[38] S. L. Olson, L. R. Kump, and J. F. Kasting. Quantifying the areal extent and dissolved oxygenconcentrations of archean oxygen oases. Chemical Geology, 362:35 – 43, 2013. Special Issue dedicatedto H.D. Holland: Evolution of the atmosphere and ocean through time.[39] K. Fennel, M. Follows, and P. G. Falkowski. The co-evolution of the nitrogen, carbon and oxygencycles in the proterozoic ocean. American Journal of Science, 305(6-8):526–545, 2005.[40] L. V. Godfrey and P. G. Falkowski. The cycling and redox state of nitrogen in the archaean ocean.Nature Geoscience, 2:725, 09/13/online 2009.[41] A. L. Zerkle and S. Mikhail. The geobiological nitrogen cycle: From microbes to the mantle. Geobiology,15(3):343–352, 2017.[42] L. V. Godfrey, S. W. Poulton, G. E. Bebout, and P. W. Fralick. Stability of the nitrogen cycle duringdevelopment of sulfidic water in the redox-stratified late paleoproterozoic ocean. Geology, 41(6):655,2013.[43] C. Thomazo, D. L. Pinti, V. Busigny, et al. Biological activity and the earth’s surface evolution: Insightsfrom carbon, sulfur, nitrogen and iron stable isotopes in the rock record. Comptes Rendus Palevol,8(7):665–678, 2009/10/01/ 2009.[44] R. A. Boyle, J. R. Clark, S. W. Poulton, et al. Nitrogen cycle feedbacks as a control on euxinia in themid-proterozoic ocean. Nature Communications, 4, Feb 2013.[45] D. E. Canfield. A new model for proterozoic ocean chemistry. Nature, 396(6710):450–453, Dec 3 1998.[46] H. D. Holland. The Chemical Evolution of the Atmosphere and Oceans. Princeton University Press, 1984.[47] D. T. Johnston, S. W. Poulton, C. Dehler, et al. An emerging picture of neoproterozoic ocean chemistry:Insights from the chuar group, grand canyon, usa. Earth and Planetary Science Letters, 290(1-2):64–73,Feb 15 2010.[48] C. Li, G. D. Love, T. W. Lyons, et al. A stratified redox model for the ediacaran ocean. Science,328(5974):80–83, Apr 2 2010.[49] C. T. Reinhard, R. Raiswell, C. Scott, A. D. Anbar, and T. W. Lyons. A late archean sulfidic seastimulated by early oxidative weathering of the continents. Science, 326(5953):713–716, Oct 30 2009.[50] M. Lliros, T. Garcia-Armisen, F. Darchambeau, et al. Pelagic photoferrotrophy and iron cycling in amodern ferruginous basin. Sci Rep, 5:13803, 2015.117[51] X. A. Walter, A. Picazo, M. R. Miracle, et al. Phototrophic fe(ii)-oxidation in the chemocline of aferruginous meromictic lake. Frontiers in Microbiology, 5:713, 12/08 09/08/received 11/30/accepted2014.[52] A-C. Lehours, I. Batisson, A. Guedon, G. Mailhot, and G. Fonty. Diversity of culturable bacteria,from the anaerobic zone of the meromictic lake pavin, able to perform dissimilatory-iron reductionin different in vitro conditions. Geomicrobiology Journal, 26(3):212–223, 2009/03/13 2009.[53] M. A. Lazzaretti, K. W. Hanselmann, H. Brandl, D. Span, and R. Bachofen. The role of sediments inthe phosphorus cycle in Lake Lugano. II. Seasonal and spatial variability of microbiological processes at thesediment-water interface, volume 54. 1992.[54] S. A. Crowe, J. A. Maresca, C. Jones, et al. Deep-water anoxygenic photosythesis in a ferruginouschemocline. Geobiology, 12(4):322–39, Jul 2014.[55] S. W. Poulton, P. W. Fralick, and D. E. Canfield. Spatial variability in oceanic redox structure 1.8billion years ago. Nature Geoscience, 3(7):486–490, Jul 2010.[56] M. M. Jensen, J. Petersen, T. Dalsgaard, and B. Thamdrup. Pathways, rates, and regulation ofn-2 production in the chemocline of an anoxic basin, mariager fjord, denmark. Marine Chemistry,113(1-2):102–113, Jan 30 2009.[57] Frost and McDowell. The dissociation energy of the nitrogen molecule. Proceedings of the Royal Societyof London. Series A. Mathematical and Physical Sciences, 236(1205):278, 1956.[58] K. L. Straub, M. Benz, B. Schink, and F. Widdel. Anaerobic, nitrate-dependent microbial oxidation offerrous iron. Appl Environ Microbiol, 62(4):1458–60, Apr 1996.[59] K. A. Weber, M. M. Urrutia, P. F. Churchill, R. K. Kukkadapu, and E. E. Roden. Anaerobic redoxcycling of iron by freshwater sediment microorganisms. Environmental Microbiology, 8(1):100–113, Jan2006.[60] T. Kalvelage, G. Lavik, P. Lam, et al. Nitrogen cycling driven by organic matter export in the southpacific oxygen minimum zone. Nature Geoscience, 6(3):228–234, Mar 2013.[61] M. Strous, J. A. Fuerst, E. H. M. Kramer, et al. Missing lithotroph identified as new planctomycete.Nature, 400(6743):446–449, 07/29/print 1999.[62] B. Kartal, N. M. de Almeida, W. J. Maalcke, et al. How to make a living from anaerobic ammoniumoxidation. Fems Microbiology Reviews, 37(3):428–461, May 2013. 123cb Times Cited:98 Cited ReferencesCount:201.[63] L. A. Codispoti, J. A. Brandes, J. P. Christensen, et al. The oceanic fixed nitrogen and nitrous oxidebudgets: Moving targets as we enter the anthropocene? Scientia Marina, 65:85–105, Sep 2001.[64] M. G. Prokopenko, M. B. Hirst, L. De Brabandere, et al. Nitrogen losses in anoxic marine sedimentsdriven by thioploca-anammox bacterial consortia. Nature, 500(7461):194–8, Aug 8 2013. Prokopenko, MG Hirst, M B De Brabandere, L Lawrence, D J P Berelson, W M Granger, J Chang, B X Dawson, S Crane,E J 3rd Chong, L Thamdrup, B Townsend-Small, A Sigman, D M eng Research Support, U.S. Gov’t,Non-P.H.S. England 2013/08/09 06:00 Nature. 2013 Aug 8;500(7461):194-8. doi: 10.1038/nature12365.[65] M. M. Jensen, P. Lam, N. P. Revsbech, et al. Intensive nitrogen loss over the omani shelf due toanammox coupled with dissimilatory nitrite reduction to ammonium. Isme Journal, 5(10):1660–1670,Oct 2011.118[66] M. M. M. Kuypers, G. Lavik, D. Woebken, et al. Massive nitrogen loss from the benguela upwellingsystem through anaerobic ammonium oxidation. Proceedings of the National Academy of Sciences of theUnited States of America, 102(18):6478–6483, 2005.[67] P. Lam, G. Lavik, M. M. Jensen, et al. Revising the nitrogen cycle in the peruvian oxygen minimumzone. Proceedings of the National Academy of Sciences of the United States of America, 106(12):4752–4757,Mar 24 2009.[68] Ph. Lam and M. M. M. Kuypers. Microbial nitrogen cycling processes in oxygen minimum zones.Annual Review of Marine Science, 3(1):317–345, 2011/01/15 2010.[69] B. B. Ward, A. H. Devol, J. J. Rich, et al. Denitrification as the dominant nitrogen loss process in thearabian sea. Nature, 461(7260):78–81, 09/03/print 2009.[70] T. Ohde, H. Siegel, J. Reimann, and M. Gerth. Identification and investigation of sulphur plumesalong the namibian coast using the meris sensor. Continental Shelf Research, 27(6):744–756, 2007/03/15/2007.[71] S. J. Weeks, B. Currie, A. Bakun, and K. R. Peard. Hydrogen sulphide eruptions in the atlantic oceanoff southern africa: implications of a new view based on seawifs satellite imagery. Deep Sea ResearchPart I: Oceanographic Research Papers, 51(2):153–172, 2004/02/01/ 2004.[72] D. Breitburg, L. A. Levin, A. Oschlies, et al. Declining oxygen in the global ocean and coastal waters.Science, 359(6371), 2018.[73] L. P. Nielsen. Denitrification in sediment determined from nitrogen isotope pairing. FEMS MicrobiologyLetters, 86(4):357–362, 1992.[74] M. Holtappels, G. Lavik, M. M. Jensen, and M. M. Kuypers. 15n-labeling experiments to dissect thecontributions of heterotrophic denitrification and anammox to nitrogen removal in the omz waters ofthe ocean. Methods Enzymol, 486:223–51, 2011.[75] J. Meija, B. Coplen Tyler, M. Berglund, et al. Atomic weights of the elements 2013 (iupac technicalreport), 2016.[76] E.A. Paul, J. Melillo, R. Knowles, and H. Blackburn. Nitrogen Isotope Techniques. Elsevier Science, 2012.[77] B. Eyre, S. Rysgaard, T. Dalsgaard, and P. Christensen. Comparison of isotope pairing and N2:Ar methodsfor measuring sediment denitrificationAssumption, modifications, and implications, volume 25. 2002.[78] J. Brandes. Isotopic effects of denitrification in the marine environment. 1996.[79] T. Yoshinari and R. Knowles. Acetylene inhibition of nitrous oxide reduction by denitrifying bacteria.Biochemical and Biophysical Research Communications, 69(3):705–710, 1976/04/05/ 1976.[80] A. C. Redfield. The biological control of chemical factors in the environment. American Scientist,46(3):230A–221, 1958.[81] N. Gruber. The Marine Nitrogen Cycle: Overview and Challenges. 2008.[82] A. Mulder, A. A. Vandegraaf, L. A. Robertson, and J. G. Kuenen. Anaerobic ammonium oxidationdiscovered in a denitrifying fluidized-bed reactor. Fems Microbiology Ecology, 16(3):177–183, Mar 1995.Qn679 Times Cited:512 Cited References Count:21.[83] B. Thamdrup and T. Dalsgaard. Production of n(2) through anaerobic ammonium oxidation coupledto nitrate reduction in marine sediments. Appl Environ Microbiol, 68(3):1312–8, Mar 2002.119[84] M. R. Hamersley, G. Lavik, D. Woebken, et al. Anaerobic ammonium oxidation in the peruvianoxygen minimum zone. Limnology and Oceanography, 52(3):923–933, 2007.[85] M. M. Kuypers, A. O. Sliekers, G. Lavik, et al. Anaerobic ammonium oxidation by anammox bacteriain the black sea. Nature, 422(6932):608–11, Apr 10 2003.[86] K.C. Cameron, H.J. Di, and J.L. Moir. Nitrogen losses from the soil/plant system: a review. Annals ofApplied Biology, 162(2):145–173, 2013.[87] S. Humbert, S. Tarnawski, N. Fromin, et al. Molecular detection of anammox bacteria in terrestrialecosystems: distribution and diversity. The Isme Journal, 4:450, 12/10/online 2009.[88] T. Rtting, P. Boeckx, C. Mller, and L. Klemedtsson. Assessment of the importance of dissimilatorynitrate reduction to ammonium for the terrestrial nitrogen cycle. Biogeosciences, 8:1779–1791, 2011.[89] B. Romain, P. W. Leadley, and B. A. Hungate. Global change, nitrification, and denitrification: Areview. Global Biogeochemical Cycles, 19(1), 2005.[90] M. Trimmer and J. C. Nicholls. Production of nitrogen gas via anammox and denitrification inintact sediment cores along a continental shelf to slope transect in the north atlantic. Limnology andOceanography, 54(2):577–589, Mar 2009.[91] L. F. Dong, M. N. Sobey, C. J. Smith, et al. Dissimilatory reduction of nitrate to ammonium, notdenitrification or anammox, dominates benthic nitrate reduction in tropical estuaries. Limnology andOceanography, 56(1):279–291, 2011.[92] W. P. Porubsky, N. B. Weston, and S. B. Joye. Benthic metabolism and the fate of dissolved inorganicnitrogen in intertidal sediments. Estuarine, Coastal and Shelf Science, 83(4):392–402, 2009/08/01/ 2009.[93] G. D. Song, S. M. Liu, H. Marchant, M. M. M. Kuypers, and G. Lavik. Anammox, denitrificationand dissimilatory nitrate reduction to ammonium in the east china sea sediment. Biogeosciences,10(11):6851–6864, 2013.[94] S. Rysgaard and R. N. Glud. Anaerobic n2 production in arctic sea ice. Limnology and Oceanography,49(1):86–94, 2004.[95] L. Bohlen, A. Dale, S. Sommer, et al. Benthic nitrogen cycling traversing the Peruvian oxygen minimumzone, volume 75. 2011.[96] K. L. Roberts, V. M. Eate, B. D. Eyre, D. P. Holland, and P. L. M. Cook. Hypoxic events stimulatenitrogen recycling in a shallow saltwedge estuary: The yarra river estuary, australia. Limnology andOceanography, 57(5):1427–1442, 2012.[97] K. L. Roberts, A. J. Kessler, M. R. Grace, and P. L. M. Cook. Increased rates of dissimilatory nitratereduction to ammonium (dnra) under oxic conditions in a periodically hypoxic estuary. GeochimicaEt Cosmochimica Acta, 133:313–324, May 15 2014.[98] L. De Brabandere, D. E. Canfield, T. Dalsgaard, et al. Vertical partitioning of nitrogen-loss processesacross the oxic-anoxic interface of an oceanic oxygen minimum zone. Environ Microbiol, 16(10):3041–54,Oct 2014.[99] A. Burgin, S. Hamilton, S. E. Jones, and J. Lennon. Denitrification by sulfur-oxidizing bacteria in aeutrophic lake, volume 66. 2012.[100] T. Dalsgaard, D. E. Canfield, J. Petersen, B. Thamdrup, and J. Acuna-Gonzalez. N2 production by theanammox reaction in the anoxic water column of golfo dulce, costa rica. Nature, 422(6932):606–8, Apr10 2003.120[101] A. Bourbonnais, M. F. Lehmann, David A. Butterfield, and S. Kim Juniper. Subseafloor nitrogentransformations in diffuse hydrothermal vent fluids of the juan de fuca ridge evidenced by theisotopic composition of nitrate and ammonium. Geochemistry, Geophysics, Geosystems, 13(2), 2012.[102] H. Schunck, G. Lavik, D. K. Desai, et al. Giant hydrogen sulfide plume in the oxygen minimum zoneoff peru supports chemolithoautotrophy. Plos One, 8(8), Aug 21 2013. 218YW Times Cited:1 CitedReferences Count:113.[103] A. Bourbonnais, M. A. Altabet, C. N. Charoenpong, et al. N-loss isotope effects in the peru oxygenminimum zone studied using a mesoscale eddy as a natural tracer experiment. Global BiogeochemicalCycles, 29(6):793–811, 2015/06/01 2015.[104] J. J. Wright, K. M. Konwar, and S. J. Hallam. Microbial ecology of expanding oxygen minimum zones.Nature Reviews Microbiology, 10(6):381, 2012.[105] A. R. Babbin, R. G. Keil, A. H. Devol, and B. B. Ward. Organic matter stoichiometry, flux, and oxygencontrol nitrogen loss in the ocean. Science, 344(6182):406–408, Apr 25 2014.[106] S. Bonaglia, I. Klawonn, L. De Brabandere, et al. Denitrification and dnra at the baltic sea oxic–anoxicinterface: Substrate spectrum and kinetics. Limnology and Oceanography, 61(5):1900–1915, 2016/09/012016.[107] C. B. Wenk, J. Blees, J. Zopfi, et al. Anaerobic ammonium oxidation (anammox) bacteria and sulfide-dependent denitrifiers coexist in the water column of a meromictic south-alpine lake. Limnology andOceanography, 58(1):1–12, Jan 2013.[108] D. E. LaRowe and P. Van Cappellen. Degradation of natural organic matter: A thermodynamicanalysis. Geochimica Et Cosmochimica Acta, 75(8):2030–2042, Apr 15 2011. 796EJ Times Cited:11 CitedReferences Count:92.[109] T. Awata, M. Oshiki, T. Kindaichi, et al. Physiological characterization of an anaerobic ammonium-oxidizing bacterium belonging to the ”candidatus scalindua” group. Applied and EnvironmentalMicrobiology, 79(13):4145–4148, Jul 2013.[110] T. Dalsgaard and B. Thamdrup. Factors controlling anaerobic ammonium oxidation with nitrite inmarine sediments. Appl Environ Microbiol, 68(8):3802–8, Aug 2002.[111] E. K. Robertson and B. Thamdrup. The fate of nitrogen is linked to iron(ii) availability in a freshwaterlake sediment. Geochimica et Cosmochimica Acta, 205:84–99, 5/15/ 2017.[112] J. C. Nicholls and M. Trimmer. Widespread occurrence of the anammox reaction in estuarinesediments. Aquatic Microbial Ecology, 55(2):105–113, 2009.[113] R. C. Brunet and L. J. GarciaGil. Sulfide–induced dissimilatory nitrate reduction to ammonia inanaerobic freshwater sediments. FEMS Microbiology Ecology, 21(2):131–138, 1996/10/01 1996.[114] M. M. Jensen, M. M. M. Kuypers, L. Gaute, and B. Thamdrup. Rates and regulation of anaerobicammonium oxidation and denitrification in the black sea. Limnology and Oceanography, 53(1):23–36,2008.[115] F. Lipschultz, S. C. Wofsy, B. B. Ward, et al. Bacterial transformations of inorganic nitrogen inthe oxygen-deficient waters of the eastern tropical south pacific ocean. Deep Sea Research Part A.Oceanographic Research Papers, 37(10):1513–1541, 1990/10/01/ 1990.[116] L. A. Bristow, T. Dalsgaard, L. Tiano, et al. Ammonium and nitrite oxidation at nanomolar oxygenconcentrations in oxygen minimum zone waters. Proceedings of the National Academy of Sciences,113(38):10601–10606, 2016.121[117] A. E. Giblin, N. B. Weston, G. T. Banta, J. Tucker, and C. S. Hopkinson. The effects of salinityon nitrogen losses from an oligohaline estuarine sediment. Estuaries and Coasts, 33(5):1054–1068,2010/09/01 2010.[118] S. Louca, A. K. Hawley, S. Katsev, et al. Integrating biogeochemistry with multiomic sequenceinformation in a model oxygen minimum zone. Proc Natl Acad Sci U S A, 113(40):E5925–E5933, Oct04 2016.[119] D. C. Reed, C. K. Algar, J. A. Huber, and G. J. Dick. Gene-centric approach to integrating environ-mental genomics and biogeochemical models. Proceedings of the National Academy of Sciences of theUnited States of America, 111(5):1879–1884, Feb 4 2014.[120] H. Stevens and O. Ulloa. Bacterial diversity in the oxygen minimum zone of the eastern tropicalsouth pacific. (1462-2920 (Electronic)), 2008.[121] O. Ulloa, D. E. Canfield, E. F. DeLong, R. M. Letelier, and F. J. Stewart. Microbial oceanography ofanoxic oxygen minimum zones. Proceedings of the National Academy of Sciences of the United States ofAmerica, 109(40):15996–16003, Oct 2 2012.[122] R. R Eady. Structure function relationships of alternative nitrogenases. Chemical reviews, 96(7):3013–3030, 1996.[123] P. M. Vitousek and R. W. Howarth. Nitrogen limitation on land and in the sea: how can it occur?Biogeochemistry, 13(2):87–115, 1991.[124] I. Berman-Frank, P. Lundgren, and P. Falkowski. Nitrogen fixation and photosynthetic oxygenevolution in cyanobacteria. Res Microbiol, 154(3):157–64, Apr 2003.[125] A. W. Thompson, R. A. Foster, A. Krupke, et al. Unicellular cyanobacterium symbiotic with asingle-celled eukaryotic alga. Science, 337(6101):1546–1550, 2012.[126] T. O. Delmont, C. Quince, A. Shaiber, et al. Nitrogen-fixing populations of planctomycetes andproteobacteria are abundant in surface ocean metagenomes. Nature Microbiology, 3(7):804–813,2018/07/01 2018.[127] R. H. Burris. Biological nitrogen fixation. Annual Review of Plant Physiology, 17(1):155–184, 1966.[128] E. S. Boyd, A. M. G. Costas, T. L. Hamilton, F. Mus, and J. W. Peters. Evolution of molybdenumnitrogenase during the transition from anaerobic to aerobic metabolism. Journal of Bacteriology,197(9):1690–1699, May 1, 2015 2015.[129] A. B. Hooper, T. Vannelli, D. J. Bergmann, and D. M. Arciero. Enzymology of the oxidation ofammonia to nitrite by bacteria. Antonie van Leeuwenhoek, 71(1-2):59–67, 1997.[130] J. I. Prosser and G. W. Nicol. Relative contributions of archaea and bacteria to aerobic ammoniaoxidation in the environment. Environmental microbiology, 10(11):2931–2941, 2008.[131] J. Simon and M. G. Klotz. Diversity and evolution of bioenergetic systems involved in microbialnitrogen compound transformations. Biochimica et Biophysica Acta (BBA)-Bioenergetics, 1827(2):114–135,2013.[132] H. Daims, E. V. Lebedeva, P. Pjevac, et al. Complete nitrification by nitrospira bacteria. Nature, 528:504,11/26/online 2015.[133] H. Daims, S. Lcker, and M. Wagner. A new perspective on microbes formerly known as nitrite-oxidizing bacteria. Trends in microbiology, 24(9):699–712, 2016.122[134] C. Moreno-Vivian, P. Cabello, M. Martinez-Luque, R. Blasco, and F. Castillo. Prokaryotic nitratereduction: molecular properties and functional distinction among bacterial nitrate reductases. JBacteriol, 181(21):6573–84, Nov 1999.[135] D. Tsementzi, J. Wu, S. Deutsch, et al. Sar11 bacteria linked to ocean anoxia and nitrogen loss. Nature,536:179, 08/03/online 2016.[136] A. Preisler, D. de Beer, A. Lichtschlag, et al. Biological and chemical sulfide oxidation in a beggiatoainhabited marine sediment. The Isme Journal, 1:341, 06/28/online 2007.[137] L. B. Maia and J. J. G. Moura. How biology handles nitrite. Chemical Reviews, 114(10):5273–5357,2014/05/28 2014.[138] D. R. H. Graf, C. M. Jones, and S. Hallin. Intergenomic comparisons highlight modularity of thedenitrification pathway and underpin the importance of community structure for n2o emissions. PloSone, 9(12):e114118, 2014.[139] R. A. Sanford, D. D. Wagner, Q. Wu, et al. Unexpected nondenitrifier nitrous oxide reductase genediversity and abundance in soils. Proc Natl Acad Sci U S A, 109(48):19709–14, Nov 27 2012.[140] J. A. Cole and C. M. Brown. Nitrite reduction to ammonia by fermentative bacteria: a short circuit inthe biological nitrogen cycle. FEMS Microbiology Letters, 7:65–72, 1980.[141] J. Simon. Enzymology and bioenergetics of respiratory nitrite ammonification. FEMS Microbiol Rev,26(3):285–309, Aug 2002. Simon, Jorg eng Research Support, Non-U.S. Gov’t Review Netherlands2002/08/08 10:00 FEMS Microbiol Rev. 2002 Aug;26(3):285-309.[142] M. G. Klotz, M. C. Schmid, M. Strous, et al. Evolution of an octahaem cytochrome c protein familythat is key to aerobic and anaerobic ammonia oxidation by bacteria. Environmental Microbiology,10(11):3150–3163, 2008.[143] O. Einsle. Structure and function of formate-dependent cytochrome c nitrite reductase, nrfa. MethodsEnzymol, 496:399–422, 2011.[144] S. B. Mohan, M. Schmid, M. Jetten, and J. Cole. Detection and widespread distribution of the nrfagene encoding nitrite reduction to ammonia, a short circuit in the biological nitrogen cycle thatcompetes with denitrification. FEMS Microbiol Ecol, 49(3):433–43, Sep 1 2004. Mohan, Sudesh BSchmid, Markus Jetten, Mike Cole, Jeff eng Research Support, Non-U.S. Gov’t England 2004/09/0100:00 FEMS Microbiol Ecol. 2004 Sep 1;49(3):433-43. doi: 10.1016/j.femsec.2004.04.012.[145] L. A. van Niftrik, J. A. Fuerst, J. S. S. Damst, et al. The anammoxosome: an intracytoplasmiccompartment in anammox bacteria. FEMS Microbiology Letters, 233(1):7–13, 2004.[146] R. Bartossek, G. W. Nicol, A. Lanzen, HP. Klenk, and C. Schleper. Homologues of nitrite reductasesin ammoniaoxidizing archaea: diversity and genomic context. Environmental microbiology, 12(4):1075–1088, 2010.[147] H. R. Harhangi, M. Le Roy, T. van Alen, et al. Hydrazine synthase, a unique phylomarker with whichto study the presence and biodiversity of anammox bacteria. Applied and environmental microbiology,78(3):752–758, 2012.[148] B. Kartal, W. J. Maalcke, N. M. de Almeida, et al. Molecular mechanism of anaerobic ammoniumoxidation. Nature, 479:127, 10/02/online 2011.[149] A. Paulmier and D. Ruiz-Pino. Oxygen minimum zones (omzs) in the modern ocean. Progress inOceanography, 80(34):113–128, 3// 2009.123[150] 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.[151] K. Grasshoff, M. Ehrhardt, K. Kremling, and L. G. Anderson. Methods of seawater analysis. Wiley-VCH,Weinheim ; New York, 3rd, completely rev. and extended edition, 1999.[152] A. Fontijn, A. J. Sabadell, and R. J. Ronco. Homogeneous chemiluminescent measurement ofnitric oxide with ozone - implications for continuous selective monitoring of gaseous air pollutants.Analytical Chemistry, 42(6):575–&, 1970.[153] E. Viollier, P. W. Inglett, K. Hunter, A. N. Roychoudhury, and P. Van Cappellen. The ferrozine methodrevisited: Fe(ii)/fe(iii) determination in natural waters. Applied Geochemistry, 15(6):785–790, Jul 2000.[154] B. Thamdrup, T. Dalsgaard, M. M. Jensen, et al. Anaerobic ammonium oxidation in the oxygen-deficient waters off northern chile. Limnology and Oceanography, 51(5):2145–2156, 2006.[155] E. K. Robertson, K. L. Roberts, L. D. W. Burdorf, P. Cook, and B. Thamdrup. Dissimilatory nitratereduction to ammonium coupled to fe(ii) oxidation in sediments of a periodically hypoxic estuary.Limnology and Oceanography, 61(1):365–381, Jan 2016.[156] 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 29 2006.[157] T. Tyrrell. The relative influences of nitrogen and phosphorus on oceanic primary production. Nature,400(6744):525–531, 08/05/print 1999.[158] C. Jones, S. Nomosatryo, S. A. Crowe, C. J. Bjerrum, and D. E. Canfield. Iron oxides, divalent cations,silica, and the early earth phosphorus crisis. Geology, 43(2):135–138, Feb 2015.[159] L. A. Derry. Causes and consequences of mid-proterozoic anoxia. Geophysical Research Letters,42(20):8538–8546, Oct 28 2015.[160] S. Zhang, X. Wang, H. Wang, et al. Sufficient oxygen for animal respiration 1,400 million years ago.Proc Natl Acad Sci U S A, 113(7):1731–6, Feb 16 2016.[161] N. J. Planavsky, C. T. Reinhard, X. L. Wang, et al. Low mid-proterozoic atmospheric oxygen levelsand the delayed rise of animals. Science, 346(6209):635–638, Oct 31 2014.[162] J. H. Martin, G. A. Knauer, D. M. Karl, and W. W. Broenkow. Vertex - carbon cycling in the northeastpacific. Deep-Sea Research Part a-Oceanographic Research Papers, 34(2):267–285, Feb 1987.[163] A. D. Anbar and A. H. Knoll. Proterozoic ocean chemistry and evolution: A bioinorganic bridge?Science, 297(5584):1137–1142, Aug 16 2002.[164] D. G. Capone. Marine nitrogen fixation: what’s the fuss? Current Opinion in Microbiology, 4(3):341–348,Jun 2001.[165] L. A. Codispoti. An oceanic fixed nitrogen sink exceeding 400 tg na-1 vs the concept of homeostasisin the fixed-nitrogen inventory. Biogeosciences, 4(2):233–253, 2007.[166] R. F. Keeling, A. Kortzinger, and N. Gruber. Ocean deoxygenation in a warming world. AnnualReview of Marine Science, 2:199–229, 2010.[167] S. Schmidtko, L. Stramma, and M. Visbeck. Decline in global oceanic oxygen content during the pastfive decades. Nature, 542:335, 02/15/online 2017.124[168] A. Mucci, B. Sundby, M. Gehlen, et al. The fate of carbon in continental shelf sediments of easterncanada: a case study. Deep Sea Research Part II: Topical Studies in Oceanography, 47(3):733–760,2000/04/01/ 2000.[169] M. Voss, H. W. Bange, J. W. Dippner, et al. The marine nitrogen cycle: recent discoveries, uncertaintiesand the potential relevance of climate change. Philos Trans R Soc Lond B Biol Sci, 368(1621):20130121,Jul 05 2013.[170] D. Fowler, M. Coyle, U. Skiba, et al. The global nitrogen cycle in the twenty-first century. PhilosophicalTransactions of the Royal Society B-Biological Sciences, 368(1621), Jul 5 2013.[171] S. E. Bulow, J. J. Rich, H. S. Naik, A. K. Pratihary, and B. B. Ward. Denitrification exceeds anammoxas a nitrogen loss pathway in the arabian sea oxygen minimum zone. Deep-Sea Research Part I-Oceanographic Research Papers, 57(3):384–393, Mar 2010.[172] T. Dalsgaard, B. Thamdrup, L. Farias, and N. P. Revsbech. Anammox and denitrification in theoxygen minimum zone of the eastern south pacific. Limnology and Oceanography, 57(5):1331–1346, Sep2012.[173] A. H. Devol. Nitrogen cycle: Solution to a marine mystery. Nature, 422(6932):575–576, 04/10/print2003.[174] W. Koeve and P. Khler. Heterotrophic denitrification vs. autotrophic anammox quantifying collateraleffects on the oceanic carbon cycle. Biogeosciences, 7(8):2327–2337, 2010.[175] B. A. S. Van Mooy, R. G. Keil, and A. H. Devol. Impact of suboxia on sinking particulate organiccarbon: Enhanced carbon flux and preferential degradation of amino acids via denitrification.Geochimica et Cosmochimica Acta, 66(3):457–465, 2/1/ 2002.[176] P. Engstrom, T. Dalsgaard, S. Hulth, and R. C. Aller. Anaerobic ammonium oxidation by nitrite(anammox): Implications for n-2 production in coastal marine sediments. Geochimica Et CosmochimicaActa, 69(8):2057–2065, Apr 15 2005.[177] J. J. Rich, O. R. Dale, B. Song, and B. B. Ward. Anaerobic ammonium oxidation (anammox) inchesapeake bay sediments. Microb Ecol, 55(2):311–20, Feb 2008.[178] C. C. Michiels, F. Darchambeau, F. A. E. Roland, et al. Iron-dependent nitrogen cycling in a ferruginouslake and the nutrient status of proterozoic oceans. Nature Geoscience, 10(3):217–U176, Mar 2017.[179] F. A. Richards. Anoxic basins and fjords. Academic press, 1965.[180] A. K. Hawley, M. Torres-Beltrn, E. Zaikova, et al. A compendium of multi-omic sequence informationfrom the saanich inlet water column. Scientific Data, 4:170160, 10/31/online 2017.[181] M. Torres-Beltrn, A. K. Hawley, D. Capelle, et al. A compendium of geochemical information fromthe saanich inlet water column. Scientific Data, 4:170159, 10/31/online 2017.[182] D. A. Walsh, E. Zaikova, C. G. Howes, et al. Metagenome of a versatile chemolithoautotroph fromexpanding oceanic dead zones. Science, 326(5952):578–82, Oct 23 2009.[183] E. Zaikova, D. F. Walsh, Claire P. Stilwell, et al. Microbial community dynamics in a seasonally anoxicfjord: Saanich inlet, british columbia. Environ Microbiol, 12(1):172–91, 2010.[184] J. J. Anderson and A. H. Devol. Deep water renewal in saanich inlet, an intermittently anoxic basin.Estuarine and Coastal Marine Science, 1(1):1–10, 1// 1973.125[185] A. H. Devol, J. J. Anderson, K. Kuivila, and J. W. Murray. A model for coupled sulfate reduction andmethane oxidation in the sediments of saanich inlet. Geochimica Et Cosmochimica Acta, 48(5):993–1004,1984.[186] C. C. Manning, R. C. Hamme, and A. Bourbonnais. Impact of deep-water renewal events on fixednitrogen loss from seasonally-anoxic saanich inlet. Marine Chemistry, 122(1-4):1–10, Oct 2010.[187] A. K. Hawley, M. K. Nobu, J. J. Wright, et al. Diverse marinimicrobia bacteria may mediate cou-pled biogeochemical cycles along eco-thermodynamic gradients. Nature Communications, 8(1):1507,2017/11/15 2017.[188] V. Shah, B. X. Chang, and R. M. Morris. Cultivation of a chemoautotroph from the sup05 clade ofmarine bacteria that produces nitrite and consumes ammonium. Isme Journal, 11(1):263–271, Jan 2017.[189] R. S. Braman and S. A. Hendrix. Nanogram nitrite and nitrate determination in environmentaland biological materials by vanadium(iii) reduction with chemiluminescence detection. AnalyticalChemistry, 61(24):2715–2718, 1989/12/01 1989.[190] T. R. Parsons, Y. Maita, and C. M. Lalli. 4.3 - fluorometric determination of chlorophylls. In A Manualof Chemical & Biological Methods for Seawater Analysis, pages 107–109. Pergamon, Amsterdam, 1984.[191] A. Bourbonnais, M. F. Lehmann, R. C. Hamme, C. C. Manning, and S. K. Juniper. Nitrate eliminationand regeneration as evidenced by dissolved inorganic nitrogen isotopes in saanich inlet, a seasonallyanoxic fjord. Marine Chemistry, 157:194–207, Dec 2013.[192] A. H. Knap, A. Michaels, A. R. Close, H. Ducklow, and A. G. Dickson. Protocols for the joint globalocean flux study (jgofs) core measurements, 1996.[193] B. Thamdrup and T. Dalsgaard. The fate of ammonium in anoxic manganese oxide-rich marinesediment. Geochimica Et Cosmochimica Acta, 64(24):4157–4164, Dec 2000.[194] J. Wright, S. Lee, E. Zaikova, D. A. Walsh, and S. J. Hallam. DNA Extraction from 0.22 M Sterivex Filtersand Cesium Chloride Density Gradient Centrifugation, volume 31. 2009.[195] W. Walters, E. R. Hyde, D. Berg-Lyons, et al. Improved bacterial 16s rrna gene (v4 and v4-5) andfungal internal transcribed spacer marker gene primers for microbial community surveys. mSystems,1(1):e00009–15, Jan-Feb 12/22 10/12/received 11/24/accepted 2016.[196] R. C. Edgar, B. J. Haas, J. C. Clemente, C. Quince, and R. Knight. Uchime improves sensitivity andspeed of chimera detection. Bioinformatics, 27(16):2194–2200, 2011.[197] J. G. Caporaso, J. Kuczynski, J. Stombaugh, et al. Qiime allows analysis of high-throughput communitysequencing data. Nature Methods, 7:335, 04/11/online 2010.[198] E. Kopylova, J. A. Navas-Molina, C. Mercier, et al. Open-source sequence clustering methods improvethe state of the art. mSystems, 1(1), Jan-Feb 2016.[199] C. Quast, E. Pruesse, P. Yilmaz, et al. The silva ribosomal rna gene database project: improved dataprocessing and web-based tools. Nucleic Acids Research, 41(D1):D590–D596, 2013.[200] C. Sancetta and S. E. Calvert. The annual cycle of sedimentation in saanich inlet, british-columbia -implications for the interpretation of diatom fossil assemblages. Deep-Sea Research Part a-OceanographicResearch Papers, 35(1):71–90, Jan 1988.[201] D. S. Grundle, D. A. Timothy, and D. E. Varela. Variations of phytoplankton productivity andbiomass over an annual cycle in saanich inlet, a british columbia fjord. Continental Shelf Research,29(19):2257–2269, Oct 30 2009.126[202] M. Takahashi, D. L. Seibert, and W. H. Thomas. Occasional blooms of phytoplankton during summerin saanich inlet, bc, canada. Deep-Sea Research, 24(8):775–&, 1977.[203] S. Christensen and J. M. Tiedje. Sub-parts-per-billion nitrate method: Use of an n(2)o-producingdenitrifier to convert no(3) or no(3) to n(2)o. Appl Environ Microbiol, 54(6):1409–13, Jun 1988.[204] D. Parsonage, A. J. Greenfield, and S. J. Ferguson. The high affinity of paracoccus denitrificans cellsfor nitrate as an electron acceptor. analysis of possible mechanisms of nitrate and nitrite movementacross the plasma membrane and the basis for inhibition by added nitrite of oxidase activity inpermeabilised cells. Biochimica et Biophysica Acta (BBA) - Bioenergetics, 807(1):81–95, 1985/04/03 1985.[205] R. E. Murray, L. L. Parsons, and M. S. Smith. Kinetics of nitrate utilization by mixed populations ofdenitrifying bacteria. Appl Environ Microbiol, 55(3):717–21, Mar 1989.[206] S. Ganesh, D. J. Parris, E. F. DeLong, and F. J. Stewart. Metagenomic analysis of size-fractionatedpicoplankton in a marine oxygen minimum zone. ISME J, 8(1):187–211, Jan 2014. Ganesh, San-gita Parris, Darren J DeLong, Edward F Stewart, Frank J eng Research Support, Non-U.S. Gov’tResearch Support, U.S. Gov’t, Non-P.H.S. England 2013/09/14 06:00 ISME J. 2014 Jan;8(1):187-211.doi: 10.1038/ismej.2013.144. Epub 2013 Sep 12.[207] O Inceoglu, M. Llirs, T. Garca-Armisen, et al. Distribution of bacteria and archaea in meromictictropical lake kivu (africa). Aquatic Microbial Ecology, 74(3):215–233, 2015.[208] K. Bergauer, A. Fernandez-Guerra, J. A. L. Garcia, et al. Organic matter processing by microbialcommunities throughout the atlantic water column as revealed by metaproteomics. Proceedings of theNational Academy of Sciences, 115(3):E400–E408, 2018.[209] K. Anantharaman, J. A. Breier, C. S. Sheik, and G. J. Dick. Evidence for hydrogen oxidation andmetabolic plasticity in widespread deep-sea sulfur-oxidizing bacteria. Proceedings of the NationalAcademy of Sciences, 110(1):330–335, 2013.[210] C. M. Callbeck, G. Lavik, T. G. Ferdelman, et al. Oxygen minimum zone cryptic sulfur cyclingsustained by offshore transport of key sulfur oxidizing bacteria. Nature Communications, 9(1):1729,2018/04/30 2018.[211] Sa. Glaubitz, K. Kielich, C. Meeske, M. Labrenz, and K. Jrgens. Sup05 dominates the gammapro-teobacterial sulfur oxidizer assemblages in pelagic redoxclines of the central baltic and black seas.Applied and Environmental Microbiology, 79(8):2767–2776, April 15, 2013 2013.[212] A. Canion, O. Prakash, S. J. Green, et al. Isolation and physiological characterization of psychrophilicdenitrifying bacteria from permanently cold arctic fjord sediments (svalbard, norway). EnvironmentalMicrobiology, 15(5):1606–1618, 2013.[213] C. O. Wirsen, S. M. Sievert, C. M. Cavanaugh, et al. Characterization of an autotrophic sulfide-oxidizing marine arcobacter sp. that produces filamentous sulfur. Applied and Environmental Microbiol-ogy, 68(1):316–325, January 1, 2002 2002.[214] M. A. Altabet, E. Ryabenko, L. Stramma, et al. An eddy-stimulated hotspot for fixed nitrogen-lossfrom the peru oxygen minimum zone. Biogeosciences, 9(12):4897–4908, 2012. BG.[215] K. R. Arrigo. Marine microorganisms and global nutrient cycles. Nature, 437:349, 09/14/online 2004.[216] P. Lam and M. M. Kuypers. Microbial nitrogen cycling processes in oxygen minimum zones. AnnRev Mar Sci, 3:317–45, 2011. Lam, Phyllis Kuypers, Marcel M M eng Research Support, Non-U.S.Gov’t Review 2011/02/19 06:00 Ann Rev Mar Sci. 2011;3:317-45.127[217] C. Thomazo and D. Papineau. Biogeochemical cycling of nitrogen on the early earth. Elements,9(5):345–351, Oct 2013.[218] C. Thomazo, M. Ader, and P. Philippot. Extreme 15N-enrichments in 2.72-Gyr-old sediments: Evidence fora turning point in the nitrogen cycle, volume 9. 2011.[219] H. Strauss. Anoxia through time. In Lev N. Neretin, editor, Past and Present Water Column Anoxia,pages 3–19, Dordrecht, 2006// 2006. Springer Netherlands.[220] H. C. Jenkyns, D. R. Grcke, and S. P. Hesselbo. Nitrogen isotope evidence for water mass denitri-fication during the early toarcian (jurassic) oceanic anoxic event. Paleoceanography, 16(6):593–603,2001.[221] M. B. Higgins, R. S. Robinson, J. M. Husson, S. J. Carter, and A. Pearson. Dominant eu-karyotic export production during ocean anoxic events reflects the importance of recyclednh¡sub¿4¡/sub¿¡sup¿+¡/sup¿. Proceedings of the National Academy of Sciences, 109(7):2269–2274, 2012.[222] M. Hannig, G. Lavik, M. M. M. Kuypers, et al. Shift from denitrification to anammox after inflowevents in the central baltic sea. Limnology and Oceanography, 52(4):1336–1345, 2007/07/01 2007.[223] C. Deutsch, J. L. Sarmiento, D. M. Sigman, N. Gruber, and J.P. Dunne. Spatial coupling of nitrogeninputs and losses in the ocean. Nature, 445:163, 01/11/online 2007.[224] B. Kartal, M. M. Kuypers, G. Lavik, et al. Anammox bacteria disguised as denitrifiers: nitratereduction to dinitrogen gas via nitrite and ammonium. Environ Microbiol, 9(3):635–42, Mar 2007.Kartal, Boran Kuypers, Marcel M M Lavik, Gaute Schalk, Jos Op den Camp, Huub J M Jetten, Mike SM Strous, Marc Journal Article Research Support, Non-U.S. Gov’t England Environ Microbiol. 2007Mar;9(3):635-42. doi: 10.1111/j.1462-2920.2006.01183.x.[225] A. E. Giblin, C. R. Tobias, B. Song, et al. The importance of dissimilatory nitrate reduction toammonium (dnra) in the nitrogen cycle of coastal ecosystems. Oceanography, 26(3):124–131, Sep 2013.[226] E. E. Roden and Q. Jin. Thermodynamics of microbial growth coupled to metabolism of glucose,ethanol, short-chain organic acids, and hydrogen. Applied and environmental microbiology, 2011.[227] D. K. Button. Kinetics of nutrient-limited transport and microbial growth. Microbiological reviews,49(3):270–297, 1985. 3930934[pmid] PMC373036[pmcid] Microbiol Rev.[228] D. E. LaRowe and J. P. Amend. Power limits for microbial life. Frontiers in Microbiology, 6:718, 07/1501/30/received 06/30/accepted 2015. 26236299[pmid] Front Microbiol.[229] D. E. LaRowe and J. P. Amend. Catabolic rates, population sizes and doubling/replacement times ofmicroorganisms in natural settings. American Journal of Science, 315(3):167–203, March 1, 2015 2015.[230] C. C. Michiels, J. A. Huggins, K. E. Giesbrecht, et al. Rates and pathways of n2 production in apersistently anoxic fjord: Saanich inlet, british columbia. Frontiers in Marine Science, 6:27, 2019.[231] F. R. Warembourg. Nitrogen fixation in soil and plant systems, 1993.[232] Tom Vtrovsk and Petr Baldrian. The variability of the 16s rrna gene in bacterial genomes and itsconsequences for bacterial community analyses. PLOS ONE, 8(2):1–10, 02 2013.[233] D. Li, C. M. Liu, R. Luo, K. Sadakane, and T. W. Lam. Megahit: an ultra-fast single-node solution forlarge and complex metagenomics assembly via succinct de bruijn graph. Bioinformatics, 31(10):1674–6,May 15 2015.128[234] K. M. Konwar, N. W. Hanson, A. P. Pag, and S. J. Hallam. Metapathways: a modular pipelinefor constructing pathway/genome databases from environmental sequence information. BMCbioinformatics, 14(1):202, 2013.[235] C. S. Miller, B. J. Baker, B. C. Thomas, S. W. Singer, and J. F. Banfield. Emirge: reconstruction offull-length ribosomal genes from microbial community short read sequencing data. Genome Biology,12(5):R44, 2011/05/19 2011.[236] J. A. Fish, B. Chai, Q. Wang, et al. Fungene: the functional gene pipeline and repository. Frontiers inmicrobiology, 4:291, 2013.[237] D. A. Benson, I. Karsch-Mizrachi, D. J. Lipman, J. Ostell, and D. L. Wheeler. Genbank. Nucleic AcidsResearch, 36(Database issue):D25–D30, 12/11 09/18/received 10/10/accepted 2008. 18073190[pmid]Nucleic Acids Res.[238] R. C Edgar. Search and clustering orders of magnitude faster than blast. Bioinformatics, 26(19):2460–2461, 2010.[239] S. Capella-Gutirrez, J. M. Silla-Martnez, and T. Gabaldn. trimal: a tool for automated alignmenttrimming in large-scale phylogenetic analyses. Bioinformatics, 25(15):1972–1973, 2009.[240] K. Katoh and D. M. Standley. Mafft multiple sequence alignment software version 7: improvementsin performance and usability. Molecular biology and evolution, 30(4):772–780, 2013.[241] A. Stamatakis. Raxml-vi-hpc: maximum likelihood-based phylogenetic analyses with thousands oftaxa and mixed models. Bioinformatics, 22(21):2688–2690, 2006.[242] A. Stamatakis and J. Hoover, P.and Rougemont. A rapid bootstrap algorithm for the raxml webservers. Systematic biology, 57(5):758–771, 2008.[243] D. Hyatt, G-L. Chen, P. F. LoCascio, et al. Prodigal: prokaryotic gene recognition and translationinitiation site identification. BMC bioinformatics, 11(1):119, 2010.[244] S. R. Eddy. Profile hidden markov models. Bioinformatics (Oxford, England), 14(9):755–763, 1998.[245] S. A. Berger, D. Krompass, and A. Stamatakis. Performance, accuracy, and web server for evolutionaryplacement of short sequence reads under maximum likelihood. Systematic biology, 60(3):291–302, 2011.[246] H. Li. Aligning sequence reads, clone sequences and assembly contigs with bwa-mem. arXiv preprintarXiv:1303.3997, 2013.[247] D. S. Grundle and S. K. Juniper. Nitrification from the lower euphotic zone to the sub-oxic waters ofa highly productive british columbia fjord. Marine Chemistry, 126(1-4):173–181, Sep 2011.[248] P. E. Galand, S. Lucas, S. K. Fagervold, et al. Disturbance increases microbial community diversityand production in marine sediments. Frontiers in microbiology, 7:1950–1950, 2016. 27994581[pmid]PMC5133735[pmcid] Front Microbiol.[249] I. Ylla, H. Peter, A. M. Roman, and L. J. Tranvik. Different diversityfunctioning relationship in lakeand stream bacterial communities. FEMS Microbiology Ecology, 85(1):95–103, 2013. 10.1111/1574-6941.12101.[250] G. Lavik, T. Sthrmann, V. Brchert, et al. Detoxification of sulphidic african shelf waters by bloomingchemolithotrophs. Nature, 457:581, 12/10/online 2008.[251] M. A. Moran, B. Satinsky, S. M. Gifford, et al. Sizing up metatranscriptomics. The Isme Journal, 7:237,08/30/online 2012.129[252] D. Y. Sorokin, T. N. Zhilina, A. M. Lysenko, T. P. Tourova, and E. M. Spiridonova. Metabolic versatilityof haloalkaliphilic bacteria from soda lakes belonging to the alkalispirillumalkalilimnicola group.Extremophiles, 10(3):213–220, June 01 2006.[253] L. Villanueva, D. R. Speth, T. van Alen, A. Hoischen, and M. S. M. Jetten. Shotgun metagenomic datareveals significant abundance but low diversity of ”candidatus scalindua” marine anammox bacteriain the arabian sea oxygen minimum zone. Frontiers in microbiology, 5:31–31, 2014. 24550902[pmid]PMC3913995[pmcid] Front Microbiol.[254] B. Kartal and J. T Keltjens. Anammox biochemistry: a tale of heme c proteins. Trends in biochemicalsciences, 41(12):998–1011, 2016.[255] D. Haase, B. Hermann, O. Einsle, and J. Simon. Epsilonproteobacterial hydroxylamine oxidoreductase(hao): characterization of a missing linkin the multihaem cytochrome c family. Molecular microbiology,105(1):127–138, 2017.[256] O. M. Neijssel and D. W. Tempest. Bioenergetic aspects of aerobic growth of klebsiella aerogenes nctc418 in carbon-limited and carbon-sufficient chemostat culture. Archives of Microbiology, 107(2):215–221,1976/03/01 1976.[257] W. Chesbro, T. Evans, and R. Eifert. Very slow growth of escherichia coli. Journal of Bacteriology,139(2):625, 1979.[258] G. Goma, R. Moletta, and M. Novak. Comments on the maintenance coefficient changes duringalcohol fermentation. Biotechnology Letters, 1(10):415–420, 1979/10/01 1979.[259] W. Beyeler, P. L. Rogers, and A. Fiechter. A simple technique for the direct determination ofmaintenance energy coefficient: An example with zymomonas mobilis. Applied Microbiology andBiotechnology, 19(4):277–280, 1984/04/01 1984.[260] P. G. Falkowski, D. Ziemann, Z. Kolber, and P. K. Bienfang. Role of eddy pumping in enhancingprimary production in the ocean. Nature, 352:55, 07/04/online 1991.[261] A. Oschlies and V. Garon. Eddy-induced enhancement of primary production in a model of the northatlantic ocean. Nature, 394:266, 07/16/online 1998.[262] W. J. Sydeman, M. Garca-Reyes, D. S. Schoeman, et al. Climate change and wind intensification incoastal upwelling ecosystems. Science, 345(6192):77, 2014.[263] P. Xiu, F. Chai, E. N. Curchitser, and F. S. Castruccio. Future changes in coastal upwelling ecosystemswith global warming: The case of the california current system. Scientific Reports, 8(1):2866, 2018/02/122018.[264] X-Y. Zheng, H. C. Jenkyns, A. S. Gale, D. J. Ward, and G. M. Henderson. Changing ocean circulationand hydrothermal inputs during ocean anoxic event 2 (cenomanianturonian): Evidence from nd-isotopes in the european shelf sea. Earth and Planetary Science Letters, 375:338–348, 2013/08/01/2013.[265] H. C. Jenkyns. Transient cooling episodes during cretaceous oceanic anoxic events with specialreference to oae 1a (early aptian). Philos Trans A Math Phys Eng Sci, 376(2130), Oct 13 2018. 1471-2962Jenkyns, Hugh C ORCID: Journal Article Review England Phi-los Trans A Math Phys Eng Sci. 2018 Oct 13;376(2130). pii: rsta.2017.0073. doi: 10.1098/rsta.2017.0073.[266] X. Lang, B. Shen, Y. Peng, et al. Transient marine euxinia at the end of the terminal cryogenianglaciation. Nature Communications, 9(1):3019, 2018/08/01 2018.130[267] B. J. Koch, T. A. McHugh, M. Hayer, et al. Estimating taxon-specific population dynamics in diversemicrobial communities. Ecosphere, 9(1):e02090, 2018.[268] K. Laufer, H. Roy, B. B. Jorgensen, and A. Kappler. Evidence for the existence of autotrophic nitrate-reducing fe(ii)-oxidizing bacteria in marine coastal sediment. Appl Environ Microbiol, 82(20):6120–6131,Oct 15 2016. Laufer, Katja Roy, Hans Jorgensen, Bo Barker Kappler, Andreas eng Appl EnvironMicrobiol. 2016 Sep 30;82(20):6120-6131. doi: 10.1128/AEM.01570-16. Print 2016 Oct 15.[269] J. M. Klatt and L. Polerecky. Assessment of the stoichiometry and efficiency of co2 fixation coupledto reduced sulfur oxidation. Front Microbiol, 6:484, 2015. Klatt, Judith M Polerecky, Lubos engSwitzerland Front Microbiol. 2015 May 21;6:484. doi: 10.3389/fmicb.2015.00484. eCollection 2015.[270] N. Gruber and J. L. Sarmiento. Global patterns of marine nitrogen fixation and denitrification. GlobalBiogeochemical Cycles, 11(2):235–266, Jun 1997. Xp160 Times Cited:551 Cited References Count:125.[271] C. J. Bjerrum and D. E. Canfield. Ocean productivity before about 1.9 gyr ago limited by phosphorusadsorption onto iron oxides. Nature, 417(6885):159–62, May 9 2002. Bjerrum, Christian J Canfield,Donald E eng Research Support, Non-U.S. Gov’t England Nature. 2002 May 9;417(6885):159-62. doi:10.1038/417159a.[272] C. Klein. Some precambrian banded iron-formations (bifs) from around the world: Their age, geologicsetting, mineralogy, metamorphism, geochemistry, and origins. American Mineralogist, 90(10):1473,2005.[273] G.R. Helz, C.V. Miller, J.M. Charnock, et al. Mechanism of molybdenum removal from the sea and itsconcentration in black shales: Exafs evidence. Geochimica et Cosmochimica Acta, 60(19):3631 – 3642,1996.131Appendix AChapter 1: supplemental materialA.1 Isotope pairing technique protocolThis method is used to measure the production of N2 or NH+4 through the addition of15N-labeled speciesto measure the rates of denitrification, anammox and DNRA. The protocol was adapted from [74, 154] butscaled to bigger incubation vessels that allowed for less heterogeneity in the samples.A.1.1 SamplingFrom a Niskin bottle, overflow the 250mL serum bottle 3 times and close it with a blue stopper to avoidoxygen contamination. Do not seal the bottle and put it in the dark until the beginning of the experiment.Avoid any big temperature variation during transportation to the lab to prevent breaking the bottles.A.1.2 Start of the incubationOnce back in the lab, start the incubations as soon as possible. First, add a 20mL Helium headspace(Helium 5.0 purity) at 1 atmosphere pressure. This can be achieved by having an outlet in the gas-line thatis open to the atmosphere. After adding a headspace, the bottle can be sealed with a crimper. The liquidobtained from the 20mL headspace can be filtered (0.2m filter) and frozen at -20◦C for later analysis of thenutrients in it. While the bottles equilibrate with their new headspace (shake gently for 15 minutes). Onceequilibrated, add the 15N-label to your bottles. The concentration added will depend on the environmentconsidered. Typically, we add 50% of the ambient concentration of the N-species naturally present. It isessential to avoid any cross-contamination of the 15NO–3 ,15NO–2 and15NH+4 while using the syringes andneedles.A.1.3 Taking time pointsTwo types of samples are taken for each time point: a gas and a liquid sample. For the gas sample, a 1mLgas-tight syringe (Hamilton company) is flushed three times with Helium at 1 atmosphere. After flushing,1mL of Helium is injected into the headspace of the serum bottle and 1mL of the headspace is taken up inthe same syringe. This constitutes the gas sample and it will be preserved in a 3mL exetainer filled withddi water. To do so, the 1mL of gas sample is inserted in the exetainer while having an output needlethat expels 1mL of ddi water replaced by the gas. It is important to have a gas-tight syringe for each ofthe 15N-labels used in the incubations in order to avoid cross-contamination that would lead to anammoxproducing 30N2 gas. The liquid sample are taken by a flushed 5mL plastic syringe, 2mL of He are inserted,similarly to the gas sample, and 2mL of liquid are taken up in the syringe. The sample is then immediatelyfrozen at -20◦C for later analysis. Time points are usually taken at 0, 3, 6, 12 and 24 hours, depending onthe activity of the microorganisms.A.1.4 Analysis of samplesGas samples are later analyzed on an Isotope Ratio Mass Spectrometer for the accumulation of 29N2 and30N2 in the headspace. In the liquid samples, 15NH+4 is transformed into15 – N2 following Warembourget al. (year) and also analyzed on the IRMS. Total NO–2 and NH+4 are analyzed spectrophotometrically132by the Griess and indophenol assay, respectively [151]. Total NOx (NO–2 and NO–3 ) is measured viachemiluminescence [189].A.2 Summary of pelagic and benthic rates of denitrification,anammox and DNRA133Table A.1: Summary of benthic rates of denitrification, anammox and DNRA The rates found in this table was summarized from the literature (seereferences column).Place/Station, Conditions, Lat, Lon, DNRA,,,AN,,,DEN,,,Reference,Laguna Madre, Texas Shallow estuary 27.279 -97.427 559.2 1843.2 0 0 256.8 902.4 (An and Gardner, 2002) Venice Lagoon, Italy Coastal seds 45.334 12.285 240 6600 0 0 240 6000 (Azzoni et al., 2014) Dorum, Denmark Intertidal flats 53.736 -8.507   120 0 0   2160 (Behrendt et al., 2013) Aarhus Bight, Denmark Coastal Bay 56.105 -10.463   120 0 0   1200 See above Mississipi, USA River delta 29.225 -83.453   240 0 0   2640 See above Limfjord , Denmark Shallow fjord 56.537 -9.370   120 0 0   1920 See above Janssand, Denmark Intertidal flats 53.735 -7.696 240 720 0 0   720 See above Little Lagoon, Alabama eutrophic estuary 30.237 -87.752 1089.6 3868.8 0 0 88.8 362.4 See above Peruvian OMZ sediments Sediments underlying OMZ -11 -78.6 480 2930 280 430 200 2020 (Bohlen et al., 2011) Baltic Sea estuary changes of oxygen conditions in sed, from hypoxia to oxidized 58.833 -17.666 0.24 720 12 48 48 408 (Bonaglia et al., 2014)  Gulf of Bothnia oligotrophic basin, cold, well-oxygenated 65.191 -23.395 10 275 10 65 50 300 (Bonaglia et al., 2017). Gulf of Mexico hypoxic sediments 29.1 -89.3 ND ND ND ND 1149.6 2594.4 (Childs et al., 2002) Western North American continental margin continental shelf sediment 28 -113.5   2660 ND ND 720 2680 (Chong et al., 2012) Lower St Lawrence Estuary estuary sediments 48.700 -68.652   0.12   132   271.2 (Crowe et al., 2011) By Fjord (Sweden) Hypoxic, euxinic sometimes basin sediements, reoxygenated 58.333 11.869 20 525 0 0 20 679 (De Brabandere et al., 2015) Mae Klong estuary (Thailand) Tropical estuary 13.411 99.997 12000 720000 0 0 2400 216000 (Dong et al., 2011) Cisadane estuary (Indonesia) Tropical estuary -6.019 106.631 216000 24000000 0 0 24000 2400000 See above Vunidawa-Rewa Estuary (Fiji) Tropical estuary -18.106 178.541 2400 288000 0 0 2400 72000 See above Florida Bay Eutrophic coastal sediments 25 -81 240 6000 ND ND 120 4200 (Gardner and McCarthy, 2009) Texas estuaries Shallow estuaries  28.5 -98.520 384 2376 ND ND 120 1128 (Gardner et al., 2006)  Arctic Fjord sediments (Svalbard, Norway) Arctic coastal sediments 79.700 11.086 ND ND 10 26 34 294 (Gihring et al., 2010)  Plum Island Sound Estuary (USA, MA) salt marsh sediments 42.724 -70.831 100 2000 0 0 50 600 (Giblin et al., 2010) Gulf of Finland, Baltic Sea Hypoxic basin 59.933 22.097 13 1060 0 0 38 1619 (Jantti and Hietanen, 2012) !134Table A.2: Summary of benthic rates of denitrification, anammox and DNRA, cont’d The rates found in this table was summarized from the literature(see references column).Lower Gt. Ouse (North Sea estuary) temperate estuarine sediments 52.816 0.383 600 26400 ND ND 168 5280 (Kelly-Gerreyn et al., 2001) Plum Island Sound Estuary (USA, MA) salt marsh sediments 42.724 -70.831 93.6 585.6 0 0 428 7971 (Koop-Jakobsen and Giblin, 2010) Ca'Stanga and Lago Verde hypolimnetic sediments of lowland lake-mesotrophic 45.054 9.796 70 120 0 0 700 4100 (Nizzoli et al., 2010) Lake - Hoffman Metropark (Ohio) Freshwater sediment - test of increase NO3 and bioturbation 39.013 -84.001 67.2 547.2 ND ND 67.2 5512.8 (Nogaro and Burgin, 2014) Colne estuary (UK) estuary sediments 51.807 0.999 ND ND ND ND 36 8848.8 (Ogilvie et al., 1997) Dover Bluff - Georgia (USA) Coastal sediments 30.416 -81.5   912   1.2 12 6720 (Porubsky et al., 2009) Grave's Dock - South Carolina (USA) Coastal sediments 32.333 -81.416   552   1.2 12.24 3984.96 See above Gulf of Mexico seds Hypoxic-sulfidic seds with Thioploca mats 25.5 -112 2500 3400   1600   1100 (Prokopenko et al., 2013) Yarra River estuary seds (Australia) River seds -37.843 145.116 480 14400 0 0 480 9600 (Roberts et al., 2012)  Yarra River estuary seds (Australia) River seds, under oxic and hypoxic conditions -37.843 145.116 160.8 808.8 0 0 3720 17760 (Roberts et al., 2014)  Arctic Sea ice - Young Sound Artic sea ice 74.309 20.250 ND ND 0 2.3 10 45 (Rysgaard and Glud, 2004) Bassin d'Arcachon sediments (France) Coastal lagoons sediments 44.699 -1.116 70 310 ND ND 20 1010 (Rysgaard et al., 1996) East China Sea sediment Shelf sediments 29.088 123.803 2600 9700 2000 5000 3000 18000 (Song et al., 2013) Banks of Weser river (Germany) River sediments 52.994 9.004   52.8 ND ND   1893.6 (Stief et al., 2010) Atlantic Ocean next to UK-Ireland Continental slope 54.119 5.569   0.024   2.64   139.92 (Trimmer and Nicholls, 2009) Atlantic Ocean next to UK-Ireland Continental shelf 48.060 9.853   0.12   60.24   4.8 (Trimmer and Nicholls, 2009) Lake Lugano (Switzerland) Freshwater sediment 46.009 9.030 16.8 148.8 16.8 91.2 141.6 1372.8 (Wenk et al., 2014) !135Table A.3: Summary of pelagic rates of denitrification, anammox and DNRA The rates found in this table was summarized from the literature (seereferences column).Place/Station Conditions Lat Lon DNRA   AN   DEN   Reference ETSP Northern Chile OMZ -20.341 -70.561 ND ND 2.5E-10 1.25E-09 5E-11 3E-10 (De Brabandere et al., 2014) Peruvian OMZ OMZ -12 -77.3 3E-09 2.2E-08 2E-08 2.50E-07 ND ND (Lam et al., 2009) ETSP  OMZ -10 -80 4.8E-10 1.74E-09 2.84E-09 2.27E-07 2.21E-09 5.42E-09 (Kalvelage et al., 2013) Juan de Fuca Ridge Sulfidic hydrothermal vents 45.92 -129.99 6E-09 1.51E-07 2E-09 5E-09 5E-09 9.77E-07 (Bourbonnais et al., 2012) Omani Shelf OMZ 18 65 2E-08 3.70E-08 2E-09 2.5E-08 ND ND (Jensen et al., 2011) Mediterranean particles Particles 43 5.5 5E-10 1.1E-08 ND ND 2E-09 1.5E-07 (Michotey and Bonin, 1997) Gotland Basin, Baltic Sea anoxic, sulfidic basin 57.3 20.5 1E-08 1.5E-07 5E-09 5E-08 2.5E-07 2.7E-06 (Hannig et al., 2007) Gotland and Westerland basin, Baltic Sea anoxic, sulfidic basin 58 18 3.6E-09 1.61E-08 1.68E-09 7.44E-09 5.52E-09 9.67E-07 (Bonaglia et al., 2016b) Wintergreen lake , MI , USA Sulphidic Eutrophic lake 42.397 -85.386 1.2E-06 3.5E-06 ND* ND 1E-06 1.7E-05 (Burgin et al., 2012) Rhone River Plume water of the river 43.4667 4.833 7E-07 2.5E-06 ND ND 1E-06 4.3E-06 (Omnes et al., 1996) Kabuno Bay, Lake Kivu (RDC) Ferruginous conditions -1.617 29.063 2.5E-08 4.8E-08 ND ND 4.8E-08 7.5E-08 (Michiels et al., 2017) Benguela upwelling system (Namibian) OMZ -22.5 13.9     1E-08 1.7E-07 ND ND (Kuypers et al., 2005)  ETSP OMZ -22 -73.5     9.6E-10 2.06E-08 2.64E-09 1.9E-07 (Dalsgaard et al., 2012)  ETNP OMZ 20.15 -106     5E-10 1.2E-08 1E-09 3E-08 (Babbin et al., 2014)  ETSP Northern Chili OMZ -20.1 -70.317     3E-09 1.68E-08 ND ND (Thamdrup et al., 2006)  Chilean OMZ OMZ -20.086 -70.336     6.96E-09 1.03E-08 1.01E-09 1.9 E-09 (Canfield et al., 2010b)  Peruvian OMZ OMZ -12 -77.5     1.5E-09 3.84E-07 ND ND (Hamersley et al., 2007)  Central Baltic Sea OMZ 57.5 18.5     ND ND 5.76E-09 3.816E-07 (Dalsgaard et al., 2003)  Arabian Sea OMZ 17.5 65     1.2E-10 4.32E-09 2.4E-10 2.54E-08 (Ward et al., 2009)  Arabian Sea OMZ 19 66      4.23E-09 1.00E-09 2.12E-08 (Bulow et al., 2010) Black sea OMZ 42.512 30.245     7E-09 7E-09 ND ND (Kuypers et al., 2003)  Golfo Dulce (Costa Rica) Anoxic basin 8.570 83.245     1.2E-07 4.8E-07 7.2E-08 2.64E-06 (Dalsgaard et al., 2003) Golfo Dulce (Costa Rica) Anoxic basin 8.6 83.267     1E-09 1.5E-08 ND ND Jensen, PhD thesis (2006) Mariager Fjord anoxic sulfidic basin 56.663 9.974     ND ND  1.86E-05 (Jensen et al., 2009)  Black Sea anoxic sulfidic basin 43.233 34     1.7E-10 1.77E-08 ND ND (Jensen et al., 2008) Lugano Lake (Switzerland) Sulfidic lake 46.009 9.031     1E-09 1.5E-08 3E-08 9E-08 (Wenk et al., 2013) Lake Cadagno (Switzerland) Sulfidic lake 46.550 8.711     ND ND 6.96E-08 7.92E-08 (Halm et al., 2009) !136Appendix BChapter 2: supplemental materialB.1 Fe-dependent NO–3 reduction – thermodynamic considerationsIn order to test the thermodynamic favourability of reactions involving the different possible intermediatesin Fe-dependent NO–3 reduction, we calculated the relevant Gibbs free energy yields (Table B.1). Insitu concentrations for the different chemical species implicated are depicted in Table B.2. Temperatureconsidered was 297◦K (i.e., 23.85◦C) and the gas constant (R) used was 0.008314 kJ K– 1 mol– 1. UnderKabuno Bay conditions (Table B.2), all the reactions outlined in Table B.1 are thermodynamically favourable.B.2 Denitrification and DNRA rates summary in Kabuno BayRates of DNRA and denitrification have been calculated by linear-regressions with the least squares methodover the time interval during which data are linear (24 or 48hrs) for 15NH+4 or30N2 production, respectively.The rates and the error associated are displayed in in Table B.3 .B.3 Dark carbon fixation in Kabuno BayRecent literature described the carbon fixation efficiency of Fe(II) dependent NO–3 reducers from coastalmarine sediments as being 1 mole of CO2 fixed per 26.5 moles of Fe oxidized [268]. The products ofNO–3 reduction were not fully known in this case but the authors hypothesized based on the reactionstoichiometry that it leads to N2 production. Therefore, because 5 moles of Fe(II) are needed to reduce1 mole of NO–3 , the carbon fixation efficiency for denitrification would be 0.18 (rC/Denitr). By applying afactor of 8/5 to rC/Denitr, we hypothesize a ratio to DNRA (rC/DNRA) of 0.3. These factors are similar tothose described for sulphide dependent NO–3 reducers by [269]. Indeed, ratios of CO2 fixed per NO–3 usedthrough sulphide dependent denitrification (to N2) vary from 0.13 to 0.36 [269]. By applying a factor of8/5 to rC/Denitri, we adapted the ratio to DNRA (rC/DNRA), which then varies from 0.21 to 0.58. In practice,growth yields for DNRA may differ from denitrification, and this stands as an important opportunity forfuture research. With 40% DNRA and 60% denitrification, the contribution of NO–3 reduction to total darkcarbon fixation [50] is 2% (summarized in Table B.4) based on the ratio inferred from [268].B.4 Box-model of C, N, S and Fe cycling for a hypotheticalProterozoic upwelling systemThe model used in the present study is based on the model developed by [5] for a modern coastal upwellingsystem. It was previously adapted to a Proterozoic upwelling system by Boyle et al. 2013 showing thateuxinia was only present when N2-fixation occurred in the photic zone. The general structure of our 5box model is briefly summarized in the main text, Figure B.1 as well as in Fig2.3a. We kept Canfield(2006)’s [5] model structure and dimensions (described in Fig.B.1), as well as most model parameters(described in Fig.B.1 and in Table B.5), but added DNRA as well as the Fe-cycle to our model. N-fixationwas not considered here as the model sustains export production without its contribution, and N-fixation iscommonly absent in modern upwelling systems. Upwelling rates are represented with the coefficient A and137B (cm hr– 1, see Fig.B.1 and in Table B.5) and vertical exchange between the different boxes is representedby the different K coefficients (cm hr– 1, see Fig.B.1 and in Table B.5). Upwelled waters bring nutrients tothe euphotic zone (here NO–3 and/or NH+4 ), settling rates of primary production are based on N limitation.Primary production, also called export production in Boyle et al. 2013 is described as follows (Eq.B.1):EP = EPNO−3 + EPNH+4 =A + B + KurN:C∗ (NO3um + NH4um) (B.1)Primary production is exported through sedimentation to the intermediate box (UM), where microbialrespiration occurs. In the UM box, part of sedimented organic matter is degraded through oxic respiration,which together with nitrification consumes oxygen. Nitrification, in turn, produces NO–3 . Oxic respiration(Raerobic) is limited by the oxygen diffusing from the surface waters (U box) into the UM box. Surfacewater oxygen was set assuming equilibrium with the atmosphere, and oxygen concentrations based on thereconstructions from the geologic record. We can calculate the rate of Raerobic, which includes nitrification,based on the flux of oxygen entering the UM box, as shown in Eq.B.2.Raerobic =Ku ∗O2u + (A + Kum)O2D + (KI + B)O2IrO2 :C(B.2)Considering that oxygen can only come from the U box, Eq.B.2 simplifies as:Raerobic =Ku ∗O2urO2 :C(B.3)All oxygen was consumed through combined respiration and nitrification directly in the UM box. NO–3reduction proceeds first using Fe(II) as an electron donor. For low upwelling rates, NO–3 limits Fe-dependentNO–3 reduction, and we can therefore calculate rates of NO–3 reduction based on the supply of NO–3 to theUM box as follows:NO−3 − limited : RNO−3 Fe = (A + Kum) ∗ NO3D + (B + KI) ∗ NO3I + rN:C ∗ Raerobic (B.4)NO–3 is supplied through upwelling from intermediate waters and is also produced through nitrification inthe UM box. Eq. B.4 implies that NO–3 in the UM box is consumed entirely and is therefore zero. If Fe(II) islimiting (instead of NO–3 ), on the other hand, we can calculate rates of Fe-dependent NO–3 reduction basedon the supply of Fe(II) to the UM box instead of the supply of NO–3 :Fe− limited : RNO−3 Fe = ((A + Kum) ∗ FeD + (B + KI) ∗ FeI) ∗ rNO3:Fe (B.5)with rNO3:Fe defined in Eq.B.11 below.In order to determine whether NO–3 or Fe(II) is limiting, we compared the supply rates of both(Eqs.B.4B.5) and considered the lowest as the actual rate of Fe-dependent NO–3 reduction (RNO–3 Fe). IfFe(II) is limiting, there will be NO–3 left in the UM box that then fuels organic matter oxidation. Thisyields both Fe and C-dependent NO–3 reduction in the UM box. The NO–3 allocated to C-dependent NO–3reduction (RNO–3 C) can be calculated by subtracting Eq.B.4-Eq.B.5. Again, the NO–3 concentration in theUM box is zero as it is all consumed through a combination of Fe and C-dependent NO–3 reduction. Insummary:Case 1: NO–3 limiting Fe-dependent NO–3 reduction RNO3−tot = RNO3−Fe (B.6)Case 2: Fe(II) limiting Fe-dependent NO–3 reduction RNO3−tot = RNO3−Fe + RNO3−C (B.7)As mentioned here above, we considered both DNRA and denitrification as part of NO–3 reduction.By doing so, we are able to evaluate the effect of the partitioning between DNRA and denitrification on138primary production, sulphate reduction rates, and the accumulation of hydrogen sulphide. We thereforevaried the relative contributions of DNRA and denitrification to overall NO–3 reduction and this ultimatelyinfluences the loss of N from the system versus recycling to NH+4 through DNRA. In order to addressthis balance between the two pathways, we reformulated the description of NO–3 reduced per molecule ofelectron donor consumed (Fe(II) or organic matter) so that it reflected the overall stoichiometry of combinedDNRA and denitrification. Denitrification consumes 5 electrons per NO–3 reduced versus the 8 electronsinvolved in DNRA. The half reactions for denitrification and DNRA are the following:Denitrification: NO−3 + 5e− + 6H+ ⇒ N2 + 3H2O (B.8)DNRA: NO−3 + 8e− + 10H+ ⇒ NH+4 + 3H2O (B.9)For Fe-dependent NO–3 reduction, we considered the following half reaction:Fe2+ + 3H2O⇒ Fe(OH)3 + 1e− + 3H+ (B.10)With x being the contribution of DNRA to NO–3 reduction and (1-x) the contribution of denitrification,we can define the number of moles of NO–3 used per mole of Fe(II) in Eq.B.11.rNO3:Fe =x8+(1− x)5(B.11)We also can define the number of moles of NH+4 released per mole of Fe(II) in Eq.B.12.rNH4:Fe =x8(B.12)Considering now C-dependent NO–3 reduction, the half reaction of organic C oxidation used here is:C106H175N16O42P+ 280H2O⇒ 106HCO−3 + 16NH+4 +HPO2−4 + 564H+ + 472e− (B.13)Eq.B.14 shows the number of moles of NO–3 consumed per mole of organic C with a varying contributionof DNRA and denitrification to NO–3 reduction.rNO3:C = 472 ∗x8 +1−x5106(B.14)Finally, Eq.B.15 was modified from [44], so that rNH4:C, accounted for both NH+4 released from remineral-ization of organic matter through NO–3 reduction as well as the production of NH+4 through DNRA permole of C oxidized. rNH4:C is therefore written as follows:rNH4:C =16+ (59 ∗ x)106(B.15)As NO–3 in the UM box equals 0, the Eq.B.1 for export production can therefore be simplified to:EP =(A + B + Ku) ∗ NH4umrN:C(B.16)With NH4UM calculated in Eq.B.17, taking into account the ammonium released from NO–3 reductionthrough DNRA (both Fe and C-dependent), we can therefore calculate Eq.B.16.NH4um =(A + Kum) ∗ NH4D + (B + KI) ∗ NH4I + rNH4:Fe ∗ RNO3−Fe + rNH4:C ∗ RNO3−C − rN:C ∗ Raerobic(Kum + KI)(B.17)139If organic matter remains after exhausting NO–3 in the UM box, the remaining amount can be oxidizedthrough iron and sulphate reduction. Iron reduction was insignificant, and wasn’t considered further.Sulfate reduction rates can be written as:RSR = EP− Raerobic − RNO3−CrNO3:C (B.18)The sulphide produced through RSR in the UM box being described in Eq.B.19:H2Sum =59106∗ RSR(A + B + KI + Kum + Ku)(B.19)With Eq.B.20 as the half reaction used for sulphate reductionSO2−4 + 8e− + 10H+ ⇒ H2S+ 4H2O (B.20)Finally, based on the rate of highly reactive Fe entering the UM box and the rate of H2S produced throughsulphate reduction, we can then infer the ratio of Fe-pyrite to highly reactive Fe used in the rock record todistinguish euxinic from ferruginous conditions.rFePY/HRFe =RSR∗59106(2 ∗ ((A + Kum) ∗ FeD + (B + KI) ∗ FeI) (B.21)Supplemental conditions from Fig.2.3b and c are displayed here below in Fig.B.2 for the Fe-pyriteto highly reactive Fe ratio under 2 atmospheric oxygen concentrations (3.8% and 6.2% PAL) and forhigher contributions of DNRA to NO–3 reduction. Results show that euxinic conditions are reached atupwelling rates lower than when the contribution of DNRA is smaller, and without the need of an increasedammonium supply from the deep ocean.The main parameters are constrained in in Table B.5. These are the benchmark values used for theruns of the model of the main text, if not stated otherwise in the text or legends of the figures. We providefurther explanation on how specific parameters were constrained below.In the main text, we explored the influence of oxygen on the model outputs from 0% to 12% PAL (Fig.2.3d in the main text). 0% PAL is a special case where oxygen is not available locally for nitrification in theupwelling zone. However, we maintained the supply of NO–3 from intermediate waters, as non-local oxygenoases are plausible in the Archean ocean, even under an ostensibly anoxic atmosphere [38]. Therefore,these oxygen oases could have enabled the local production of NO–3 through nitrification in other partsof the Archean ocean and supplied the NO–3 to intermediate waters as considered in our box model. Wealso tested a broad range in deep ocean Fe(II) and NH+4 concentrations, as mentioned in the main text.Indeed, these parameters are poorly constrained in the literature and we therefore studied the influenceof likely ranges on our model outputs. Fe(II) concentrations are commonly thought to be controlled byequilibrium with siderite (FeCO3), which yields between 40 to 120µM deep ocean Fe(II) [46]. However,[159] suggests siderite formation was kinetically limited and Fe(II) concentrations may have been muchhigher (< 3mM). Assuming upwelled P is needed to fuel oxygenic photosynthesis and sustain appreciableatmospheric O2 in the Proterozoic Eon, Fe(II) concentrations must then have been less than 424 timesdeep water P concentrations [158]. Indeed, at Fe(II):P ratios greater than 424, upwelling P is consumedthrough photoferrotrophy and would therefore not reach the surface waters to support appreciable oxygenicphotosynthesis. Based on these arguments, we chose an Fe(II) concentration 42µM (based on 424 x 0.1 µMP) for the benchmark in our model runs presented in the main text. However, we also tested a range ofconcentrations (from 10 to 500µM) that encompasses the values described by [46]. We focused on the role ofFe(II) concentrations in dictating the FePY/FeHR ratio across a suite of different model conditions in Fig.B.3(a-h). Overall, and as expected, without DNRA and at low deep ammonium concentrations (2µM, Fig. B.3a to d), ferruginous conditions tend to prevail as Fe(II) concentrations increase. However, with DNRA andat 10µM Fe(II) (Fig. B.3 e), Fe(II) is limiting and euxinic conditions (FePY/FeHR>0.7) develop at relativelylow upwelling rates. Above 42µM on the other hand (Fig. B.3 b-d, f-h), with or without DNRA, Fe(II) is140supplied in excess and effectively titrates any sulphide produced through sulphate reduction, invariablyyielding ferruginous conditions (FePY/FeHR<0.7). Figure B.3 (j to l) depicts different concentrations of deepFe(II) versus a wide range of deep NH+4 concentrations (between 0 and 15µM) without the contribution ofDNRA (0% DNRA). This shows that euxinic conditions could also be reached without the contribution ofDNRA, but mainly under low deep Fe(II) concentrations (between 10 and 42µM) and under relatively highNH+4 concentrations (>13µM), as mentioned in the main text. Our benchmark model runs described in themain text invariably consider NO–3 to be present in the intermediate waters. We then assumed that theseintermediate waters would be Fe(II) free as it would have been consumed through NO–3 reduction. Theopposite could be true, on the other hand, and so we also tested this here to evaluate the effect of Fe(II)bearing NO–3 free intermediate waters. When intermediate waters contain Fe(II), we added equimolar NH+4instead of NO–3 , accordingly. Results of this test are depicted in Figure B.4 and show that, although exportproduction is very high compared to the benchmark model scenarios, euxinic conditions (FePY/FeHR>0.7)do not occur in the water column. Without a supply of NO–3 through the intermediate waters, NO–3reduction is fuelled only through nitrification and therefore by the oxygen supply from the surface water(3.8% PAL in this test). This being minimal, NO–3 reduction and N-loss are highly restricted.B.5 Global N-fixation and N-loss in the Archean and ProterozoicAnnual rates of marine N-fixation for the Proterozoic Eon are estimated based on the modern ratesdescribed in [270]. In order to scale the modern rates to the Proterozoic Eon, we assumed that N-fixationwas ultimately limited by P supply [157] and was therefore proportional to deep ocean P concentrations.We thus divided the modern rates of 135 50 Tg N yr– 1 (encompassing both pelagic and benthic N-fixation)by the modern phosphorous concentration (2.3µM, [271]) in the deep ocean and multiplied this by thehighest estimates for the phosphorous concentration (0.13µM) described for Paleoproterozoic oceans [158].The 4.8 Tg N yr– 1 we report in the main text is our lowest estimate if we consider the error on the N-fixationestimate. To assess the extent to which we could apply the 0.13 µM deep water P concentration from [158]across the Proterozoic Eon we took values of Fe/Si from the Rapitan iron-formation [272] and applied theseto [158] model to infer phosphorous concentrations for the Neoproterozoic oceans. Values found for theRapitan were within the range of those calculated by [158]. We also considered trace metal limitation ofN-fixation very unlikely. The most likely metal to limit N-fixation in the Proterozoic is molybdenum [163].However, [273] showed that high levels of sulphide (between 50 and 250µM) are necessary to effectivelystrip Mo from seawater under euxinia. Our model implies that under most reasonable scenarios sulphideconcentrations do not exceed about 20µM and are therefore insufficient to trigger effective Mo removal.Conservative rates of global N-loss were inferred from our box-model when denitrification contributes100% of NO–3 reduction (no DNRA, therefore higher N-loss) under 6.2% PAL, low ammonium conditions(2µM) and deep ocean Fe(II) concentrations of 42µM. The highest rates of N-loss were found with thehighest upwelling rate explored in this model (3 cm hr– 1). We then extrapolated N-loss from our model toan area equivalent to upwelling regions in the modern ocean (0.36 10 12 m2) as indicated in the main text.In comparison, rates of N-loss under the lowest upwelling rate considered in our model (0.5 cm hr– 1) are 4times lower than with an upwelling rate of 3 cm hr– 1.141Table B.1: Free Gibbs Energy yield under standard conditions (4G◦) and for Kabuno Bay concentrations(4G). Values for 4G◦ can be found in [4]Reactions ∆G˚ (kJ /mol N) ∆G (kJ /mol N) 8 Fe2+ + 21 H2O + NO3- => NH4+ + 8 Fe(OH)3 + 14 H+ 51.67 -272.41 5 Fe2+ + 12 H2O + NO3- => 0.5 N2 + 5 Fe(OH)3 +9 H+ -143.19 -336.97 6 Fe2+ + 16 H2O + NO2- => NH4+ + 6 Fe(OH)3 + 10 H+ 31.93 -189.81 3 Fe2+ + 7 H2O + NO2- => 0.5 N2 + 3 Fe(OH)3 +7 H+ -162.93 -328.28 2 Fe2+ + 4.5 H2O + NO2- => 0.5 N2O + 2 Fe(OH)3 + 3 H+ -83.85 -137.86 4 Fe2++ 11.5 H2O + 0.5 N2O => NH4+ + 4 Fe(OH)3 + 7 H+ 115.78 -51.95 Fe2+ + 2.5 H2O + 0.5 N2O => 0.5 N2 + Fe(OH)3 + 2 H+  -158.16 -186.59 2 Fe2+ + 5 H2O + NO3- => NO2- + 2 Fe(OH)3 + 4 H+ 19.74 -82.60 !Table B.2: Chemical species concentrations (in µM) representative for the chemocline in Kabuno Bay! Chemical species Concentration (µM) NO3- 1 NO2- 1 N2O 0.01 NH4+ 100 Fe (II) 100 Fe(OH)3 106  activity=1 as a pure solid pH 6.5 142Table B.3: Summary of DNRA and denitrification rates for KBs water column. Rates were calculated over48 hours unless stated otherwise next to the calculated rates.Depth (m) 30N2 production  (nmol N d-1)  30N2 production  (nmol N d-1) with Fe added  15NH4+ production (nmol N d-1) 15NH4+ production (nmol N d-1) with Fe added 9.5 0 ± 0 0 ± 0 0 ± 0 0 ± 0 10 0 ± 0 0 ± 0 0 ± 0 0 ± 0 10.5 0 ± 0 0 ± 0 0 ± 0 0 ± 0 11 80 ± 10 230 ± 40 20 ± 0 (24h) 70 ± 20 (24h) 11.5 50 ± 10 140± 20 50 ± 10 (24h) 70 ± 10 !Table B.4: Rates and ratio considered for calculationsMicrobial process Process rate measurements Reference Dark Carbon fixation 1.49µmol C L-1 d-1 Lliros et al. 2015 DNRA 70 nmol N L-1 d-1 This paper Denitrification 230 nmol N L-1 d-1 This paper !143Table B.5: Description of the different parameters used in the current modelParameters Value/ Units Description Reference Kum 0.2 cm h-1 Vertical exchange Canfield 2006 Ku 0.1 cm h-1 Vertical exchange Canfield 2006 Ki 0.4 cm h-1 Vertical exchange Canfield 2006 A 0 cm h-1 Upwelling Rate Canfield 2006 B 0.5-3 cm h-1 Upwelling Rate Canfield 2006 O2u 9.5-15.5 µM (3.8% to 6.2% PAL) O2 concentration in U box Zhang et al. 2016  O2D 0 µM O2 concentration in D box Boyle et al. 2013 O2I 0 µM O2 concentration in I box Boyle et al. 2013 r O2:C 170/117  Ratio of molecule of O2 consumed for 1 molecule of carbon oxidized Boyle et al. 2013 r N:C  16/106  Ratio based on Redfield ratio  Redfield, 1934 EP nmol C cm-2 h-1 Rate of export production  Raerobic nmol C cm-2 h-1 Rate of aerobic respiration  RNO3-tot nmol N cm-2 h-1 Rate of nitrate reduction through Fe dependent nitrate reduction (if case 1) or through Fe and C dependent NO3– reduction (if case 2)  RNO3-Fe nmol N cm-2 h-1 Rate of nitrate reduction through Fe dependent NO3– reduction  RNO3-C nmol N cm-2 h-1 Rate of nitrate reduction through C dependent NO3– reduction (only case 2)  RSR nmol C cm-2 h-1 Rate of sulphate reduction  RFe-ox nmol Fe cm-2 h-1 Rate of Fe oxidation  NH4UM, NH4D, NH4I TBD, 2, 0 µM Ammonium concentration in UM, D, I box Deep water concentrations based on Jones et al. 2015 Phosphorous concentration estimates. NO3UM, NO3D, NO3I TBD, 0, 1 µM Nitrate concentration in UM, D, I box Deep water concentrations based on Jones et al. 2015 Phosphorous concentration estimates. FeUM, FeD, FeI TBD, 42, 0 µM Iron (II) concentration in UM, D, I box Deep water concentrations based on Jones et al. 2015 Phosphorous concentration estimates. H2SUM TBD µM Sulfide concentration in UM box  !144Figure B.1: Box-model for C, N, S and Fe cycling in hypothetical Precambrian upwelling system adapted from [5]Notation is as follows: Upwelling coefficients (A+B) from intermediate and deep waters (boxes I and D)as well as horizontal (KI) and vertical mixing (Ku and Kum) between the UM box and the other boxesconsidered (I, U and D respectively). Box S represents ocean surface waters away from the upwelling zone.The parameter values are listed in Table B.5. Organic matter produced in the euphotic zone (box U) asexport production settles to box UM where it is partially (in [5]) or entirely ([44] and this paper) degraded.The order of the pathways through which it is degraded is oxic respiration, followed by nitrate reduction([5] and [44]), and finally by sulphate reduction.0.5 1.5 2.5Upwelling rate (cm hr-1) HRRatio O2 1_50%Ratio O2 2_50%Ratio O2 2_100%Ratio O2 1_100%Treshold ratio pyrite0.5 1.5 2.5Upwelling rate (cm hr-1) HRRatio O2 1_50%Ratio O2 2_50%Ratio O2 2_100%Ratio O2 1_100%Treshold ratio pyrite50% DNRA, 3.8% PAL 50% DNRA, 6.2% PAL  100% DNRA, 3.8% PAL 100% DNRA, 6.2% PALFigure B.2: Fe-pyrite to highly reactive Fe ratio for 50 and 100% DNRAFePY/FeHR for 50 and 100% DNRA (ingreen and orange respectively) with varying surface waters oxygen (3.8% PAL in solid lines and 6.2%PALin dashed lines) and for different upwelling rates. These model runs are for deep NH4+ concentrations of2µM.1450.5 1.0 1.5 2.0 2.5 3.0Upwelling rate (cm hr-1) Fe(II) (µM)05101520253035Export production (nmol C cm-2 hr-1)NH4+Fe(II)NH4+ UM_1Fe_1Export productionEP_10.5 1.0 1.5 2.0 2.5 3.0Upwelling rate (cm hr-1)051015NH4+and Fe(II) (µM)05101520253035Export production (nmol C cm-2 hr-1)NH4+ UM_3_1Real FeUM_3_1NH4+ UM_4Real FeUM_4EP_3_1EP_40.5 1.0 1.5 2.0 2.5 3.0Upwelling rate (cm hr-1)0255075100NH4+and Fe(II) (µM)05101520253035Export production (nmol C cm-2 hr-1)NH4+ UM_7Real FeUM_7NH4+ UM_8Real FeUM_8EP_7EP_80.5 1.0 1.5 2.0 2.5 3.0Upwelling rate (cm hr-1) Fe(II) (µM)05101520253035Export production (nmol C cm-2 hr-1)NH4+ UM_2Real FeUM_2NH4+ UM_3Real FeUM_3EP_2EP_30.5 1.0 1.5 2.0 2.5 3.0Upwelling rate (cm hr-1)051015NH4+and Fe(II) (µM)05101520253035Export production (nmol C cm-2 hr-1)NH4+ UM_5Real FeUM_5NH4+ UM_6Real FeUM_6EP_5EP_60.5 1.0 1.5 2.0 2.5 3.0Upwelling rate (cm hr-1)0255075100NH4+and Fe(II) (µM)05101520253035Export production (nmol C cm-2 hr-1)NH4+ UM_9Real FeUM_9NH4+ UM_10Real FeUM_10EP_9EP_100.5 1.0 1.5 2.0 2.5 3.0Upwelling rate (cm hr-1) Fe(II) (µM)05101520253035Export production (nmol C cm-2 hr-1)NH4+Fe(II)NH4+ UM_1Fe_1Export productionEP_10.5 1.0 1.5 2.0 2.5 3.0Upwelling rate (cm hr-1) Fe(II) (µM)05101520253035Export production (nmol C cm-2 hr-1)0.5 1.5 2.5Upwelling rate (cm hr-1) pyrite/reactive FePyrite ratio_1Treshold ratio pyrite0.5 1.5 2.5Upwelling rate (cm hr-1) pyrite/reactive Fe_3_1Ratio pyrite/reactive Fe_4Treshold ratio pyrite0.5 1.5 2.5Upwelling rate (cm hr-1) pyrite/reactive Fe_7Ratio pyrite/reactive Fe_8Treshold ratio pyrite0.5 1.5 2.5Upwelling rate (cm hr-1) pyrite/reactive Fe_2Ratio pyrite/reactive Fe_3Treshold ratio pyrite0.5 1.5 2.5Upwelling rate (cm hr-1) pyrite/reactive Fe_5Ratio pyrite/reactive Fe_6Treshold ratio pyrite0.5 1.5 2.5Upwelling rate (cm hr-1) pyrite/reactive Fe_9Ratio pyrite/reactive Fe_10Treshold ratio pyrite0.5 1.5 2.5Upwelling rate (cm hr-1) 1.5 2.5Upwelling rate (cm hr-1) 3 5 7 9 11 13 15NH4+ concentration in deep ocean (µM) 1 cm hr-1_1Upwelling 2 cm hr-1_1Upwelling 3 cm hr-1_1Possible euxinia_11 3 5 7 9 11 13 15NH4+ concentration in deep ocean (µM) 1 cm hr-1Upwelling 2 cm hr-1Upwelling 3 cm hr-1Possible euxinia1 3 5 7 9 11 13 15NH4+ concentration in deep ocean (µM) 1 cm hr-1_1_1Upwelling 2 cm hr-1_1_1Upwelling 3 cm hr-1_1_1Possible euxinia_11 3 5 7 9 11 13 15NH4+ concentration in deep ocean (µM) 1 cm hr-1_1_2Upwelling 2 cm hr-1_1_2Upwelling 3 cm hr-1_1_2Possible euxinia_1a. b. c. d.e. f. g. h.i. j. k. l.Figure B.3: role of Fe(II) concentrations in dictating the FePY/FeHR ratio across a suite of different model conditionsGraphs (a-d) represents model runswith 0% DNRA; (e-h) represents model runs with 30% DNRA. Solid lines represent model runs with the surface water oxygen concentrationsof 3.8% PAL, whereas dashed lines represent runs at 6.2% PAL [blue=export production, orange= NH+4 concentrations, and black=Fe(II)concentrations, insets show the Fe-pyrite to highly reactive Fe ratio (FePY/FeHR) where the grey line delineates plausible euxinic conditions]; (i-l)represent model runs of FePY/FeHR ratios for a range of deep ocean NH+4 concentrations at 0% DNRA with surface water oxygen concentrationsof 3.8% PAL (orange=upwelling rate of 1cm hr– 1, blue=upwelling rate of 2cm hr– 1, and green=upwelling rate of 3cm hr– 1). The first column ofthese graphs are for deep Fe(II) concentrations of 10µM, the second column is 42µM (as represented in the main text), the third is 120µM andthe fourth is 500µM.1460.5 1.0 1.5 2.0 2.5 3.0Upwelling rate (cm hr-1)020406080100120Export Production (nmol C cm-2 hr-1) HR ratio0%40%FePY/FeHR 0%FePY/FeHR 40%Figure B.4: Run of the model with 20µM Fe(II) in the intermediate box (I) but no NO–3 . Instead, upwelled watersfrom box I are bringing 1µM of NH+4 to the upwelled zone. Deep waters are bringing 42µM Fe(II) and 2µMNH+4 . Solid lines represent export production whereas dashed lines represent FePY/FeHR ratio. Case with0% DNRA is in orange and case with 40%DNRA is in green, however, as the two cases yield very similarresults, the orange case is hidden by the green.147Appendix CChapter 3: supplemental materialC.1 NH+4 sediment fluxes in SaanichBased on the measured Sulphate Reduction rates (SRR) in SI [185], we considered SRR to be 2.97 to 10.44mmol m– 2 d– 1. We consider the ratio 53:16 to convert SRR into NH+4 release from carbon degradation,based on the Redfield Ratio. Therefore, when scaled to NH+4 release from sediments and a 50% desorptionoff the sediments, we obtain fluxes of NH+4 out of the sediments of 2.97 to 10.44 mmol N m– 2 d– 1. Based onour depth-integrated rates, NH+4 requirements for anammox range between 0.15 to 3.36 m– 2 d– 1. Therefore,NH+4 fluxes from the sediments in Saanich Inlet can support 90 to 100% of the NH+4 requirements foranammox.C.2 Microbial communities in SI148Table C.1: Summary of samples and the number of sequences and OTUs observed in each sample, as wellas bacterial small subunit ribosomal RNA (SSU or 16S rRNA) gene abundance obtained through qPCR.The chao diversity index was also calculated for each sample based on OTUs.Sample chao Observed otus # sequences #16S L-1 ALL / 28946 6638571 / JAN15.10m 7020 3303 73265 4.42E+09 JAN15.100m 8442 3816 111088 7.95E+09 JAN15.120m 8229 3935 132852 6.71E+09 JAN15.135m 5619 2540 92149 1.68E+09 JAN15.150m 4979 2218 84718 1.23E+10 JAN15.200m 6467 3032 67297 7.10E+09 FEB15.10m 5531 2491 55585 1.57E+10 FEB15.100m 6515 2951 71687 5.67E+09 FEB15.120m 6176 2769 106204 2.34E+09 FEB15.135m 5781 2775 87220 5.29E+09 FEB15.150m 6248 2673 97350 1.01E+10 FEB15.200m 7836 3437 85859 3.49E+08 MAR15.10m 4708 2248 35602 7.78E+08 MAR15.100m 6685 3090 81536 3.42E+09 MAR15.120m 5532 2283 78958 1.57E+09 MAR15.135m 6104 2898 104198 5.52E+09 MAR15.150m 6556 3137 94787 2.30E+09 MAR15.200m 7733 3418 87169 1.20E+09 APR15.10m 3799 2079 119314 8.06E+09 APR15.100m 6761 3314 86563 4.52E+09 APR15.120m 5342 2435 89452 1.68E+10 APR15.135m 3600 1665 70646 6.82E+09 APR15.150m 4408 1946 63358 4.56E+09 APR15.200m 6428 2986 76538 5.11E+08 MAY15.10m 2863 1660 81768 4.75E+09 MAY15.100m 7208 3245 71663 1.26E+09 MAY15.120m 7129 3504 88110 3.66E+10 MAY15.135m 7095 3157 92109 1.05E+09 MAY15.150m 8002 3756 102431 1.47E+09 MAY15.200m 8137 3900 116481 1.48E+09 JUN15.10m 1700 1110 102049 4.79E+08 JUN15.100m 7592 3648 95726 2.74E+09 JUN15.120m 6867 3191 98121 9.55E+09 JUN15.135m 6713 3038 102559 5.81E+09 JUN15.150m 9148 4140 111352 3.02E+09 JUN15.200m 7371 3682 91422 1.90E+09 JUL15.10m 4899 3506 100021 3.05E+09 JUL15.100m 7345 3647 95214 3.51E+09 JUL15.120m 5893 2731 75056 3.98E+09 JUL15.135m 4830 2257 80596 9.12E+09 JUL15.150m 5585 2517 91495 1.15E+09 JUL15.200m 5989 2720 83155 1.03E+10 AUG15.10m 3742 2043 74247 2.44E+10 AUG15.100m 8367 3472 103099 3.27E+09 AUG15.120m 7878 3438 99616 1.43E+09 AUG15.135m 5656 2607 94232 3.01E+09 AUG15.150m 5927 2548 100997 2.47E+09 AUG15.200m 6756 3097 110607 1.73E+09 SEP15.10m 7302 3612 93822 1.29E+08 SEP15.100m 9417 4110 105994 1.62E+09 SEP15.120m 6888 3093 114820 2.51E+09 SEP15.135m 4893 2335 72435 6.93E+08 SEP15.150m 5107 2360 81670 8.30E+09 SEP15.200m 4101 1942 99747 3.44E+10 OCT15.10m 5490 2856 84800 8.65E+08 OCT15.100m 4931 2389 76400 1.20E+10 OCT15.120m 5740 2657 90584 1.20E+10 OCT15.135m 5456 2525 92290 6.84E+09 OCT15.150m 4444 1986 96184 7.90E+09 OCT15.200m 6690 2866 89212 1.17E+09 NOV15.10m 8909 4599 111511 1.40E+10 NOV15.100m 8559 3973 117157 1.30E+10 NOV15.120m 6565 3075 119680 3.11E+08 NOV15.135m 6428 3107 117327 6.57E+08 NOV15.150m 7554 3638 88536 1.36E+08 NOV15.200m 7602 3499 108763 5.59E+07 DEC15.10m 9686 5304 125996 1.19E+09 DEC15.100m 5828 2974 76294 3.56E+08 DEC15.120m 4929 2359 81375 3.25E+08 DEC15.135m 5895 2569 81654 2.96E+08 DEC15.150m 6710 2982 97678 6.21E+07 DEC15.200m 5521 2864 99121 4.45E+07 !149●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●2500500075001000010m 100m 120m 135m 150m 200mSample2chao1Sample10m135m200m150m120m100mChao diversity indexSample depth●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●2500500075001000010m 100m 120m 135m 150m 200mSample2chao1Sample10m135m200m150m120m100●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●2500500075001000010m 100m 120m 135m 150m 200mSample2chao1Sample10m1352 050120●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●2500500075001000010m 100m 120m 135m 150m 200mSample2chao1Sample0m35252100m●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●2500500075001000010m 100m 120m 135m 150m 200mSample2chao1Samplem352052100m●●●●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●2500500075001000010m 100m 120m 135m 150m 200mSample2chao1Samplem35252100mFigure C.1: Chao1 diversity index from iTags sequencing1500 20 40 60JAN15.10mFEB15.10mMAR15.10mAPR15.10mMAY15.10mJUN15.10mJUL15.10mAUG15.10mSEP15.10mOCT15.10mNOV15.10mDEC15.10mSampleSequence_proportion0204060JAN15.10mFEB15.10mMAR15.10mAPR15.10mMAY15.10mJUN15.10mJUL15.10mAUG15.10mSEP15.10mOCT15.10mNOV15.10mDEC15.10mSampleSequence_proportionTaxonomyAcidimicrobiales;OM1Alteromonadales;AlteromonadaceaeCellvibrionales;PorticoccaceaeChloroplast;UnculturedFlavobacteriales;Flavobacteriaceae_1Flavobacteriales;Flavobacteriaceae_2Flavobacteriales;Flavobacteriaceae_3Flavobacteriales;Flavobacteriaceae_4Marine_Group_1;Unknown_FamilyMethylophilales;MethylophilaceaeOceanospirillales;SUP05Oceanospirillales;UnclassifiedRhodobacterales;RhodobacteraceaeSAR11;Surface1SAR11;Surface2Relative abundance (%)0 20 40 600 20 40 60JAN15.10mFEB15.10mMAR15.10mAPR15.10mMAY15.10mJUN15.10mJUL15.10mAUG15.10mSEP15.10mOCT15.10mNOV15.10mDEC15.10mSampleSequence_proportion0204060JAN15.10mFEB15.10mMAR15.10mAPR15.10mMAY15.10mJUN15.10mJUL15.10mAUG15.10mSEP15.10mOCT15.10mNOV15.10mDEC15.10mSampleSequence_proportionTaxonomyAcidimicrobiales;OM1Alteromonadales;AlteromonadaceaeCellvibrionales;PorticoccaceaeChloroplast;UnculturedFlavobacteriales;Flavobacteriaceae_1Flavobacteriales;Flavobacteriaceae_2Flavobacteriales;Flavobacteriaceae_3Flavobacteriales;Flavobacteriaceae_4Marine_Group_1;Unknown_FamilyMethylophilales;MethylophilaceaeOceanospirillales;SUP05Oceanospirillales;UnclassifiedRhodobacterales;RhodobacteraceaeSAR11;Surface1SAR11;Surface210mJDFMAMJJASONFigure C.2: Relative abundance of 15 most abundant taxa for the surface waters of SI in 2015 (10m). Thesetaxa were the most abundant ones found in average throughout the samples.151Relative abundanceDEC15.200mDEC15.150mDEC15.135mDEC15.120mDEC15.100mE11NOV15.200mNOV15.150mNOV15.135mNOV15.120mNOV15.100mE10OCT15.200mOCT15.150mOCT15.135mOCT15.120mOCT15.100mE9SEP15.200mSEP15.150mSEP15.135mSEP15.120mSEP15.100mE8AUG15.200mAUG15.150mAUG15.135mAUG15.120mAUG15.100mE7JUL15.200mJUL15.150mJUL15.135mJUL15.120mJUL15.100mE6JUN15.200mJUN15.150mJUN15.135mJUN15.120mJUN15.100mE5MAY15.200mMAY15.150mMAY15.135mMAY15.120mMAY15.100mE4APR15.200mAPR15.150mAPR15.135mAPR15.120mAPR15.100mE3MAR15.200mMAR15.150mMAR15.135mMAR15.120mMAR15.100mE2FEB15.200mFEB15.150mFEB15.135mFEB15.120mFEB15.100mE1JAN15.200mJAN15.150mJAN15.135mJAN15.120mJAN15.100m0 20 40 60 80Sequence_proportionSampleTaxonomyBacteroidetes_VC2.1_Bac22;Uncultured_FamilyCampylobacterales;Campylobacteraceae;ArcobacterChromatiales;EctothiorodospiraceaeFlavobacteriales;FlavobacteriaceaeMarine_Group_1;Unknown_FamilyMarinimicrobia;Uncultured_Family_1Marinimicrobia;Uncultured_Family_2Marinimicrobia;Uncultured_Family_3Marinimicrobia;Uncultured_Family_4Methylophilales;MethylophilaceaeNitrospinales;NitrospinaceaeOceanospirillales;SUP05Oceanospirillales;UnclassifiedSAR11;Surface_1SAR324;Uncultured_FamilyDEC15.200mNOV15.200mOCT15.200mSEP15.200mAUG15.200mJUL15.200mJUN15.200mMAY15.200mAPR15.200mMAR15.200mFEB15.200mJAN15.200mE4DEC15.150mNOV15.150mOCT15.150mSEP15.150mAUG15.150mJUL15.150mJUN15.150mMAY15.150mAPR15.150mMAR15.150mFEB15.150mJAN15.150mE3DEC15.135mNOV15.135mOCT15.135mSEP15.135mAUG15.135mJUL15.135mJUN15.135mMAY15.135mAPR15.135mMAR15.135mFEB15.135mJAN15.135mE2DEC15.120mNOV15.120mOCT15.120mSEP15.120mAUG15.120mJUL15.120mJUN15.120mMAY15.120mAPR15.120mMAR15.120mFEB15.120mJAN15.120mE1DEC15.100mNOV15.100mOCT15.100mSEP15.100mAUG15.100mJUL15.100mJUN15.100mMAY15.100mAPR15.100mMAR15.100mFEB15.100mJAN15.100m0 20 40 60 80Sequence_proportionSample100m120m135m150m200mJDFMAMJJASONJDFMAMJJASONJDFMAMJJASONJDFMAM JJASONJDFMAM JJASONFigure C.3: Relative abundance of 15 most abundant taxa for the deeper waters of SI in 2015 (100, 120, 135,150, 200m). These OTUs were the most abundant on s found in average throughout the samples.152C.3 Model of NO–2 competition between anammox and completedenitrificationThis model describes how anammox bacteria and denitrifiers compete against each other in response tovarying metabolite input fluxes. The reaction rates are described using metabolite concentrations ([C]),cell abundance (X), as well as kinetic parameters (Km and Vmax) determined for microbial communitiesin Saanich Inlet (SI) or found in the literature when missing. In turn, metabolite concentrations (Eq. C.1)vary depending on the metabolite input fluxes (Rin) as well as the rates of metabolite consuming reactionsdescribed (Eqs. C.2 to C.7). Here, we choose 3 metabolic reactions: NO–3 reduction to NO–2 (NO3R, Eqs.C.2 and C.3), complete denitrification (DEN, NO–2 to N2, Eqs. C.4 and C.5) as well as anammox (AN,Eqs. C.6 and C.7). In this model, DEN and AN both compete for the NO–2 produced through NO3R.We also consider both NO3R and DEN to use sulphide (HS– ) as an electron donor as implied frommetagenomic information [33, 187] and our process rate measurements. The stoichiometric relationshipsfor the 3 metabolic reactions and their metabolites are shown in Eqs. C.2 C.5 C.7 as well as in table C.2 andconsidered in the model on a per mole of electron donor (ED) basis.d[C]dt= RIN − Rxn (C.1)NO–3 reduction to NO–2 (NO3R): 4NO−3 +HS− ⇒ 4NO−2 + SO2−4 +H+ (C.2)NO3R = Vmax,NO3R ∗ XNO3R ∗ [HS−][HS−] +Km,NO3R,HS∗ [NO−3 ][NO−3 ] +Km,NO3R,NO3(C.3)DEN: 8/3NO−2 +HS− + 5/3H+ ⇒ 4/3N2 + SO2−4 + 4/3H2O (C.4)DEN = Vmax,DEN ∗ XDEN ∗ [NO−2 ][NO−2 ] +Km,DEN,NO2∗ [HS−][HS−] +Km,DEN,HS(C.5)AN: NH+4 +NO−2 ⇒ N2 + 2H2O (C.6)AN = Vmax,AN ∗ XAN ∗ [NH+4 ][NH+4 ] +Km,AN,NH4∗ [NO−2 ][NO−2 ] +Km,AN,NO2(C.7)The change in metabolite concentrations is followed in our model for NH+4 , NO–2 , NO–3 and HS– , anddepends on the input fluxes (Rin), as well as the production and consumption of the metabolites throughTable C.2: Stoichiometric coefficients for the metabolites considered in the model and their respectivereactions.M\RXN AN NO3R DEN NH4+ -1 0 0 NO2- -1 +4 -8/3 NO3- 0 -4 0 HS- 0 -1 -1 !153the reaction rates AN, DEN and/or NO3R. For example, NH+4 concentrations only depend on the input ofNH+4 to the system and the consumption of NH+4 through AN. NO–2 concentrations, on the other hand,depend on the input fluxes of NO–2 and the production of NO–2 through NO3R, as well as on consumptionthrough AN and DEN. A stoichiometric coefficient is added in front of the reaction rate when needed asthe reaction rates are described in [mol ED L– 1 d– 1] (Eqs. C.8 to C.11).d[NH+4 ]dt= RNH4I N − AN (C.8)d[NO−2 ]dt= RNO2I N − AN − (8/3)DEN + 4NO3R (C.9)d[NO−3 ]dt= RNO3I N − 4NO3R + 4NO3R (C.10)d[HS−]dt= RHSI N − NO3R− DEN (C.11)Finally, the change in cell abundance (cell L– 1), described in Eqs. ??, depends on the respective reactionrates (AN, DEN or NO3R), the biomass yield (Y [cells (moles of ED)– 1]), and the death rate ( [d– 1]). It isimportant to note that we differentiate here between the growth of cells through NO3R and DEN as theenergy yield can be different.d[XAN ]dt= AN ∗YAN − XAN ∗ λAN (C.12)d[XDEN ]dt= DEN ∗YDEN − XDEN ∗ λDEN (C.13)d[XNO3R]dt= NO3R ∗YNO3R − XNO3R ∗ λNO3R (C.14)The biomass yield (Y) can be calculated according to Eq. C.15 and is based on the coefficient ofthe reaction (γ), the free gibbs energy (∆Gr), and the quotient of the concentrations of the chemicalspecies involved in the reaction (Qr, Eq. C.16). However, for the sake of simplifying this model and foreasier comparison, we calculated this biomass yield and concluded that it ranges on the order of 10 13(cells mole ED)– 1. Changes of the biomass yield were tested below in this supplement.Y = (2.08 ∗ γ− 0.0211 ∗ ∆Gr) ∗ 1weight− o f − 1− cell (Roden and Jin, 2011) (C.15)With: ∆Gr = ∆G◦r + RT ∗ ln(Qr) (C.16)We solved the multiple differential equations (Eqs. C.8 to C.14) numerically using Eulers techniquewith a step of 0.01 days (Eq. C.17) and build the equations in Matlab for the different simulations. Weconsider the model to have reached steady-state when the changes in metabolite concentrations, reactionrates and cell concentrations reaches an asymptote or a constant increase or decrease (see general remarks).Ifdydt= f (y, t)Then, yt = yt−1 + dt ∗ f (yt−1, tt−1) (C.17)This model was created to test whether we could reproduce the stagnation phenotype with low N2production rates and the renewal phenotype with higher N2 production rates. This was accomplished byvarying the kinetic parameters of the underlying phenotypes and the input nutrient fluxes. While the mainpurpose of the model is highlighted developed in the main text, we also conducted a stability analysis of themodel here in the supplement. We then looked at what controls the partitioning of N2 production between154anammox and complete denitrification in a stagnation phenotype setting as the partitioning appear to varythroughout the year during this phenotype.C.3.1 General remarks• The code was developed in MATLAB (version R2015b) and is available online and in the supple-ment.below.• All the rates reported for the model are in moles N L– 1 d– 1.• Nutrient concentration changes are not constrained by any output fluxes that would naturally occurin the environment such as advection and diffusion out of the depth studied. Therefore, if unusedby the microbial reaction described in the mass balance of this model, nutrient can accumulate tounrealistic concentrations for SI. Also, because the model is dynamical, if a cell population wasnot sustained by the conditions set at the beginning of the stimulation, the cell population woulddecrease until reaching 0. Therefore, in these conditions, a constant development is reached but noso-called steady-state.• Initial concentrations influenced how fast the model would get to a constant development. We thuschose cell abundance and nutrient concentrations that are closer to the end of the simulation.• The figures for the model are constituted of 3 panels (figure C.4 to C.8). The first panel shows theevolution of the different substrates over the time of the simulation studied in the model in molesL– 1 (NH+4 , NO–3 , NO–2 , HS– ). The second panel shows the cell abundance of the three microbialpopulations studied in the model in cell L– 1 (NO3R, DEN, and AN). The third panel shows the ratesof NO–3 reduction to NO–2 (NO3R), anammox (AN) and complete denitrification (DEN) in moles NL– 1 d– 1.C.3.2 Stability of the modelThe stability of the model was tested for a confined range of kinetic parameters, as these constants couldbe constrained through incubation experiments or taken out of the existing literature (i.e. anammox kmconstants were taken out of [109]). Therefore, we only varied the input nutrient fluxes to find the workinglimits of the model. The parameters used in the stability analysis can be found in table C.3. The lowestlimit was determined by nutrient input fluxes that could not sustain any of the 3 microbial populationsdescribed in this model. Thus, we show the lower input nutrient fluxes to be of 10 – 11 moles of substrateL– 1 d– 1 (figure C.4). From the start of this stimulation (figure C.4), these microbial populations decrease toconcentrations as low as 100 cells L– 1 after running the model for 10 000 days. On the other side of therange, input nutrient fluxes of 5 · 10 – 5 moles of substrate L– 1 d– 1 rendered the model to be unstable, withno solution found for the set conditions (table C.3). Thus, with input nutrient fluxes ranging between 10 – 11and 5 · 10 – 5 moles of substrate L– 1 d– 1, it is possible to obtain a constant development from the model.With this model, we could test whether we could reproduce two environmental phenotypes in SI (lowN2 production during stagnant phase of SI and high N2 production Test 1 produces a first set of conditionsthat shows dominance of anammox in the system (figure C.4,tables C.3 and C.4). Among the modelparameters, initial metabolite concentrations are low and metabolite input fluxes are stoichiometricallybalanced aside from the NO–2 input flux, which is 0. This was set based on the fact that NO–2 concentrationsare relatively low at all times in SI and therefore, input fluxes from below and above depths should benegligible. Most of the kinetic parameters were chosen from the literature (see table C.3) and some weredetermined here.155Table C.3: Kinetic parameters used in the stability analysis of the model.Parameter Value Unit Reference km_NH4_AN 3 10-6 Moles L-1 (Thamdrup and Dalsgaard, 2002) km_NO2_AN 0.45 10-6 Moles L-1 (Thamdrup and Dalsgaard, 2002) km_NO2_DEN 5 10-6 Moles L-1 This paper and (Jensen et al., 2009) km_NO3_NO3R 5 10-6 Moles L-1 This paper and (Jensen et al., 2009) km_HS_DEN 12 10-6 Moles L-1 This paper km_HS_NO3R 12 10-6 Moles L-1 This paper Vmax_AN, Vmax_DEN, Vmax_NO3R 2 10-14 Moles cell-1 d-1 This parameter is based on cell specific rate for anammox and used at the same value for DEN and NO3R for comparison (Jensen et al., 2008) YAN 1.5 1013 Moles ED cell-1 Calculated from (Roden and Jin, 2011) and used as a reference. YDEN, YNO3R 1.5 1014 Moles ED cell-1 Biomass yield for denitrifiers appear to be higher as they constitute a larger part of the microbial population with similar rates as anammox dAN, dDEN, dNO3R 0.001 d-1 (Louca et al., 2016; Whitman et al., 1998) !0 100 200 300days10-910-810-7Nutrients(molesL-1)NH4+NO3-HS-NO2-0 100 200 300days104105Cell abundance (cells L-1)ANNO3RDEN0 100 200 300days10-1710-1610-1510-1410-1310-12Reactionrate(molesL-1 d-1)ANNO3RDENFigure C.4: Lower limit of the stability of the model for the set conditions found in table C.3.156C.3.3 Stagnation phenotype: partitioning of N2 production through anammox andcomplete denitrificationSimilar rates of anammox and denitrification were found in SI during most of the year, with a partitioningof N2 production between anammox and denitrification close to 50% if the highest denitrification ratesare excluded (during renewal phenotype). The alternation between the dominance of anammox ordenitrification plausibly lies in slight changes in the capacity of the microbial populations to use thesubstrates (i.e. changes in kinetic parameters). Although input nutrient fluxes are also likely to influencethe partitioning of N2 production by limiting one process over the other, the inherent capacity of themicrobial populations to process substrates more efficiently than others (based on their respective kineticparameters) is likely to be the main determinant on the dominance of one process over the other.Anammox was found to dominate N2 production under the set conditions for the model found in tableC.3 and with the following nutrient input fluxes: RNO3 and RNO2 = 5 · 10 – 9 moles L– 1 d– 1, and RNH4 andRHS = 5 · 10 – 8 moles L– 1 d– 1 (figure C.5). These nutrient input fluxes represent stagnation periods in SI,as described in the main text, with high NH+4 and HS– fluxes coming from the sediments, and low NO–3and NO–2 present in the anoxic water column. Indeed, the range of the rates of the described processeshere correspond to those found in SI under the stagnation period. The high affinity of anammox for NO–2likely gave the advantage to anammox bacteria to process lower NO–2 concentrations, despite completedenitrifiers possessing a higher biomass yield (table C.3). When the NO–2 dependency constant waslowered for complete denitrifiers below 1.5µM (km,NO2,DEN), table C.4), the complete denitrifiers populationappear to grow faster (figure C.6). After a stabilization period of 100 days in the simulation, the rates ofdenitrification were shown to be higher than those of anammox. Similarly, when the Vmax or the biomassyield for complete denitrifiers was increased (Vmax ≥ 6 · 10 – 14 moles L– 1 d– 1 and YDEN ≥ 7 · 10 – 14 cell(moles ED)– 1, table C.4), rates of denitrification also ended up higher than the rates of anammox (figures??) after a stabilization period of 100 days as well. Thus, within a same order of magnitude, variation inthe kinetic parameters can highly influence the outcome of the partitioning of N2 production betweenanammox and denitrification. It is thus highly relevant to constrain these parameters as well as identifyingspecific kinetic parameters for specific phenotypes in order to refine the modeling of the N-cycle.Table C.4: Kinetic parameters for complete denitrifiers.Parameter Value Simulation where tested Unit km_NO2_DEN <1.5 10-6 Figure S6 Moles L-1 Vmax_DEN >6 10-14 Figure S7 Moles cell-1 d-1 YDEN >7 1014 Figure S8 Moles ED cell-1 !1570 100 200 300days10-910-810-710-610-510-4Nutrients(molesL-1)NH4+NO3-HS-NO2-0 100 200 300days123456789Cell abundance (cells L-1)#107ANNO3RDEN0 100 200 300days10-1010-910-810-7Reactionrate(molesL-1 d-1)ANNO3RDENFigure C.5: Simulation of the model for a stagnation phenotype that shows anammox dominating N2production. See table C.3 for kinetic parameters used here.1580 100 200 300days10-910-810-710-610-510-4Nutrients (moles L-1)NH4+NO3-HS-NO2-0 100 200 300days107108109Cell abundance (cells L-1)ANNO3RDEN0 100 200 300days10-910-810-7Reaction rate (moles L-1 d-1)ANNO3RDENFigure C.6: Decrease of km,NO2 (see table C.4) for complete denitrification shows rates of denitrificationdominating over anammox after 100 days.1590 100 200 300days10-910-810-710-610-510-4Nutrients (moles L-1)NH4+NO3-HS-NO2-0 100 200 300days107108109Cell abundance (cells L-1)ANNO3RDEN0 100 200 300days10-910-810-7Reaction rate (moles L-1 d-1)ANNO3RDENFigure C.7: Increase of Vmax,DEN (See table C.4) for complete denitrification shows rates of denitrificationdominating over anammox after 100 days.1600 100 200 300days10-910-810-710-610-510-4Nutrients (moles L-1)NH4+NO3-HS-NO2-0 100 200 300days107108109Cell abundance (cells L-1)ANNO3RDEN0 100 200 300days10-910-810-7Reaction rate (moles L-1 d-1)ANNO3RDENFigure C.8: Increase of YDEN (see table C.4) for complete denitrification shows rates of denitrificationdominating over anammox after 100 days.C.3.4 Matlab code for model NO–2 competition1 clc2 clear all34 %parameters used5 %67 %nutrient fluxes in (moles L-1 d-1)89 RNH4 IN=5∗10ˆ-8;10 RNO3 IN=5∗10ˆ-6;11 RHS IN=5∗10ˆ-7;12 RNO2 IN=5∗10ˆ-7;131415 %t0 input in ODE, 7 variables - nutrient concentrations (moles L-1) and16 %cells concentrations (cells L-1)17 %x(1)=NH4 x(2)=NO3 x(3)=HS x(4)=NO2 x(5)=X AN x(6)=X DEN x(7)=X NO3R1819 x0=[10ˆ-6 10ˆ-6 10ˆ-6 10ˆ-9 10ˆ7 10ˆ7 10ˆ7];20 %time range21 tspan=0:0.01:10000;2223 %death rate (d-1)2416125 d AN=0.001;26 d NO3R=0.001;27 d DEN=0.001;2829 %Km parameters (moles L-1)3031 km NH4 AN=0.000003;32 km NO2 AN=0.00000045;33 km HS DEN=0.00001;34 km NO2 DEN=0.000005;35 km HS NO3R=0.00001;36 km NO3 NO3R=0.000005;3738 %Vmax (moles cells-1 d-1)3940 vm AN=2∗10ˆ-14;41 vm DEN=2∗10ˆ-13;42 vm NO3R=2∗10ˆ-14;4344 %biomass yield (cell mol Electron Donor-1)4546 Y AN=5∗10ˆ13;47 Y DEN=1.5∗10ˆ15;48 Y NO3R=5∗10ˆ14;4950515253 %f = system of equations to elucidate the following variables over time:54 %x(1) = NH455 %x(2) = NO356 %x(3) = HS57 %x(4) = NO258 %x(5) = AN cells59 %x(6) = DEN cells60 %x(7) = NO3R cells616263 f= @(t,x) [RNH4 IN-(vm AN∗x(5)∗x(1)∗x(4)/((x(1)+km NH4 AN)∗(x(4)+km NO2 AN)));64 RNO3 IN-4∗(vm NO3R∗x(7)∗x(2)∗x(3)/((x(2)+km NO3 NO3R)∗(x(3)+km HS NO3R)));65 RHS IN-(vm NO3R∗x(7)∗x(2)∗x(3)/((x(2)+km NO3 NO3R)∗(x(3)+km HS NO3R)))66 -(vm DEN∗x(6)∗x(4)∗x(3)/((x(4)+km NO2 DEN)∗(x(3)+km HS DEN)));67 RNO2 IN-(vm AN∗x(5)∗x(1)∗x(4)/((x(1)+km NH4 AN)∗(x(4)+km NO2 AN)))68 -(8/3)∗(vm DEN∗x(6)∗x(4)∗x(3)/((x(4)+km NO2 DEN)∗(x(3)+km HS DEN)))69 +4∗(vm NO3R∗x(7)∗x(2)∗x(3)/((x(2)+km NO3 NO3R)∗(x(3)+km HS NO3R)));7071 ((vm AN∗x(5)∗x(1)∗x(4)/((x(1)+km NH4 AN)∗(x(4)+km NO2 AN)))∗Y AN)-(x(5)∗d AN);72 ((vm DEN∗x(6)∗x(4)∗x(3)/((x(4)+km NO2 DEN)∗(x(3)+km HS DEN)))∗Y DEN)-(x(6)∗d DEN);73 ((vm NO3R∗x(7)∗x(2)∗x(3)/((x(2)+km NO3 NO3R)∗(x(3)+km HS NO3R)))∗Y NO3R)-(x(7)∗d NO3R);];747576 %solution for the 7 equations/variables with Euler s technique77 options = odeset( NonNegative ,1);78 [t,x] = ode15s(f,tspan,x0,options);798081 if(x<0)16282 x=0;83 end8485 %moles ED L-1 d-1, is in moles N L-1 d-1 already because it is anammox!86 AN=vm AN.∗x(:,5).∗x(:,1).∗x(:,4)./((x(:,1)+km NH4 AN).∗(x(:,4)+km NO2 AN));8788 %moles ED L-1 d-189 %DEN=vm DEN.∗x(:,6).∗x(:,4).∗x(:,3)./((x(:,4)+km NO2 DEN).∗(x(:,3)+km HS DEN));9091 %if needs to be in moles N L-1 d-192 DEN=(8/3)∗vm DEN.∗x(:,6).∗x(:,4).∗x(:,3)./((x(:,4)+km NO2 DEN).∗(x(:,3)+km HS DEN));939495 %moles ED L-1 d-196 %NO3R=vm NO3R.∗x(:,7).∗x(:,2).∗x(:,3)./((x(:,2)+km NO3 NO3R).∗(x(:,3)+km HS NO3R));9798 %if needs to be in moles L-1 d-199 NO3R=4∗vm NO3R.∗x(:,7).∗x(:,2).∗x(:,3)./((x(:,2)+km NO3 NO3R).∗(x(:,3)+km HS NO3R));100101102103 %plotting of solutions104 figure105 set(gcf, Position , [100, 100, 800, 400])106 subplot(1,3,1);107108 semilogy(t,x(:,1),t,x(:,2),t,x(:,3),t,x(:,4), LineWidth ,1.5);109 xlabel( days );110 ylabel( Nutrients (moles Lˆ{-1}) );111 legend( NH {4}ˆ{+} , NO {3}ˆ{-} , HSˆ{-} , NO {2}ˆ{-} );112 %the following line can be commented so the figure shows time for 10000 days113 %xlim([0 1000])114115116 subplot(1,3,2);117118 semilogy(t,x(:,5),t,x(:,7),t,x(:,6), LineWidth ,1.5);119 xlabel( days );120 ylabel( Cell abundance (cells Lˆ{-1}) );121 legend( AN , NO3R , DEN );122 %the following line can be commented so the figure shows time for 10000 days123 %xlim([0 1000])124125126 %plot reaction rates127 subplot(1,3,3);128129 semilogy(t,AN,t,NO3R,t,DEN, LineWidth ,1.5);130 legend( AN , NO3R , DEN );131 xlabel( days );132 ylabel( Reaction rate (moles Lˆ{-1} dˆ{-1}) );133 %the following line can be commented so the figure shows time for 10000 days134 %xlim([0 1000])135136137138 figure163139 set(gcf, Position , [100, 100, 800, 400])140 subplot(1,3,1);141142 semilogy(t,x(:,1),t,x(:,2),t,x(:,3),t,x(:,4), LineWidth ,1.5);143 xlabel( days );144 ylabel( Nutrients (moles Lˆ{-1}) );145 legend( NH {4}ˆ{+} , NO {3}ˆ{-} , HSˆ{-} , NO {2}ˆ{-} );146 %the following line can be commented so the figure shows time for 10000 days147 xlim([0 60])148149150 subplot(1,3,2);151152 semilogy(t,x(:,5),t,x(:,7),t,x(:,6), LineWidth ,1.5);153 xlabel( days );154 ylabel( Cell abundance (cells Lˆ{-1}) );155 legend( AN , NO3R , DEN );156 %the following line can be commented so the figure shows time for 10000 days157 xlim([0 60])158159160 %plot reaction rates161 subplot(1,3,3);162163 semilogy(t,AN,t,NO3R,t,DEN, LineWidth ,1.5);164 legend( AN , NO3R , DEN );165 xlabel( days );166 ylabel( Reaction rate (moles Lˆ{-1} dˆ{-1}) );167 %the following line can be commented so the figure shows time for 10000 days168 xlim([0 60])169 %print( 5by3DimensionsFigure , -dpdf , -r0 )164Appendix DChapter 4: supplemental materialD.1 Geochemical profiles in Saanich Inlet (SI)165a.b.c.d.e.J     F     M    A    M    J     J     A     S    O    N     D    J     F     M    A    M    J     J     A     S    O    N     D2015                                                                          2016Figure D.1: Nutrient concentrations in SI for the years 2015-2016 at station S3. (a) O2 concentrations (µM), (b)NO–3 concentrations (µM), (c) NO–2 concentrations (µM), (d) NH+4 concentrations (µM), (e) HS– concentra-tions (µM).166D.2 Potential and scaled rates of anaerobic N-metabolisms167!"#$% &'($% )"*$% +,*$% )"-$% !.#$% !./$% +.0$% 1',$% 234$% 567$% 8'3$% !"#$9 &'($9 )"*$9 +,*$9 )"-$9 !.#$9 !./$9 +.0$9 1',$9 234$9 567$9 8'3$9$::$;:$<:$9:$=:;::8',4>?@AB:C%$$:%:$::%::$:::85D+?@#)?>*E$B!"#$% &'($% )"*$% +,*$% )"-$% !.#$% !./$% +.0$% 1',$% 234$% 567$% 8'3$% !"#$9 &'($9 )"*$9 +,*$9 )"-$9 !.#$9 !./$9 +.0$9 1',$9 234$9 567$9 8'3$9$::$;:$<:$9:$=:;::8',4>?@AB:C%$%$:$%+#"AA6D*"4'E#)>*F$!"#$% &'($% )"*$% +,*$% )"-$% !.#$% !./$% +.0$% 1',$% 234$% 567$% 8'3$% !"#$9 &'($9 )"*$9 +,*$9 )"-$9 !.#$9 !./$9 +.0$9 1',$9 234$9 567$9 8'3$9$::$;:$<:$9:$=:;::8',4>?@AB:C%$%$:$%+#"AA6D?*"4'E?#)?>*F$!"#$% &'($% )"*$% +,*$% )"-$% !.#$% !./$% +.0$% 1',$% 234$% 567$% 8'3$% !"#$9 &'($9 )"*$9 +,*$9 )"-$9 !.#$9 !./$9 +.0$9 1',$9 234$9 567$9 8'3$9$::$;:$<:$9:$=:;::8',4>?@AB:C%$%$:%:$::;::8'#4D*DED3"4D6#?@#)?>*F$Ba.b.c.d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e.f.g.h.15NO3-15NO3-15NO3-15NH4+ +15NO3-J     F     M    A    M    J     J     A     S    O    N     D    J     F     M    A    M    J     J     A     S    O    N     D2015                                                                          2016Figure D.2: Potential and scaled rates of denitrification, anammox and DNRA. (a) rates of anammox based on the accumulation of 29N2 from 15NH+4+ 14NO–3 incubations (nM hr– 1), (b) rates of anammox based on the accumulation of 29N2 from 15NO–3 incubations (nM hr– 1), (c) rates ofdenitrification based on the accumulation of 30N2 from 15NO–3 incubations (nM hr– 1), (d) rates of denitrification based on the accumulation of30N2 from 15NO–3 incubations (nM hr– 1), (f-h) scaled rates168To produce depth-integrated rates of denitrification, anammox and DNRA (figure 1 in main text), thevolumetric potential rates were scaled to in situ levels of NO–3 for denitrification and DNRA or NH4+for anammox, based on their half-saturation constant (km) found in Michiels et al. (2018, submitted) foranammox and denitrification and km=5µM for DNRA (data not shown). This was done according to Eq.D.1:Rscaled = Rpot ∗ [S](km + [S])(D.1)The scaled rates of denitrification, anammox and DNRA were then integrated over the anoxic watercolumn ( 90m to 200m) and is summarized in table D.1.Table D.1: Partitioning of N-loss and N-retention through NOx reduction in moles m– 2 d– 1.Year Month DNRA Denitrification Anammox Total N-loss Total NOx reduction % N-retention 2015 January / 4E-05±1E-05 1.1E-03±2E-04 1.20E-03 / / 2015 March / 2.2E-03±1E-04 1.8E-03±1E-04 6.27E-03 / / 2015 April 4.88E-05±0 2.1E-05±6E-06 6.9E-04±7E-05 7.28E-04 7.76E-04 6.29 2015 May / 2.0E-03±3E-04 2.6E-03±1E-04 6.49E-03 / / 2015 June 1.17E-02±2E-05 5E-04±1E-04 10.0E-04±7E-05 2.04E-03 1.37E-02 85.15 2015 July / 4.4E-03±5E-04 1.6E-03±2E-04 1.04E-02 / / 2015 August 3.23E-01±3E-03 1.4E-02±2E-03 1.5E-04±3E-05 2.90E-02 3.52E-01 91.77 2015 September 1.85E-02±5E-04 1.6E-03±2E-04 1.51E-03±9E-05 4.78E-03 2.33E-02 79.46 2015 October 3.09E-02±2E-04 8E-04±3E-04 1.5E-03±4E-04 3.02E-03 3.39E-02 91.09 2015 November 2.13E-01±4E-05 1.8E-04±2E-05 3.4E-03±3E-04 3.73E-03 2.17E-01 98.28 2015 December / 2.5E-03±6E-04 1.2E-03±2E-04 6.21E-03 / / 2016 February 1.67E-01±4E-03 5.6E-03±1E-04 8E-04±1E-04 1.20E-02 1.79E-01 93.26 2016 March 9.4E-03±1E-04 5.5E-02±6E-03 9.6E-04±6E-05 1.10E-01 1.19E-01 7.89 2016 April 2.71E-02±0 7.8E-03±3E-04 1.74E-04±4E-06 1.59E-02 4.29E-02 63.07 2016 May 1.33E-01±8E-04 3.0E-02±1E-03 1.20E-03±9E-05 6.07E-02 1.93E-01 68.58 2016 June 1.68E-02±1E-04 2.19E-02±3E-04 2.59E-03±0 4.64E-02 6.32E-02 26.53 2016 July 4.36E-03±0 4.13E-03±8E-04 4.0E-06±2E-07 8.27E-03 1.26E-02 34.53 2016 August 2.33E-03±3E-05 7.6E-02±3E-03 1.6E-03±1E-04 1.53E-01 1.55E-01 1.50 2016 September 2.11E-02±3E-04 1.3E-02±2E-03 4.0E-03±2E-04 2.93E-02 5.05E-02 41.86 2016 October 1.38E±0 6.49E-03±0 1.62E-03±0 1.46E-02 1.39E+00 98.95 2016 November 1.82E-01±1E-04 1.9E-03±6E-04 5.4E-03±2E-04 9.20E-03 1.92E-01 95.20 2016 December 1.22E-03±2E-06 1.7E-03±5E-04 8.3E-03±3E-04 1.17E-02 1.30E-02 9.44  D.3 Taxonomy and functional gene abundances1690 25 50 75Abundance100m120m135m150m200mAprilAugustSeptemberOctoberAprilAugustSeptemberOctoberAprilAugustSeptemberOctoberAprilAugustSeptemberOctoberAprilAugustSeptemberOctoberTaxonFigure D.3: Taxonomic composition of microbial communities at the OTU level in SI. Taxonomic composition of SIs microbial communities at the OTUlevel based on Metagenomic data, through EMIRGE analysis (extraction and reconstruction of 16S rRNA gene sequences out of metagenomes)and classified using the latest SILVA database (v132) - relative abundance (higher than 1%) of OTUs present in samples for SI at 5 depths and 4months in 2016.170OCT_16SEPT_16AUG_16APR_16CO_dh CitSyn OGFOxyRBsCOABCtrans dsrA dsrB p_amoA hao hzo hzs napA narG nirS norB norC nosZ nrfA nirAGene2Month2●●●●●●●●●●●●●●●●●●CitSynOGFOxyRBsCOABCtransdsrAdsrBp_amoAhaohzohzsnapAnarGnirSnorBnorCnosZnrfAnirAOCT_16SEPT_16AUG_16APR_16CO_dh CitSyn OGFOxyRBsCOABCtrans dsrA dsrB p_amoA hao hzo hzs napA narG nirS norB norC nosZ nrfA nirAGene2Month20.90.9Gene2●●●●●●●●●●●●●●●●●●●CO_dhCitSynOGFOxyRBsCOABCtransdsrAdsrBp_amoAhaohzohzsnapAnarGnirSnorBnorCnosZnrfAnirAOCT_16SEPT_16AUG_16APR_16CO_dh CitSyn OGFOxyRBsCOABCtrans dsrA dsrB p_amoA hao hzo hzs napA narG nirS norB norC nosZ nrfA nirAGene2Month2●●●●50100500100040000.90.9Gene2●●●●●●●●●●●●●●●●●●●CO_dhCitSynOGFOxyRBsCOABCtransdsrAdsrBp_amoAhaohzohzsnapAnarGnirSnorBnorCnosZnrfAnirAOCT_16SEPT_16AUG_16APR_16CO_dh CitSyn OGFOxyRBsCOABCtrans dsrA dsrB p_amoA hao hzo hzs napA narG nirS norB norC nosZ nrfA nirAGene2Month2RPKM●●●●●050100500100040000.90.9Gene2●●●●●●●●●●●●●●●●●●●CO_dhCitSynOGFOxyRBsCOABCtransdsrAdsrBp_amoAhaohzohzsnapAnarGnirSnorBnorCnosZnrfAnirAOCT_16SEPT_16AUG_16APR_16CO_dh CitSyn OGFOxyRBsCOABCtrans dsrA dsrB p_amoA hao hzo hzs napA narG nirS norB norC nosZ nrfA nirAGene2Month2RPKM●●●●●050100500100040000.90.9Gene2●●●●●●●●●●●●●CO_dhCitSynOGFOxyRBsCOABCtransdsrAdsrBp_amoAhaohzohzsnapAnarGOCT_16SEPT_16AUG_16APR_16CO_dh CitSyn OGFOxyRBsCOABCtrans dsrA dsrB p_amoA hao hzo hzs napA narG nirS norB norC nosZ nrfA nirAGene2Month2RPKM●●●●●050100500100040000.90.9Gene2●●●●●●●CO_dhCitSynOGFOxyRBsCOABCtransdsrAdsrBOCT_16SEPT_16AUG_16APR_16TotalGene2Month2RPKM●●●04000140000.90.9OCT_16SEPT_16AUG_16APR_16TotalGene2Month2Gene2● TotalRPKM●●●04000140000.90.9OCT_16SEPT_16AUG_16APR_16TotalGene2Month2Gene2● TotalRPKM●●●04000140000.90.9OCT_16SEPT_16AUG_16APR_16TotalGene2Month2Gene2● TotalRPKM●●●04000140000.90.9OCT_16SEPT_16AUG_16APR_16TotalGene2Month2Gene2● TotalRPKM●●●04000140000.90.9OCT_16SEPT_16AUG_16APR_16TotalGene2Month2Gene2● TotalRPKM●●●04000140000.910m100m120m135m150m200mAprilAugustSeptemberOctoberOCT_16SEPT_16AUG_16APR_16CO_dh CitSyn OGFOxyRBsCOABCtrans dsrA dsrB p_amoA hao hzo hzs napA narG nirS norB norC nosZ nrfA nirAGene2Month2●●●●●●norCnosZnrfAOGFOxyp_amoARBsCORPKM●●●●●●050100500100040000.70.7OCT_16SEPT_16AUG_16APR_16CO_dh CitSyn OGFOxyRBsCOABCtrans dsrA dsrB p_amoA hao hzo hzs napA narG nirS norB norC nosZ nrfA nirAGene2Month2●●●●●●●●●●●●hzsnapAnarGnirAnirSnorBnorCnosZnrfAOGFOxyp_amoARBsCORPKM●●●●●●050100500100040000.70.7OCT_16SEPT_16AUG_16APR_16CO_dh CitSyn OGFOxyRBsCOABCtrans dsrA dsrB p_amoA hao hzo hzs napA narG nirS norB norC nosZ nrfA nirAGene2Month2●●●●●●●●●●●●●●●●●●CitSynCO_dhdsrAdsrBhaohzohzsnapAnarGnirAnirSnorBnorCnosZnrfAOGFOxyp_amoARBsCORPKM●●●●●●050100500100040000.70.7OCT_16SEPT_16AUG_16APR_16CO_dh CitSyn OGFOxyRBsCOABCtrans dsrA dsrB p_amoA hao hzo hzs napA narG nirS norB norC nosZ nrfA nirAGene2Month2Gene●●●●●●●●●●●●●●●●●●●ABCtransCitSynCO_dhdsrAdsrBhaohzohzsnapAnarGnirAnirSnorBnorCnosZnrfAOGFOxyp_amoARBsCORPKM●●●●●●050100500100040000.70.7OCT_16SEPT_16AUG_16APR_16CO_dh CitSyn OGFOxyRBsCOABCtrans dsrA dsrB p_amoA hao hzo hzs napA narG nirS norB norC nosZ nrfA nirAGene2Month2Gene●●●●●●●●●●●●●●●●●●●ABCtransCitSynCO_dhdsrAdsrBhaohzohzsnapAnarGnirAnirSnorBnorCnosZnrfAOGFOxyp_amoARBsCORPKM●●050100500OCT_16SEPT_16AUG_16APR_16CO_dh CitSyn OGFOxyRBsCOABCtrans dsrA dsrB p_amoA hao hzo hzs napA narG nirS norB norC nosZ nrfA nirAGene2Month2Gene●●●●●●●●●●●●●●●●●●●ABCtransCitSynCO_dhdsrAdsrBhaohzohzsnapAnarGnirAnirSnorBnorCnosZnrfAOGFOxyp_amoARBsCOAprilAugustSeptemberOctoberAprilAugustSeptemberOctoberAprilAugustSeptemberOctoberAprilAugustSeptemberOctoberAprilAugustSeptemberOctoberrAcCoA   3HP-    rTCA     CBB       ABCt     dsrA      dsrB       pmo/      hao       hzo       hzs       napA     narG      nirS       norB     norC      nosZ      nrfA       nirA     Total RPKM     4HB                                                                           OCT_16SEPT_16AUG_16APR_16CO_dh CitSyn OGFOxyRBsCOABCtrans dsrA dsrB p_amoA hao hzo hzs napA narG nirS norB norC nosZ nrfA nirAGene2Month2RPKM●●●●●●050100500100040000.90.9Gene2●●●●●●●●●●●●●●●●●●●CO_dhCitSynOGFOxyRBsCOABCtransdsrAdsrBp_amoAhaohzohzsnapAnarGnirSnorBnorCnosZnrfAnirAO T_16SEPT_16AUG_16APR_16TotalGene2Month2Gene2● TotalRPKM●●●04000140000.90.9Total RPKMFigure D.4: RPKM counts for functional genes. RPKM for genes involved in C-fixation (rAcCoA=CO dehydrogenase, 3HP-4HB= acetyl CoAcarboxylase in archaea, rTCA= 2-oxoglutarate synthase, CBB=Ribulose-1,5-bisphosphate carboxylase/oxygenase), C-degradation (ABCt=ABCtransporter), sulphide oxidation (dsrA and B= reversible dissimilatory sulphate reductase), nitrification (pmo/amoA=ammonia monooxygenase,hao=hydroxylamine oxidoreductase), anammox (hzo=hydrazine dehydrogenase, hzs=hydrazine synthase), NO–3 reduction (napA=periplasmicdissimilatory nitrate reductase, narG=membrane-bound dissimilatory reductase), denitrification (nirS=nitrite reductase, norBC=nitric oxidereductase, nosZ=nitrous oxide reductase), DNRA (nrfA=dissimilatory periplamic cytochrome c nitrite reductase, nirA=assimilatory nitritereductase) and total RPKM for all the ORFs detected in the metagenomic samples.171Table D.2: List of genes and their acronyms used in this paper for the metagenomic analysis.Gene name Abbreviation Process associated Metacyc number CO dehydrogenase rAcCoA C-fixation Acetyl CoA carboxylase in archaea 3HP-4HB C-fixation, 2-oxoglutarate synthase rTCA C-fixation Ribulose-1,5-bisphosphate carboxylase/oxygenase CBB C-fixation ABC transporter ABCt C-degradation K12536, K05648 (KEGG) Reversible dissimilatory sulphate reductase dsrA and B HS- oxidation Ammonia monooxygenase pmo/amoA Nitrification Hydroxylamine oxidoreductase hao Nitrification Hydrazine dehydrogenase hzo Anammox Hydrazine synthase hzs Anammox Periplasmic dissimilatory nitrate reductase napA NO3- reduction, Membrane-bound dissimilatory reductase narG NO3- reduction Nitrite reductase nirS Denitrification Nitric oxide reductase norBC Denitrification Nitrous oxide reductase nosZ Denitrification, Dissimilatory periplamic cytochrome c nitrite reductase nrfA DNRA Assimilatory nitrite reductase nirA DNRA  172D.4 Energy availability and power supplyTable D.3: Examples of free energy yields calculated for 2 months anoxic water column in SI (in kJ moles N – 1).Year Month Depth ∆G_DEN ∆G_AN ∆G_DNRA 2016 9 90 -422.13 -427.65 -404.17 2016 9 100 -420.29 -427.03 -402.80 2016 9 120 -428.78 -417.42 -412.20 2016 9 135 -430.34 -417.97 -413.99 2016 9 150 -427.63 -417.91 -411.01 2016 9 165 -425.28 -433.30 -407.80 2016 9 200 -424.51 -435.57 -404.57 2016 10 90 -421.49 -413.11 -400.99 2016 10 100 -423.08 -436.70 -403.01 2016 10 120 -428.24 -418.04 -411.24 2016 10 135 -429.13 -418.00 -412.58 2016 10 150 -430.51 -439.02 -413.21 2016 10 165 -427.44 -415.54 -409.01 2016 10 200 -420.95 -415.35 -398.23  173D.5 Methods supplementTable D.4: Sampling dates and type of 15N-labeled incubations. Month Exact date of sampling Type of 15N labeled-incubation January 2015 14 January 2015 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM) February 2015 11 February 2015 / March 2015 11 March 2015 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM) April 2015 8 April 2015 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM) May 2015 13 May 2015 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM) June 2015 3 June 2015 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM),  July 2015 8 July 2015 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM) August 2015 12 August 2015 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM) September 2015 9 September 2015 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM) October 2015 22 October 2015 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM) November 2015 18 November 2015 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM) December 2015 9 December 2015 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM) January 2016 13 January 2016 / February 2016 4 February 2016 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM) March 2016 16 March 2016 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM) April 2016 13 April 2016 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM) May 2016 11 May 2016 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM) June 2016 1 June 2016 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM) July 2016 13 July 2016 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM) August 2016 10 August 2016 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM) September 2016 14 September 2016 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM) October 2016 12 October 2016 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM) November 2016 9 November 2016 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM) December 2016 14 December 2016 15NO3- (10µM), 15NH4+ & 14NO3- (10µM&10µM)  174Table D.5: Accession numbers for NCBI raw reads of samples.Sample name Accession number SI_118_April2016_10_MG PRJNA468231 SI_118_April2016_100_MG PRJNA468232 SI_118_April2016_120_MG PRJNA468233 SI_118_April2016_135_MG PRJNA468234 SI_118_April2016_150_MG PRJNA468235 SI_118_April2016_200_MG PRJNA468236 SI_122_August2016_10_MG PRJNA468237 SI_122_August2016_100_MG PRJNA468238 SI_122_ August2016_120_MG PRJNA468239 SI_122_ August2016_135_MG PRJNA468240 SI_122_ August2016_150_MG PRJNA468241 SI_122_ August2016_200_MG PRJNA468242 SI_123_September2016_10_MG PRJNA468243 SI_123_ September2016_100_MG PRJNA468244 SI_123_ September2016_120_MG PRJNA468245 SI_123_ September2016_135_MG PRJNA468246 SI_123_ September2016_150_MG PRJNA468247 SI_123_ September2016_200_MG PRJNA468248 SI_124_October2016_10_MG PRJNA468249 SI_124_ October2016_100_MG PRJNA468250 SI_124_ October2016_120_MG PRJNA468251 SI_124_ October2016_135_MG PRJNA468253 SI_124_ October2016_150_MG PRJNA468254 SI_124_ October2016_200_MG PRJNA468255  175


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items