UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Concentrations and turnover rates of reduced sulfur compounds in polar and sub-polar marine waters :… Asher, Elizabeth Colleen 2015

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

Item Metadata


24-ubc_2015_september_asher_elizabeth.pdf [ 14.88MB ]
JSON: 24-1.0166363.json
JSON-LD: 24-1.0166363-ld.json
RDF/XML (Pretty): 24-1.0166363-rdf.xml
RDF/JSON: 24-1.0166363-rdf.json
Turtle: 24-1.0166363-turtle.txt
N-Triples: 24-1.0166363-rdf-ntriples.txt
Original Record: 24-1.0166363-source.json
Full Text

Full Text

 CONCENTRATIONS AND TURNOVER RATES OF REDUCED SULFUR COMPOUNDS IN POLAR AND SUB-POLAR MARINE WATERS: FIELD APPLICATION OF NOVEL ANALYTICAL AND EXPERIMENTAL TECHNIQUES  by  ELIZABETH COLLEEN ASHER   A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY  in  THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Oceanography)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  July 2015  © Elizabeth Colleen Asher, 2015        ii ABSTRACT Dimethylsulfide (DMS), dimethylsulfoniopropionate  (DMSP) and dimethyl sulfoxide (DMSO) are ubiquitous in surface marine environments.  These reduced sulfur compounds are crucial to the physiological ecology of bacteria and phytoplankton.  DMS also has a role in climate regulation, as a source of aerosols that back-scatter incoming solar radiation.  This thesis aims to characterize the processes driving DMS/P/O accumulation in polar and sub-polar marine waters. Chapters two and three document DMS/P/O concentrations in surface waters of the Subarctic Northeast Pacific, using automated measurement systems.  These studies employed an existing system based on membrane inlet mass spectrometry (MIMS) and a novel automated system for the sequential analysis of DMS/P/O (OSSCAR; see chapter three).  DMS/P/O concentrations demonstrate significant spatial variability over a range of scales in both coastal and open ocean waters, revealing relationships with key oceanographic variables.  Chapter four describes the first application of a recently developed, stable isotope tracer technique using purge and trap capillary inlet mass spectrometry (PT-CIMS) in Antarctic sea-ice.  This chapter documents extremely rapid DMS/P/O turnover in sea-ice brines and demonstrates a previously unrecognized role for DMSO, as well as DMSP, as important sources of DMS in these environments. Chapters five and six use MIMS, PT-CIMS, and OSSCAR in parallel to examine DMS/P/O cycling in the Subarctic Northeast Pacific, and in coastal waters of the Western Antarctic Peninsula (WAP).  Chapter five focuses on the spatial distribution of DMS accumulation and net production in the Subarctic Pacific, while chapter six follows the seasonal changes in DMS/P/O concentrations and its production in the WAP, and highlights short-scale temporal variability of DMS/P/O.  Results demonstrate strong spatial gradients in DMS production and consumption terms (higher values in near-shore waters) in the Subarctic Pacific, and showed that net DMS production predicts DMS accumulation in surface waters.  Over the seasonal cycle in the WAP, zooplankton grazing and DMSP cleavage dominated DMS production, and bacterially mediated DMS consumption controlled the removal of DMS in surface waters.  iii PREFACE Chapter 2.  This material has been published in the journal Marine Chemistry1.  Ancillary thermo-salinograph (TSG), conductivity, temperature, depth (CTD) and nutrient data were provided complimentary of the Institute of Ocean Sciences (IOS).  Membrane inlet mass spectrometer (MIMS)2 measurements of DMS were collected by co-author Anissa Merzouk.  As with all subsequent data chapters all additional research questions, data analysis and writing are my own, with the help of Dr. Tortell. Chapter 3. A version of this chapter was recently accepted for publication in Limnology and Oceanography: Methods3.  Data for ancillary phytoplankton pigments and acoustic measurements were processed by the Institute of Ocean Sciences (IOS) marine technicians, whose contribution is acknowledged. Underway TSG and chlorophyll fluorescence measurements were also provided by IOS.  I collected the samples for phytoplankton pigments, and made all additional measurements with the help of a fellow Ph.D. student, Tereza Jarniková.  The method for the sequential chemical analysis of sulfur compounds is my own, although it was conceived jointly with Dr. John Dacey (Woods Hole Oceanographic Institute) and Philippe Tortell, and utilizes a purge and trap system developed and built by John Dacey.   Chapter 4. This chapter has been published in the journal Geophysical Research Letters Asher et al.4.  Ancillary chlorophyll measurements and logistical support were provided by Zachary Brown, Matt Mills, and Kevin Arrigo from Stanford University.  I made all additional measurements, analyzed all of the resulting data and wrote the paper with significant input from Dr. Tortell.  The stable isotope tracer method was developed by co-author John Dacey5, to which I contributed minor modifications.  Chapter 5.  Rosette, CTD, and TSG data were provided by IOS and the Line P program courtesy of Dr. Ianson.  I made all additional measurements, conducted all shipboard experiments, and analyzed these data.  I wrote the paper with significant input from Dr. Tortell. Chapter 6.  Environmental data (e.g. surface irradiance levels) and logistic field support were provided by the Palmer Station Antarctic Research station.  CTD data, nitrate, chlorophyll a data, ancillary phytoplankton pigment data, zooplankton abundance  iv data, were collected and analyzed with the help of the following students and scientists: Nicole Couto, Dr. Michael Stukel, Johanna Goldmann, Stefanie Strebel, Dr. Sven Kranz, Dr. Jodi Young, Dr. John Dacey, and Dr. Philippe Tortell.  In particular, grazing experiments were conducted with the help of Dr. Michael Stukel, Dr. John Dacey, and DMS/P/O measurements were conducted primarily by myself, but also by Dr. John Dacey and Dr. Philippe Tortell.    v TABLE OF CONTENTS ABSTRACT	  ..................................................................................................................................	  ii	  PREFACE	  ...................................................................................................................................	  iii	  TABLE	  OF	  CONTENTS	  .............................................................................................................	  v	  LIST	  OF	  TABLES	  .......................................................................................................................	  ix	  LIST	  OF	  FIGURES	  ......................................................................................................................	  x	  LIST	  OF	  SYMBOLS	  AND	  ABBREVIATIONS	  .......................................................................	  xii	  ACKNOWLEDGEMENTS	  ........................................................................................................	  xv	  1	   Introduction	  .......................................................................................................................	  1	  1.1	   The	  Role	  of	  DMS	  in	  the	  Global	  Sulfur	  Cycle	  ....................................................................	  1	  1.2	   The	  Marine	  DMS	  Cycle	  ..........................................................................................................	  1	  1.3	   Global	  DMS/P/O	  Distributions	  ..........................................................................................	  3	  1.4	   Emerging	  Methodological	  Approaches	  ...........................................................................	  6	  1.5	   Thesis	  Objectives	  ....................................................................................................................	  7	  1.6	   Thesis	  Overview	  .....................................................................................................................	  8	  2	   Fine-­‐Scale	  Spatial	  and	  Temporal	  Variability	  of	  Surface	  Water	  Dimethylsulfide	  (DMS)	  Concentrations	  and	  Sea-­‐Air	  Fluxes	  in	  the	  Subarctic	  Northeast	  Pacific	  ...................................................................................................................	  14	  2.1	   Introduction	  ..........................................................................................................................	  14	  2.2	   Methods	  ..................................................................................................................................	  16	  2.2.1	   Underway	  DMS	  Measurements	  ...............................................................................................	  16	  2.2.2	   Remote	  Sensing	  Data	  ...................................................................................................................	  17	  2.2.3	   Separation	  of	  Coastal	  and	  Oceanic	  Regimes	  ......................................................................	  18	  2.2.4	   Evaluating	  Global	  Empirical	  Algorithms	  .............................................................................	  19	  2.2.5	   Stepwise	  Multiple	  Regression	  ..................................................................................................	  19	  2.2.6	   Length	  Scales	  of	  Variability	  .......................................................................................................	  20	  2.2.7	   DMS	  at	  Ocean	  Station	  Papa	  .......................................................................................................	  21	  2.3	   Results	  .....................................................................................................................................	  22	  2.3.1	   Spatial	  and	  Seasonal	  Patterns	  in	  Surface	  DMS	  Concentrations	  .................................	  22	  2.3.2	   Length	  Scales	  of	  Spatial	  Variability	  .......................................................................................	  24	  2.3.3	   Sea-­‐Air	  DMS	  Fluxes	  .......................................................................................................................	  24	   vi 2.3.4	   Empirical	  Predictive	  Algorithms	  ............................................................................................	  25	  2.3.5	   Variability	  at	  Ocean	  Station	  Papa	  ...........................................................................................	  26	  2.4	   Discussion	  ..............................................................................................................................	  28	  2.5	   Conclusions	  ...........................................................................................................................	  34	  3	   Measurement	  of	  DMS,	  DMSO,	  and	  DMSP	  in	  Natural	  Waters	  by	  Automated	  Sequential	  Chemical	  Analysis	  ...........................................................................................	  48	  3.1	   Introduction	  ..........................................................................................................................	  48	  3.2	   Materials	  and	  Procedures	  ................................................................................................	  50	  3.2.1	   Overview	  ...........................................................................................................................................	  50	  3.2.2	   Sample	  Collection	  ..........................................................................................................................	  50	  3.2.3	   DMS	  Stripping	  and	  Analysis	  ......................................................................................................	  51	  3.2.4	   Catalytic	  Reduction	  of	  DMSO	  ....................................................................................................	  52	  3.2.5	   DMSP	  Fast	  Hydrolysis	  .................................................................................................................	  53	  3.2.6	   Sample	  Rinsing	  ...............................................................................................................................	  53	  3.2.7	   Automatic	  Calibration	  .................................................................................................................	  54	  3.3	   Assessment	  ............................................................................................................................	  55	  3.3.1	   Optimizing	  Sample	  Analysis	  .....................................................................................................	  55	  3.3.2	   DMSO	  and	  DMSP	  Sample	  Recovery	  .......................................................................................	  55	  3.3.3	   Calibrations	  ......................................................................................................................................	  56	  3.3.4	   Accuracy	  and	  Precision	  ...............................................................................................................	  56	  3.3.5	   Underway	  Analysis	  .......................................................................................................................	  57	  3.4	   Discussion	  ..............................................................................................................................	  59	  3.5	   Comments	  and	  Recommendations	  ................................................................................	  60	  4	   High	  Concentrations	  and	  Turnover	  Rates	  of	  DMS,	  DMSP,	  and	  DMSO	  in	  Antarctic	  Sea	  Ice	  ....................................................................................................................	  75	  4.1	   Introduction	  ..........................................................................................................................	  75	  4.2	   Methods	  ..................................................................................................................................	  76	  4.3	   Results	  .....................................................................................................................................	  79	  4.4	   Discussion	  ..............................................................................................................................	  81	  4.5	   Conclusions	  ...........................................................................................................................	  82	  5	   Concentrations,	  Cycling	  of	  MDS,	  DMSP,	  and	  DMSO	  in	  Coastal	  and	  Offshore	  Waters	  in	  the	  Subarctic	  Northeast	  Pacific	  During	  Summer,	  2010-­‐2011	  ...........	  91	   vii 5.1	   Introduction	  ..........................................................................................................................	  91	  5.2	   Methods	  ..................................................................................................................................	  94	  5.2.1	   Field	  Sampling	  ................................................................................................................................	  94	  5.2.2	   Surface	  Water	  Gas	  Measurements	  .........................................................................................	  94	  5.2.3	   DMS/P/O	  Concentrations	  Measurements	  ..........................................................................	  95	  5.2.4	   Rate	  Experiments	  ..........................................................................................................................	  96	  5.2.5	   Ancillary	  Measurements	  ............................................................................................................	  99	  5.2.6	   Sea-­‐Air	  Flux	  ...................................................................................................................................	  100	  5.2.7	   Empirical	  Algorithms	  ................................................................................................................	  100	  5.3	   Results	  ...................................................................................................................................	  101	  5.3.1	   Hydrography	  and	  Plankton	  Biomass	  in	  Coastal	  and	  Open	  Ocean	  Waters	  .........	  101	  5.3.2	   High	  Frequency	  Measurements	  of	  DMS	  Concentrations	  ..........................................	  102	  5.3.3	   High	  Resolution	  Survey	  of	  DMS	  Across	  the	  Shelf-­‐Break	  ...........................................	  103	  5.3.4	   Empirical	  Algorithms	  and	  DMS	  Distributions	  ...............................................................	  103	  5.3.5	   Discrete	  DMS/P/O	  Measurements	  .....................................................................................	  104	  5.3.6	   Sea-­‐Air	  fluxes	  ...............................................................................................................................	  105	  5.3.7	   Rate	  Constants	  of	  DMS	  Production	  and	  Consumption	  ...............................................	  105	  5.3.8	   Patterns	  in	  DMS	  Production	  and	  Consumption	  Terms	  ..............................................	  106	  5.4	   Discussion	  ............................................................................................................................	  107	  5.4.1	   Strong	  Spatial	  Gradients	  in	  DMS	  Concentrations	  and	  Sea-­‐Air	  Fluxes	  .................	  107	  5.4.2	   DMS/P/O	  Concentrations	  in	  the	  Subarctic	  Northeast	  Pacific	  .................................	  108	  5.4.3	   Trends	  in	  DMS	  Production/Consumption	  .......................................................................	  109	  5.4.4	   A	  Comparison	  of	  Rate	  Measurements	  from	  the	  Subarctic	  Northeast	  Pacific	  ...	  110	  5.5	   Conclusions	  and	  Future	  Outlook	  ..................................................................................	  111	  6	   Biogeochemical	  Controls	  on	  the	  Seasonal	  and	  Temporal	  Variability	  of	  DMS,	  DMSP	  and	  DMSO	  Concentrations	  in	  Coastal	  Waters	  of	  the	  Western	  Antarctic	  Peninsula	  ...............................................................................................................................	  130	  6.1	   Introduction	  ........................................................................................................................	  130	  6.2	   Methods	  ................................................................................................................................	  133	  6.2.1	   Sampling	  Overview	  ....................................................................................................................	  133	  6.2.2	   Sulfur	  Concentration	  Measurements	  .................................................................................	  133	  6.2.3	   Process	  Studies	  and	  Rate	  Experiments	  .............................................................................	  136	  6.2.4	   Grazing	  Studies	  ............................................................................................................................	  140	   viii 6.2.5	   Inferring	  Total	  DMS	  Production/Consumption	  Terms	  from	  Specific	  Rate	  Measurements	  .............................................................................................................................................	  142	  6.2.6	   Ancillary	  Measurements	  .........................................................................................................	  143	  6.3	   Results	  ...................................................................................................................................	  144	  6.3.1	   Surface	  Water	  Hydrography	  and	  Plankton	  Biomass	  ..................................................	  144	  6.3.2	   Seasonal	  DMS/P/O	  Distributions	  ........................................................................................	  144	  6.3.3	   High	  Frequency	  DMS/P/O	  Measurements	  and	  Diel	  Cycles	  .....................................	  146	  6.3.4	   Rate	  Measurements	  and	  Process	  Studies	  ........................................................................	  147	  6.3.5	   Grazing	  Experiments	  ................................................................................................................	  148	  6.3.6	   Trends	  in	  DMS	  Production/Consumption	  .......................................................................	  149	  6.3.7	   Seasonal	  DMS	  Budget	  ...............................................................................................................	  149	  6.4	   Discussion	  ............................................................................................................................	  150	  6.4.1	   Concentrations	  ............................................................................................................................	  150	  6.4.2	   Rate	  Measurements	  ...................................................................................................................	  152	  6.4.3	   Conclusions	  and	  Future	  Outlook	  .........................................................................................	  155	  7	   Conclusions	  ....................................................................................................................	  174	  7.1.1	   Major	  Findings	  and	  Contributions	  ......................................................................................	  174	  7.1.2	   Future	  Directions	  .......................................................................................................................	  175	  Bibliography	  .........................................................................................................................	  178	      ix LIST OF TABLES Table 2.1. Concentrations and Air-Sea Fluxes ................................................................. 36	  Table 3.1. Optimizing Parameters for OSSCAR .............................................................. 63	  Table 3.2. Comparison of OSSCAR and IOS Instruments. .............................................. 64	  Table 5.1. Comparison of CI and Tracer Methods ......................................................... 113	  Table 5.2. Concentrations of DMS/P/O .......................................................................... 114	  Table 5.3. Air-Sea Fluxes of DMS ................................................................................. 115	  Table 5.4. Rates of DMS Production and Consumption in CI Experiments .................. 116	  Table 5.5. Rates of DMS Production and Consumption in Tracer Experiments ............ 117	  Table 6.1. Equations for DMS Rate Measurements ....................................................... 157	  Table 6.2. Correlation Matrix for DMS/P/O and Ancillary Variables ........................... 158	  Table 6.3. DMSP Production Measurements .................................................................. 159	  Table 6.4. DMS Production and Removal Terms ........................................................... 160	  Table 6.5. Microzooplankton Grazing and DMS Production ......................................... 161	      x LIST OF FIGURES Figure 1.1 Diagram of the Role of Dimethyl Sulfide (DMS) in the Sulfur Cycle ............ 12	  Figure 1.2 Diagram of Important Rate Processes in the Marine DMS Cycle ................... 13	  Figure 2.1. Definition of Coastal and Oceanic Domains .................................................. 37	  Figure 2.2. Spatial Distributions of DMS ......................................................................... 38	  Figure 2.3. Underway Measurements in 2007 .................................................................. 39	  Figure 2.4. Deriving the DMS Length Scale of Variability .............................................. 40	  Figure 2.5. Comparison of Length Scales of Variability .................................................. 41	  Figure 2.6. Prediction Using Multiple Linear Regression ................................................ 42	  Figure 2.7. Results from Multiple Regression Analysis Across Cruises .......................... 43	  Figure 2.8. Regional Empirical Algorithms ...................................................................... 44	  Figure 2.9. Seasonal Empirical Relationships .................................................................. 45	  Figure 2.10. Temporal Changes in DMS at Ocean Station Papa ...................................... 46	  Figure 2.11. Calculated Rates of Air-Sea Flux and Photo-Oxidation ............................... 47	  Figure 3.1. Schematic Diagram of OSSCAR System ....................................................... 65	  Figure 3.2. Sampling Time-Line for OSSCAR System .................................................... 66	  Figure 3.3. Photograph of OSSCAR Instrument. ............................................................. 67	  Figure 3.4. Chromatogram of a Single DMS/O/P Sample Using OSSCAR .................... 68	  Figure 3.5. Calibration Curve for DMS/O/P ..................................................................... 69	  Figure 3.6. Repeated Calibration Curves for DMS ........................................................... 70	  Figure 3.7. Underway Measurements of DMS/O/P and Ancillary Parameters ................ 71	  Figure 3.8. Relationships between DMSO/P and Photosynthetic Pigments ..................... 73	  Figure 3.9. Relationship between DMS and DMSO ......................................................... 74	  Figure 4.1. Map of Sampling Stations in the Southern Ocean .......................................... 84	  Figure 4.2. Schematic Diagram of Tracer Method ........................................................... 85	  Figure 4.3. Boxplots of DMS/P/O Concentrations ........................................................... 86	  Figure 4.4. Depth Profiles of DMS/P/O Concentrations .................................................. 87	  Figure 4.5. Time Course Data of Tracer Experiment ....................................................... 88	  Figure 4.6. Summary of Tracer Experiments .................................................................... 90	  Figure 5.1. Map of Study Area in the Northeast Pacific ................................................. 118	   xi Figure 5.2. Mixed Layer Depths, Chla, and Si: N Drawdown ....................................... 119	  Figure 5.3. Spatial Distribution of DMS Concentrations and Surface Salinity .............. 121	  Figure 5.4. Box Plots of DMS Concentrations by Oceanographic Region. ................... 122	  Figure 5.5. Detailed View of DMS and Surface Salinity at Brooks Peninsula .............. 123	  Figure 5.6. Example of CI Experiment ........................................................................... 124	  Figure 5.7. Example of Tracer Experiment. ................................................................... 125	  Figure 5.8. Spatial Distribution of DMS Consumption and Production Measurements. 126	  Figure 5.9. Relationship between Net DMS Production and DMS Concentrations. ...... 127	  Figure 5.10. Comparison of DMS Production and Consumption Along Line P ............ 128	  Figure 6.1. Map of Study Area ....................................................................................... 162	  Figure 6.2. Seasonal Changes in Ancillary Measurements ............................................. 163	  Figure 6.3. Seasonal Changes in DMS/P/O Concentrations ........................................... 164	  Figure 6.4. High-Frequency Measurements of DMS by MIMS ..................................... 165	  Figure 6.5. High Frequency Measurements of DMSP/O by OSSCAR .......................... 166	  Figure 6.6. Example of Tracer Experiment .................................................................... 167	  Figure 6.7. Example of DMS CI Experiment ................................................................. 168	  Figure 6.8. Example of Dilution Experiment .................................................................. 169	  Figure 6.9. Relationship between Phaeocystis and DMSP Removal ............................. 170	  Figure 6.10. Krill Grazing and DMS Production ............................................................ 171	  Figure 6.11. DMS Cleavage, DMS Consumption, and Bacterial Respiration ................ 172	  Figure 6.12. Seasonal Summary of DMS Production and Removal Terms ................... 173	      xii LIST OF SYMBOLS AND ABBREVIATIONS aq: aqueous BATS: Bermuda Atlantic Time Series  BC: British Columbia 13C: stable isotope with an atomic mass of 13 Chla: chlorophyll a pigment Chla/MLD: chlorophyll a pigment to mixed layer depth ratio CI: competitive inhibition rate measurements CTD: instrument to measure conductivity, temperature, and depth CO2: carbon dioxide gas d: day D-3: stable isotope compound with three deuterium atoms in the place of hydrogen atoms in a single methyl group D-6: stable isotope compound with six deuterium atoms in the place of hydrogen atoms in a single methyl group DMS: dimethyl sulfide DMSP: dimethylsulfoniopropionate DMSO: dimethyl sulfoxide DMSPd: dissolved dimethylsulfoniopropionate DMSOd: dissolved dimethyl sulfoxide DMSPt: total dimethylsulfoniopropionate DMSOt: total dimethyl sulfoxide DMSPp: particulate dimethylsulfoniopropionate DMS/P/O: dimethyl sulfide, dimethylsulfoniopropionate, and dimethyl sulfoxide  DMS/P: dimethyl sulfide and dimethylsulfoniopropionate DMSP/O: dimethylsulfoniopropionate and dimethyl sulfoxide DMSOr: dimethyl sulfoxide reductase enzyme  EDTA: ethylenediaminetetraacetic acid  FMN: Flavin mononucleotide Fe: iron  xiii FPD: flame photometric detector H2SO4: sulfuric acid HNLC: high nutrient, low chlorophyll hr: hour GC: gas chromatography  GOTM: General Ocean Turbulence Model k: rate constant kd: extinction coefficient LN: natural log M: moles per liter or molar (concentration) MIMS: membrane inlet mass spectrometry MLD: mixed layer depth MSA: methane sulfonic acid MSE: mean squared error m/z: mass to charge ratio (of ions detected in mass spectrometry) N: nitrogen N2: nitrogen gas NaOH: sodium hydroxide (used to hydrolyze DMSP) NE: northeast nM: nanomoles per liter or nanomolar (concentration) NO3: nitrate O2/Ar: the ratio of oxygen to argon gas, measures biological supersaturation of O2 OSSCAR: organic sulfur sequential chemical analysis robot P: phosphorous PAL-LTER: Palmer station long term ecological research  PO4: phosphate PAR: photosynthetically active radiation PFPD: pulse flame photometric detector PT-CIMS: purge and trap capillary inlet mass spectrometry PT-GC: purge and trap gas chromatography S: sulfur  xiv SCD: sulfur chemiluminescence detector Si: silica SO4: sulfate SO2: sulfur dioxide SRD: solar radiation dose SST: sea surface temperature TiCl3: titanium chloride solution (used to chemically reduce DMSO to DMS) UV: ultraviolet light UVT: ultraviolet transparent WAP: western Antarctic peninsula WCVI: west coast of Vancouver island (British Columbia) Δ  change (indicates subtraction of one value from another) Δσt  change in seawater density from the surface     xv ACKNOWLEDGEMENTS I would like to thank the members of my supervisory committee, Kristin Orians, Susan Allen, Stephen Hallam and in particular, my Ph.D. advisor Dr. Philippe Tortell. They have helped me to understand rigorous quantitative methods and interdisciplinary studies, have always encouraged a close attention to detail, and have coached my academic writing.  Without their guidance and patience, I would not have embarked on this particular adventure.  I feel lucky to have had so many amazing research opportunities and an extensive support network during my Ph.D. I would also like to thank everyone with whom I have collaborated—most of all I would like to thank Dr. John Dacey and Dr. Matt Long.  Without Dr. Dacey’s collaboration, technical expertise, and wealth of knowledge, this thesis would have a different focus.  In the past year and a half, I have also welcomed Dr. Long’s guidance and support on topics ranging from how to establish an efficient workflow to discussions to the scientific questions that haunt us in our dreams. Of course, I am extremely grateful for the logistical support and delightful company of the Institute of Ocean Sciences staff and the crew of the JP Tully, whom I admire very much and will always appreciate.  Similarly, I am indebted to the Palmer LTER science team and Raytheon Polar Services for two successful Antarctic deployments. Finally, I would like to thank the lab mates and friends I have met at UBC and doing fieldwork.  I would like to single out Dave, Nina, Ania, Tereza, Elaina, Dave, Jason, Nicole, Stef, Kim, Domi, Mike, yet another Dave, Sven, Jodi, Johanna, Shelley, Lindsey, Mike, and Chance who have inspired me to play hard, work harder, and relish every moment of each.  I cannot express how much I appreciate the love and support of my fiancée Ben, my sister Alexis, my mother Sheila, and my father Nicholas here in one sentence.    1 1 Introduction 1.1 The Role of DMS in the Global Sulfur Cycle The trace gas dimethyl sulfide (DMS) is an important source of atmospheric sulfate aerosols6, which back-scatter incoming solar radiation and thus influence Earth’s radiative balance7 (Fig. 1.1).  Notwithstanding its climatic importance, atmospheric DMS represents a small fraction of the total oceanic DMS pool.  This compound and the related molecules, dimethylsulfoniopropionate (DMSP) and dimethyl sulfoxide (DMSO), are primarily produced by microscopic algae (phytoplankton) in the surface ocean, and they have been suggested to function as cellular osmo-regulators8, cryoprotectants9 and anti-oxidants10 (Fig. 1.1).  Although the physiological roles of DMS, DMSP and DMSO (DMS/P/O) remain unclear, research over the past two decades has established the essential role of these compounds marine ecosystem dynamics.  DMSP and DMS are substrates for growth; DMSP serves as an important source of carbon, sulfur and energy for many bacteria11, while some bacteria also use DMS as an energy source12 (Fig. 1.1).  These compounds can also act as signaling molecules connecting trophic levels across the marine ecosystem. DMSP, for example, is a chemo-attractant for bacteria and non-DMSP producing phytoplankton13, while seabirds use DMS as a foraging cue14.  With increasing research, the depth and complexity of the oceanic DMS/P/O cycle becomes ever more apparent.  Using new experimental and analytical methods, this thesis sought to provide insights into the marine DMS/P/O cycle, and specifically the ecosystem dynamics and environmental conditions that lead to DMS accumulations in surface waters.  1.2 The Marine DMS Cycle Below, I frame aspects of the marine DMS/P/O system, which are well understood, and highlight a few key active areas of research.  DMSP, the main precursor of DMS, is produced by a variety of marine phytoplankton in the surface ocean.  Haptophytes (e.g. and E. Huxleyi) dinoflagellates (e.g. P. Minimum) are prolific DMSP producers15  and a number of species in these groups contain the enzyme DMSP lyase, which catalyzes the cleavage of DMSP to DMS and acrylate16 (Fig. 1.2).  In the absence of DMSP-lyase containing  2 phytoplankton, DMS is produced from the dissolved DMSP pool in seawater through a series of complex ecosystem interactions.  First, DMSP is released from phytoplankton into the dissolved pool via direct exudation, or during zooplantkon grazing17 and/or viral lysis18 (Fig. 1.2). Thereafter, dissolved DMSP (DMSPd) is taken up by bacteria and incorporated into biomass via the demthylation pathway to satisfy cellular metabolic demands for protein synthesis and growth (Fig. 1.2).  Although DMSP can also be broken down to produce DMS via the DMSP-lyase dependent cleavage pathway11, relatively few bacteria have the genes necessary for DMSP cleavage19,20 (Fig. 1.2).  This restricted capacity for DMSP cleavage, coupled with the use of DMSP for cellular metabolism, results in low DMS yields, with DMS production typically representing a small fraction of DMSPd uptake.   Under certain conditions, however, DMS yields can increase significantly.  For example, intensive zooplankton grazing and viral lysis can lead to the release of large quantities of phytoplankton-derived intracellular DMSP, exceeding microbial sulfur (S) demands and resulting in high DMS production.  Moreover, a number of studies have demonstrated direct DMS production within the guts of feeding zooplankton (Fig. 1.2).  Enhanced DMS production from DMSP is particularly important in the presence of high levels of DMSP-lyase.  However, due to the spatiotemporal “patchiness” of grazing and notable viral lysis events (and the time-consuming, laborious nature of these measurements), predicting DMS accumulations based on the interactions between these different trophic levels (i.e. phytoplankton, zooplankton and viruses) proves difficult17.  An atmospheric oxidation product of DMS that is subject to dry deposition and wet deposition, DMSO represents a second potential source of DMS in marine surface waters.  Both algae and bacteria are capable of DMSO reduction to DMS21,12,22 (Fig. 1.2).  In suboxic environments, this reaction is energetically favorable, leading to widespread DMS production23.  The extent to which this reaction may occur in well-oxygenated surface ocean environments has not been extensively studied, and is an important question in DMS/P/O research.  In general, the marine dynamics of DMSO remain poorly studied, relative to DMSP.  It is clear, however, that this molecule is ubiquitous in surface ocean waters24, and it could thus play a significant role in the DMS cycle.   Processes governing DMS removal are better constrained than the myriad of processes that control DMS production.  DMS removal from the surface ocean is comprised  3 of biological DMS consumption and abiotic photo-oxidation, and sea-air flux25 (Fig. 1.2).  It is generally believed that bacteria are exclusively responsible for biological DMS/P consumption, and a number of studies have reported correlations between DMS/P consumption and bacterial production26.  Photo-oxidation and sea-air flux typically represent smaller terms in the mixed layer DMS budget, and are only relevant in the surface mixed layer (Fig. 1.2).  High levels of UV light, and elevated nitrate concentrations and dissolved organic matter drive DMS photochemical oxidation27,28, while wind speeds and gas solubility exert a primary control on gas exchange across the sea-air interface.  DMS loss due to photo-oxidation and sea-air flux can be parameterized given environmental variables (e.g. wind speed, sea surface temperature, down-welling UV irradiance), and readily incorporated into biogeochemical models29.  Similarly, the contribution of mixing and dilution to surface ocean DMS budgets can be estimated from derived transport terms and concentration gradients.  By comparison, quantifying the biological production / consumption terms and their response to environmental variability remains a central challenge in DMS/P/O research.  1.3 Global DMS/P/O Distributions A global database of DMS observations has been compiled with ~50,000 data points (http://saga.pmel.noaa.gov/dms/), which has been used to identify oceanographic regions of persistent DMS accumulations, and develop empirical relationships between DMS and hydrographic variables.  Based on this database, the global average concentration is ~2 nM, with considerable seasonal variability in high latitude regions (0 -20 nM).  Although DMS concentrations are highest during the spring / summer phytoplankton growing season, concentrations do not scale linearly with chlorophyll a (chla), a widely used proxy for phytoplankton biomass.  Several empirical algorithms have been developed to predict DMS distributions on regional and global scales30,31.  Simo and Dachs30 found a strong global relationship between DMS concentrations and the ratio of chla to mixed layer depth (chla/MLD) or between DMS and MLD below a 0.02 chla/MLD threshold, when data were averaged to 1° x 1° resolution.  Similarly, Vallina and Simo31 reported a strong regional relationship between DMS and the solar radiation dose (a measure of the ultraviolet (UV)  4 light experienced in the mixed layer).  Although not mechanistic in nature, these observed relationships can identify oceanographic conditions that promote high DMS accumulation. Using the global DMS database, researchers have identified two primary oceanographic regions with particularly high DMS accumulation: the Southern Ocean, and the Subarctic Northeast Pacific32,33.  We discuss key similarities between DMS/P/O cycles and environmental conditions in these oceanographic regions.  DMS accumulation in these regions has been linked to the abundance of high DMSP producing algae, and with algal physiology and environmental stressors.  Both the Southern Ocean and Subarctic Pacific are regions of known iron (Fe) limitation, where low Fe concentrations limit phytoplankton growth and nutrient consumption.  These two regions exist in a chronically high nutrient low chlorophyll (HNLC) state.  This is significant in the context of DMS/P/O cycling, since iron limitation has been shown to significantly enhance the production of these compounds.  Indeed, open ocean iron enrichment experiments in both the Southern Ocean34 and Subarctic Pacific35,36 have demonstrated significant effects of Fe limitation on DMS/P concentrations and turnover rates. The concentrations of DMS in surface waters of the Subarctic Pacific have been measured for over two decades as part of the Line P time-series program, providing a baseline for DMS (and more recently DMSP) concentrations in these waters37,38.  The frequency and length of the time-series collections have provided insight into both the seasonal and inter-annual variability of these reduced sulfur compounds.  Results from this work show that DMS/P exhibit a wide concentrations range (~1 - 25 nM and 10 - ~70 nM, respectively), with maximum values in coastal waters observed in spring, and peak concentrations in offshore regions during late summer35,38,39.  The spring maximum in coastal DMS/P concentrations is believed to be linked to the annual phytoplankton production cycle, where the appearance of large phytoplankton blooms is tied to regional upwelling of nutrients into surface waters.  By comparison, late summer DMS accumulation in offshore waters is likely tied to increased irradiance levels and density stratification of the surface mixed layer, in conjunction with the onset of iron limitation.  These environmental conditions may select for high DMS/P producing species, by creating conditions of oxidative stress10. Beyond these seasonal cycle of DMS/P, considerable inter-annual variability has been observed in the Subarctic Northeast Pacific.  This longer term variability is related, at  5 least in part, to regional climate forcing, including the presence of strong El Niño / La Nino cycles38. Unlike waters of the Subarctic Pacific, there are no long-term time-series of DMS concentrations in the Southern Ocean.  However, there have been a number of studies examining DMS/P dynamics across this vast oceanic region.  Much of the focus to date has been on studies of polynyas (seasonally open waters surrounded by permanent sea ice), where exceptionally high (> 100 nM) DMS concentrations have been observed.  In these systems, high DMS concentrations appear to be tied to massive blooms of Phaeocystis antarctica, a colony forming phytoplankton (haptophyte), which is a prolific producer of DMSP and DMSP lyase.  Exceptionally high DMS/P concentrations have also been observed in Antarctic sea ice25,40,41, where a variety of phytoplankton grow under very challenging environmental conditions42.  For both sea ice habitats and open waters of the Southern Ocean, little information is available on the underlying processes driving high DMS accumulation. Recent work has sought to understand the dynamics of DMS/P in Southern Ocean waters, by examining the seasonal processes in the coastal West Antarctic Peninsula (WAP) region.  This work has exploited the Long Term Ecological Research (LTER) time-series at Palmer Station to facilitate biogeochemical studies.  Over the past two decades, the Palmer LTER has documented significant changes in ecosystem dynamics in the WAP, associated with regional warming and sea ice retreat.  The extent to which these changes influence the cycles of DMS and other climate-active gases remains largely unknown.  Results presented based on previous field studies43 at Palmer Station demonstrated that DMS concentrations peak several times throughout the austral summer, ranging between 0 – 20 nM, while DMSP concentrations vary between ~50 - >350 nM.  The authors showed that biological DMS consumption dominated the removal of this trace gas in surface waters and maintained a tight balance with gross biological DMS production (inferred from this model).  This work opened many questions regarding the dominant DMS production processes driving high DMS accumulation in polar regions of the Southern Ocean.   6 1.4 Emerging Methodological Approaches Over the past decade, a number of powerful new methodologies have been developed to study the distribution and cycling of DMS and related compounds in marine surface waters.  Whereas traditional DMS measurements have relied on labor intensive and time-consuming analyses of discrete bottle samples, a new generation of autonomous 'underway' systems have been developed for high frequency ship-board measurements.  The first of these systems was based on membrane inlet mass spectrometry (MIMS)2, and a number of additional systems have since been developed44.  These methods facilitate the collection of data at unprecedented spatial resolution, and have provided new insight into the mesoscale (eddy resolving, ≥ 100 km) and sub-mesoscale (1-100 km) variability of surface ocean DMS concentrations.  For example, repeat MIMS surveys in various Southern Ocean waters have demonstrated strong DMS variability across hydrographic frontal zones and in regions of localized sea ice melt.  As of yet, however, deployment of MIMS and other related high resolution measurement systems has been limited to a handful of marine environments.  Expansion of MIMS surveys would significantly improve the global coverage of existing DMS measurements.   Moreover, there is a significant need to develop automated systems for DMSP and DMSO analysis, in order to increase the coverage and spatial resolution of observations in surface ocean waters. Beyond the need for additional concentration measurements, we also require information on the rate of key production and consumption processes in the DMS/P/O cycle.  To date, the majority of DMS/P consumption measurements have been made using a 35S radio-isotope technique developed by Kiene et al.45, in which DMS/P consumption is measured as the disappearance of radio-labled DMS/P (Fig. 1.2).  Thereafter, the appearance of radio-isotope labels can be traced to other sulfur pools (e.g. DMS, DMSO, sulfate, or macromolecules).  DMS or DMSO production is calculated indirectly from the final fraction of radiolabeled molecules in the DMS or DMSO pool and the rate of DMS or DMSP consumption.  An alternative approach has been used to quantify gross DMS or DMSP production rates, based on competitive inhibition experiments.  In these experiments, glycine betaine and dimethyl disulfide (DMDS) have typically been used as competitive inhibitors of DMSP and DMS, respectively.  These inhibitors are added in large quantities (> 20 times the size of the DMS or DMSP pool) in order to temporarily block the consumption of either  7 DMS46 or DMSP47.  In the absence of a significant consumption term, the change in DMS or DMSP concentrations over time represents a gross production rate.  Radio-isotope and competitive inhibitor experiments have provided fundamental new insight into the dynamics of DMS and related compounds in seawater.  However, the methods are time-consuming and labor intensive, and have not yet been widely applied by the majority of research groups. Recently, a new stable isotope technique has recently been developed5 to directly track simultaneous DMS production from multiple sources (DMSP and DMSO) and gross DMS consumption (Fig. 1.2).  This method could provide insights in our understanding of oceanic DMS dynamics, by providing simultaneous quantitative estimates of multiple production and consumption terms.  To date, the method has not been adopted by the oceanographic community.  Thus, this thesis demonstrates that 1) the widespread application of MIMS methods in polar and sub-polar DMS studies to document spatiotemporal DMS distributions, 2) the development of a new automated system for DMSP/O measurements in surface seawater to increase the spatiotemporal coverage of these observations, and 3) the application of the stable isotope tracer technique are useful to quantify DMS dynamics in polar and sub-polar waters.  1.5 Thesis Objectives The current understanding of the marine DMS/P/O cycle will benefit from new technologies and the improved spatiotemporal coverage of DMS/P/O concentration measurements and simultaneous rate measurements.  The main objective of this thesis was to develop and deploy new technologies for concentration and rate measurements in the DMS/P/O cycle, including 1) the application of automated MIMS to quantify mesoscale and sub-mesoscale variability in DMS concentrations in polar and sub-polar waters; 2) the development of an automated system for sequential analysis of DMS/P/O concentrations; and 3) the refinement and application of a stable isotope tracer technique to simultaneously measure DMS consumption and its production from DMSP and DMSO.  These technologies were developed and deployed to answer a number of key research questions including:  8 1. How do DMS/P/O concentrations vary over space and time across multiple spatial/ temporal scales in polar and sub-polar marine waters, and how are these compounds inter-related? 2. How do DMS/P/O concentrations correlate with biological and hydrographic parameters? 3. What biological processes govern DMS production (i.e. DMSP cleavage, DMSOd reduction, grazing or direct DMS production from phytoplankton), and how do these processes respond to environmental variability?  1.6 Thesis Overview My study of the marine DMS/P/O cycle in the Subarctic Northeast Pacific and Southern ocean was motivated by the need to predict heterogeneous DMS concentrations in space and time, and guided by the development and deployment of emerging technologies.  First, I used MIMS measurements collected over 2 year in the Subarctic Northeast Pacific (by a postdoctoral researcher) to describe DMS concentrations at three distinct times in the seasonal cycle (i.e. winter, spring and summer) with unprecedented spatial resolution.  Thereafter, I developed and validated a new method for the sequential analysis of DMSO and DMSP in surface seawaters to document the spatial distributions of these compounds and study the relationships between DMS, DMSP, and DMSO concentrations in the Subarctic Northeast Pacific in summer.  To further explore the relationships between reduced S compounds (and the role of DMSP and DMSO as important sources of DMS) in high DMS marine environments, I employed a novel stable-isotope tracer technique in Antarctic sea ice.  Thereafter, in the Subarctic Northeast Pacific, I used MIMS measurements and the stable isotope tracer (and competitive inhibition) experiments to study the spatial distributions of DMS concentrations and DMS production and consumption terms.  I sought to determine if simultaneous stable isotope tracer measurements could be used to predict sub-mesoscale DMS variability.  Finally, I employed MIMS, stable isotope tracer and competitive inhibition experiments, automated DMSP and DMSO measurements, and grazing experiments to study the seasonal and temporal DMS/P/O accumulations and production terms in the coastal waters of the WAP.  9 This thesis is organized into 5 primary data chapters, each one examining one or more aspects of the DMS/P/O cycle in polar or sub-polar marine waters.  A brief outline of these chapters is given below.  Following the data chapters, a short concluding chapter is presented, summarizing key findings and outlining some directions for future research.    Chapter 2: Fine-Scale Spatial and Temporal Variability of Surface Water Dimethylsulfide Concentrations and Sea-Air Fluxes in the Subarctic Northeast Pacific  In this chapter, I used novel MIMS technology to measure DMS concentrations across coastal and oceanic waters along Line P (Subarctic Northeast Pacific) with unprecedented spatial resolution.  I used data previously collected on several research cruises to study relationships between DMS and oceanographic variables in order to generate a predictive regional empirical algorithm for DMS concentrations.  In addition, I calculated sea-air DMS flux across our transect, and monitored changes in surface water DMS concentrations over ~12- 24 hours at Ocean Station Papa (145°W 50°N) to understand diel (i.e. day - night) cycles.  The results of this chapter show that 1) DMS concentrations increased from ~1 – 2 nM in February to ~8 nM in August, and that in June the highest DMS concentrations were observed in coastal waters, whereas in August, the highest DMS concentrations were observed in open ocean waters.   This chapter was published in Marine Chemistry1.    Chapter 3: Measurement of DMS, DMSO, and DMSP in Natural Waters by Automated Sequential Chemical Analysis  In this study, I outlined the development of an automated sequential analysis system (OSSCAR) for DMS/P/O concentrations in oceanic waters, and provided a proof of concept based on fieldwork conducted in the Subarctic Northeast Pacific.  I conducted a series of laboratory studies, and a ship-based inter-calibration exercise to demonstrate good analytical performance of the system.  Field data collected with the new analytical system demonstrated strong variability in DMS, DMSP and DMSO concentrations in the Subarctic Pacific, and yielded new insight into the relationship of these compounds with other hydrographic and  10 biological variables in marine surface waters.  This chapter was recently accepted for publication in Limnology and Oceanography: Methods4.  Chapter 4: High Concentrations and Turnover Rates of DMS, DMSP, and DMSO in Antarctic Sea Ice I conduced isotope tracer experiments in Antarctic sea-ice environments using a novel experimental approach, and an analytical system based on purge and trap capillary inlet mass spectrometry (PT-CIMS).  This method enabled me to detect multiple isotopic species of DMS, and thus simultaneously examine a variety of production / consumption processes in the DMS/P/O cycle.  I measured extremely high concentrations of DMS/P/O in sea-ice brines and rapid DMS/P/O turnover resulting from biological activity.  These measurements demonstrated the potential for DMSO and DMSP to act as an important source of DMS.  In addition, these data indicated that DMSOd reduction may be an important removal term for DMSO.  This chapter was published in Geophysical Research Letters3, and was also featured as an Editor's Highlight in Nature Geosciences48.  Chapter 5: Concentrations, Cycling of DMS, DMSP, and DMSO in Coastal and Offshore Waters in the Subarctic Northeast Pacific During Summer, 2010-2011  Using the stable isotope tracer technique deployed in Asher et al.3, I measured DMS/P/O concentrations, sea-air fluxes and turnover rates in the Northeast Subarctic Pacific.  I compared the stable isotope tracer method to competitive inhibitor assays, and found good agreement, suggesting that the various techniques provide consistent data.  I characterized the spatial variability in DMS concentrations using MIMS, and used PT-CIMS3 to quantify DMS/P/O turnover rates.  This study showed regional (i.e. coastal vs. open ocean) and inter-annual differences in DMS concentrations, sea-air fluxes and turnover rates, and suggests that DMSP cleavage, as opposed to DMSO reduction was the dominant DMS production term.  Measured rates of net DMS production were significantly correlated with observed surface water DMS concentrations, suggesting that our tracer studies can captured the most significant processes in the DMS cycle.  This paper will be submitted for publication to in the next one or two months (pending comments from collaborators).  11  Chapter 6: Biogeochemical Controls on the Seasonal and Temporal Variability of DMS, DMSP, and DMSO Concentrations in the Coastal Waters of the Western Antarctic Peninsula Using a variety of analytical and experimental approaches, I characterized the seasonal dynamics of DMS/P/O in high latitude Antarctic waters of the WAP.  I quantified the variability in DMS/P/O concentrations on various time-scales using MIMS, and a prototype of OSSCAR.  Using stable isotope tracer and DMS competitive inhibition experiments, I examined changes in underlying DMS production and consumption terms through various biological and abiotic processes.  This study demonstrated significant seasonal variability in DMS/P/O concentrations, and highlighted the importance of micro-zooplankton grazing and bacterial DMSP cleavage as dominant mechanisms for DMS production in this region.  In addition, this study revealed diel cycles of DMS/P/O superimposed on the seasonal variability.  This paper will be submitted for publication in the next one or two months (pending comments from collaborators).    12  Figure 1.1 Diagram of the Role of Dimethyl Sulfide (DMS) in the Sulfur Cycle This diagram illustrates the cycling of DMS in the surface ocean and in the atmosphere and indicates that DMS emissions influence climate.  DMS is ultimately derived from the biological reduction of sulfate to form dimethylsulfoniopropionate (DMSP) or dimethyl sulfoxide (DMSO) in the surface ocean, influenced by mixing, incoming solar radiation, and the availability of inorganic nutrients shape.  The release of these compounds into the dissolved pool followed by subsequent DMSP cleavage or DMSO reduction, as well as direct DMS exudation from algae constitute important sources of DMS in the surface ocean.  Zooplankton grazers that feed on algae may also contribute to DMS production in two ways.  Once ventilated to the atmosphere, DMS is oxidized to form sulfur dioxide (SO2), sulfuric acid (H2SO4), and eventually sulfate aerosols (SO4-) aerosols that scatter and reflect incoming solar radiation.  In addition, sulfate aerosols and other DMS oxidation products, such as methane sulfonic (MSA) contribute to cloud formation, which increase the earth’s natural albedo.    13   Figure 1.2 Diagram of Important Rate Processes in the Marine DMS Cycle This diagram shows how competing rate processes control the marine DMS cycle and accumulations of DMS in the surface ocean.  Removal terms of DMS (highlighted in red) are calculated using wind speed parameterization in the case of sea-air flux, and measured directly in radiolabeled and stable-isotope tracer experiments.  Production terms (highlighted in blue) include direct algal exudation of DMS, the enzymatic algal and bacterial cleavage of DMSP to DMS, the reduction of DMSO to DMS, and its excretion by zooplankton.  These rate processes are simultaneously quantified using the stable-isotope tracer technique (and its variations) described in this thesis. Processes in black, such as the release of DMSP or DMSO into the dissolved pool are not captured by this method.  The radiolabeled tracer technique has been used to measure gross DMS consumption and demethylation extensively (underlined in black) in the field and can be used to infer DMSP cleavage.    14 2 Fine-Scale Spatial and Temporal Variability of Surface Water Dimethylsulfide (DMS) Concentrations and Sea-Air Fluxes in the Subarctic Northeast Pacific 2.1 Introduction Over the past several decades, research on the trace gas dimethylfulfide (DMS) has significantly advanced our understanding of the marine sulfur cycle and the ocean's influence on climate. This compound is primarily derived from the algal metabolite dimethylsulfoniopropionate (DMSP), which is produced to different extents by various phytoplankton taxa15. While some phytoplankton can directly release DMS from the cellular cleavage of DMSP16, much of the oceanic DMS production is associated with bacterial degradation of dissolved DMSP45, which is released into the water column by a variety of processes including zooplankton grazing and viral lysis18. The accumulation of DMS in oceanic waters thus depends upon complex trophic interactions that are, in turn, sensitive to a number of environmental factors such as solar radiation and mixing depth49,50.  Sea-air fluxes contribute 28 Tg DMS yr-1 to the atmosphere, which represents ~ 25% of emissions in the global sulfur budget32. The atmospheric oxidation products of DMS scatter sunlight and enhance cloud formation, thereby reducing the incoming short wave radiation and increasing Earth's albedo51. Recent efforts to predict surface ocean DMS concentrations have exploited global climatological databases and derived empirical relationships between DMS and various predictor variables at broad spatial scales (e.g. global scale or oceanographic basin scale). For example, Simo and Dachs30 generated a predictive global DMS algorithm based on a combination of the ratio of chlorophyll a (chla) to mixed layer depth and the mixed layer depth alone, while Vallina and Simo31 reported a strong relationship between surface ocean DMS concentrations and mixed layer solar radiation dose (SRD). These empirical algorithms have been valuable both in identifying the oceanographic conditions promoting high DMS concentrations at a global and basin-wide scale and in generating models linking the ocean's biosphere to global climate52. However, global-scale models typically do not resolve variability in regional processes that may control DMS cycling, and the accuracy of predictive algorithms varies significantly across oceanographic regimes52. By averaging over  15 large spatial and temporal scales, global-scale models do not fully capture the complex underlying processes driving DMS variability in the oceans53. Among all open ocean regions, some of the highest DMS concentrations are found in the Subarctic NE Pacific. DMS time series measurements in this region began in 1996 with discrete measurements at oceanographic stations along Line P, a transect stretching between the southern tip of Vancouver Island, British Columbia, and Ocean Station Papa (OSP, 145° W, 50° N). Results from this work37 show that average summer time DMS concentrations in the NE Subarctic Pacific exceed 95% of global DMS measurements54, despite a small stock of primary producers and relatively low primary productivity55. These high DMS levels may result from the dominance of high DMSP producing phytoplankton species, including various nanoflagellates and coccolithophores56.  Moreover, the region is subject to chronic Fe limitation34,57, which has been hypothesized to increase DMS/P production10. Analogous Fe-limited conditions are found over large regions of the Southern Ocean34, where elevated DMS concentrations are also observed54. The most comprehensive study of DMS cycling in the Subarctic Pacific Ocean occurred during the Subarctic Ecosystem Response to Iron Enrichment Study (SERIES), in which DMS/P concentrations and a wide suite of other biogeochemical parameters were measured following a mesoscale Fe enrichment34,35. Over the course of the Fe-induced phytoplankton bloom, DMS was not linearly correlated with its precursor, DMSP. Instead, the highest DMS concentrations corresponded with a mixed layer-shoaling event resulting from high irradiance and low wind speeds35. Thereafter, a peak in bacterial growth, coinciding with iron enrichment, led to a sharp decline in DMS concentrations36. It was thus suggested that DMS accumulation in surface waters reflected the impact of shallow mixed layers on DMSP production, bacterial growth and sea–air fluxes, as well as the relative demand of bacteria for reduced C and S. The SERIES results demonstrated the potential for short-term temporal and sub-mesoscale spatial variability in NE Pacific DMS concentrations, resulting from complex food web interactions and physical dynamics. Given the complexity of the oceanic DMS cycle, the spatial heterogeneity of surface concentrations is difficult to resolve using conventional analytical methods and ship-based surveys of discrete sampling stations. To address this limitation, we utilized a technique for high-resolution underway analysis of surface ocean DMS concentrations based on shipboard  16 membrane inlet mass spectrometry (MIMS)2. Using this method, it has been possible to resolve spatial DMS variability on sub-km spatial scales and document strong concentration gradients associated with mesoscale and sub-mesoscale variability in the underlying physical and biological fields. To date, MIMS-based DMS measurements have been conducted in several oceanic regions, including the Southern Ocean58–61. High-resolution aqueous DMS surveys using other mass spectrometric methods have also been conducted Equatorial Pacific, the western North Pacific, and the Southern Pacific62–64. In this article, we present new results from a 2-year survey of DMS concentrations in the coastal and open ocean Subarctic Pacific, as part of the Line P time series program65. Our research objective was to characterize spatial and temporal variability in surface water DMS concentrations and sea-air fluxes, and examine the relationship between DMS concentrations and other biological and hydrographic variables at various spatial scales. In this respect, we exploited a number of ship-based underway measurements and satellite-derived data products (e.g. chla and calcite) to test several empirical algorithms for DMS prediction in the Subarctic NE Pacific. Our results demonstrate strong spatial patterns in surface water DMS concentrations, and significant inter-annual, seasonal, and short-term temporal variability. While basin-wide seasonal variability across cruises follows established empirical relationships, simple algorithms fail to explain the strong mesoscale (eddy resolving, ~100’s km) and sub-mesoscale (1- 100 km) variability we observed, suggesting that other complex underlying processes are critical.  2.2 Methods 2.2.1 Underway DMS Measurements DMS was measured on the CCGS John P Tully along Line P (48°34.5’N 125° 30’ to 50°N 145W) on 6 oceanographic cruises between February 2007 and August 2008. Cruises occurred in February, June and August, spanning ~ 2 weeks in the winter and ~ 3 weeks in the spring and summer. Ship-board measurements of surface water (~ 5 m) DMS concentrations were made using MIMS as previously described2. Prior to measurement, surface seawater was passed through 20 ft of stainless steel tubing immersed in a 10 C water bath to achieve temperature equilibration. Measurement frequency was approximately 2–3  17 times per minute, yielding an effective spatial resolution of ~ 200 m along the cruise track for typical cruising speeds of ~ 10 kn. The raw DMS signal (ion current at m/z62) was calibrated daily using DMS standards derived from the alkaline hydrolysis of DMSP66 following the protocol of Nemçek et al.61.  The working detection limit of the instrument for DMS was ~ 1 nM. In practice, however, calibrated data can yield final concentrations below 1 nM, which exhibit spatial variability that is consistent with other ancillary measurements. These low concentration data are thus likely meaningful in a biogeochemical sense, even if the absolute concentrations are subject to some uncertainty. The mass spectrometer control software was coupled to the ship's GPS acquisition system so that each DMS measurement was associated with a date/time stamp and a geo-referenced position. These GPS data were then used to align DMS measurements with ancillary hydrographic and satellite-based data. An in-line Seabird SBE45 MicroTSG thermosalinograph and a WETLabs WETstar chla fluorometer recorded temperature, salinity, and chlorophyll a fluorescence measurements with a sampling frequency of ~ 0.25 Hz. DMS data collected along the cruise track were averaged into 1 min sampling bins and aligned with other hydrographic data.  2.2.2 Remote Sensing Data We used Quickscat scatterometer data to obtain daily averaged wind fields67.  Level 3 data were obtained from ftp://ftp.ssmi.com/qscat/bmaps_v03a/, with a spatial resolution of 0.25° × 0.25°.  Daily ascending and descending passes were averaged, excluding missing data and flagged data due to land or heavy rain. Mean daily wind speeds were generated along the cruise track by averaging wind speed data for 2 days (the day of sampling and 1 day prior), and linearly interpolating the 0.25° ×0.25° spatial grid to the cruise track. The piston velocity was calculated using average wind speeds and the empirical formulation of Ho et al.68, with a Schmidt number derived from the sea surface temperature and salinity69. The Ho et al. parameterization of the gas transfer coefficient was used because it was developed using a dual gas tracer method and higher wind speeds (> 16 m s− 1) than many previous studies. High wind speeds are common in the NE Pacific, particularly during winter. The Ho et al. formulation also agrees well with the parameterization of Nightingale et  18 al.70 and the calculated values of the DMS gas transfer coefficient using eddy correlation62,64. Sea–air flux (µmol m-2 d-1) was calculated as the product of the piston velocity and the aqueous DMS concentration, assuming negligible atmospheric DMS concentrations. Remotely sensed level 3 processed chla (Aqua-MODIS), calcite (Aqua-MODIS) and photosynthetically active radiation (PAR; SeaWiFS) data were obtained from NASA (http://oceancolor.gsfc.nasa.gov/cgi/l3) in 9 × 9 km latitude by longitude grids. This resolution was chosen since PAR data were not available at 4 × 4 km resolution. The ship-based underway measurements of DMS, temperature, salinity, and chla fluorescence were averaged into 9 × 9 km sampling bins along the cruise track for comparison with these remotely sensed data.  2.2.3 Separation of Coastal and Oceanic Regimes We used a statistical approach to separate our sampling transects into distinct open ocean and coastal hydrographic regimes, based on a computation of a normalized χ2 value. This normalized χ2, known as Cramer's V, provides a measure of statistical independence between two groups, and is inversely proportional to the likelihood that differences observed in two (or more) sample groups is attributable to chance71.  To compute Cramer's V, the χ2 statistic is normalized to the number of observations per sample to account for differences in sample sizes. Underway data from each cruise were separated into two groups, using arbitrary cut offs along the continental slope generated in 50 km increments, (the slope was designated as the change in depth from 120 m to 2700 m or ~ 25° W and 30° W on line P). Cramer's V was calculated for chla, calcite, PAR, and SST at all of these arbitrary boundaries to identify the longitude at which the region-specific variables became most statistically different. The Cramer's V statistics were averaged between June and August of 2007 and 2008, and the boundary between the open-ocean and coastal regimes was selected as the average distance where the maximum Cramer's V statistic occurred. Open-ocean and coastal regimes were treated separately in all subsequent statistical analyses.   19 2.2.4 Evaluating Global Empirical Algorithms Data from the open-ocean regimes on line P in 2007 and 2008 were used to test two global empirical DMS prediction algorithms in the NE Subarctic Pacific.  The first of these algorithms, proposed by Simo and Dachs30, predicts surface water DMS concentrations as a linear function of the chla to mixed layer depth ratio (chla/ MLD).  When the chla/ MLD ratio is < 0.02, the algorithm predicts DMS using a logarithmic function of the MLD.  The algorithm was developed using mixed layer depths (minimum depth where density > 0.125 kg m-3 greater than surface values) calculated from a monthly climatology, and satellite-derived chla.  For our study, we calculated the MLD using ship-based CTD profiles collected at 26 stations along the line P cruise track and applying the Δσt 0.125 kg m-3 criterion of Simo and Dachs30.  Recent work31 has applied a finer criterion to define the MLD, based on temperature differences of 0.1 °C from 5 m depth.  This MLD formulation is not appropriate for our study region given the importance of salinity in determining surface water density, and hence stratification intensity.  We also examined the relationship between DMS and the solar radiation dose (SRD) as proposed by Vallina and Simo31. These authors calculated SRD with the surface irradiance obtained from meteorological stations, an assumed underwater extinction coefficient of 0.06 m-1, and the mixed layer depth determined using CTD data with a threshold of 0.2 °C difference from the surface temperature. To approximate the seasonal irradiation dose, we used remotely sensed (SeaWiFS) 9 × 9 km PAR, the mixed layer depth calculated every 0.5° × 0.5°, and an average open ocean extinction coefficient of 0.083 m-1 estimated from CTD PAR profiles at Station P. For comparison with original publications, all measurements were binned to a spatial resolution of 1° latitude × 1° longitude. We also explored seasonal trends between the predictor variables and mean DMS concentrations from each cruise.  2.2.5 Stepwise Multiple Regression In addition to examining the published empirical algorithms discussed above, we used multiple linear regression analysis to generate the best-fit predictions of DMS concentrations from a suite of hydrographic and remotely sensed variables. We ran a stepwise multiple linear regression using a combination of underway measurements and remotely sensed  20 parameters for data from both coastal and open ocean regimes of every summer cruise. Predictor variables were sequentially added to the statistical model to obtain a minimum root mean square error. The analysis was run using native statistical functions in MATLAB.  2.2.6 Length Scales of Variability  To characterize length scales of DMS spatial variability, we followed a heuristic approach similar to that presented by Hales and Takahashi72.  This approach involves calculating interpolation errors associated with low-resolution sampling of a data time-series. Briefly, a high-resolution data set is sub-sampled with increasingly coarse resolution (representing longer distances between samples) and linearly interpolated back to the resolution of the original measurements. An interpolation error is then computed as the mean square error (MSE) between the original and interpolated data sets. The MSE increases predictably in proportion to the interpolation distance since closely spaced samples have surface properties that are more similar than distant samples. This relationship breaks down at some interpolation distance; the change in variance with distance (i.e., slope) generally decreases and the relationship between variance and distance becomes weaker, resulting in a lower correlation. This distance is defined as a characteristic length scale of variability, which is designated by a break in slope on log–log plot of MSE vs. interpolation distance. The calculated length scales of DMS variability were compared with the Baroclinic Rossby radii (R1) from the major oceanographic sampling stations, (P4, P8, P12, P16, P20, and P26). The Rossby radius defines the scale at which the ocean's rotational forces balance its pressure gradient forces in a stratified system, and this parameter controls the length scales of many physical oceanographic properties.  To calculate the local Baroclinic Rossby radius, the first wave mode phase speed c1 was divided by the Coriolis parameter (F) at each station.  The first modal speed m s-1 was calculated using the Dynmodes MATLAB code obtained from (http://woodshole.er.usgs.gov/operations/sea-mat/klinck-html/dynmodes.html), CTD-derived Brunt–Vaisala buoyancy frequencies (as a measure of stratification intensity), and CTD-obtained pressures with depth.   21 2.2.7 DMS at Ocean Station Papa To asses short-term temporal DMS variability, continuous DMS data were collected while the research vessel conducted sampling operations for > 12 h at Ocean Station P. Ancillary data on sea surface temperature (SST) and salinity, wind speed and surface irradiance (short wave radiation, SWR) were obtained from a NOAA mooring deployed at OSP (http://www.pmel.noaa.gov/stP/data.html), in addition to the ship-based thermosalinograph and chla fluorescence data described above, and data were binned 10 min intervals to match the temporal resolution of the mooring wind speed data. To ensure that mooring data and ship-based measurements were derived from the same water mass, we include in our analysis only data where the shipboard SST measurements agree to within 0.1 C of the mooring data. Although we are confident that mooring data and ship-based observations were derived from a common water mass, lateral advection of different water masses through our study site during the time of our extended occupation could have obfuscated observations of temporal and spatial variability at Station P. To examine the temporal dynamics of surface water DMS concentrations at Station P, we first used a stepwise multiple regression to assess the relationship between DMS concentrations and the ancillary mooring and underway measurements. Thereafter, we estimate the DMS sinks attributable to sea–air flux and photo-oxidation. We calculated both the instantaneous turnover rate of aqueous DMS concentrations, as well as the net daily average turnover rate. Instantaneous turnover rates were computed from the change in DMS concentrations for each 10 min interval our sampling period. Net daily turnover rates were estimated by taking the difference between the minimum and maximum DMS concentrations over a 24 h observation period). In addition to these simple computations, we also calculated the DMS burden (µmolm-2d-1), which is the DMS concentration multiplied by depth of the mixing layer of interest (~ 5 m), in order to assess the contributions of sea–air flux and photo-oxidation to temporal DMS variability in the top ~ 5 m. DMS sea–air flux (µmolm-2d-1) was calculated as the product of the aqueous DMS concentrations and the piston velocity using the wind speeds, SST and salinity described above. We estimated the rate of photo-oxidation in the top 5 m (µmolm-2d-1) according to the pseudo-first order relationship DMS/dt = [DMS]aq × KΓ. Photolysis constants (KΓ) were  22 calculated at OSP in July as reported in Bouillon et al.73, using data outside the Fe-enrichment zone during the SERIES experiment.  Photolysis constants were scaled to account for differences in mooring data SWR and the modeled SWR presented in Steiner et al.74, namely 65-level ECMWF model SWR generated using observed UVA, UVB, and PAR irradiance components, and differences in the extinction coefficient Kd between cruises. Because 90% of the photo-oxidation occurs at UV wavelengths27,73, the Kd values used here, which were calculated using wavelengths of PAR, were scaled to represent the extinction coefficients in the UV spectrum using the exponential fit of Kd across wavelengths presented in Bouillon et al.73.  Final photo-oxidation rates (µmol m-2 d-1) were corrected for temperature according to Toole et al.27.  While it has been shown that the quantum yield of DMS photo-oxidation is directly proportional to nitrate concentrations75, NO3− concentrations remain uniformly high throughout the summer in the NE Pacific55 and thus are not considered here.  2.3 Results 2.3.1 Spatial and Seasonal Patterns in Surface DMS Concentrations Our statistical analysis revealed a clear separation of variables between oceanic and coastal domains. All variables showed statistically significant coastal vs. oceanic differences (V > 0.50, p < 0.005), and maximum values for Cramer's V statistic were observed for all parameters along the continental slope (Fig. 2.1), corresponding to the approximate location of P5 on the Line P transect (~ 127.2° W), following the 2000 m isobath. Based on this result, we used the 2000 m isobath to define the boundary between the oceanic and coastal regimes. The ratio of DMS/calcite yielded the highest average Cramer's V statistics (V = 0.67 p < 0.005), indicating that this variable showed the largest coastal-oceanic difference. Surface seawater DMS maps (Fig. 2.2) further illustrate the distinction between the coastal and open-ocean regimes. There was a strong spatial concentration gradient from coastal to open ocean waters during the summertime cruises (we focus on the June and August cruises because DMS concentrations in February were uniformly low). The nature of this spatial gradient differed significantly, however, between June and August. In June of  23 2007 and 2008, average DMS concentrations were highest in the coastal zone (> 5 nM), while oceanic waters had the highest mean DMS concentrations in August (up to ~ 8 nM; Table 2.1). Despite these consistent trends, there was no statistically significant difference in the mean oceanic and coastal DMS concentrations, due to high spatial variability within each region (and correspondingly large standard deviations; see Table 2.1). Isolated high DMS features were apparent near shelf-breaks (~ 200 m isobath) in the coastal regime in both June and August (Fig. 2.2). In August of 2007 and 2008, high DMS regions also developed offshore with elevated DMS concentrations extending from ~ 130 to 145°W (Fig. 2.2). During August 2007, the highest concentrations were observed in the vicinity of Ocean Station P, while the transitional waters from the costal to the open ocean (~ 126–136° W) had the highest DMS concentrations in August 2008. The high spatial variability within the coastal and oceanic domains is illustrated more clearly in Figure 2.3, which presents line plots of DMS concentration along the cruise tracks in February, June, and August of 2007. During February 2007, the overall pattern of surface DMS distributions across the coastal–oceanic transect largely followed chla fluorescence, with several notable exceptions (e.g. ~ 2700 km along the cruise track; Fig. 2.3a). For both June and August 2007, DMS distributions were significantly more variable and only loosely followed chla fluorescence. The dominant feature of these summer DMS data are the high degree of mesoscale and sub-mesoscale spatial heterogeneity. For example, we observed a number of cases in which surface DMS concentrations changed ~ 4-fold (± 10 nM) over a distance of < 10 km. As expected, surface water DMS concentrations exhibited significant seasonal differences in both the coastal and oceanic regimes (Table 2.1). However, the apparent seasonal cycle was significantly larger in oceanic waters and temporally offset from the coastal seasonal cycle. From February to August, open-ocean aqueous DMS increased by nearly an order of magnitude (from ~ 1 nM to ~ 8 nM), peaking in August (Table 2.1). In contrast, coastal DMS concentrations reached a smaller maximum in June, and declined slightly through August (Table 2.1).  Overall, we observed the highest aqueous DMS concentration in the open-ocean regime during August.   24 2.3.2 Length Scales of Spatial Variability The six cruises with highly resolved measurements document mesoscale and sub-mesoscale heterogeneity in aqueous DMS. Using a simple statistical technique, we estimated the characteristic length scale for the four summer cruises, corresponding to the minimum spatial resolution necessary to fully describe DMS distributions. This characteristic length-scale is derived from the break in slope on a log–log plot of interpolation errors vs. interpolation distance (Fig. 2.4). For comparative purposes, we calculated variability length scales for several hydrographic variables (i.e. salinity and temperature) as well as for chla. As can be seen in Figure 2.5, the mean length scales of variability for temperature and salinity were very similar to each other (11 ± 2.5 km and 11.2 ± 2.6 km, respectively), and also to the calculated Rossby radius of deformation. This latter parameter provides a measure for the length scale of variability driven by physical dynamics. Superimposed upon this physically induced variability, biological processes produce additional heterogeneity in surface water properties, leading to shorter length scales of variability. Indeed, we found that the mean length scale for chla variability (3.3 ± 0.77 km) was ~ 3-fold lower than that of temperature or salinity. The length scale for DMS variability (7.4 ± 2.2 km) was intermediate between that of chla and the physical variables. In addition, the length scales for DMS variability during both August cruises (range = 4.8–6.6 km) were shorter than those in June (range = 8.4–9.9 km), which was not true for the other variables.  2.3.3 Sea-Air DMS Fluxes Across all of our cruises, mean sea air-fluxes, calculated with the formulation of Ho et al.68, ranged from ~ 5 to 25 (µmolm-2d-1) (Table 2.1). The fluxes did not exhibit the same clear seasonal cycle or coastal–oceanic differences observed for aqueous DMS concentrations (Table 2.1). The lowest sea–air fluxes were observed in February and June, although relatively high values were identified in the oceanic regime during February 2007. We observed the highest average sea-air DMS flux in August 2008 (24 ± 22 µmolm-2d-1, range = 1.5–110). Calculations using different parameterizations of the piston coefficient  25 yielded similarly large fluxes for August 2008; according to Liss and Merlivat76 (16 ± 16), Wanninkhof77 (22 ± 27), and Nightingale et al.70 (23 ± 21) µmolm-2d-1.  2.3.4 Empirical Predictive Algorithms We examined the relationship between DMS concentrations, and several biological and hydrographic variables by comparing ship-based thermosalinograph measurements and remotely sensed data interpolated to the cruise track with our underway DMS distributions. DMS concentrations did not exhibit strong pair-wise Pearson correlations with any of the variables we examined (r2 < 0.1, data not shown). For some cruises, however, we were able to obtain a strong empirical relationship between surface DMS concentrations and several predictor variables using a multiple linear regression.  Figure 2.6 illustrates the results of such an analysis for the August 2007 data. During this cruise, a combination of underway temperature and salinity measurements, remotely sensed PAR, chla, calcite, and calculated SRD and chla/MLD explained ~ 78% of observed open-ocean DMS measurements. In contrast, the same predictor variables explained much less of the DMS variance for other cruises, and regression coefficients for the August 2007 data were significantly different from those for other cruises. This indicates that the empirical relationship shown in Figure 2.6 has little general applicability as a predictive tool. Nonetheless, we did observe some general consistency in the relationship between underway DMS concentrations and several predictor variables in our regression analysis. For example, we observed a consistent positive DMS correlation with both chla/MLD and calcite in the oceanic region (Fig. 2.7a).  In the coastal region, we observed a positive relationship between underway chla fluorescence and DMS as well as a negative correlation between calcite and DMS for all cruises (Fig. 2.7b).   We further explored the trends in surface DMS concentrations across our six cruises using two published empirical algorithms. When measurements were examined in 1° × 1° bins (as done in the original publications) the empirical chla/MLD30 and SRD31 formulations did not provide a good fit to the data (r2 < 0.1 and 0.15 respectively; Fig. 2.8a and b). In contrast, MLD alone showed a significant negative relationship with DMS at this spatial scale (r2 = 0.41; Fig. 2.8c). However, when data were further aggregated into cruise averages, mean DMS concentrations were strongly related to chla/MLD (Fig. 2.9a).  We found that the  26 coefficients of the Simo and Dachs's global algorithm provided a good fit to the observed seasonal variance along line P and captured the overall regional trend across cruises (Fig. 2.9a; r2 = 0.47), although our regression equation DMS = 35 ⁎ chla/MLD + 1.8 provided an even better fit (r2 = 0.79). Note that our linear chla/MLD fit includes winter DMS data even though the chla/MLD is < 0.02.  Simo and Dachs30 suggest using a separate, logarithmic equation for data with chla/MLD in this range. The relationship we observed between DMS and chla/MLD (Fig. 2.9a) was mostly driven by variability in MLD since surface chla concentrations changed relatively little across cruises compared to MLD. Indeed, MLD alone was able to explain a large fraction of the seasonal variability in mean oceanic DMS concentrations among cruises, with a logarithmic relationship between these variables (Fig. 2.9b) (r2 = 0.82). Together, chla/MLD and MLD, provide a good prediction of all monthly cruise averages.  However, the coefficients of the DMS–MLD relationship were significantly different than the global empirical algorithm30. We also observed a strong seasonal relationship between the solar radiation dose31 and average aqueous DMS concentration for each cruise (Fig. 2.9c, r2 = 0.80). However, our data fit a trend line with a significantly higher slope than the empirical fit proposed by this second global empirical algorithm31. As discussed above for the CHL/MLD relationship, the relationship between surface DMS and SRD appeared to be driven primarily by seasonal changes in MLD. The open ocean region MLD decreased from June to August in both 2007 and 2008 by 56% and 60%, while changes in PAR levels were not statistically significant.  2.3.5 Variability at Ocean Station Papa We documented variability in DMS concentrations at OPS using high-resolution measurements (two-three data points per minute) for several days in June and > 20 h in August 2007. The high-resolution data demonstrate the rapid accumulation and disappearance of DMS over short time-scales (several hours) during both cruises (Fig. 2.10). The instantaneous rates of DMS turnover, calculated over successive 10 min intervals ranged from 10 nMd-1 to 18 nMd-1, whereas the net daily turnover rates (averaged over 24 h) ranged from 9.3 nMd-1 and 3.0 nMd-1 in June 2007 and August 2007, respectively.  27 In both June and August, the apparent temporal variability in surface DMS concentrations matched the daily cycle of incoming short wave solar radiation. In June, DMS concentrations rose sharply following the onset of daylight and declined rapidly prior to mid-day. However, this strong temporal pattern was only observed once during our multi-day occupation. In August of 2007 we observed a much smaller amplitude in the apparent temporal DMS cycle, with concentration changes on the order of ~ 3 nM occurring around mid-day. An important caveat in this analysis is the possibility that we sampled several different water masses during our occupation of Station P due to lateral advection. In this case, variability that is ascribed to the temporal dynamics of DMS production and consumption may result, in part, from spatial heterogeneity.  To fully separate spatial vs. temporal components of variability, it is necessary to conduct a LaGrangian drift study where a single water mass can be continually tracked. In the absence of such an approach, the apparent temporal variability we observed at Station P cannot be conclusively ascribed to a diel cycle. We note that that same limitation applies to the historic temporal sampling37 at Station P. Despite the caveat discussed above, we can still draw important conclusions from our time-series measurements at Station P.  We observed, for example, that a combination of available mooring data (wind speeds, irradiance, shown in Figure 2.10, and SST-data not shown) and underway data (chla fluorescence, SST, and salinity-data not shown) explained 62% of the observed change of DMS concentrations in the mixed layer in June 2007 and 35% in August 2007. However, these step-wise regression coefficients varied considerably between June and August, and thus offered little general predictive power.  Nevertheless, wind speed was consistently anti-correlated with DMS, and high wind speeds appear to have contributed to temporal dynamics of DMS, particularly in June 2007 (Fig. 2.10). We used simple calculations to examine the relative importance of sea–air flux and photo-oxidation in driving short-term DMS variability. Our calculations suggest that the sea–air flux exceeded photo-oxidation as a DMS sink term (Fig. 2.11).  During the sampling period in June 2007, sea-air flux removed total of 9.4 µmolm-2 from the DMS burden in the top ~ 5 m, while DMS photo-oxidation removed only 2.7 µmolm-2.  In August 2007, the two DMS sinks were 8.0 µmolm-2 and 5.1 µmolm-2, for sea–air flux and photo-oxidation, respectively.  Total DMS loss attributed to these processes was summed from rates calculated  28 in 10 min intervals. The difference between these two sampling periods reflects the more significant cloud cover, lower SST, and greater percentage of nighttime sampling in June relative to August, 2007.  2.4 Discussion Time series observations along line P suggest that long-term (decadal) trends in regional hydrography have influenced ecological dynamics in the NE Pacific. Changes in primary productivity and nutrient drawdown have accompanied warming surface seawater, increased stratification, and reduced vertical nutrient fluxes78. At present, the implications of these changes on surface ocean carbon balance are better understood than the potential impacts on the sulfur cycle and DMS emissions. Understanding the factors driving DMS variability in marine surface waters is critical for a comprehensive understanding of ocean-climate feedbacks. Our study contributes to a growing body of work documenting high DMS concentrations and sea-air fluxes in the Subarctic Pacific. The spatial and temporal patterns we observed in surface water DMS concentrations are consistent with those reported recently in a 5 year survey of the Line P sampling stations37, and with the recently updated global DMS climatology32.  Wong et al.37 reported mean DMS concentrations along the entire sampling transect of ~ 2, 6 and 10 nM for winter, spring and summer, respectively, in good agreement with our observations (Table 2.1). The dominant spatial gradient reported by Wong et al.37 was an increase in DMS concentrations from the coastal stations (P4 and P12) to the open ocean regime. Our data, which are derived from highly resolved spatial measurements, suggest that the longitudinal concentration gradient actually reverses seasonally, with the highest concentrations accumulating in offshore waters during late summer.  Lana et al.32,33 also report a similar pattern for the NE Subarctic Pacific. The DMS concentrations we measured, and those previously reported32,37, are somewhat higher than the values reported for the Subarctic Pacific in the global DMS compilation of Kettle et al.54. This suggests that previous authors54,79,80 have underestimated the importance of the Subarctic NE Pacific as a source of DMS to the atmosphere. Updated model results and observations suggest that the Subarctic Pacific contributes 15–30   29 µmolm-2d-1 for a prolonged period from June until September32,37. We observed maximum summertime fluxes of ~ 25 µmolm-2d-1 which are in very good agreement with these values. Using monthly averaged wind speeds in the Gulf of Alaska and our mean open-ocean DMS concentrations, we estimate the minimum regional flux with the Ho et al.68 formulation in June as ~ 4.8 and 6.2 in 2007 and 2008, respectively, while minimum values for August are ~ 13.7 and 14.9 µmolm-2d-1 in 2007 and 2008, respectively. These minimum values are comparable to summertime fluxes in the NE Atlantic, which peak in May and June32,79. Because the piston velocity depends on the square (or in some parameterizations, the cube) of wind speed, estimates based on aggregated wind speeds, such as those presented in Table 2.1, probably underestimate average flux81. We thus conclude that the NE Pacific is indeed a significant contributor to global sea–air DMS fluxes. The apparent spatial patterns in DMS concentrations along the Line P sampling transect result from underlying variability in physical and biological driving forces. In many respects, the Subarctic Pacific is largely characterized by two contrasting oceanographic regimes.  Summer upwelling fuels high primary productivity in the coastal waters, while chronic Fe deficiency limits primary productivity in the open ocean34,57. As a result, pronounced differences in phytoplankton species composition exist between these two regimes; diatoms are typically most abundant in the coastal regime82, while a combination of coccolithophores (particularly E. Huxleyi), other small haptophytes and dinoflagellates typically dominate the open-ocean regime56,83. It has been suggested that a wide continental shelf can be one important source of iron for the Northeast Pacific84, making mid shelf-break waters, a natural transition zone between an iron-limited oceanic regime and the productive coastal regime. Indeed we found that the 2000 m isobath (corresponding roughly to a mid-way position along the continental slope) was a defining boundary of the coastal and oceanic DMS regimes. It has been argued that DMS, DMSO, and other products of the DMS synthesis relieve cells of oxidative stress caused by Fe limitation, high UV radiation and other factors10. Fe limitation is most severe in offshore waters during August and September, leading to lower primary productivity and nutrient drawdown relative to the early summer55. Decreasing dissolved Fe levels, in combination with enhanced vertical stratification, also influence phytoplankton community structure, shifting the species composition towards  30 dominance by coccolithophorids and other small nanoflagellates that produce high intracellular DMS/P. Together, these physiological and ecological effects of Fe limitation are likely responsible for the accumulation of high DMS concentrations in the oceanic region during late summer (August). Increased surface water irradiance also likely influenced the high DMS levels in the late summer. Laboratory studies have demonstrated that increased UV light and photo inhibition resulted in elevated DMS(P) production (and higher rates DMSP cleavage) in the regionally prominent coccolithophore, E. huxleyi85,86.  In support of these observations, we observed a strong correlation between average DMS concentrations and the solar radiation dose in Fe limited open ocean waters (Fig. 2.9b) where haptophytes, including E. huxleyi, are abundant.  We thus suggest that a combination of increased iron stress and elevated mixed layer irradiance throughout the summer (related to increasing stratification) contribute to the strong seasonal cycle of DMS in open ocean waters of the Subarctic Pacific. The higher slope of our DMS vs. SRD relationship (Fig. 2.9b), compared to that reported by Vallina and Simo31, is consistent with the idea of additional concurrent oxidative stressors such as Fe-limitation contributing to the elevated aqueous DMS concentrations in the late summer. Coastal waters are not Fe limited and appear to follow a seasonal cycle akin to other phytoplankton bloom dominated regimes (chla concentrations peak in June, as oppose to August). In contrast to the oceanic regime, underway chla fluorescence, which is used as a proxy for phytoplankton biomass, was consistently correlated with DMS in the coastal region (Fig. 2.7). The localized high DMS spatial features observed in the coastal regime may thus reflect localized blooms of high DMS/P producing phytoplankton species, most often observed near the shelf break. Indeed, recent work87 has documented significant blooms of small (≤ 7 µm) phytoplankton in the transition zone waters, occurring near P4 in June 2008 (126.7° W). These blooms occur at the mesoscale and in some cases the sub-mesoscale, and appear to be caused by the mixing of nutrient rich open-ocean waters with Fe-rich coastal waters. The locations of these blooms correspond well with both the location and scale of DMS frontal structures in coastal waters (Fig. 2.3 and Fig. 2.7).  Spatial heterogeneity in coastal waters thus appears to be driven primarily by the physical dynamics that produce heterogeneity in nutrient supply and thus primary productivity.  Marandino et al.64 recently also conducted a high-resolution survey of DMS, documenting mesoscale spatial variability  31 in three distinct oceanographic regimes: upwelling, gyre, and subpolar; during their study, concentrations ranged 0–25 nM L-1. The highest DMS concentrations corresponded with the transitional boundary open-ocean sub-polar, and the productive coastal upwelling waters off the coast of Chile during a coccolithophore bloom, as indicated by remotely sensed maps of chla concentrations. Predicting global and regional DMS concentrations from simple hydrographic measurements is of significant importance for biogeochemical models including climate feedbacks88.  In our analysis, we were unable to produce a single relationship to explain the full DMS variability we observed in the NE Subarctic Pacific. Nonetheless, there were a number of statistical relationships between DMS and several predictor variables that warrant further discussion. The relationship we observed between aqueous DMS and calcite in the oceanic regime is consistent with previous suggestions that coccolithophores are high DMS and DMSP producing species. However, phytoplankton assemblages dominated by coccolithophores often include other non-calcifying nanoflagellate species, which also produce high DMS and DMSP concentrations, but cannot be readily detected by remote sensing. This may explain the inconsistency of the observed correlations between calcite and DMS concentrations. The development of taxon-specific satellite chla algorithms based on differential wavelength absorption signatures may prove useful in further elucidating the relationship between DMS concentrations and plankton assemblage composition. Preliminary work suggests that DMS:chla ratios are not closely coupled to phytoplankton taxonomic composition on a global scale89.  However, a better relationship may be observed within individual oceanographic regions. The chla/MLD ratio is correlated with average aqueous DMS because high DMS accumulation is associated with elevated chla in the coastal productive regime and with shallow mixed layers in the late summer in the offshore waters. Due to relatively low salinity of surface waters in the NE Pacific, a permanent pycnocline exists ~ 200 m, resulting in chla/MLD ratios that fit the linear trend proposed by Simo and Dachs. In the NE Subarctic Pacific open ocean regime, however, MLD appears to be a stronger predictor of surface water DMS concentration than chla/MLD (Fig. 2.8c and Fig. 2.9c).  A significant DMS–MLD correlation has also been reported by Simo and Dachs30 for > 80% of global DMS distributions, particularly for unproductive regions, such as the NE Subarctic Pacific.  32 Stratification controls nutrient and light availability, which in turn exert a strong influence on phytoplankton community structure. All of these variables directly or indirectly influence DMS emissions in the NE Pacific. Our results suggest that predictive algorithms derived from global databases can reproduce seasonal changes in DMS concentrations at an oceanographic basin-wide scale in the Subarctic Northeast Pacific. At finer spatial resolution, however, these algorithms do not accurately predict aqueous DMS variability in our study region.  Derevianko et al.53 have recently highlighted the limitations of global empirical algorithms for predicting regional DMS concentrations, and our analysis further underscore this point. The failure of empirical algorithms to accurately predict surface ocean DMS concentrations likely results from the underlying complexity of the DMS cycle, with complex biophysical driving forcing and trophic interactions. A number of studies have documented rapid turnover rates for surface reduced sulfur compounds in marine surface waters26,45,90–93 and diel cycling has been implicated specifically in driving DMS cycling at Ocean Station P. Our temporal observations support findings of rapid turnover and corroborate reported maximum daily changes in surface DMS concentrations on the order of ~ 10 nM37. Whereas the results of Wong et al.37 were based on a few measurements (usually three) made over a 24-hour period, our MIMS-based analysis of DMS variability at Station P is derived from extremely high-resolution data, permitting us to uncover unprecedented temporal dynamics.  One important caveat to our results is the possibility that advection of distinct water masses contributed to the observed variability that we ascribe to temporal dynamics. We documented net daily DMS turnover of 3–9 nM with turnover times ranging from < 1 day to > 3 days. This range of values is in good agreement with the results of Merzouk et al.36 who calculated an average turnover outside of the Fe enriched patch of 6 ± 3 nMd-1 at Station P and a turnover times between < 1 day and < 6 days (n = 4). Although we observed strong changes in DMS concentrations during our occupations of Station P, our data suggest that temporal variability does not adhere to predictable or recurring patterns. Indeed, over multiple days of sampling in June 2007, we did not observe repeated temporal patterns in surface water DMS concentrations (Fig. 2.10a).  Similarly, Wong et al.37 reported that increased DMS concentrations between pre-dawn and mid-day  33 were not observed on all cruises. A combination of predictor variables including wind speed, solar radiation, fluorescence, temperature and salinity partially explained observed temporal dynamics of surface water DMS concentrations, but accurate prediction of DMS temporal dynamics from simple hydrographic variables appears unlikely. As expected due to DMS loss in surface waters from atmospheric flux, higher wind speeds corresponded with lower DMS concentrations. Air-sea flux and photo-oxidation vary between being minor and significant terms in short-term DMS cycling, as noted during the SERIES experiment36,73. Our calculations suggest that, in June, sea–air flux and photo-oxidation only accounted for ≤ 25% of the large DMS ‘diel’ cycle. However, in August 2007, when we observed a smaller DMS cycle of 15 µmol m-2, a combination of sea–air flux and photo-oxidation explained > 87% of the DMS removal within the top ~ 5 m of the mixing layer. Both atmospheric flux and photo-oxidation were implicated in the removal of DMS during our sampling at Station P, although sea-air flux appeared to be a more consistent contributor to DMS removal in the top ~ 5 m. These contributions are subject to change even daily, however, based on fluctuations in both wind speed and incoming solar radiation. After accounting for air-sea gas exchange and photolysis, the remainder of DMS turnover must therefore be attributed to net effects of biological DMS production and consumption. To predict DMS at small spatial and temporal scales, the relationship between the biological turnover rates of DMSP and DMS and potential physical and biological driving forces merit further exploration. Phytoplankton growth and bacterial production are tied to changes in MLD and SRD, which influence DMS production and consumption92,94, but the quantitative nature of these relationships has been rarely studied.  To date, relatively few studies have quantified the kinetics of these biological processes26,46,95, and only one study has done so in the NE Pacific36.  Initial work has resulted in successful biogeochemical models of DMS emissions resulting from natural iron fertilization in the NE Pacific74. At present, inter-annual variability and short-term temporal variability at OSP remain difficult to model accurately. Our time series observations did not include several key components of the complex ecosystem dynamics between phytoplankton, zooplankton, and microbes that drive DMS dynamics, such as zooplankton and microbial species composition and zooplankton grazing and bacterial production rates. Although the MIMS provides a snapshot of DMS  34 distributions, rates of DMS production and consumption remain unknown during the 6 cruises. Because DMS dynamics are rapid and highly variable over both time and space, biological DMS production and consumption rates may provide the key understanding to successful DMS predictions at small spatial and temporal scales. The rate measurements needed to collect these data are laborious and time-consuming, however, and cannot be conducted with high frequency. Underway MIMS data could provide a useful tool for real-time monitoring of surface DMS concentrations to identify ‘hotspots’ for opportunistic sampling. In the future, the simultaneous application of real-time underway monitoring and focused process studies should provide significant insight into the factors driving DMS accumulation and cycling in marine surface waters.  2.5 Conclusions Our results document the basin-wide seasonal and spatial patterns in surface water DMS concentrations across the Subarctic NE Pacific, and provide new insight into the sub-mesoscale spatial variability and short-term temporal dynamics of DMS in this region. Data collected over 6 cruises confirms that both coastal and open ocean waters of the Subarctic NE Pacific are characterized by high DMS concentrations in the spring and summer, and that this region contributes significantly to global oceanic DMS fluxes. The regional seasonal and spatial patterns of DMS accumulation appear to be tied to Fe supply and mixed layer dynamics, which control the development of phytoplankton blooms across our study region. The combination of high solar radiation dose and Fe limitation may explain the particularly high DMS concentrations observed in off shore waters during late summer.  At a basin-wide scale, DMS concentrations the NE Pacific behave in a manner that is consistent with that observed in other oceanic regimes, and global empirical algorithms can potentially be applied in this system, with regionally specific coefficients. Superimposed on the regional spatial and seasonal temporal patterns in DMS concentrations, a variety of physical and biological factors drive significant variability at small length scales (i.e. < 10 km) and short-term timescales (i.e. hours). While sea–air flux and photo-oxidation can contribute substantially to the rapid apparent turnover of DMS and significant mesoscale and sub-mesoscale spatial heterogeneity in our study region, it is clear that biological processes are also important.  35 Quantifying the rate of biological DMS production and consumption in the Subarctic NE Pacific and examining the sensitivity of these processes to environmental conditions is a high priority for future research.    36 Table 2.1. Concentrations and Air-Sea Fluxes Average DMS concentrations and sea-air fluxes for each cruise.  Piston velocities were derived using daily averaged Quickscat satellite winds and the formulation of Ho et al. (2006). Errors represent standard deviations and ranges are shown in parentheses to show the natural variability in sea-air fluxes.  The separation between coastal and oceanic regimes is roughly defined by the 2000 m isobath (see methods for additional details).  Note that DMS concentrations in Feb. 2008 were in many cases below the MIMS detection limit, and are thus, not presented here.  Cruise Coastal  DMSaq (nM) Oceanic  DMSaq (nM) Coastal flux (µmol m-2d-1) Oceanic flux (µmol m-2d-1) February 2007 1.6±0.20 (1.0-2.7) 1.6±0.22 (1.2-2.6) 4.8±3.2 (1.1-12) 8.0±3.5 (1.2-22)  June 2007 5.7±2.2 (1.0-14) 2.7±1.8 (1.0-18) 8.1±5.3 (2.7-45) 5.7±5.8 (0.21-57)  August 2007 3.8±3.5 (1.0-17) 7.6±5.3 (1.0-20) 5.73±6.40 (0.15-36) 9.43±10.72 (0.13-80)  June 2008 5.9±4.0 (1.0-25) 3.6±1.3 (1.0-13) 16±13 (1.0-72) 4.9±6.3 (0.15-93)  August 2008 5.1±3.1 (1.0-29.4) 8.1±6.3 (1.0-69) 25±20 (2.6-130) 24±22 (1.5-110) 	     37   Figure 2.1. Definition of Coastal and Oceanic Domains Definition of coastal and oceanic domains based on the computation of normalized c2 values (Cramer’s effect size, V).  Larger values of V indicate a greater probability of statistical independence between two groups of observations.  We used this approach to determine the position of the coastal-oceanic boundary which best separated the data into distinct groups.  The figure shows an example of the results obtained for calcite, showing a maximum V value at ~ 127.2 ˚W, corresponding approximately to the 2000 m isobath.  Bathymetry along the transect is represented by the dashed line on the figure.    38  Figure 2.2. Spatial Distributions of DMS Spatial distribution of surface water DMS concentrations in (a) June 2007 (b) June 2008 (c) Aug 2007 and (d) August 2008.  The grey lines on the figure represent the 200 m and 2000 m isobaths.  The 2000 m is the approximate location of the coastal – oceanic boundary (see Fig. 1).     39  Figure 2.3. Underway Measurements in 2007 Underway surface measurements of DMS and chla fluorescence during a) Feb. 2007, b) June 2007 and c) Aug. 2007.  This figure demonstrates the mesoscale and sub-mesoscale spatial variability in these variables, and the seasonal accumulation of DMS in offshore waters.  Note that y scales are different in the three panels.  40  Figure 2.4. Deriving the DMS Length Scale of Variability Measurement interpolation errors as a function of sampling resolution along the August 2007 cruise track from coastal waters to Ocean Station P.  Underway DMS data were sub-sampled at increasingly coarse resolution (x axis) and root mean square errors (y axis) derived as the difference between interpolated data and observed measurements.  The break in slope observed at ~ 7 km (denoted by a vertical line on the figure) represents the characteristic variability length scale.    41  Figure 2.5. Comparison of Length Scales of Variability Variability length scales for DMS and other biological / hydrographic variables along Line P in June and Aug. of 2007, 2008.  Length scales were derived from the analysis of interpolation errors vs. sampling distance (see Fig. 2.4), and values represent means ± standard deviations to show the natural variability over seasons (i.e. spring vs. summer) and year.  42  Figure 2.6. Prediction Using Multiple Linear Regression Results of a multiple linear regression model predicting DMS concentrations from other biological and hydrographic measurements.  Data are from the August 2007 cruise.  The best-fit derived relationship is:  DMS = 6.3 * calcite -0.05 * SRD + 34 * CHL/MLD -3.2 * CHL+ 0.01 * PAR -2.8 * SST +2.3 * salinity -0.02 * fluorescence (r2 = 0.78).  43  Figure 2.7. Results from Multiple Regression Analysis Across Cruises Results from multiple regression analysis for all summer cruises (June / August, 2007/2008).  Plotted values represent the mean regression coefficient normalized to 1 standard deviation for each predictor variable for the a) oceanic and b) coastal domains to show the natural variability in the system. Although the absolute value of the regression coefficients differed significantly between cruises, calcite and the SRD were consistently positively correlated with DMS in the oceanic regime, while chla was positively correlated with DMS in the coastal regime. 44  Figure 2.8. Regional Empirical Algorithms Relationship between DMS and a) chla/MLD, b) solar radiation dose and c) mixed layer depth.  Data from the oceanic regime were averaged into 1° latitude by 1° longitude bins.  Data with chla /MLD not within 0.02-0.2 are excluded from panel (a) as suggested by Simo and Dachs (2002).  No strong correlation was observed between DMS and either a) chla /MLD (r2 < 0.1) or b) solar radiation dose (r2 < 0.15).  MLD explained significantly more of the variance in DMS (r2 = 0.41).    45  Figure 2.9. Seasonal Empirical Relationships As in Fig. 8, but data from were averaged for each cruise to show the relationship between seasonal changes in DMS and a) chla/MLD (r2 = 0.47), b) SRD (r2 = 0.80) and c) MLD (r2 = 0.82).  Error bars represent standard deviations of the means, showing the natural variability in DMS concentrations during each oceanographic cruise.  The dotted lines represent global algorithms of Simo and Dachs (2002) and Vallina et al (2007) predicting the relationship between DMS, chla/MLD and SRD.    46  Figure 2.10. Temporal Changes in DMS at Ocean Station Papa Temporal changes in surface water DMS concentrations at Ocean Station Papa in a) June 2007 and b) Aug 2007.  Data were collected several times per minute while the ship was holding station for extended sampling operations.  Also shown are measurements of incoming short wave solar radiation, and wind speed derived from a meteorological surface buoy deployed at station P (http://www.pmel.noaa.gov/stnP/data.html).  Note that the buoy was first deployed in June 2007 and no data were thus available during the early portion of our sampling on that cruise.    47  Figure 2.11. Calculated Rates of Air-Sea Flux and Photo-Oxidation Calculated rates of air-sea flux and photo-oxidation compared with the temporal changes in surface water DMS burden (µmolm-2) at Ocean Station Papa in a) June 2007 and b) August 2007.   48 3 Measurement of DMS, DMSO, and DMSP in Natural Waters by Automated Sequential Chemical Analysis 3.1 Introduction The reduced organic sulfur compounds dimethyl sulfide (DMS), dimethylsulfoniopropionate (DMSP) and dimethyl sulfoxide (DMSO) constitute an integral part of the marine sulfur cycle96, influencing microbial metabolism11, food web dynamics14, and (in the case of DMS) the atmospheric radiative balance97.  Many species of marine phytoplankton produce DMSP, DMSO, and DMS, and these compounds are subsequently released, degraded and recycled through a variety of ecosystem interactions between phytoplankton, zooplankton, and bacteria.  Uptake of DMSP, DMSO and DMS (abbreviated here as DMS/O/P) can fulfill bacterial metabolic needs for sulfur and carbon11, and DMSP may be incorporated by non-DMSP producing phytoplankton cells98,99.  Several studies have proposed that DMSP may function as an osmolyte8, an anti-oxidant10 and a cryoprotectant9. In addition, DMSP and DMS act as signaling molecules, attracting reduced sulfur seeking bacteria13, guiding seabirds to zooplankton aggregations14 and alerting coral pathogens to nearby stressed coral hosts100. Once ventilated to the atmosphere, DMS contributes to the formation of sulfur aerosols that directly reflect incoming solar radiation and promote cloud formation, thereby influencing regional atmospheric chemistry and climate101.  In the atmosphere, DMSO is an important intermediate of DMS oxidation, linking DMS emissions to climate feedbacks. The role of DMSO in the marine sulfur cycle remains poorly understood.  This compound is believed to be a dominant sink for DMS in the surface ocean, via biological and photochemical oxidation, and also a potential source of DMS through chemical and biological reduction3,21,22.  For over three decades, the use of purge and trap gas chromatography, coupled with flame photometric detection has enabled measurements of DMS, DMSP and DMSO at nanomolar concentrations in aquatic environments102–104. Although sensitive, accurate and precise, this method, as traditionally applied, is time-consuming and requires the collection of discrete water samples.  This inherently limits the spatial and temporal resolution of measurements, and makes it challenging to fully resolve mesoscale and sub-mesoscale variability in DMS/O/P concentrations.  In order to achieve higher spatial resolution data,  49 several shipboard systems have been recently developed for continuous high-frequency measurements of DMS.  These methods are based on the extraction of DMS from seawater across a semi-permeable membrane prior to analytical detection.  The sampling frequency of these methods ranges from several times per minute44,60 to several times per hour ,with detection limits from 0.2 to ~1 nM.  The growing database of high frequency DMS measurements from these instruments has provided new insight into the heterogeneity of surface ocean DMS concentrations on short temporal and spatial scales1,2,44,60,64,105–107. Despite significant advances in oceanic DMS measurements, no current systems exist for automated, high-throughput analysis of DMSP and DMSO.  The analysis of these two compounds is usually based on the treatment of samples with reagents to convert the solutes to DMS prior to gas-phase detection.  Simo et al.108 first introduced a method for sequential DMSP hydrolysis and DMSO reduction with borohydride for the determination of DMS/O/P in single samples.  This method requires ~7 hours per sample, including the complete hydrolysis of DMSP and the careful adjustment of sample pH, and is thus not well suited to high throughput analysis.  In general, the chemical conversion of DMSP and DMSO to DMS is typically conducted off-line, thereby increasing the analysis time for discrete samples.  Moreover, the reagents used for the chemical conversion (e.g. strong NaOH) may be incompatible with some membrane-based gas extraction systems.  The result is that DMSP and DMSO measurements are conducted independently of DMS, and are not readily amenable to high frequency, automated analysis. Here we describe the development of the Organic Sulfur Sequential Chemical Analysis Robot (OSSCAR), an automated method for field-based analysis of total DMS (DMSt), total DMSP (DMSPt) and total DMSO (DMSOt) in natural waters.  This method improves sampling resolution by combining the sensitivity and versatility of GC-based S detection methods with automated, sequential delivery of reagents needed for DMS/O/P analysis.  We show that OSSCAR functions well under oceanic field conditions, and is uniquely able to provide information on coupled DMS/O/P variability across significant biological and hydrographic features in surface marine waters.   50 3.2 Materials and Procedures 3.2.1 Overview Our analytical system is based on a combination of three existing methods for total (i.e. dissolved and particulate) DMS, DMSO and DMSP analysis (DMSt, DMSPt and DMSOt) using a purge and trap GC (PT-GC), with a pulsed flame photometric detector (PFPD).  We have automated the chemical treatment of samples to use these methods sequentially on a single water sample, converting DMSP and DMSO to DMS for analysis.  The system, (see schematic Fig. 3.1), uses a programmable syringe pump and valve (Kloehn, VersaPump 3) to deliver water and chemical reagents to its various components.  The syringe pump consists of a precise stepper motor connected with a gas tight syringe, and an internal 8-position sampling valve.  To ensure chemical compatibility with 10N NaOH, all system parts in contact with liquid are exclusively glass, Kel-F polychlorotrifluoroethylene, ceramic, or Teflon.   The typical analysis of a water sample proceeds as shown in Figure 3.2, and the individual steps are discussed fully below.  The analysis sequence consists of: 1) collection of a sample, DMS stripping out of solution and PT-GC analysis, 2) enzymatic reduction of DMSO to DMS and PT-GC analysis, and 3) DMSP hydrolysis to DMS and PT-GC analysis.  These three steps are carried out prior to acquiring another seawater sample and require ~1.5 hours.  Following DMS/O/P analysis on a single water sample, the syringe pump empties and rinses the contents of the sparge vessel to flush out any residual reagents (Fig. 3.2).  The entire analytical sequence is controlled by custom LabVIEW software.  3.2.2 Sample Collection The continuous analysis of surface seawater DMS/O/P relies on collecting samples from a ship's underway seawater supply, which is typically drawn from an intake at ~5 – 10 m depth.  In our system, we use ¼” polypropylene tubing connected to the ship’s seawater system as a continuous sampling line.  This sampling line is plumbed into a Swagelok stainless steel T-junction that is connected to one of the valve ports on the syringe pump.  The syringe pump withdraws samples from this T-junction and dispenses seawater into a sparging vessel (described below), where gas sparging and all chemical reactions take place.   51 In addition to the continuous underway-sampling mode, the system can sub-sample from discrete water samples (i.e. depth profiles).  In this discrete analysis mode, the syringe pump can be interfaced to an electronic port selector valve for automated batch processing of multiple samples.  3.2.3 DMS Stripping and Analysis The Kloehn syringe pump dispenses a 2.5 mL seawater sample (or standard / blank) into a custom-built glass gas tight sparge vessel (14cm high and 2cm diameter, beveled glass) fitted with a crimp topped Teflon-faced septum (Wheaton PN 224100-175), which is exchanged every ~2-4 weeks.  After dispensing water, the syringe pump flushes the transfer tubing with air to ensure that full volumes are dispensed into the sparge vessel and that the tubing is cleared of residual water or reagents.  Ultra high purity N2 (grade 5.0) is bubbled into the sparge vessel through 1/16" teflon tubing at a rate of 30 mL min-1 for ~11 minutes to strip DMS and other volatile gases from the sample.  DMS that is stripped out of the sparge chamber is adsorbed onto Carbopack-X packed in a 1/8” stainless steel trap held at room temperature (~ 24°C) by cooling with a 24VDC fan.  In the load configuration, DMS adheres onto the trap, while the effluent vents to the room, and the detector receives an N2 carrier flow of 2 mL min-1 (Fig. 3.1).  After the sparge is completed, a series of high current pulses within ~2 sec rapidly heats the 1/8” stainless steel trap to ~260°C to rapidly desorb gasses from the heated trap (Fig. 3.1).  Tests with a variety of blanks and standards demonstrate negligible adsorption of sulfur onto the heated trap.  High current is delivered to the trap from a transformer (Triad Magnetics model no. VPT12-20800, nominally 6 VAC 25 amp) connected to AC wall power via a solid state relay.  After the heating cycle is completed, the trap cools to ~30°C within ~20 seconds.  Gases eluting from the heated trap are passed onto a capillary column (Restek SS MXT, 15 m) that separates DMS from other volatile sulfur gases prior to detection by PFPD (see below).  The analytical GC column is maintained at 80°C using a custom-built column oven consisting of a temperature controller (Omega CN8200), an AC heating pad (Rapid Industrial Supply SH-3D-5-115) and a 12 VDC brushless fan, mounted in a box insulated with 1/8” thick tightly woven fiberglass material (FRP Woven Fiberglass GLWR4SR).  With this custom-built system, we are able to hold  52 column temperature to within ± 1°C of the set point, while minimizing bench space requirements. We use an OI Analytical PFPD (Model 5380; Fig. 3.3b) to quantify the sulfur eluting off of the GC column.  The detector is operated using the manufacturer’s recommended settings for sulfur analysis.   The PFPD has three operating gain ranges (1, 10 and 100), which offer different levels of signal amplification and analytical sensitivity.  We used range 10 to measure seawater DMSt and DMSOt, and range 100 (lower sensitivity) to measure DMSPt, which typically exists at higher concentrations in natural waters.  The range setting can be adjusted prior to a sample run, using an executable AutoIt script, which controls the PFPD software.  A custom LabVIEW 8.6 program runs this script and controls all sample handling, hardware operations, data collection and processing, including the preliminary analysis of the raw chromatograms.  Each chromatogram is archived by file name, and a log file records the date and time-stamp, and various ancillary information about each sample (i.e. compound analyzed, gain setting, sample volume, trap temperatures, peak area, height and elution time).  3.2.4 Catalytic Reduction of DMSO Following the sparging and analysis of DMS in the sample, DMSO is reduced to DMS using the DMSO reductase method of Hatton et al.109.  We add (via the syringe pump) a DMSO reductase (DMSOr) solution containing 25 µg DMSOr enzyme, 30mM ethylenediaminetetraacetic acid (EDTA) and 540 µM Flavin mononucleotide (FMN) supplied by Glycomar (DMSO Reductase Kit). Briefly, EDTA radicals (formed when the solution is illuminated) reduce FMN to FMNH2, which under semi-anaerobic conditions supplies DMSOr with the electrons necessary to catalyze the reduction of DMSO to DMS.  The transfer lines are flushed with air to ensure that the full volume of enzyme is dispensed into the sparge vessel and that enzyme solution does not remain in the lines.  For DMSO-reductase catalysis, the sparge chamber is lit with four high-powered (~750 Lumens) white light LEDs (Cree XML2) during the 20-minute reaction, under continuous N2 sparging (30 mL min-1) in the gas tight sparge-vessel.  The N2 sparging transfers all the DMS produced in the reaction chamber onto the trap, and also maintains the low O2 conditions necessary for the assay.  Because the DMSO reductase will denature at room temperature, we hold the  53 working enzyme solution at 2-3°C, using a Peltier air cooler (TE Technology, AC-027) connected to an Omega (CN7500) temperature controller.  The Peltier cooler is mounted on top of a custom-built aluminum enclosure where the enzyme solution is secured, and the Omega controller regulates power to the cooler in order to maintain the temperature set point.  Our cooling system eliminates the need for ice storage or a water bath to maintain the enzyme solution at low temperatures.  As discussed below, standard curves of DMSO reduction to DMS demonstrated complete conversion of DMSO in our system, indicating that we are able to maintain active enzyme for up to 5 days of the system’s unattended deployment.  3.2.5 DMSP Fast Hydrolysis DMSPt is analyzed in each sample after the DMSO reduction is complete.  For DMSPt analysis, the syringe pump removes half the liquid in the sparge vessel and dispenses it to waste.  In typical seawater samples, the concentrations of DMSPt are significantly higher than that of DMS or DMSO94,110, and this procedure is used to avoid saturating the detector with high DMS concentrations derived from DMSPt.  The addition of 3mL of 10N NaOH hydrolyzes the DMSP in solution to DMS in ~14 minutes according to Dacey and Blough66.  We allow ~22 min for the conversion of DMSPt in samples (and standards) to DMS.  A continuous sparge rate of 30 mL min-1 efficiently removes the DMS produced from this reaction and transfers it onto the trap for subsequent detection (Fig. 3.1).  We completed several tests to show that fast DMSP hydrolysis could be completed within thirteen minutes by testing reaction times between 1 and 25 minutes.  3.2.6 Sample Rinsing The sparge vessel needs to be rinsed free of any residual reagent, so that DMSO and DMSP are not converted to DMS in earlier steps of the DMS/O/P analysis.  Using the syringe pump, we rinse the sparge vessel twice with 14mL of MilliQ-water, and 5 times with sample (in the underway mode) or MilliQ-water (in the automated calibration mode, described below).   54 3.2.7 Automatic Calibration We calibrate OSSCAR using liquid standards prepared from commercially available DMS (Sigma-Aldrich PN 274380), DMSO (Sigma Aldrich D8418) and custom-synthesized DMSP derived from DMS and Fluka 3-bromopropionic acid using the method of Challenger and Simpson111 as employed by Dacey and Stefels5.  A two-step dilution in MilliQ-water facilitates the production of liquid standards in the nM concentration range.  Primary stock solutions contain 9.5 mM DMSP, 272 mM DMSO and 136 mM DMS in separate 100 mL crimped serum vials, which are stored in a fridge at 4 ˚C.  Prior to each full calibration curve (~ once per day), these primary stocks are used to make fresh intermediate stock solutions, containing 4.88 mM DMSP, 14.1 mM DMSO and 6.8 mM DMS, diluted in MilliQ in separate 100 mL crimped serum vials.  From these intermediate stock solutions, we prepare a working 500 ml standard solution containing 20 nM of DMS, DMSP and DMSO, which is plumbed into one of the sampling ports on the syringe pump.  Intermediate and working stocks were prepared in a well-ventilated fume hood, thereby minimizing exposure to DMS vapors.  During calibration, the working standard solution is automatically diluted, by varying the ratio of stock solution and Q-water added to the sparge vessel, to produce a range of DMS/O/P concentrations (0 - 20 nM).  Previous authors have reported occasional background DMSO contamination upwards of ~10nM in ship-based MilliQ-water systems109.  We found no DMSO contamination (i.e. less than 0.5 nM) in MilliQ-water from our UBC laboratory supply, and we thus prepared all intermediate and working standards using a 50L carboy of this UBC MilliQ-water. Full calibration curves are produced to check the detector response to sulfur, and also to quantify system blanks (the sixth standard in the calibration sequence consists of Q-water without any added standard).  In addition to these full calibration curves, we also run more frequent single point standards to monitor the efficiency of chemical conversion and DMS extraction.  After every 6 samples, we introduce a 20 nM standard for sequential DMS/O/P analysis as described above.  Full extraction / conversion of DMSO and DMSP is gauged by the similarity of peak heights for DMS, and DMSO in the single point standards.  Peak heights of DMS and DMSO are two-fold higher than the peak height of DMSP due to the lower volumes used for DMSP analysis.  55 We used a square root function (built into the manufacturer's PFPD control soft-ware) to process peak heights in calibration curves, in order to account for the detector's quadratic response in sulfur mode analysis.  3.3 Assessment 3.3.1 Optimizing Sample Analysis Table 3.1 shows the optimal settings we derived for various parameters in our analytical system, based on a variety of experimental trials.  OSSCAR operates best in temperature-controlled environments.  A low initial trap temperature (< 24°C) is essential to concentrate DMS onto the trap.  We have found that maintaining significant airflow over the trap when loading DMS (by keeping the fan running) is important when working in warm shipboard laboratories.  In contrast, final trap temperatures (i.e. post-heating) in excess of 300°C lead to degradation of the Carbopack trapping material and decrease the trap’s longevity.  We thus limit the total heating of the trap, by controlling the number of short (50 ms) electrical pulses used to heat the trap uniformly and desorb DMS.  In our system, we have found that the use of 6 pulses, separated by 100 ms produces optimal trap heating.   Given our small sample sizes (2.5 mL), we have found that 1.5mL of enzyme solution is sufficient to ensure 100% yield of DMS from DMSO (discussed below), although Hatton et al.109 recommend using 2 mL of DMSOr enzyme solution for samples less than 20 mL. Proper operation of the PFPD requires specific flow rates of H2, Air, and N2 carrier.  A constant N2 sparge rate and sufficient chromatographic separation times are also required for DMS/O/P sample recovery (Table 3.1).  In-line regulators maintain constant flow rates to the PFPD after initial set-up of gas flows.  Peak elution times, sample volumes and temperatures are set and recorded automatically through LabVIEW software.  3.3.2 DMSO and DMSP Sample Recovery A typical chromatogram of a DMS/O/P standard (Fig. 3.4) demonstrates full recovery of DMSO and DMSP.  Since all three compounds are converted to DMS prior to detection, the elution time remains consistent for all peaks (~5.3 min; Fig. 3.3).  Peaks are roughly symmetrical and are well resolved using a PFPD data acquisition rate of 5Hz (Fig. 3.4).  For  56 the sample shown in Fig. 3.4, quantification of peak heights (and the associated mol of sulfur) demonstrated DMSO-DMS and DMSP-DMS conversion efficiency of 100% and 97%, respectively.  Due to very small and inconsistent secondary peaks (e.g. Fig. 3.4), we used peak heights rather than peak integrals to measure sample recovery.  Additional tests were conducted to compare the DMSO and DMSP conversion efficiency of OSSCAR with standard methods based on overnight DMSP hydrolysis and TiCl3 reduction of DMSO to DMS.  Results of these tests showed good agreement between the methods, replicates within < 5% of each other (n = 6).  Repeated measurement of single point standards in the underway mode (see below) demonstrates that the system maintains a similarly high extraction efficiency during prolonged use.  3.3.3 Calibrations Full calibration curves, plotting the concentration of standards vs. peak height for all three compounds, were linear over two orders of magnitude (r2 = 0.99; Fig. 3.5).  As discussed above, calibration curves based on peak heights (r2 = 0.99) provided a better fit for our data than curves based on peak integrals (r2 = 0.97).  We fit a single calibration curve to DMS/O/P data.  Initially, we derived separate calibration curves for each compound; however, the slopes and intercepts across compounds were not statistically significantly different (Fig. 3.5).  We also calculated the yield (conversion efficiency) of DMSO and DMSP to DMS as the ratio of DMSO/P to DMS for each standard (0-24 nM).  Across all standards, we calculated a mean DMSO yield of 102% ± 3.7% and a mean DMSP yield of 101% ± 3.6% in our standards.  Full DMS calibrations on each gain setting (range 100, range 10 and range 1) were performed every 2-3 days during our field work to test for detector drift, which proved negligible (Fig. 3.6).  Replicate calibration curves, performed on different days for both gain settings used during this cruise (R10 and R100), showed slopes and y-intercepts that varied by less than 10% (Fig. 3.6).  3.3.4 Accuracy and Precision During this cruise, we tested the accuracy of our system by inter-calibrating DMS standards on OSSCAR with the shipboard flame photometric detector (FPD) operated by the  57 Institute of Ocean Science (IOS).  IOS participated in the Assessment and Qualifications Alliance (AQA) 12-23 DMS IN SEAWATER Proficiency Study in 2013, using this instrument, scoring top 10% for accuracy and top 5% for precision112.  DMS standards were prepared fresh by IOS marine technicians onboard the J.P. Tully and analyzed as described in Steiner et al.38.  Duplicate standards were run within 30 minutes on OSSCAR.  Prior to analysis, standards were stored in crimped glass serum vials without headspace and kept refrigerated at 4 C.  The comparison between our two systems yielded very good agreement, with a mean difference of 0.4 nM.  Although the values derived from OSSCAR were slightly higher than those measured by IOS, (Y = 1.01X + 0.44; Table 3.2), the difference was not statistically significant.  Precision of our system was based on analysis of replicate standard measurements, calculated as the standard deviation of replicates.  The mean standard deviation of all 12 replicate standard samples was 10%, with no standard deviations of individual duplicates exceeding 20%.  Unfortunately, a direct comparison for DMSP and DMSO standards was not possible because IOS does not measure DMSO, and small-volume 3.5mL DMSPt samples, stored for subsequent laboratory analysis are run using a separate instrument at IOS (M. Arychuk, pers. comm.).  3.3.5 Underway Analysis Using OSSCAR, we obtained automated, near real-time measurements total DMS/O/P along the Line P transect from open ocean waters at Station Papa (50 ˚N 145 ˚W) to the northern most point of Vancouver Island (50 ˚N 127 ˚W; Fig. 3.7a).  The fieldwork was conducted on board the CCGS J.P. Tully in August 2014 (cruise no. 19-2014).   We obtained ancillary measurements of temperature, salinity and chlorophyll a fluorescence from a shipboard thermosalinograph system (Fig. 3.7b), as well as acoustic measurements at a frequency of 120 kHz (Stephane Gautier and Chelsea Stanley, personal communication), and samples for photosynthetic pigment analysis (analyzed by HPLC at IOS).  We made additional high frequency measurements of surface DMS concentrations using our existing membrane inlet mass spectrometer (MIMS2; Fig. 3.7c). As shown in Figure 3.7c, we found good agreement between DMS measurements made with OSSCAR and those obtained using MIMS.  The exception to this occurred at one particular location (~128 ˚W) where MIMS-derived DMS concentrations were significantly  58 lower than those obtained by the PFPD (potential reasons for this are discussed below).  Across the transect, we observed significant hydrographic variability (Fig. 7b) and associated gradients in DMS/O/P (Fig. 3.7d).  We measured consistently high DMSPt (90.0 ±19.3 nM), DMSt (10.6 ± 6.0 nM) and DMSOt (7.0 ± 3.4 nM) concentrations in the oceanic waters of our transect (west of 138 ˚W).  Transitional waters between 138 ˚W and 131 ˚W contained moderate levels of DMSPt (43.2 ± 12.0 nM), and low DMSt (2.3 ± 1.6) and DMSOt concentrations (3.8 ± 2.8 nM).  The concentrations of all three sulfur compounds were most variable in coastal waters east of 131 ˚W (Fig. 3.7).  DMS (11.2 ± 13.7 nM), DMSOt (11.0 ± 11.2 nM) and DMSP (49.3 ± 33.4 nM) all showed maximum concentrations in the vicinity of Queen Charlotte sound, a region that is influenced by the Vancouver Island Counter Current (VICC; 50 ˚N~127 ˚W Fig. 3.7c).  This coastal region is typically characterized by strong gradients in sea surface temperature, salinity and Chla (Fig. 3.7), and has been identified as a DMS hot-spot in previous studies60,61.  Interestingly, an area of elevated Chla in the offshore waters at ~ 138 ˚W was not associated with significant increases in the concentration of sulfur compounds.   Phytoplankton species composition, phytoplankton physiology, and zooplankton grazing may help explain the variability in DMS/P/O concentrations across the oceanic-coastal transect.  We observed a correlation between photosynthetic pigment concentrations and DMS/O/P.  Between ~140 ˚W to ~128 ˚W the sum of peridinin and 19-Hex pigments correlated with DMSPt concentrations (r = 0.89, p <0.001; Fig. 3.8a).  Peridinin and 19-Hex are good indicators of dinoflagellate and haptophyte phytoplankton groups, many species of which produce high cellular DMSP concentrations15.  In addition, the sum of photo-protective pigments diadinoxanthin and diatoxanthin was correlated with DMSt (r = 0.87, p< 0.001, n = 10) and DMSOt (r = 0.88, p< 0.001; n = 10 Fig. 3.8b).  The concentrations of these two xanthophyll cycle pigments respond to increased irradiance, and have been used as an indicator for cellular photo-protective mechanisms113.  In addition to previous studies114, the correlation observed here indicates that total DMS/O concentrations may respond to light stress and is consistent with a role for DMS/O in photophysiology10.  Finally, acoustic measurements demonstrated a significant zooplankton migration event (shoaling of a strong 120 kHz acoustic signal) on the morning of August 30th that paralleled an offshore DMS  59 accumulation (144 ˚W – 140 ˚W).  The 120 kHz frequency signal represents aggregations of macro-invertebrates and micro-zooplankton (S. Gautier and C. Stanley, pers. comm.).  Along the entire transect, DMSPt accounted for the largest reduced sulfur pool.  Elevated concentrations of DMSt and DMSOt only occurred in regions with high background DMSPt concentrations (>80nM).  We found that DMSPt was correlated with both DMS (Type II regression, r = 0. 58, p < 0.01, n = 28) and DMSOt (r = 0.38, p < 0.05, n = 28), while DMS concentrations were well correlated with DMSOt (r = 0.9, p < 0.001, n = 30; Fig. 3.9).  3.4 Discussion We present the first shipboard method for automated sequential DMS/O/P analysis.  Our system includes automated calibration routines and blank determinations, such that underway sampling requires little attention, and data can be collected in near real-time with essentially no operator involvement.  Despite a considerably lower temporal sampling resolution than gas phase analysis systems such as MIMS, OSSCAR captures major DMS features, while also providing valuable information about DMSPt and DMSOt concentrations.  Values reported here (11.0 ± 11.2 nM) agree with existing published values115, and with DMSOt concentrations measured along line P in 2011 using the standard analytical TiCl3 method (10.5-35nM) (see chapter 5).  DMSPt concentrations (90.0 ±19.3 nM) are also well within the range of published values (50 to >100nM) at station Papa and along Line P, according to previous studies35,39.   At one location with high DMS concentrations, we observed a substantial difference between MIMS and OSSCAR measurements of DMSt (Fig. 3.7c).  In previous MIMS studies, we have noted the possibility of biological DMS consumption in the system’s Teflon tubing and standards, which could lead to a potential decrease in observed DMS concentrations.  In contrast, since the OSSCAR system involves sparging whole seawater, live material may release cellular DMS into the dissolved pool during analysis, leading to an increase in measured DMS concentrations. We found that acoustics-based observations of macro-invertebrate aggregations and HPLC pigment concentrations help explain the spatial variability in DMS/O/P, but we  60 observed no correlation between DMS/O/P and temperature or chla.  Previous studies have shown that copepods release DMS through phytoplankton grazing, leading to DMS accumulation17.  Our data suggest that concurrent DMS/O/P measurements, photosynthetic pigment samples and acoustics data may offer insight into different processes driving DMS/O/P accumulation.   Sequential DMS/O/P analysis on a single water sample facilitates inferences based on statistical correlations.  Our field data show, for example, that DMSt and DMSOt exhibit a positive correlation in both coastal and open ocean waters of the Subarctic Pacific Ocean in August.  At present, little is known about DMSOt distributions or seasonality in the Subarctic NE Pacific, with only a handful of published DMSO measurements currently available for this region115.  Nonetheless, previous studies36,73 have reported high photochemical and microbial turnover rates of DMS to DMSO in the Subarctic Pacific, suggesting that DMSO is a major sink for DMS near station P, and potentially, a minor source for DMS as well (see chapter five).   The tight coupling between DMS and DMSO we observed (Fig. 3.7) is consistent with a dominant production of DMSO from DMS.  To our knowledge, the data shown in Figure 3.7 represent the most comprehensive DMSOt measurements in the Subarctic NE Pacific to date. DMS and DMSPt are a core part of the Line P time-series program37 in the Subarctic Pacific region.  Using our new system, surface DMSOt measurements can now also be routinely added to this measurement program.  In addition to spatial surveys, OSSCAR is a powerful analytical tool for examining temporal variability in DMS/O/P concentrations over diel cycles in LaGrangian experiments.  Automated data collection facilitates the efficient use of ship-board personnel, and allows researchers to focus their efforts on more complex measurements, such as isotope-based tracer studies of S cycling3,95.  3.5 Comments and Recommendations Our measurements represent operationally-defined concentrations of total DMS/O/P.  The bulk of DMS and DMSO concentrations exist in the dissolved pool, with a small, but potentially variable, particulate pool.  In our system, sparging of live material may lead to cell breakage and the partial release of any intracellular DMS or DMSO concentrations.  To  61 the extent that cells remain intact (and particular DMS/O are not released into seawater), our measurements will represent something closer to the dissolved DMS/O pools.  Cell disruption could help to ensure complete release of particulate pools, but the requirement for an active DMSOr enzyme limits the methods available to accomplish this (e.g. heat would denature the enzyme).  The majority of the DMSP pool resides in the particulate phase.  Treatment of samples with strong NaOH (pH >14.2) converts dissolved DMSP to DMS, and should also lead to the lysis of phytoplankton cells.  Our DMSPt concentrations showed good agreement with previous samples (based on longer-term storage in NaOH), suggesting that we did recover the bulk of particulate pool.  However, it remains possible that we missed some fraction of the cellular DMSP in our relatively short alkaline hydrolysis.  Not withstanding these sources of our uncertainty, we believe that our measurements represent reasonable operational definitions of DMS/O/Pt concentrations.   Future method development will focus on decreasing the amount of time needed for sample analysis and thus increasing the temporal resolution of measurements.  In addition, we are currently exploring ways to automate the measurement of dissolved DMS/O/P pools, which are particularly critical to understanding the turnover of reduced sulfur compounds in marine environments.  Though technically challenging to implement, a slow on-line gravity filtration could separate particulate from dissolved DMS/O/P.  At present, analyzing DMS, DMSP and DMSO requires ~1.5 hours (Fig. 3.2).  Our instrument can be used, however, to measure any combination of DMS, DMSO and DMSP.  Simple analysis of DMS alone requires ~20 minutes, and the combination of either DMS or DMSP or of DMS and DMSO requires ~1 hour.  Analysis times for sequential DMS/O/P are limited by the reaction times required for DMSO catalysis, fast DMSP hydrolysis and rinsing.  The use of two parallel sparge vessels and purge and trap systems interfaced to a single GC and PFPD detector would increase our sampling frequency considerably.  Such an approach could decrease the analysis time for the full DMS/O/P cycle to ~50 minutes (i.e. ~ two-fold increase in measurement frequency). While the construction and operation of the OSSCAR system requires some technical proficiency, we hope that our detailed explanations and diagrams will be of use to those in the field who wish to pursue high throughput DMS/P/O analysis on research vessels.  We would emphasize that the front-end liquid sampling system can be readily interfaced to  62 existing, commercial GCs that investigators may be currently operating.  The essential aspect of our system is the use of a syringe pump and various electronic valves to automate the collection of samples, addition of chemical reagents and processing of standards and blanks.  This automated approach helps to resolve the surface water distributions of several key compounds in the reduced sulfur cycle.    63 Table 3.1. Optimizing Parameters for OSSCAR Optimal analysis times, volumes, flows and temperatures for OSSCAR.    Parameter   Set Point  Time (minutes) 12 for DMS sparge 25 for DMSO reduction 20 for DMSP hydrolysis  10 for chromatography   Volume (ml) 2.5 for DMS and DMSO 1.25 for DMSP 1.5 for DMSOr 3 for NaOH (10M) 14ml per rinse  (2 x 7 ml)  Gas flows (ml min-1)  11.2 H2 9.0 air 1.8 N2 carrier 30 N2 sparge Temperature (˚C)  Trap Initial < 24   Trap final 260   GC Column 80   Enzyme chamber (2 – 3)     64 Table 3.2. Comparison of OSSCAR and IOS Instruments. Comparison between DMS concentration measurements made by OSSCAR and by the Institute of Ocean Sciences (IOS). The mean value and standard deviation are reported to show the variability for OSSCAR measurements of IOS prepared standards.  IOS (nM) OSSCAR (nM) 2.0 2.4 ± 0.20 4.0 4.5 ± 0.27 8.0 8.5 ± 0.8 16.0 16.2 ± 1.0    65  Figure 3.1. Schematic Diagram of OSSCAR System Schematic diagram of the OSSCAR system.  A sample is loaded, either from an underway seawater supply or from a discrete sample, and DMS, DMSO and DMSP are extracted individually from the sample in the sparge vessel and analyzed sequentially using a GC - PFPD.  The Kloehn syringe drive ports 1-8 are used to pick up and dispense liquid to various parts of the system.   66  Figure 3.2. Sampling Time-Line for OSSCAR System Time-line of the steps in DMS/O/P analysis using OSSCAR.  Note that rinsing begins during the GC analysis of DMSP.   67  Figure 3.3. Photograph of OSSCAR Instrument. Photograph of the PFPD detector and custom-built GC on board the research vessel J.P.Tully in August, 2014 (a), a detailed photo of the PFPD detector set-up (b), and a photo of the liquid handling flow through system (c).    68   Figure 3.4. Chromatogram of a Single DMS/O/P Sample Using OSSCAR A chromatogram of an automated underway DMS/O/P calibration standard containing 20nM DMS, 20nM DMSO and 20nM DMSP.  Sample sizes followed our typical underway procedure for DMS/O/P analysis (Table 1).  The smaller volume of the DMSP standard analyzed explains the 2-fold difference in peak size.  To facilitate a visual comparison of peak sizes, all data were collected using the gain setting range 10.    69  Figure 3.5. Calibration Curve for DMS/O/P A calibration curve of DMS/O/P completed onboard the J.P. Tully during the August 2014 Line P cruise.  Calibration curves were linear across a range of concentrations (r2 ≥0.99), and standards show >98% yield from DMSP and DMSO.   70  Figure 3.6. Repeated Calibration Curves for DMS Repeated calibration curves of DMS for two PFPD range settings (R10 and R100) during the August line P cruise illustrate a drift of 5.6% in the slope of sensitive R10 calibration curves and 8.7% in the slope of R100 calibration curves over our 2-week cruise.  As a result, routine single-point standards and full calibration curves every 3-4 days are required.  R10 was used to calibrate concentrations up to 20nM, while R100 was used to calibrate concentrations up to 40nM in these standard curves.  Legends show the date of each calibration curve (month / day).   71  Figure 3.7. Underway Measurements of DMS/O/P and Ancillary Parameters Distribution of DMS/O/P along the Line P transect in the Subarctic Pacific Ocean in Aug, 2014.  Panel (a) shows the location of our transect between Ocean Station Papa (50 ˚N 145 ˚W) to the northernmost point of Vancouver Island (50 ˚N 126 ˚W).  Data were collected over a period of 2.5 days, and the location of individual samples for DMS/O/P is shown by the symbols in panel (a). Panel (b) shows the distribution of sea surface temperature (˚C) and chlorophyll fluorescence (mg L-1).  DMS concentrations, measured using MIMS and the OSSCAR (PFPD), are shown in panel (c), while DMS/O/P concentration measurements,  72 derived from OSSCAR, are shown in panel (d).  Total DMS/O/P concentrations measured in internal standards are also shown in panel (d).   73  Figure 3.8. Relationships between DMSO/P and Photosynthetic Pigments a) A plot of photosynthetic pigments vs. DMSPt and DMSOt along our transect.  DMSPt was significantly correlated with a combination of 19’-Hex and Peridinin (DMSPt = 713 x (19’-Hex + Peridinin) – 6.0; type II regression, r = 0.89, p <0.001, n = 10), while  b) DMSOt was significantly correlated with the Diadino and Diato pigments (DMSOt = 143 x (Diadino+Diato) – 2.6; r = 0.88, p < 0.001, n = 10).  The relationship in Fig. 8b remains significant (0.78, p<0.05, n = 9) despite the exclusion of a high point at ~128W, which exerts considerable leverage on the fit, and without which (DMSOt = 91 x (Diadino+Diato) + 0.3).  Note the relationship between Diadino and Diato pigments also remained significant, despite the exclusion of this point (r= 0.69, p<0.05, n=9).   74  Figure 3.9. Relationship between DMS and DMSO A plot of DMSt vs. DMSOt along our transect.  These concentration of these compounds were significantly correlated (DMSO = 0.81 x DMS + 0.6; type II regression, r = 0.9, p <0.001, n = 30).  Again, exclusion of the high point at ~128W modified the relationship (DMSO = 0.75 x DMS + 0.9; r = 0.75, p<0.001, n= 29) but did not alter the conclusion that DMS and DMSO were significantly correlated along the transect.   75 4 High Concentrations and Turnover Rates of DMS, DMSP, and DMSO in Antarctic Sea Ice 4.1 Introduction The biogenic gas dimethylsulfide (DMS) is produced by various biotic and abiotic processes in the surface ocean and ventilated to the overlying atmosphere where it undergoes oxidation to form aerosols that backscatter incoming solar radiation, influence atmospheric acidity and serve as cloud condensation nuclei116–118.  It has been suggested that oceanic DMS emissions may act as a biological climate feedback mechanism97 and recent modeling studies have predicted climate-dependent changes in the marine DMS cycle31,88,119–122.  The Southern Ocean is the largest natural source of DMS to the atmosphere32, contributing more than 50% of the total S aerosols in the Southern Hemisphere6,123.  Recent work has also documented high concentrations of DMS and the related compounds dimethylsulfoniopropionate (DMSP) and dimethyl sulfoxide (DMSO) in the sea-ice zone (SIZ) adjacent to polynya waters40,41,124,125.  At present, the spatial and temporal coverage of sea ice sulfur measurements remains severely limited. The dominant marine source of DMS is believed to be the enzymatic cleavage of algal-derived DMSP.  While some marine algae (including P. antarctica) can directly catalyze the breakdown of intracellular DMSP to DMS, field studies indicate that the majority of DMSP cleavage may be derived from bacterial metabolism of the dissolved DMSP pool45,92.  A number of studies have also demonstrated the potential for biological reduction of DMSO by marine bacteria and phytoplankton as a source of DMS12,21,22,126, though direct measurements of this process in oceanic waters are lacking.  Physiochemical conditions in Antarctic sea ice (near freezing temperatures, high UV, strong salinity and nutrient gradients) may lead to elevated DMS/P/O production by algae, given the hypothesized role of these compounds in cryoprotection, osmo-regulation, and as cellular antioxidants10,24,50.  However, the connections between these physiochemical conditions and high sea ice DMS/P/O concentrations in situ remain obscure.  76  4.2 Methods To examine the spatial distribution of DMS, DMSP, and DMSO across the Antarctic SIZ, we sampled 16 sea ice stations and one station in open waters of the Ross Sea polynya on a transit through the Amundsen and Ross Seas on board the Ice Breaker Oden during the middle of the Austral summer (December 2010 – to January 2011) (Fig. 4.1).  Stations in first-year pack ice, multiyear pack ice, and land-fast ice were sampled to determine DMS, total DMSO (DMSOt; i.e. dissolved and particulate forms), and total and dissolved DMSP (DMSPt and DMSPd) in sea ice brines, ice covered seawater, surface slush and melt ponds.   Samples were collected in Teflon PFA bags and measurements were made using a custom-built gas chromatographic system and capillary inlet quadrupole mass spectrometer. Samples for the measurement of DMS/P/O concentrations and turnover rates were collected from sack hole brines42, ice covered seawater, surface slush, and melt ponds using a Rule Pump (02 bilge pump 1500GPH).  Briefly, sack holes were drilled using an electric ice auger with a coring diameter of ~ 30 cm.  The sack holes were covered for 10-20 minutes, so that brine could percolate in from the surrounding ice walls while minimizing exposure full solar irradiance and inhibiting gas exchange.  For depth profiles, the auger was lowered to several pre-determined depths below the ice surface.  Snow and brine depth (cm) were measured from the sea ice surface with a meter stick.  After complete drilling to the bottom of the sea ice, we lowered the pump to the seawater – ice interface to sample seawater immediately below the ice.  In one case, we also collected surface seawater (~5 m depth) using the research vessel’s underway seawater system.  All samples were transferred into UV transparent (UVT), gas tight Welch Fluorocarbon 0.005” PFA bags, the headspace was removed and the bags were clamped shut with Teflon closures.  Samples were placed in opaque plastic bags and taken back to the shipboard laboratory for analysis within one hour of collection.  Prior to analysis, samples were stored at 4˚C. Ancillary measurements were made on ice cores drilled within <1m from DMS/P/O depth profile cores.  Chla was measured fluorometrically127 using a Turner Fluorometer 10-AU (Turner Designs, Inc.) in triplicate subsamples from 10cm ice core sections (7cm diameter), which were thawed in the dark in 2L of 0.2 mm filtered seawater128.  Samples were filtered onto 25mm GF/Fs, and the filters were placed in 5mL of 90% acetone for 24 h  77 extraction in the dark at 4ºC.  Filtration volumes ranged from ~25ml to 250ml, depending on the biomass of samples.  Chl a concentrations from depths ≤ the sack hole brine depth were averaged for comparisons between Chl a levels with DMSPt because sackhole brine percolates from overlying and adjacent sea-ice brine channels.  Ice temperature was measured immediately following core extraction on the ice and brine salinity was measured on board using a refractometer. The analysis of DMS, DMSPt, DMSPd and DMSOt concentrations was conducted using a purge and trap gas chromatographic separation method, coupled to a capillary inlet mass spectrometer (PT-CIMS). Briefly, 30 ml subsamples were transferred from the Teflon bags into gas tight vials with Teflon-faced butyl seals.  The vials were connected to a 16-position manifold valve (VICI Valco Instruments) and sequentially sparged for 5 min at a rate of 450 ml min-1 with UHP He to extract DMS onto a Carbopack-X trap held at room temperature.  When the sparging was complete, the trap was rapidly heated to ~210 ˚C to desorb DMS onto a fused capillary column.  The effluent from the column was introduced, via capillary bypass inlet, into the electron impact ion source of a quadrupole mass spectrometer (Hiden Analytical HAL 301), for detection using a secondary electron multiplier, with a voltage gain of 950 V.  DMSP samples were analyzed after purging background DMS out of solution for ~10 min with 450 ml min-1 of UHP He (until no DMS remained) and a subsequent > 6 hour alkaline hydrolysis to DMS in 1M KOH.  Samples for DMSPd determination were gently (~15 ml min-1) syringe filtered (Acrodisc 0.2µm) following the procedures described by Kiene and Slezak129 to minimize cell lysis.  DMSOt concentrations were measured using the TiCl3 reduction method to convert DMSO to DMS on samples stored at -20 ˚C for 6 weeks according to Kiene and Gerard47. Isotope tracer studies were conducted at five stations in sack hole brines to measure the production and consumption rates of DMS through various pathways.  Our tracer approach uses the simultaneous addition of DMS, DMSP and DMSO with different 2H and/or 13C signatures, which can be individually tracked during a short-term incubation experiment (Fig. 4.2).  This technique enables the simultaneous quantification of DMS derived from DMSP cleavage and DMSO reduction, gross DMS loss, and the net change in DMS concentrations (i.e. gross production-gross consumption).  DMS production from DMSP cleavage was measured as the rate of accumulation of 2H6-DMS from dissolved 2H6- 78 DMSPd, while DMS production from DMSO reduction was measured as the accumulation of 13C2-DMS derived from 13C2-DMSOd (Fig. 4.2).  Gross DMS consumption was measured by following the rate of depletion of deuterated 2H3-DMS (Fig. 4.2).  Rates of tracer production and consumption were scaled to the concentrations of total DMS, DMSPd and DMSOd in samples to approximate the in situ turnover rates of these compounds.   DMS produced from DMSP and DMSO in these experiments was analyzed as described above by PT-CIMS.  Concentrations of isotopically-labeled species (13C2-DMS, 2H3-DMS, and 2H6-DMS) were calculated by integrating chromatogram signals of different isotopic DMS species measured by peak jumping the mass spectrometer between m/z 62, 64, 65 and 68.  Final concentrations were calculated from standard curves using known concentrations of both unlabeled and labeled commercially available DMS with a working background detection of 0.1 nM for 30 ml samples.  Integrated peak areas were converted to DMS concentrations using standard additions of DMS prepared in deep seawater (>1500 m), with intermediate dilutions prepared in Milli-Q water.  Tracer production and consumption rates were quantified with linear regressions of averaged concentrations for 2-3 replicates (one for each bag that was incubated) over 4-5 time points and converted to nM d-1.  Rates that were not statistically different from 0 (with p > 0.1) were considered below the detection limit.  To scale the rates of tracer consumption and production to in situ values, the calculated rates were divided by the concentration of added tracers (yielding first order rate constants, d-1) and multiplied by the concentration of the natural dissolved DMS/P/O pools respectively.  Concentrations of unlabeled DMSOd and DMS were measured at the start of each incubation and DMSPd concentrations were inferred from our depth profile measurements.  Net DMSO turnover rates were calculated from the difference in 13C2-DMSO concentrations at the beginning and end of the incubations, and converted into nM d-1.  DMSO-DMS yields were then estimated as the fraction of DMSO converted to DMS.  The calculated DMSO-DMS yield are subject to significant uncertainty due error propagation and analytical variability.   The rates of change of deuterated and 13C-labeled stable isotope tracers are assumed to be representative of natural DMS cycling because isotope discrimination between species of DMS, favoring lighter isotopes, appears minimal < 10% (Asher unpublished data). We constructed a simple mass balance equation, which assumes that the observed change in the DMS pool equals the amounts produced by DMSP cleavage and DMSO  79 reduction minus the loss due to gross DMS consumption (losses due to gas exchange did not occur in our incubation experiments).  Thereafter, any imbalances resulting in unexplained net DMS production point to other unspecified (and difficult to trace) processes, most likely tied to DMS release from the particulate pool.  These processes include direct DMS excretion from particles, as well as the conversion of unlabeled DMSP or DMSO that has leaked from cells into the dissolved pool.  Rearranging the mass balance to solve for the unknown net source yields Eq.1. (1) d[DMSex]/dt = d[1H612C2-DMS]/dt + d[2H3-DMS]/dt – d[2H6-DMS]/dt –  d[13C2-DMS]/dt  The left hand term in the equation represents an unspecified source attributed to particulate DMS release.  The right hand terms in the equation represent the net change in DMS concentrations, gross DMS consumption, the production of DMS from DMSP cleavage, and the production of DMS from DMSO reduction.  4.3 Results Our measurements revealed considerable variability in the concentrations of DMS, DMSP and DMSO both across sampling sites (Fig. 4.3), and with depth at individual sampling locations (Fig. 4.4).  Mean DMS, DMSPt, and DMSOt concentrations in sea ice brines were 33.2 ± 4.78 nM (std. err.), 305 ± 54.8 nM and 138 ± 14.5 nM, with maximum values of 277nM, 2990nM, and 471nM, respectively.  The concentrations of DMS/P/O were significantly (at least five-fold) higher in sea-ice brines than in the ice-covered seawater (Fig. 4.3; t-test, t-stat ≥ 9.5, p < 0.0001).  Given our sampling technique, our values may underestimate the particulate DMS/P/O fraction present in ice brines due to the adhesion of particulates on brine channel walls40,42, leading to even greater differences between brine and seawater concentrations, which were uniformly low. We observed significant phytoplankton biomass in sea-ice and brines typically exceeding 10 µgL-1 Chla and, in several cases, exceeding 100 µgL-1 in conjunction with high concentrations of methylated S compounds (Fig. 4.3).  The concentrations of DMS/P/O were generally highest in brines collected near the upper ice surface, decreasing with ice core depth (r2 = 0.14, p < 0.005; Fig.4.3, Fig. 4.4).  This depth-dependent distribution was  80 particularly evident for DMSOt and DMSPt, which is consistent with increased photo-oxidation of DMS to DMSO in high light regimes and the hypothesized role of reduced S compounds in UV protection of ice algae24,50.  In multiple regressions, DMSPt and DMSOt concentrations explained the majority of variance in DMS across our samples (r2 = 0.66, p < 0.001; data not shown), while DMSPt concentrations were best predicted by a combination of Chla concentration and core depth (r2 = 0.51, p < 0.001; data not shown). Using our isotope tracer approach, we observed significant (p < 0.05) rates of gross DMS production from multiple sources in sea-ice brines, concurrent with net production of unlabeled DMS.  Fig. 4.5 shows time-course data for one experiment (station 27) in which we measured a net DMS production rate of 17 ± 3.5 nMd-1 in a sea ice brine sample (Fig 4.5a).  Gross DMS consumption, which includes the biotic consumption of DMS as well as the photo-oxidation of DMS to DMSO (measured the disappearance of 2H3-DMS) reached 88 ± 20 nMd-1 (Fig. 4.5b).  Summing net DMS production and gross DMS consumption yielded a gross DMS production rate of 105 ± 24 nMd-1 (i.e. 17 + 88).   Our direct measurements show that 25 ± 6.9 nMd-1 DMS was produced from DMSP cleavage at this station (Fig. 4.5c), while 96 ± 26 nMd-1 was produced from DMSO reduction (Fig. 4.5d).  In this experiment, the measured rates yield a mass balance of production and consumption terms within the error of our measurements.  As discussed below, results from other sampling stations did not yield an exact mass balance, suggesting that direct DMS release from the particulate biological pool (which we did not measure directly) contributed significantly to gross DMS production. Similar results to Figure 4.5 were observed in sea-ice brines sampled across our study region, i.e. rapid DMS consumption and production through various pathways (Fig. 4.6).  In many cases, DMSO reduction rates (range 75-210 nMd-1; mean 150 ± 22 nMd-1) dominated DMS production in brine samples (Fig. 4.6), with rates > 3-fold higher than DMSP cleavage (range 21-62 nMd-1; mean 37 ± 6 nMd-1).  DMSO yields varied from 20%-100% with a mean of 45%, although these values are subject to significant uncertainty (see supplement for details).  Although DMSO reduction was the dominant measured source of DMS in most of our tracer experiments, we estimated a high (> 50 nMd-1) apparent particulate release of DMS at several ice stations based on the mass balance approach presented in Eq. 1.  These high DMS release rates, which may reflect algal lysis, account for the large variability seen  81 in the inferred biological DMS release in Figure 4.6.  Rates of gross DMS consumption and production were significantly higher in sea ice brines (range 57-250; mean 160 ± 33) than in underlying ice-covered seawater.  Intense DMS cycling resulted in higher net DMS production in the sea ice brines (93 ± 48 nMd-1; Fig. 4.6) relative to the ice covered seawater samples (3.5 ± 13 nMd-1). Although the highest rates of DMSO reduction were observed in sea ice slush and brines, we also observed significant DMSO reduction rates (34 ± 16 nMd-1) in ice-free waters of the Ross Sea polynya (Station 41) sampled in mid January.  These rates of DMSO reduction were considerably higher than the rates of DMSP cleavage (8.4 ± 1.1 nMd-1).  We estimated a ~30% DMSO yield at this site due to high DMSO turnover (~100 nMd-1), providing further evidence for the widespread importance of DMSO as a source of DMS.  In contrast to the ice brine samples, however, we observed very small net changes in DMS concentrations in our Ross Sea samples.  Correlations between rate constants (d-1) of key DMS production pathways in the Ross Sea polynya and sea ice brines indicate rapid DMS/O cycling and ties between particulate release of DMS and DMSP cleavage in both environments.  Rate constants of DMSO reduction and DMS consumption were closely correlated (r2 = 0.92, p < 0.01) as were DMSP cleavage and the apparent release of DMS from the particulate pool (r2 = 0.77, p < 0.05).  4.4 Discussion Our results indicate that active microbial cycling in Antarctic sea ice leads to the accumulation of high DMS concentrations due to rapid DMSO reduction, DMSP cleavage, and in some cases the likely release of DMS from the particulate biological pool.  We present novel DMS turnover rate measurements in sea-ice samples, and the first direct quantification of oceanic DMSO reduction to DMS (as opposed to net changes in DMSO concentrations).  While DMSP has typically been considered to be the main oceanic source of DMS, our data suggest that rapid biological DMSO reduction dominates DMS production in varied Antarctic sea-ice environments.  The abiotic disproportionation of DMSO, which forms DMS and dimethyl sulfone, has been observed in aerobic freshwater systems.  However, the maximum rates of this reaction (<1 nMd-1) are considerably lower than the DMSO reduction  82 rates reported here, and negligible in comparison to the biological reduction of DMSO to DMS observed in freshwater systems, where DMSO competes with oxygen as an electron acceptor and DMSO reduction rates are well correlated with bulk bacterial respiration130–132.  Bacterial DMSO-reducing enzymes are widespread and appear to be bound to the cellular membrane12,23,133,134, facilitating rapid biological DMSO reduction130,131.  Recent work with laboratory cultures also suggests that phytoplankton may also be an important source of biological DMSO reduction to DMS in aerobic marine environments21. Our data indicate an important role of DMSO reduction as a major source of DMS in marine systems and rapid DMS/DMSO cycling where DMSO concentrations are high.  At present, we are unable to evaluate the relative contributions of sea ice bacteria and algae to DMSO reduction. DMSO concentrations are particularly high in the Antarctic sea ice zone, likely due to high rates of biotic production, photo-oxidation of DMS, and possible deposition in snow90,135,136.  Although in situ biological production of particulate DMSO has thus far not been examined in the Southern Ocean, this process has been observed in a variety of phytoplankton and bacterial species as well as in natural plankton communities110,137,138.  Both DMS and DMSO have been suggested to function as cellular anti-oxidants, and it is hypothesized that DMS oxidation to DMSO may play a role in cellular photo-protection10,24.  We thus suggest that rapid redox cycling between DMS and DMSO plays an important role in photo-protective mechanisms of Antarctic microbes, and accounts for the exceptionally high concentrations of these compounds in the Southern Ocean SIZ.  Future work may determine if rapid DMS/P/O cycling occurs in slushes and melt ponds as well as sea ice brines, as suggested by similarly high DMS/P/O concentrations we observed in these samples.  Additionally, our work and that of others24,90,139 has documented high DMS/O concentrations and turnover rates in ice-free Antarctic polynya waters, particularly under late summer conditions of high solar irradiance and mixed layer stratification, suggesting that our observations may be relevant beyond the SIZ.  4.5 Conclusions Recent modeling studies reporting large climate-dependent changes in Southern Ocean DMS emissions have not explicitly included the potential influence of the vast SIZ.   83 Our results and those of several recent studies indicate, however, that the SIZ may contribute significantly to DMS cycling in the Southern Ocean.  While it has typically been assumed that gas exchange is severely limited in sea ice, significant DMS fluxes have recently been measured over snow and ice covered waters125.  The SIZ, by virtue of its active microbial populations, high DMS concentrations and vast areal extent could play a significant role in climate-dependent DMS feedback mechanisms.    Furthermore, melt-induced surface water stratification could lead to greater DMS/DMSO cycling in high irradiance surface waters as we observed in the Ross Sea polynya.  In order to predict the climate-sensitivity of the Southern Ocean DMS cycle, it is thus critical to understand the factors controlling DMS cycling in the Antarctic water column and the SIZ, and their sensitivity to various environmental perturbations.    84  Figure 4.1. Map of Sampling Stations in the Southern Ocean Map of sampling stations across an east to west transect through the SIZ of the Amundsen Sea and Ross Sea between December 16, 2010 and January 10, 2011.  Station numbers appear next to station locations.  Green triangles denote stations where isotope tracer experiments were conducted in addition to DMS/P/O measurements.  Red circles denote stations where only concentration measurements were made. The background black and white color scale shows the mean sea ice concentrations during the period of our survey derived from the AMSR-E satellite.   85  Figure 4.2. Schematic Diagram of Tracer Method Schematic diagram of the DMS tracer method.  Tracers, shown in gray, green, and purple are added to the ambient DMS, DMSOd and DMSPd pools, respectively, and detected in the in DMS pool.  Labeled black arrows denote measured production and consumption processes affecting each isotopic mass of DMS.   86  Figure 4.3. Boxplots of DMS/P/O Concentrations Box plot summarizing the DMS, DMSPt, DMSPd, and DMSOt concentrations based on sample type across 16 ice and seawater sampling stations.  Sample types (shown on the y-axis) are in order of relative sample depth.  On each box, the central mark represents the data median, while the edges represent the 25th and 75th percentiles (n ≥ 92).  The error bars represent the range of data, with ‘+’ signs denoting outliers 1.5 times beyond the inner quartile range (IQR).     87  Figure 4.4. Depth Profiles of DMS/P/O Concentrations Depth profiles of methylated S compounds at 12 sea ice stations with chlorophyll a concentrations (in red) for comparison.  Blue squares and lines denote DMSOt concentrations, while black stars represent DMSPt (plotted along the top x-axis).  Green diamonds and lines represent DMS and magenta circles represent DMSPd, which are plotted along the bottom x-axis.  Dashed horizontal lines on the plots show the sea-ice/seawater interface, and station numbers are marked in the bottom right of each subplot.  Note that all subplots do not have identical x-axes.  Error bars represent the standard error.   88  Figure 4.5. Time Course Data of Tracer Experiment Time course data showing the change in concentrations of different isotopically-labeled DMS species used to calculate consumption and production rates at station 27.  The net change of unlabeled DMS (mass 62) is shown in panel (a).  Gross consumption is calculated from the 2H3-DMS (mass 65) tracer depletion rate (Fig 2b).  DMSP cleavage is calculated from the accumulation of 2H6-DMS (mass 68) (Fig 2c), and DMSO reduction is calculated from the accumulation of 13C2-DMS (mass 64) (Fig 2d).   Error bars represent the standard error of triplicates.  These tracer turnover rates were subsequently scaled using natural DMS/P/O concentrations to calculate in situ production and consumption rates.  At this station 25±6.9 nM d-1 DMS was produced from DMSP cleavage, 96±26 nM d-1 DMS was produced from the DMSO reduction, and DMS was consumed at the rate of 88±20 nM d-1.   89 The net DMS production rate was 17±3.5 nM d-1.  Error bars represent the standard error from the mean.   90  Figure 4.6. Summary of Tracer Experiments Summary of DMS production and consumption rates measured in isotope tracer experiments in sea ice brines (station 27, 30, 31, 43), ice-covered seawater (ICSW) (at stations 31 and 43) and ice-free seawater from the Ross Sea polynya (Station 44). DMS release from the particulate biological pool is calculated from the mass balance of measured production and consumption rates as in Eq. 1.  Bars show the mean of all rate measurements and error bars represent standard errors of the means.   91 5 Concentrations, Cycling of MDS, DMSP, and DMSO in Coastal and Offshore Waters in the Subarctic Northeast Pacific During Summer, 2010-2011  5.1 Introduction Dimethylsulfide (DMS) is a biogenic sulfur compound derived from the algal metabolite dimethylsulfoniopropionate (DMSP) in marine surface waters.  Lovelock et al.96 revealed an important role for dimethyl sulfide (DMS) in the global sulfur budget, stimulating decades of subsequent research into the oceanic cycling of this compound.  A number of studies have addressed the potential role of DMS in climate regulation72,97,140, as a source of sulfate aerosols that promote cloud formation and backscatter incoming solar radiation.  While the importance of DMS in the global radiative budget remains under debate, it is now firmly established that this compound (along with dimethylsulfoniopropionate, DMSP, and dimethyl sulfoxide, DMSO) plays an important role supporting the metabolism of many marine microbes, as a key source of reduced carbon and sulfur11,141,142.  DMS and DMSP may also play important roles in chemotactic attraction for predators13,14, thus providing a biogeochemical link across different trophic levels of the marine ecosystem.  Biogeochemical processes and ecological dynamics influence surface ocean DMS concentrations over a range of spatial and temporal scales.   Global-scale oceanographic databases have been used to develop empirical algorithms correlating DMS concentrations with a variety of biophysical variables, such as the ratio of chlorophyll to mixed layer depths30 (MLD), ultraviolet radiation (UV) and the solar radiation dose10.  On a regional scale, time-series observations have increased our understanding of the temporal dynamics of DMS/P/O with respect to environmental forcing.  The most comprehensive DMS/P/O time-series observations have been conducted in the Sargasso Sea as part of the Bermuda Atlantic Time Series (BATS) program102,143, which has documented moderate seasonality and inter-annual variability in surface DMS concentrations (range 1-7 nM) in sub-tropical waters of the N. Atlantic.  These dynamics appear to reflect changes in the net balance of bacterial and algal DMS consumption, and in the activity of DMSP lyase, an enzyme that produces DMS  92 during the cleavage of DMSP.  Levine et al.143 proposed that UV light functions as an environmental switch regulating biological DMS production and consumption at BATS.  The apparent UV-dependent cycling of DMS at BATS is consistent with global-scale relationships between DMS and solar intensity and mixed layer stratification.  Despite the significant insight gained from time-series work at BATS, processes driving the seasonal cycle of DMS and related compounds in sub-tropical waters are not necessarily indicative of those occurring in other oceanic regions.  In particular, several oceanic DMS 'hot-spots' have been reported in polar and sub-polar waters, where extremely high DMS concentrations (> 20 nM) are observed during the phytoplankton growing season.  These high DMS regions, which include a number of Antarctic polynyas107,139 and open ocean waters of the Subarctic Pacific37,38, contribute disproportionately to global sea-air DMS fluxes32.   Over the past decade, the Line P time-series program has documented seasonal and inter-annual variability in surface water DMS concentrations in the Subarctic Pacific.  Surface waters in this region are characterized by significant spatial variability (over a range of length scales) and large potential inter-annual variability38.  The region comprises two distinct ecological provinces; a high productivity coastal regime, and an iron (Fe)-limited offshore regime, with persistently high nutrient and low chlorophyll (HNLC) concentrations55,57,144.  Maximum concentrations of DMS in excess of ~ 20 nM (~ 10-fold higher than the global average) have been observed in HNLC open ocean waters along Line P during the late summer and early fall.  Results from the SERIES in situ Fe enrichment35,36 suggested that iron limitation is a dominant factor driving high DMS concentrations in these offshore HNLC waters.  Results from this experiment showed that Fe enrichment indirectly decreased DMS concentrations by altering the community composition of phytoplankton from DMS/P producing nanoflagellates to large diatoms.  This result is consistent with laboratory studies showing elevated DMS/P levels in Fe-limited phytoplankton cultures10.  Moreover, the SERIES iron fertilization stimulated bacterial consumption of DMS and DMSP as sources of sulfur, carbon, and energy, leading to a sharp decrease in bacterial release of DMS.  In conjunction with rapid biological cycling, physical processes, such as entrainment, sea-air exchange and photo-chemical oxidation of DMS to DMSO have been  93 shown to drive temporal changes in DMS at Ocean Station Papa145, the western most station along the Line P transect.   Continental shelf waters along the British Columbia coast have also attracted recent attention as a seasonally strong source of DMS61,106,146. These productive coastal waters are characterized by complex shelf bathymetry, summertime wind-driven upwelling and the presence of several strong near shore current systems, namely, the southbound California Current and northbound Vancouver Island Coastal Current (VICC)147.  In this coastal region, physical dynamics lead to strong spatial and temporal variability in nutrient supply, which drives significant variability in phytoplankton biomass and productivity.  This variability can, in turn, lead to strong temporal and spatial gradients in the surface water concentrations of DMS and other biogenic gases61,106,148.  Despite recent progress towards documenting the spatial and temporal variability of DMS concentrations in the Subarctic NE Pacific1,61,106, fundamental gaps remain in our understanding of the underlying processes of DMS production and consumption.  In general, dissolved DMSP (DMSPd) cleavage is considered to be the main pathway for oceanic DMS production, and this process can be mediated either by phytoplankton that possess extracellular DMSP lyase, or by bacteria acting on the dissolved DMSP pool in seawater.  Abiotic photo-oxidation of DMS to DMSO, and biological (bacterial) DMS consumption can both contribute significantly to DMS removal, with their relative importance depending on solar (UV) intensity, mixed layer depth and the activity, taxonomic composition and sulfur requirements of bacterial assemblages.  To date, two studies have employed radio-isotope 35S labeling methods to quantify biological DMS consumption36 at Station Papa (P26) and DMS production from DMSP cleavage39 at major stations along Line P.  These studies have demonstrated a considerable range of DMSP cleavage and DMS consumption rates, with lower DMS production and consumption observed in offshore HNLC waters.  While this work has provided important information on DMS production / consumption processes, the key rates of the DMS cycle (DMS production from DMSPd cleavage and DMS consumption) have not been measured simultaneously in Subarctic Pacific waters.  Moreover, no studies to date have examined the potential contribution of DMSO reduction to DMS cycling in this region.  Recent work in Antarctic polynyas3, suggests that this process may be important in at least some marine environments.   94 In this article, we present new observations of the distribution and turnover rates of DMS in the Subarctic Pacific.  Working in both coastal waters around Vancouver Island, and along the Line P sampling transect from coastal waters to offshore HNLC regions, we conducted high spatial resolution surface DMS surveys using membrane inlet mass spectrometry (MIMS), and also employed a new tracer-based method to simultaneously quantify key processes in the DMS cycle, namely gross DMS consumption, DMSP cleavage and DMSO reduction3.  Our goal was to characterize the spatial distribution of DMS/P/O concentrations along Line P and to quantify key production and consumption rates in the DMS cycle in contrasting coastal and open ocean waters.  Our results validate and extend previous studies of DMS/P/O cycling in Subarctic Pacific waters.  5.2 Methods 5.2.1 Field Sampling We conducted two research cruises on board the CCGS John P. Tully, surveying coastal and open ocean regions of the Subarctic NE Pacific (Fig. 5.1).  Coastal waters of British Columbia were surveyed between July 20th to August 15th, 2010 on the West Coast Acidification Cruise (cruise IOS2010-10).  The cruise track covered much of the outer coast of Vancouver Island, the Queen Charlotte Islands and the central British Columbia coast.  Stations were sampled along the continental shelf, and on a number of cross-shelf transects.  During this cruise, we also had the opportunity to sample a detailed grid near the Brooks Peninsula (~ 148 W, 50 N; see Fig. 5.5), during a 2-day search and rescue (SAR) mission.  During the summer of 2011, we sampled along the Line P time-series transect (from Vancouver Island to Ocean Station Papa, 145 °W, 45 °N) from August 16th to September 1st 2011 (cruise 2011-27).  The westward and returning eastward ship tracks for this cruise were conducted along the same survey line, providing us with an opportunity to sample similar waters masses approximately 2 weeks apart.  5.2.2 Surface Water Gas Measurements Surface water DMS was measured using membrane inlet mass spectrometry2 (MIMS), following the procedures outlined in Tortell et al.149.  Briefly, surface seawater  95 obtained from the ship’s intake (~5m depth) was pumped through a sampling cuvette at 500 ml min-1, allowing gasses to permeate across a ~ 0.25 mm thick dimethylsilicone membrane into a Hiden Analytical quadrupole mass spectrometer.  Sample water temperature was maintained at 15 ˚C, using ~20ft of stainless steel tubing immersed in a water bath immediately upstream of the sampling cuvette.  Measurements were made every ~30 seconds, which is equivalent to ~ 200 m spatial resolution at cruising speeds of 5-10 kn.   DMS partial pressures were calibrated from raw ion current intensities every ~ 1 to 2 days using known DMS standards made from liquid DMS diluted in deep (> 200 m) seawater.  5.2.3 DMS/P/O Concentrations Measurements For the analysis of total DMSt, DMSPt and DMSOt concentrations, we collected 1L samples in UV transparent (UVT) bags from Niskin bottles and dispensed 10mL duplicate subsamples into 20 mL acid-cleaned, transparent serum vials with Teflon faced caps (Wheaton PN 224100).  DMS was sparged out of solution under N2 flow and analyzed as described below.  After DMS sparging and analysis, samples were treated with either 2 mL of 10N NaOH (for DMSP samples) or 2mL of TiCL3 (for DMSO samples), capped with teflon faced caps and allowed to sit for 12-24 hours.  Prior to analysis, duplicate subsamples for dissolved DMSP and DMSO were gently filtered through 0.2 µm acrodisc syringe filters according to Kiene and Slezak129. All discrete concentration measurements, including those for rate measurements (described below) were conducted using a purge and trap capillary inlet mass spectrometer (PT-CIMS) to follow the concentrations of isotopically-labeled S compounds.  The PT-CIMS detector is a Hiden Analytical triple filter quadrupole mass spectrometer (HAL 3F) with an electron impact ion source set to 750 mA emission current, and a secondary electron multiplier operated at a gain of 900V to detect ions.  The PT-CIMS was interfaced to a custom built purge and trap gas extraction system (described in detail in chapter three) coupled with a gas chromatograph.  DMS was purged from each 60 mL seawater sample under a 500 mL min-1 He flow for 14 minutes (or 7 minutes from 30 mL seawater samples), and loaded onto a stainless steel trap packed with carbopack at room temperature.  After the sparge was completed, the trap was electrically heated to 200 °C (with short pulses of high current) to elute analytes under a He carrier flow (10 ml min-1) through a Chromasil 330  96 column for separation of gases prior to mass spectrometric analysis.  The quadrupole mass spectrometer selectively analyzed different isotopic species of DMS eluting from the GC column on the basis of their mass to charge ratio (m/z).  In situ DMS/P/O concentrations were measured at the dominant naturally occurring DMS mass/ charge ratio (m/z) of 62.  During rate experiments (described further below), isotopically-labeled tracers, including double 13C-labeled DMS (m/z 64) and DMS with either three or six deuterium atoms (m/z 65 and 68, respectively) were also detected by mass spectrometry. Sample handling and data collection were automated using custom LabVIEW software.  Thereafter, peak areas for every m/z signal were integrated using a Simpson’s 3/8 rule, and the baseline for each isotopic species was subtracted in Matlab.  Prior to peak integration, data were treated with a notch filter using the native Matlab idealfilter function to attenuate specific frequencies associated with electrical noise and occasional carrier flow fluctuations.  Concentrations were calculated based on the linear relationship between peak area and known concentrations of standards made from diluted stocks of liquid DMS, DMSP and DMSO every ~3-4 days (r2 ≥ 0.98 ).  5.2.4 Rate Experiments Rate experiments were conducted using either a competitive inhibition (CI) approach, or with isotope tracer additions, as previously described by Asher et al.3.  While both approaches offer valuable information on the balance of gross DMS production and gross DMS consumption, tracer experiments have the added advantage of quantifying the relative importance of DMS production from different reduced sulfur pools (i.e. DMSP vs. DMSO). For all rate experiments, seawater samples were collected from 10m depth with Niskin bottles and homogenized inside a 20L carboy by inverting the container ~10 times.  Three-liter volumes were dispensed into triplicate UV transparent FEP plastic bags (Welch Fluorocarbon – P00020-1) using a graduated cylinder.  Bags were incubated in deck-board seawater tanks, maintained at situ surface temperature using flowing seawater.  Ambient light levels were reduced to ~30% of surface values using 2 layers of neutral density screening.  Tracers and competitive inhibitors were prepared in two dilution steps using a gas tight syringe to handle pure, volatile DMS liquids.  Liquid DMS was diluted in Milli-Q water, which was well mixed hours before the experiment. Final tracer and competitive inhibitor  97 solutions were added (with a pipette) and bags were gently inverted 10-15 times to mix the tracer and competitive inhibitor additions in order to homogenize samples before initial T0 subsamples were removed within 20 minutes.  Sub-samples were removed from the bags, using a 60mL syringe via a luer lock port, and loaded onto a rack for automated sampling and PT-CIMS analysis every 1.5 –2 hours. Competitive inhibition experiments were used to measure gross production rates of DMS, by inhibiting DMS consumption in the presence of a competing analog compound.  In previous work, Dimethyl Disulfide46 (DMDS) has been used to competitively inhibit DMS consumption.  For our work, we used D6-DMS as competitive inhibitor for DMS consumption, adding this labeled compound at concentrations ~ 20x higher than background unlabeled DMS levels.  In a number of previous experiments (Asher et al., unpublished data), we have confirmed that the D6-tracer provides similar results to DMDS, with < 20% difference between the two gross DMS production estimates.  Results from one such comparative experiment are shown in Table 5.1.  In the presence of excess commercially available D6-DMS, consumption of unlabeled DMS is inhibited, and the net change in unlabeled DMS thus represents gross DMS production.  In control experiments without D6-DMS additions, the change in DMS concentrations over time reflects net DMS production rates.  In a subset of the competitive inhibitor experiments, a commercially available D3-DMS tracer was added to the control treatments to directly measure the loss of DMS due to combined abiotic photo-oxidation and biological DMS consumption.  Since there is no background production of D3-DMS, loss rates of this added tracer are used to quantify gross DMS consumption.  This direct measurement of gross DMS consumption can be compared against gross consumption rates derived from the difference of measured net and gross DMS production rates in competitive inhibitor experiments, as described above.   At some sampling stations, gross DMS consumption was measured with these two independent methods; using stable isotope tracers (following D3-DMS disappearance) and as the difference between gross DMS production and net DMS change in competitive inhibitor experiments.  We found that values derived from these different methodological approaches agreed well (Table 5.1).  Small discrepancies (~5%) between experimental approaches suggest that our estimates of gross DMS consumption are robust.  98 Several post-processing steps were required to obtain rate constants from the raw data from tracer and competitive inhibitor experiments.  Rate constants for gross DMS production in CI experiments were derived from the slope of the natural logarithm of unlabeled DMS/P concentrations over time (equations 1 and 2).  First, however, unlabeled (i.e. m/z 62) DMS concentrations in DMS CI experiments were corrected for ~1% ion source mass fragmentation from the m68 DMS (http://webbook.nist.gov/), which was added as a competitive inhibitor at >20 times the concentration of background DMS. 1) n𝑎𝑡𝑢𝑟𝑎𝑙  𝐷𝑀𝑆 = 𝑚62  𝐷𝑀𝑆 − 0.01  𝑋  𝑚68  𝐷𝑀𝑆 2) ln ™????      ™?™?␥??    ™? ? = −𝑘™? _ ™?␥?????   𝑡 For isotope tracer experiments, we amended four replicate bags with tracer level (i.e. < ~ 20% of ambient) additions of D-3 deuterated DMS, D-6 deuterated DMSP, and 13C labeled DMSO to achieve final concentrations of 1, 0.7 and 0.5 nM, respectively.  DMS consumption was calculated as a pseudo first-order reaction: 3)	  ln ???     ™????   ™? ? = −𝑘™? _ ™??   𝑡	  where kdms_cons is the observed rate constant, t is time, and [D-3 DMS] is the concentration of the added tracer (equation 1).  The m/z 64 signal, indicative of C-132 labeled DMS derived from DMSOd, was corrected for the background pool of S34-containing DMS and ion source fragmentation; 4)  [𝐶13?DMS] = [𝐷𝑀𝑆™ ] − 0.3  [DMS ™ ] − 0.043  [𝐷𝑀𝑆™ ]	  To calculate a gross DMSOd reduction rate, we corrected the measured concentration of C-132 labeled DMS for gross DMS consumption, as measured using D-3 DMS tracer; 5)  [𝐶13?DMS] ™?? = [𝐶13?DMS] +    [? ™ ? ™? ]????™ ? ™⌤ ?	  Similarly, we corrected the D-6 DMS concentrations for gross DMS consumption (equation 1) to calculate a gross DMSPd cleavage rate as: 6)   D− 6  DMS ? ™? = D− 6  DMS + [???   ™? ]???? ™? _ ™?? ?	  Gross DMSOd reduction and DMSPd cleavage were computed in a manner similar to gross DMS consumption (though opposite in sign).  We used the slope of the natural logarithm of corrected C-132 and D-6 labeled DMS concentrations over time to calculate the respective rate constants. 7)  ln [? ™ ?? ™? ] ™ ⌣[? ™ ?? ™? ]?   ™ ⌣ = 𝐾™?? _ ™?????⠩ 𝑡	   99 8)  ln ???   ™? ™ ⌣???   ™? ? = 𝐾™??? 𝑡	  If the initial T0 13C2 or D-6 DMS concentrations were below the 0.1nM detection limit, we assumed an initial concentration of 0.1nmol L-1.  For all rate constant calculations, we report uncertainty as one standard error from the mean slope in these experiments.  To determine natural consumption and production terms (nmol L-1 d-1 or nMd-1), rate constants measured in the experimental bags were multiplied by in situ concentrations of DMS, DMSP and DMSO.  Standard error propagation was used to extend the uncertainty of rate constant measurements and in situ DMS/P/O concentrations to natural production and consumption terms (nMd-1).   5.2.5 Ancillary Measurements We used a series of additional measurements to provide a biogeochemical and biophysical context for our DMS data.  All ancillary data described below for the Line P cruise are publicly available (http://www.pac.dfo-mpo.gc.ca/science/oceans/data-donnees/line-p/2011-27/index-eng.htm).  Mixed layer depths were calculated using a 0.125 kg/m-3 density difference (Dst) criteria.  Density profiles were calculated using conservative temperature and absolute salinity.  These variables were derived from CTD-measurements of temperature and practical salinity using the Gibbs Oceanographic Toolbox for MATLAB (http://www.teos-10.org/pubs/gsw/html/gsw_contents.html).  Chlorophyll a (Chla) and nutrient concentrations were measured according to standard procedures127,150 at all stations on both cruises.  Nutrient drawdown ratios of NO3- + NO22- to PO4 (N: P) and silicate to nitrogen (Si: N) were calculated from the Niskin bottle data.  Nutrient drawdown was calculated as the difference between nutrient concentrations in the mixed layer and sub-surface (100m) waters.  For stations along Line P that did not have nutrient measurements from 100m depth, the sub-surface nutrient concentration was calculated as an average concentration of 100 m nutrient concentrations along the transect.   100 5.2.6 Sea-Air Flux DMS sea-air fluxes along the cruise tracks were calculated using aqueous MIMS DMS concentrations and piston velocities (kw), derived from wind speeds, surface temperature and salinity (equation 9): 9)  𝐷𝑀𝑆  𝐹𝑙𝑢𝑥 = 𝑘𝑤  [𝐷𝑀𝑆] ™ 	  The Schmidt number used for these calculations was derived using the equation of Saltzman et al.69.  Ship’s wind speed data were not available due to a malfunctioning of the anemometer.  Thus, daily wind speeds for the 2010 WCVI cruise were obtained from environment Canada buoys located between 48N and 55N and 124W and 140W http://www.ndbc.noaa.gov/index.shtml).  For the 2011 Line P cruise, NCEP Reanalysis II 10m daily winds from 2011 were used (http://www.esrl.noaa.gov/psd/data/gridded/data.ncep.reanalysis2.gaussian.html).  DMS fluxes in 2010 and 2011 were calculated using three wind-speed parameterizations68,70,151.  DMS removal rates (nM d-1) in the mixed layer due to air-sea flux were computed by dividing flux (µmol m-2 d-1) by the mixed layer depth (m) to yield µmol m-3d-1, or equivalently, nmol L-1 (nMd-1).  For comparison, these rates (nMd-1) were divided by the in situ concentration of DMS (nM) to yield rate constants of (d-1) of DMS removal in surface waters due to air-sea flux.  5.2.7 Empirical Algorithms Using the algorithms of Simo and Dachs30 and Vallina and Simo31, we examined the relationship between DMS concentrations and the ratio of chlorophyll to the mixed layer depth (CHL/MLD), and between DMS and the solar radiation dose (SRD).  The SRD was calculated as a function of incoming solar radiation (I0), the mixed layer depth (MLD) and the extinction coefficient (Kd) according to Vallina and Simo31: 10)  𝑆𝑅𝐷 =    ??™?   ??     (1− 𝑒???)	  	  	  To estimate I0, we used 8-day rolling composites of level 3 (4km resolution) gridded PAR from the AquaModis sensor http://oceancolor.gsfc.nasa.gov/.  Kd was calculated from CTD profiles of PAR (measured with a Biospherical QSP-400 mounted on the ship’s CTD) at different depths (e.g. Z1 and Z2) where I is the irradiance:  101 11) 𝐾? = ™ ???????  ?????   5.3 Results 5.3.1 Hydrography and Plankton Biomass in Coastal and Open Ocean Waters Our sampling region (Fig. 5.1) encompassed several distinct oceanographic regimes, from high productivity coastal upwelling waters, to Fe-limited HNLC regions.  A transitional region, subject to coastal and offshore influences, separated these waters.  For the purpose of this study, we define the boundary between coastal waters and transitional waters as the 2000m isobath, according to Asher et al.1.  Similarly, we define the boundary between transitional waters and open-ocean HNLC waters as a threshold value in summer time surface nitrate concentrations >2 µM (see Fig. 5.1).  The presence of excess nutrients in late summer offshore surface waters can be used as a proxy for iron limitation39,57.  Based on surface water concentrations of nitrate, it appears that Fe limitation was apparent west of station P15, (i.e. ~133°W).   Coastal BC waters (shore-ward of the 2000 m isobath) exhibited shallow average mixed layer depths (mean 16.8 ± 13.3, range 5.1 – 75 m), with minimum values of < 10m observed in regions influenced by a near surface halocline (Fig. 5.2a).  Deeper mixed layer depths corresponded with localized coastal upwelling.  By comparison, open-ocean HNLC waters and transitional waters showed higher surface water salinity, and deeper mixed layer depths (mean 26.9 ± 12.6 m, range 13 – 77m; Fig. 5.2a).  Phytoplankton biomass (measured as total chla concentrations) was highest in coastal waters, averaging 3.35 ± 3.45 µg L-1 with maximum values ≥ 5µg L-1 (Fig. 5.2b).  In the offshore HNLC waters, maximum chla concentrations were generally low (mean 0.35 ± 0.17 µg L-1), although a small apparent phytoplankton bloom was observed in the vicinity of Ocean Station P, where chla concentrations approaching 1 µgL-1 were observed.   Phytoplankton biomass was also low in the transition waters, with chla concentrations averaging 0.91 ± 0.87 µgL-1.  We used Si:N drawdown ratios as an indicator of the relative contribution of diatoms to nutrient consumption.  As shown in Fig. 5.2c, high Si:N drawdown (≥ 1; Fig. 5.2c) was observed in coastal waters, suggesting diatom-dominance of phytoplankton assemblages.  With the exception of P26, Lower Si:N drawdown (< 0.6) was observed in transition and offshore  102 waters where non-diatom groups (including haptophytes and dinoflagellates) have been shown to be an important component of phytoplankton assemblages152.  The region of elevated chla near station P26 showed very high Si:N drawdown >1.5, suggesting that enhanced diatom growth in this region, possibly due to a short-lived mesoscale iron fertilization event.  It is important to note, however, that the high apparent Si: N drawdown at P26 could also be related to the presence of Si rich sub-surface water (~100-200 m) in this particular region.  Inspection of recent Line P data (2007-2009) reveal a similarly high Si: N drawdown at P26, suggesting that this may be a persistent feature, even in the absence of elevated chla.  5.3.2 High Frequency Measurements of DMS Concentrations Using automated MIMS analysis of surface waters, we were able to map, with high spatial resolution, the distribution of DMS concentrations in coastal and open ocean waters of the Subarctic Pacific.  Our data revealed a wide range of surface water DMS concentrations, with significant mesoscale and sub-mesoscale variability (Fig. 5.3a). During the 2010 WCVI cruise, DMS concentrations ranged from 0.5 to ~40 nM, with average values of 10.2 ± 7.4 nM  (mean ± standard deviation) in coastal waters and 5.2 ± 2.6 nM in transitional waters, respectively (no measurements were made in open ocean waters on this cruise).  During the 2011 Line P cruise, concentrations ranged between ~0 – 29 nM, with mean DMS concentrations of 3.1 ± 1.1 nM in coastal waters, 3.9 ± 1.7 nM in transitional waters, and 8.9 ± 5.4 nM in open ocean waters.  Variability in DMS concentrations across the different oceanographic regions (with data combined from the two cruises) is illustrated in Figure 5.4.  Overall, DMS concentrations averaged 9.8 ± 7.3 nM in coastal waters, 4.9 ± 2.5 nM in transitional waters and 8.6 ± 5.4 nM open-ocean waters (Fig. 5.4).  Due to more than ~4000 independent measurements in each oceanographic region, small standard errors from the mean, namely ± 0.064 nM in coastal waters, ± 0.022 nM in transitional waters, and ± 0.086 nM in open ocean waters, suggest that these differences between the mean DMS concentrations are in fact statistically significant.  This is the case although standard deviations presented in Figure 4 and discussed above are quite large.  Thus, higher DMS concentrations occurred in coastal and open ocean waters with moderate DMS concentrations and lower variability in transitional waters.  103 Across our survey region, strong gradients in surface salinity (Fig. 5.3b) corresponded with elevated DMS concentrations in a number of locations, particularly in transitional waters.  Along the shelf break adjacent to the WCVI, high salinity waters exhibited the highest DMS concentrations.  In addition, we observed a localized region of DMS accumulation (>10 nM) associated with a strong hydrographic frontal zone (salinity gradient 32.2. to 32.6 PSU) in the transitions waters at ~130.7° W (Fig. 5.3b).  5.3.3 High Resolution Survey of DMS Across the Shelf-Break Additional DMS measurements were made in coastal BC waters in 2010, as part of a search and rescue exercise along the Brooks Peninsula (Fig. 5.5).  During this survey, we measured strong DMS concentration gradients (i.e. ~ 0.5 – ~40 nM) (Fig. 5.5a), and significant variability in surface temperature (~9 – 15 °C) (Fig. 5.5b) and salinity (~31.7 – 32.7 PSU) (Fig. 5.5c).  The highest DMS concentrations (>30nM) were observed in upwelling regions with high salinity, cold waters north of the Brooks Peninsula (Fig. 5.5).  Conversely, lower DMS concentrations (10 – 18 nM) were observed south of the Brooks Peninsula, associated with fresh, cold water from riverine inputs.  Measurements of major dissolved gasses in surface waters, namely ΔO2/Ar and CO2 (data not shown), suggest that these continental shelf waters were the most biologically productive of the survey region.  DMS concentrations appeared to be highly dynamic over the short time of our survey, with repeated measurements showing significant differences in DMS levels in cross-over transects.  As noted above for the WCVI region, high DMS concentrations in the coastal SAR grid were associated strong gradients in surface salinity (Fig. 5.5c).  However, high DMS concentrations occurred in both low salinity waters on the shelf, and in saline, recently upwelled waters north of Brooks Peninsula. 	  5.3.4 Empirical Algorithms and DMS Distributions As in previous studies1,106, we found few correlations between surface water DMS concentrations and other ancillary oceanographic variables.  Despite some regional coherence between DMS concentrations and surface salinity (see above), there was no overall correlation between these variables along the entire cruise track.  Similarly, we did  104 not find any statistically significant correlations between surface DMS concentrations and Si:N drawdown, MLD, CHL, or sea surface temperature (data not shown).  We did, however, observe a weak (r = 0.60, p< 0.001) correlation between DMS and the CHL/MLD ratio binned to 1x1 degree as per the algorithm of Simo and Dachs30.  The slope of the DMS vs. CHL/MLD relationship we found for our data set was virtually identical to that reported previously by Tortell et al.106 for the coastal Subarctic Pacific (12 ± 1).  In the original work of Simo and Dachs30, CHL/MLD values greater than 1 were excluded from the analysis.  Indeed, we found that the apparent correlation broke down at CHL/MLD ratios > 0.8.  Relative to the CHL/MLD ratio, we found a much weaker (non-significant) relationship between surface DMS concentrations and solar radiation dose31.  5.3.5 Discrete DMS/P/O Measurements In addition to our MIMS-based continuous DMS observations, we also made discrete measurements of DMS, DMSP and DMSO concentrations at a number of sampling stations in coastal and open ocean waters along the Line P transect in 2011.  (Unfortunately, concentrations for DMSP and DMSO were not measured on the WCVI cruise in 2010, as we did not have the appropriate methods available at that time).  These Line P DMS/P/O measurements can be paired with rate constants (see below) to calculate in situ values of gross DMS consumption, DMSPd cleavage and DMSOd reduction in (nMd-1).  Based on this more limited data set of discrete observations, we observed higher DMS concentrations in open-ocean (7.8 ± 4.3 nM) (mean ± standard deviation of DMS/P/O at stations in each oceanographic region) and transitional waters (7.9 ± 4.5 nM) than in coastal waters (2.7 ± 1.5 nM) (Table 5.2).  These mean values are in reasonably good agreement with average DMS concentrations derived from MIMS underway analysis (Fig. 5.4).  In contrast to the observed DMS distribution, average DMSPd concentrations did not show apparent differences between coastal, transitional and open ocean waters, due in part to significant variability within each region.  In contrast, DMSPt concentrations (i.e. the sum of dissolved and particulate DMSP) differed between offshore (35 ± 15 nM), transitional (19 ± 15) and coastal waters 52 ± 2.8 nM (Table 5.2).  Due to a bottle labeling problem, we were unable to distinguish DMSOd samples derived from the coastal and transitional stations (P4 - P12).  The samples were thus pooled to obtain an average concentration for all of these stations.  This limits our ability to  105 discern DMSOd variability among coastal / transitional waters, but our observations suggest that mean DMSOd concentrations varied little between the coastal / transitional stations (22.9 ± 13.4 nM) and the oceanic stations (17 ± 4 nM).  To summarize, the interesting differences between DMS and DMSPd concentrations in coastal, transitional, and open ocean waters based on a very limited data set merit further study, with improved sampling resolution of DMS/P/O concentrations in surface waters.  5.3.6 Sea-Air fluxes We calculated DMS fluxes in open-ocean, transitional, and coastal waters (Table 5.3), pooling data from 2010 and 2011 cruises.  We observed a strong oceanic-coastal gradient in DMS air-sea flux (with higher values in open ocean waters) derived from three wind speed parameterizations.  Across the three parameterizations, mean coastal DMS flux was ~20-70% (mean ~50%) of the mean oceanic DMS flux.  The oceanic-coastal gradient sea-air flux was driven by higher wind speeds in the open ocean region (6.0 ± 2.6 ms-1) and the transitional region (5.3 ± 2.3 ms-1) than in coastal waters (3.9 ± 1.2 ms-1).  As noted above, both coastal waters and open ocean waters contained high aqueous DMS concentrations.  In general, we found good agreement between the fluxes derived from the different wind speed parameterizations.  The exception to this was a significantly lower coastal flux derived from the Wanninkhof and McGillis151 cubic formulation (Table 5.3).  This formulation is particularly sensitive to variable (low) wind speeds that occurred in the coastal regions.  This discrepancy notwithstanding, these results suggest that the waters of the Subarctic Northeast Pacific are a significant summer-time source of atmospheric DMS.  5.3.7 Rate Constants of DMS Production and Consumption Figure 5.6 shows the time course of DMS concentrations in a typical competitive inhibitor experiment, used to derive rate constants for gross and net DMS production, and gross DMS consumption.  Data in this figure were obtained at station 6 in the DE in the northernmost transect of the WCVI 2010 cruise (see Fig. 5.8 for station location).  In this experiment, we measured gross DMS production of 0.44 ± 1.0 d-1 (5.8 nMd-1), gross consumption of 0.48 ± 0.30 d-1 (6.2 nMd-1) and a net production of -0.039 ± 0.015 d-1 (-0.40  106 nMd-1).  In Table 5.4, we present rate constants of gross DMS production, gross DMS consumption, and net DMS production derived from stable isotope CI at all of the sampling stations on the WCVI cruise.  Across all of the sampling stations, gross DMS production rate constants averaged 0.76 ± 0.83 d-1 (range n.d. – 2.6 d-1) and were generally lower than gross DMS consumption (mean -1.1 ± 0.90 d-1; Table 5.4).  As a result, the average rate constant for net DMS production (-0.34 ± 0.64 d-1) was less than zero for most stations, indicating net loss of DMS from surface waters. In 2011, we made additional measurements of gross DMS consumption, net DMS production, and DMS production from different pathways (DMSPd cleavage and DMSOd reduction).  Figure 5.7 shows the results from a typical tracer experiment conducted at the open ocean station P12 at ~130.7 °W (located along the transitional-open ocean boundary).  In this experiment, we calculated rate constants for gross DMS consumption of -1.4 ± 0.38 d-1, DMSPd cleavage of 1.7 ± 0.14 d-1, DMSOd reduction of 0.47 ± 0.22 d-1 and net DMS production of 0.47 ± 0.28 d-1.    The sum of the gross production and consumption rate terms is consistent with the measured net production term, within the measurement errors.  Rate constants of gross DMS consumption, DMSP cleavage, DMSO reduction, and net DMS production along the Line P transect are summarized in Table 5.5.  We observed considerable variability in rate constants of gross DMS consumption (mean -1.0 d-1 ± 0.68 d-1), DMSPd cleavage (1.4 ± 0.31 d-1) and DMSOd reduction (0.12 ± 0.20 d-1) and net DMS production (-0.17 ± 0.50 d-1).  Turnover times for DMS averaged 0.5 – 4.5 days.  DMSPd cleavage was higher than DMSO reduction between P8 – P26, and the highest rates of DMSPd cleavage were observed in the offshore waters.  In three out of five stations sampled, net DMS production was negative, indicating that DMS consumption exceeded production.  5.3.8 Patterns in DMS Production and Consumption Terms The spatial distribution of gross DMS consumption, gross DMS production and net DMS production rate constants are shown in Fig. 5.8 for the stations surveyed in 2010 and 2011.  While there were no clear spatial patterns in gross or net DMS production, rate constants for net DMS production across the full survey region were positively correlated with surface water DMS concentrations (r = 0.62, p < 0.05, n = 14; Fig. 5.9).   107 5.4 Discussion To our knowledge, the results presented here constitute the most complete regional survey of DMS/P/O concentrations and consumption / production rates in surface waters of the Subarctic NE Pacific.   Our measurements corroborate the few existing data on gross DMS consumption and DMSP cleavage, and represent the first regional measurements of DMS production from DMSOd reduction.  Below, we discuss our results in terms of currently existing data for the Subarctic Pacific, and show how our new data extend existing knowledge of DMS dynamics in this region.  5.4.1 Strong Spatial Gradients in DMS Concentrations and Sea-Air Fluxes DMS concentrations appeared higher and more heterogeneous in coastal waters and open ocean waters than in transitional waters (Fig. 5.3; Fig. 5.4).  DMS data obtained near the Brooks Peninsula (Fig. 5.5) exemplified the strong spatial variability we observed in near-shore coastal waters.  Overall, data from the 2010 WCVI cruise agree well with the observations of Tortell et al.106 and Nemçek et al.61, who documented high DMS concentrations and spatial variability along the West Coast of Vancouver Island and in the Queen Charlotte Sound.  In contrast with the 2010 WCVI data, coastal DMS concentrations exhibited less mesoscale and sub-mesoscale spatial variability during the 2011 Line P cruise, due in large part to the more limited sampling area.  The dominant spatial gradient along the Line P cruise track was associated with the transition from iron-replete to iron-limited waters west of ~ 137 ° (as judged from maximum surface NO3- data).  In addition, a sub-mesoscale DMS ‘hotspot’ was observed at ~130.7° W, corresponding with a hydrographic front, as indicated by a strong surface salinity gradient.  This region has previously been identified as a biological productivity hotspot87, influenced by the mixing of high Fe coastal waters, with high NO3- offshore waters.  Our results from the 2011 Line P cruise are consistent with the observations of Steiner et al.38, who documented persistently high late summer (August) DMS concentrations in open-ocean waters between 1996 - 2010.  These high DMS concentrations result in large sea-air DMS fluxes (Table 5.3), with values ranging between ~ 1 and 74 mmol m-2d-1 (mean ~15 ± 12 mmol m-2d-1).  These mean sea-air fluxes are within  108 the range of the long-term climatological means derived by Lana et al.32, who reported summer-time DMS fluxes from the Subarctic Pacific on the order of ~ 30 mmol m-2d-1.  5.4.2 DMS/P/O Concentrations in the Subarctic Northeast Pacific Discrete measurements of total DMSP (DMSPt) from this study are consistent with published values of particulate DMSP (DMSPp) obtained from stations between P2 and P26 in 200739 (33.9 ± 14.8).  The similarity between these total and particulate DMSP measurements reflects the dominance of the DMSPp pool, and the small contribution of the dissolved DMSP (DMSPd) pool.  Our DMSPd measurements (3.5 ± 2.7 nM) in 2011 were slightly higher than values from measurements from the line P transect June in 200739 (2.2 ± 0.6).  Apparent differences in DMSPd concentrations between datasets may be attributed to inter-annual variability, or spatial variability among sampling sites in the various sub-regions.  An additional source of variability in DMSPd concentrations could also be attributable to artifacts related to sample handling.  Phytoplankton cell lysis during filtration has been shown to release substantial amounts of particulate DMSP into the dissolved pool, leading to artificially high DMSPd concentrations.  While we cannot rule out potential sampling artifacts, we and the previous authors working along Line P have all followed the protocol recommended by Kiene and Slezak129 to minimize cell disruption during the filtration process. To our knowledge only two other studies have measured DMSO concentrations in the Subarctic Pacific.  Bates et al.115 measured DMSOt at five depths down to 60m at the PSI-3 time series station (~128° W 48° N) in April 1991, and reported maximum concentrations of 5nM.  By comparison, DMSOd values in all of the regions we sampled (i.e. coastal, transitional and open-ocean waters) were substantially higher than this value.  Although we have a limited data set, our results suggest little difference in DMSOd concentrations between coastal / transitional waters (22.9 ± 13.4 nM) and open-ocean waters (17 ± 4 nM).  The higher concentrations we observed relative to Bates et al.115 likely reflect the difference in sampling time (mid-summer vs. early spring), and the role of DMS photo-oxidation in the formation of DMSOd27,73.  Our own recent measurements (see chapter three) demonstrate high DMSOd/t concentrations along Line P during August of 2014, with a mean value of ~11 nM.  These additional data suggest persistently high summer-time DMSO concentrations in  109 the NE Subarctic Pacific.  As discussed below, the role of the DMSO pool in DMS dynamics requires further study.  5.4.3 Trends in DMS Production/Consumption We observed a positive correlation between surface water DMS concentrations and the rates of net DMS production across our study area (Fig. 5.9).  This suggests that our rate measurements capture the majority of processes driving DMS accumulation in the Subarctic NE Pacific.  Overall, net DMS production increased with distance from the coast and with latitude, suggesting that DMS accumulation can persist longer in these waters.  We observed the opposite pattern in gross DMS production and consumption (i.e. highest rates in more southern, near-shore waters).  These higher rates could be related to higher rates of bacterial metabolism in coastal waters, driven by greater phytoplankton biomass (chla) and productivity (biological O2 supersaturation, DO2/Ar - data not shown).  Across our study area, DMS turnover times ranged between < 1day to ~4 days and averaged 0.87 ± 0.34 days in coastal waters, 2.0 ± 1.4 days in transitional waters, and 3.5 ± 2.1 days in open ocean waters, respectively.  These high turnover rates, particularly in coastal waters, are sufficiently rapid to remove any signature of DMS accumulation, thus effectively 'resetting' the mixed layer DMS budget on short time-scales (i.e. hours – days).  Shorter DMS turnover times in coastal waters help to explain the greater spatial variability we observed in near shore DMS concentrations.  Our measured gross consumption and production rates represent a combination of biological and abiotic (e.g. photo-oxidation) processes, but they do not take into account physical processes such as mixing and sea-air exchange.  Calculations of sea-air flux (Table 5.3) demonstrate that this is generally a minor term in the DMS budget of the Subarctic Pacific, with maximum removal of DMS averaging 0.64 ± 0.68 nMd-1 and equivalent rate constants of 0.10 ± 0.084 d-1.  The rate constants associated with sea-air flux represents an average of ~8% of total DMS consumption we measured in the surface layer.   There were, however, two instances (station DE9, Table 5.4; Table P20, Table 5.5) where sea-air flux accounted for between 30 - 50% of measured gross consumption rates.   We estimated maximum DMS photo-oxidation rate constants by scaling published values from Bouillon et al.73 from the Subarctic Northeast Pacific by in situ nitrate concentrations and short wave  110 radiation measurements, obtained from http://www.pmel.noaa.gov/OCS/data/disdel_v2/disdel_v2.html.  Bouillon et al.75 showed that the DMS quantum yield is well correlated with nitrate concentrations (y = 0.15x + 0.41), and Bouillon et al.73 demonstrated that changes in nitrate concentrations and solar radiation resulted in the largest changes in DMS photo-oxidation rate constants (± ~50%) in surface waters during the SERIES experiment.  Our calculations yielded rate constants between ~0.027-0.15 d-1.  On average, these values represent ~20% of gross DMS consumption across our study area, though at some high NO3- stations (e.g. P20, Table 5.5), photo-oxidation can account as much as ~70% of measured gross DMS consumption.  We thus conclude that biological processes dominated DMS consumption at most of the stations we sampled. Given the strong density stratification of surface waters across the offshore Subarctic Pacific (due to a low salinity surface layer), vertical entrainment of sub-surface waters is not likely a strong DMS source.  In coastal waters, however, upwelling could potentially supply DMS to the surface, although depth profile measurements37 suggest low concentrations of DMS in sub-surface waters.  Our results suggest that upwelling is not a strong DMS source to surface waters.  In particular, our detailed 2010 survey of the Brooks Peninsula (Fig. 5.5) showed that DMS accumulation began some time after upwelling and was then attenuated, likely by rapid biological consumption.  Although we did not conduct any rate measurements during the SAR transects survey, the short turnover time of DMS at other coastal stations (residence time < 1 day) suggests that DMS accumulation can be attenuated over short temporal and spatial scales.  5.4.4 A Comparison of Rate Measurements from the Subarctic Northeast Pacific  To date, three different methods have been used to examine DMS cycling in the NE Subarctic Pacific; competitive inhibition assays, stable isotope tracer additions and radio isotopes36,39.  To examine the consistency of our data with these previous rate measurements, we compiled the available gross DMS consumption and DMSPd cleavage rates from previous studies36,39, for comparison with our new observations (Fig. 5.10).   As shown in Fig. 5.10a, the available data show very good agreement in terms of DMS consumption rates at Station P26, despite differences in sampling years and methodology.  Indeed, the agreement is rather striking, and provides some confidence in the  111 available measurements.   Merzouk et al.36 used radio-tracers to measure biological DMS consumption rates in Fe-limited waters of 6.5 ± 3.8 nM over a ~20 day LeGrangian experiment at Station P26 in 2002.  Using the D-3 DMS tracer method, we derived a nearly identical rate of gross DMS consumption (5.8 nMd-1) at P26 (Fig. 5.10a).  Previous measurements of DMSP cleavage rates in spring (May/June) along Line P39 have produced estimates averaging ~0.5 ± 0.3 d-1, with values ranging from 0.14 – ~0.83 d-1 (Fig. 5.10b).  By comparison, we measured an average DMSPd cleavage rate of ~1.4 ± 0.4 d-1, with values ranging from 1.0 – 2.0 d-1 (Fig. 5.10b) in late summer.  Overall, these rates agree within a factor of two-four, which seems reasonable considering the comparison across seasons and years.   Moreover, we expect higher rates of DMSPd cleavage in summer than spring, to sustain the elevated open-ocean DMS concentrations frequently observed in August in the Subarctic Northeast Pacific.  There is thus some reason to believe that DMSP cleavage rates would be elevated in August relative to June in this region. Our results provide the first DMSOd reduction rate constants in the Subarctic Pacific.  We found that rate constants of DMSOd reduction were significantly lower than DMSPd cleavage and gross DMS consumption except in coastal waters (Table 5.5, Fig. 5.10c).  However, since DMSOd pool sizes are comparable or larger than DMSPd pool sizes (Table 5.2), DMSOd reduction can, at times, provide an important source of DMS in the Subarctic Northeast Pacific.  Additional DMSO concentration and reduction rate measurements are thus needed (both in coastal and open ocean waters) to better characterize the importance of this compound in the DMS cycle of the Subarctic Pacific.  As both an important sink of DMS through photo-oxidation and biological oxidation, and as a source of DMS from DMSO reduction, DMSO measurements should be included future DMS research programs.  We have recently developed an automated method for surface DMSO (and DMS/P) measurements, which should enable future research into the spatial distribution of this compound in marine surface waters.  5.5 Conclusions and Future Outlook  This study provides extensive coverage of DMS/P/O concentrations and cycling in open-ocean and coastal regions in the Subarctic Northeast Pacific.  We observed high DMS  112 concentrations and sea-air fluxes in both open ocean and coastal oceanographic regions, with lower values observed in the transitional waters.  Open ocean DMS concentrations and sea-air fluxes further support the importance of the Subarctic Northeast Pacific as an important source of DMS to the atmosphere.  While DMS concentrations may be linked to changes in measured oceanographic variables (e.g. Chla and mixed layer depths) on global and regional scales, predicting fine-scale (<1 x 1˚) DMS variability requires simultaneous measurements of DMS production and consumption processes.  Our results, and those of previous studies, show that rates of DMS production and consumption vary between coastal, open-ocean HNLC, and transitional waters, likely in response to a range of oceanographic and physiological / ecological processes.  The elevated DMS turnover times we measured in coastal waters may help explain the presence of localized DMS ‘hotspots’ in these waters, as compared to the more homogeneous DMS accumulation in the HNLC open-ocean.  In general, our measurements suggest that surface water patterns in DMS concentrations can be understood in terms of the underlying rate processes.  Future studies are needed to link physical and biological oceanographic parameters to individual production and consumption terms in the Subarctic Pacific DMS cycle, and to document the seasonality of DMS production and consumption terms in these contrasting oceanographic regimes.  These data sets will help parameterize regional biogeochemical models29 examining the response of Subarctic Pacific surface waters to on-going climate-dependent perturbations.    113 Table 5.1. Comparison of CI and Tracer Methods Comparison of DMS gross consumption and production rates derived from stable isotope tracer experiments (D3-DMS uptake), stable isotope competitive inhibition (CI) assays (D6-DMS addition) and dimethyl disulfide (DMDS) CI assays.  The results show good agreement between gross DMS consumption rates derived from D3-DMS uptake and D6-DMS CI assays, and good agreement between DMS production rates derived from D6-DMS and DMDS CI assays. See methods for details of these assays and the derivation of rate constants.  Values represent the mean ± the standard error.  Station D3-DMS Consumption  (d-1) D6-DMS CI  Inferred Cons. (d-1) D6-DMS CI  Production (d-1) DMDS CI Production (d-1) QCS1 -3.1 ± 0.73 -3.2 ± 2.2 2.6 ± 0.54 - QCS7 -0.80 ± 0.22 -0.70 ± 0.47 n.d. - HSI -1.7 ± 0.52 -1.6 ± 1.2 1.5 ± 1.3 1.3 ± 0.9 ML1 -1.1 ± 0.034 -1.1 ± 1.4 0.87 ± 0.42 -  	   	   114 Table 5.2. Concentrations of DMS/P/O Concentrations of reduced-sulfur compounds in surface waters (~5m) along Line P in Aug. 2011.  Mean concentrations ± one standard error.  Due a labeling error, DMSOd samples from stations P2-P12 could not be distinguished from one another.  These samples were thus pooled and the values represent the mean ± standard deviation, showing the natural variability in DMSO concentrations between P2-P12.  DMSP and DMSO measurements were made on duplicate samples, while DMS measurements were made on single samples.  STATION Region DMS (nM) DMSPd(nM) DMSPt(nM) DMSOd (nM) P2 Coastal 1.6 2.4±0.4 50±1.1 23 ±13 P4 Coastal 3.8 1.4±1.8 54±6.1 23 ±13 P9 Transitional 4.7 2.1±0.10 8.6±0.18 23 ±13 P12 Transitional 11 8.3±3.5 30±4.9 23 ±13 P16 Oceanic 3.5 3.0±0.14 51±0.4 19.6±3.0 P20 Oceanic 8.0 1.3±0.32 22±1.7 13.4±2.8 P26 Oceanic 12 6.3±3.5 31±4.4 20.9±5.8 	     115 Table 5.3. Air-Sea Fluxes of DMS DMS sea-air fluxes by oceanographic region (i.e. open ocean, transitional, and coastal), calculated as the product of aqueous DMS concentrations and the piston velocity (kw) derived from several wind speed parameterizations.  Data from 2010 and 2011 were pooled together.  Separation by oceanographic region was based on bathymetry and summer time nitrate concentrations (see methods).  All fluxes are reported as means ± standard error, with the range in brackets.  Units are µmol m-2 d-1.  Parameterization Open ocean Transitional   Coastal   Ho et al. 2006 14 ± 12 (0.71 – 74) 11 ±11 (0.16 – 73) 7.2 ± 6.7 (0.0076 - 100)  Wanninkhof and McGillis 1999 12 ± 13 (0.40 – 72) 9.2 ± 12 (0.048 – 72) 3.3 ± 3.3 (0.033-198)   Nightingale 2000 15 ± 12 (0.70 – 74) 12± 11 (0.21 – 74) 8.2 ± 7.5 (0.047 - 220)      116 Table 5.4. Rates of DMS Production and Consumption in CI Experiments Mean rates of gross DMS production, gross DMS consumption and net DMS production (mean value ± standard error) derived from DMS CI experiments on the WCVI cruise in 2010.     Station Region Lat. (°N) Long. (°E) Gross DMS Prod. (d-1) Gross DMS  Cons. (d-1) Net Production (d-1) ML1 Coastal 47.86 -125.05 0.87 ± 0.42 -1.1 ± 1.4 -0.18 ± 0.17 BP3 Coastal 50.05 -127.92 0.45 ± 0.22 -1.1 ± 0.59 -0.62 ± 0.21 QCS1 Coastal 52.25 -129.05 2.6 ± 0.54 -3.2± 2.2 -0.52 ± 0.47 QCS7 Coastal 51.62 -130.77 n.d. -0.70 ± 0.47 -0.67 ± 0.47 HSI Coastal 52.63 -131.37 1.5 ± 1.3 -1.6 ± 1.2 -0.080 ± 0.16 DE1 Coastal 54.42 -132.30 0.75 ± 4.4 -1.1± 2.9 -0.35 ± 0.48 DE6 Transition 53.81 -134.25 0.44 ± 0.10 -0.48 ± 0.30 -0.039 ± 0.015 DE9 Transition 53.11 -135.60 n.d. -0.27 ± 0.83 -0.27 ± 0.7 QCS10 Transition 51.02 -132.18 0.24 ± 0.21 -0.33 ± 0.70 -0.085 ± 0.075     117 Table 5.5. Rates of DMS Production and Consumption in Tracer Experiments Mean rates (± one standard error) of gross DMS consumption, DMSP cleavage, DMSO reduction and net DMS production at the major stations along the Line P transect in Aug., 2011.   Rates below detection (i.e. no measurable changes in tracer concentrations) are listed as n.d.  Station Name Region Lat. (°N) Long. (°E) Gross DMS Consump. (d-1) DMSP Cleavage (d-1) DMSO Reduction (d-1) Net DMS Production (d-1) P4 Coastal 48.6 126.7 -1.1 ± 0.4 1.0 ± 0.72 0.11 ± 1.0 -0.50 ± 0.7 P9 Trans. 48.9 129.2 -1.9 ± 1.0 1.7 ± 1.4 n.d. -0.31 ± 0.9 P12 Trans. 49 130.7 -1.3 ± 0.78 1.7± 0.14 0.47 ± 0.22 0.47 ± 0.20 P20 Oceanic 49.6 138.7 -0.22 ± 0.03 1.3 ± 0.61 n.d. -0.71 ± 0.7 P26 Oceanic 50 145 -0.48 ± 0.2 1.2 ±0.60 n.d. 0.20 ± 0.7     118    Figure 5.1. Map of Study Area in the Northeast Pacific Surface water (~5m depth) nitrate concentrations in coastal and open ocean waters of the Subarctic Pacific in late summer of 2010 and 2011. Bold line along the coast represents the 2000m isobath, which is taken as the boundary between coastal and transitional waters (the 500m isobath is also shown for reference).   A nitrate threshold of >1µM is used as the boundary between transitional and open ocean waters, and denoted by the strait diagonal line on the figure.  Boxes and arrows denote the following geographic regions; Queen Charlotte Sound (QCS), the west coast of Vancouver Island (WCVI), the strait of Georgia (SoG), and the Juan de Fuca Strait (JdF).  The size of the symbols represents nitrate concentrations, as shown on the figure legend.     119  Figure 5.2. Mixed Layer Depths, Chla, and Si: N Drawdown Spatial map of CTD-derived mixed layer depth (a), chla measured from rosette bottle casts from 10m depth (b) and the drawdown ratio of Si:N between 10m depth and 200m nutrient  120 replete waters (c). Note the logarithmic color scale for Chlorophyll.  Labels on panel (a) denote the location of the major stations along the line P transect.   121  Figure 5.3. Spatial Distribution of DMS Concentrations and Surface Salinity Spatial distribution of surface (~5m depth) DMS concentrations (a) and salinity (b) in coastal and offshore waters of the NE Subarctic Pacific during late summer 2010 and 2011.    122  Figure 5.4. Box Plots of DMS Concentrations by Oceanographic Region. Box plots denoting the median, the upper and lower quartile values, and density distribution (grey patch) of DMS concentrations by oceanographic region (i.e. coastal, transitional, and open ocean waters). Each white box spans the 25th to the 75th percentile of the DMS observations, with the solid horizontal line representing the median (50th percentile). For reference, the means (solid circles) ± standard deviation (diamonds) are also plotted. Data from both cruises were pooled together.    123  Figure 5.5. Detailed View of DMS and Surface Salinity at Brooks Peninsula Detailed view of DMS concentrations (a) temperature (b) and salinity (c) in surface waters (~5m) along the West Coast of Vancouver Island near the Brooks Peninsula in July 2010.  The box denotes an area of upwelling as indicated by low temperature, high salinity waters.     124  Figure 5.6. Example of CI Experiment Results of a competitive inhibition experiment at station DE6 (~53N 134W) in transitional waters showing gross DMS production and net DMS production derived in the presence and absence of D6-DMS (see methods for experimental details).  Note the logarithmic y scale.  Error bars represent the standard error from the mean.  125  Figure 5.7. Example of Tracer Experiment. Results from an isotope tracer experiment at station P12 (49 °N 130.7 °W). Net DMS production was measured by following the time-course of background DMS concentrations as m62 (a), DMSPd cleavage was measured by following D6-DMS concentrations (m68) after the addition of D6-labelled DMSP (b), DMSOd reduction was measured using 13C labeled DMS (m64) after the addition of 13C-labelled DMSO (c), and gross DMS consumption was measured using isotopically labeled DMS (3D, m65)  (d).  Note the different y-scales, all of which are presented as the natural logarithm of DMS concentrations.  Error bars represent the standard error from the mean.   126  Figure 5.8. Spatial Distribution of DMS Consumption and Production Measurements. Spatial distribution of gross DMS consumption rate constants (d-1) (a), Gross DMS production (d-1) (b) and Net DMS production rate constants (d-1) (c) rate constants in coastal and offshore waters of the Subarctic Pacific during 2010 and 2011.  Station names are labeled in panel a.  The 500m and 2000m isobaths are included for reference.    127  Figure 5.9. Relationship between Net DMS Production and DMS Concentrations. Relationship between rate constants of net DMS production (d-1) and surface water DMS concentrations in the Subarctic NE Pacific. The line represents the best-fit linear regression (r2 = 0.62, p <0.05, n= 14), with a slope of 10.57 and y-intercept of 5.26.  Error bars represent the standard error from the mean.   128  Figure 5.10. Comparison of DMS Production and Consumption Along Line P Comparison of DMS consumption rate constants derived using D-3 tracer (this study) and by Merzouk et al. (2006) using radio-labeled tracers at P26 in 2002. b) Comparison of DMSPd cleavage rate constants using D-6 tracer in August 2011 (this study) against results obtained with a radio-labeled tracer study in June 2007 (Royer et al. 2010).  c) DMSOd reduction rate  129 constants in August 2011 (this study). Note the different scales on the Y-axes. Error bars represent one standard error from the mean.     130 6 Biogeochemical Controls on the Seasonal and Temporal Variability of DMS, DMSP and DMSO Concentrations in Coastal Waters of the Western Antarctic Peninsula 6.1 Introduction The trace gas dimethyl sulfide (DMS) is the main source of natural, non-sea-salt sulfate to the atmosphere6,153, a key player in the global sulfur cycle and atmospheric radiative balance7,96,97, and an important compound for the metabolism of several marine trophic groups.  The gas is ultimately derived from the algal metabolite dimethylsulfoniopropionate (DMSP), which serves a number of physiological functions, including potential roles as an osmolyte8, cryoprotectant9, and anti-oxidant10.  Particulate DMSP (DMSPp) in phytoplankton is released into the dissolved pool (DMSPd) through phytoplankton mortality, and is actively taken up by non-DMSP producing autotrophs98,99 and heterotrophic bacteria45. The uptake and assimilation of (DMSPd) can satisfy the energy, C and S demands of entire marine bacterial communities11, and this compound also serves as a chemo-attractant for a wide array of micro-organisms13.  By comparison with DMS and DMSP, the physiological and ecological function of dimethyl sulfoxide (DMSO) remains less well studied.  This compound is the main product of biological and photochemical DMS oxidation, and is ubiquitous in surface ocean waters.  It has been suggested to function as an intracellular cryo-protectant and anti-oxidant24. A primary focus of DMS, DMSO, and DMSP (DMS/P/O) research is to understand the spatial and temporal variability of these compounds in oceanic waters, and quantify the underlying production and consumption processes.  Rapid biological production and consumption of DMS/O/P has been documented in many marine environments, yielding short (i.e. <1-5 day) turnover times, and significant mesoscale and sub-mesoscale spatial and hourly or daily temporal variability in concentrations.  In the surface ocean, DMS is produced through the cleavage of DMSPd by phytoplankton and bacteria45,155 and by the intercellular cleavage of DMSPp in algae.  DMS may also be produced through the photochemical and biological reduction of DMSO to DMS3,21,22, and as a result of grazing by micro-grazers and larger zooplankton17,156–158.  The removal of DMS and DMSP occurs via  131 bacterial consumption12,91,136, lateral advection and vertical mixing, and, in the case of DMS, sea-air flux and photochemical oxidation to DMSO, sulfate or possibly dimethyl sulfone27,90. Whereas many of the physical processes influencing DMS/P/O concentrations in seawater can be quantified with environmental variables (e.g. wind speeds and surface irradiance) and well represented in ocean models, biological cycling processes have proven difficult to quantify directly and to parameterize mechanistically.  Over the past decade, however, the development of new experimental approaches has improved the quantification of several biological production/ consumption processes in the oceanic DMS/P/O cycle.  In particular, radioactively labeled DMS/P26,91,95,136,159 has proven to be an invaluable tool for quantifying DMS/P biological consumption rates and production yields (i.e. net DMS yield from DMSP-consumption).  More recently, a new stable isotope tracer method has been developed to simultaneously track the production and consumption of various compounds in the S cycle, and this method has revealed rapid turnover of DMS/P/O in Antarctic sea-ice3.  To date, these new experimental approaches have only been applied in a handful of oceanic environments. The Southern Ocean contains the world’s highest marine surface water DMS concentrations and contributes significantly to global oceanic DMS sea-air flux32.  Strong seasonal variability in sea-ice cover, solar irradiance, wind-speed, and mixed layer depth result in a highly dynamic biological production cycle160, which drives DMS/P/O accumulation in surface waters59,139,149,161.  Disentangling spatial variability from temporal variability in such dynamic systems, is challenging using ship-based surveys at discrete hydrographic sampling stations.  To address this problem, several LaGrangian studies have been conducted in the S. Ocean34,162,163, tracking individual water masses to isolate temporal variability from the underlying spatial gradients.  These studies have demonstrated a strong (and often dominant) imprint of biological production and consumption on DMS cycling, and have examined how DMS and DMSP concentrations respond to Fe manipulations.  While these results have provided invaluable information on DMS dynamics in Southern Ocean waters, their short time frame provides little insight into longer-term seasonal processes. A deeper understanding of the seasonal dynamics of S. Ocean DMS requires observations on time-scales that are beyond the scope of ship-based programs.  Fortunately, there are several well-equipped Antarctic research stations with access to coastal seawater.   132 The Palmer Station Long Term Ecological Research Site (PAL-LTER) is located on the northwest coast of the Antarctic Peninsula (Fig. 6.1).  For over two decades, Palmer’s LTER program has documented the seasonal and inter-annual variability of microbial populations and productivity (http://pal.lternet.edu), demonstrating strong changes in microbial processes both within and between years, as well as long-term ecological changes along the WAP shelf associated with rapid regional warming164.  The extent to which these shifts in the marine food web influence the biogeochemical cycling of trace gasses such as DMS has not been extensively studied. Thus far, two dedicated DMS research programs have been conducted at Palmer Station.  Berresheim et al.165 focused on quantifying the seasonal cycle of atmospheric DMS concentrations, and reported a close relationship between atmospheric and sea surface DMS levels.  In a subsequent study, a group of international investigators examined the seasonal and depth-dependent variability in DMS concentrations and loss pathways for DMS/P during austral spring and summer at two stations on the Palmer Station LTER grid.  Results from this field program43 showed that DMS concentrations tracked phytoplankton biomass, exhibited considerable seasonal variability and reached values as high as ~20 nM.  During the study period, microbial consumption dominated DMS removal, overshadowing DMS losses attributed to photochemical oxidation and sea-air flux in surface waters.  Hermann et al.43 used an inverse 1-dimensional General Ocean Turbulence Model (GOTM) to determine gross DMS production rates between January 1, 2006 and March 1, 2006.  Based on their model, and semi-weekly in situ DMS concentration and consumption measurements, an average gross DMS production rate of ~3nM day-1 was derived, with vertical mixing and entrainment suggested as important mechanisms for DMS removal from surface waters. This work provided important new insight into the seasonal dynamics of DMS in coastal WAP waters, but it did not assess the relative importance of various processes driving the biological production of DMS (e.g. grazing, DMSPd cleavage, DMSOd reduction), or the magnitude of higher-frequency temporal DMS/P/O variability. Following up on the studies discussed above, we present results from a recent field sampling program at Palmer Station, aimed at documenting temporal variability in surface seawater DMS/P/O concentrations and turn-over rates during the spring and summer phytoplankton growth season.  We sampled DMS/P/O in conjunction with the semi-weekly  133 LTER sampling program, and made additional high temporal resolution measurements of DMS/P/O to examine variability in these compounds on hourly and/or daily time-scales.  To quantify the production and consumption of DMS through various biotic and abiotic pathways, we used competitive inhibitor assays and recently developed stable isotope labeling assays3.  We also conducted experiments to examine the influence of microzooplankton and krill grazing on DMS production.  Our results reveal several scales of temporal DMS/P variability in the shelf waters of the WAP, and indicate that grazing processes are essential to the DMS seasonal cycle.  We drew on our results to suggest priorities for future studies.  6.2 Methods 6.2.1 Sampling Overview We conducted a field campaign between October 2012 and March 2013, studying the DMS/P/O dynamics at Palmer Station. Our objective was to characterize the temporal variability of DMS/P/O concentrations on various time-scales (hours to months) and to quantify DMS/P production and loss terms.  In order to achieve our research objectives, we used a variety of instruments and experimental methods, which are described below.   In addition, we obtained a variety of ancillary measurements (e.g.phytoplankton / zooplankton biomass, hydrography and meteorological data), as described in the following sections.  Working in collaboration with scientists in the Palmer LTER program, we collected discrete semi-weekly Zodiac samples from Station B (64.78 S 64.07 W), situated between the mouth of Arthur Harbor and a deep bathymetric trough (Fig. 6.1).  These discrete samples were used to measure the seasonal evolution of DMS/P/O concentrations in coastal WAP waters, and to conduct process studies examining the underlying production and consumption rates through various pathways.  The semi-weekly samples were supplemented with continuous automated analysis from the station's unfiltered seawater supply at Station A (Fig. 6.1).  6.2.2 Sulfur Concentration Measurements Semi-weekly samples for the determination of surface water DMS/P/O concentrations were collected from 10m depth at Station B, using a Monsoon bilge pump or Niskin bottles  134 deployed by Zodiac.  Samples for concentration measurements were handled according to Asher et al.3.  Briefly, clean Welch Fluorocarbon UV transparent (UVT) PFA bags were rinsed with seawater and filled with 1 liter of seawater from ~10m depth.  The excess headspace was removed using a gas tight bag clip closure.  These samples were maintained in the dark (under opaque bags) during sampling and transit to shore, and held in a dark 4 °C cold room on station until analysis (<1 hr total storage time). For the analysis of total DMS (DMSt), DMSP (DMSPt) and DMSO (DMSOt), 10mL duplicate subsamples were removed from the UVT bags using a 60 mL plastic syringe and dispensed into 20 mL acid-cleaned, transparent serum vials with Teflon faced caps (Wheaton PN 224100).  All bottles, tubing and caps used for samples were pre-cleaned by thorough rinsing with 10% HCl and distilled water.  Serum bottles used for DMSO samples were pre-heated in a muffle furnace for ~2 hours at 125°C and capped, to remove DMSO that might adhere to the glass walls166.  After transfer of samples to the serum vials, DMS was sparged out of solution under N2 flow and analyzed as described below.  After DMS sparging and analysis, samples were treated with either 2 mL of 10N NaOH (for DMSP samples) or 2mL of TiCL3 (for DMSO samples), capped with teflon faced caps and allowed to sit for 12-24 hours prior to subsequent analysis of the DMS produced through chemical conversion of DMSP and DMSO.  We used the same procedure for the analysis of dissolved DMSP and DMSO in 10 mL duplicate subsamples, except that subsamples were gently filtered through 0.2 µm acrodisc syringe filters according to Kiene and Slezak129.  In tests conducted over a three-week period, we found no significant difference between our DMSPd measurements made using the syringe-based method and the small volume gravity filtration. Particulate DMSP (DMSPp) was derived from the difference between DMSPt and DMSPd measurements. All semi-weekly DMS/P/O samples were analyzed on an Agilent 355 sulfur chemiluminescence detector (SCD; model no. G6603A), in conjunction with a custom-built purge and trap (PT) system, controlled by LabVIEW software.  The PT system (described in detail in chapter three) was set up to sparge samples with N2 at 400 mL min-1, thereby stripping volatile DMS from solution onto a Tenax (Restek model no. 25701) packed stainless steel trap (~6” long ~1/4” o.d.) held at room temperature.  Following completion of the sample sparging (10 minutes), the trap was heated electrically, by passing high current  135 onto the stainless steel.  Analytes were released from the heated trap into a ~10 ml min-1 N2 carrier flow connected to a 2.4m chromatographic packed column (Supelco 330), where DMS was separated from other S compounds.  The SCD has a detection limit of ~0.1 nM when 10mL samples are sparged. We used a custom LabVIEW program to interface the PT system with the SCD, and to read in raw voltage output from the detector as a continuous analog signal.  Custom MATLAB scripts were used to automatically identify peak locations and compute integrals in raw output files.  The average baseline prior to each peak was subtracted from the peak signal, and blank-corrected peak areas were integrated using a Simpson’s 3/8 rule.  Uncertainty for all concentration measurements was calculated as one standard error from the mean of the replicate samples. Calibrations were performed 1-2 times per week using known standards of commercially available DMS (Sigma-Aldrich 471577, DMSO (Sigma-Aldrich D8418) and DMSP.  DMS and DMSO standards were prepared by gravimetric dilutions of pure liquids, which are commercially available, while DMSP was synthesized from DMS and Fluka 3-bromopropionic acid with the method of Challenger and Simpson111 employed more recently by Dacey and Stefels5, and used to generate stock solutions in methanol.  Dilutions of primary stock solutions into de-ionized Milli-Q water were used to create final DMS/O/P standards spanning a wide range of concentrations (0-500nM).  Calibration curves were linear (r2≥0.97) over 3 orders of magnitude. Continuous measurements of dissolved DMS, DMSP and DMSO at Station A were made over a ~24hr period on December 25th and January 20th using an early prototype of the Organic Sulfur Sequential Chemical Analysis Robot (OSSCAR; see chapter three).  This system was designed to automate the collection of water samples and sequential analysis of DMS, DMSO and DMSP.  Seawater from the continuous Station seawater supply (with an intake at ~ 6 m depth in Arthur Harbor) was filtered through a 1.2µm pore in-line filter cartridge (Parker 22-A0R10-012-5) and analyzed for DMS, DMSPd and DMSOd every ~2 hours.  A programmable syringe pump was used to withdraw sub-samples from a continuously flowing seawater supply line.  DMS was first stripped from solution and analyzed by SCD as described above, with DMSO and DMSP sequentially converted to DMS using enzymatic reduction109 and rapid alkaline hydrolysis66, respectively.  The sparge  136 vessel and tubing were rinsed thoroughly with Milli-Q water between samples.  Full details of the method are described elsewhere (see chapter three). Additional high-frequency DMS measurements were conducted at station A from mid December to March using MIMS2. For these measurements, water from the station's continuous unfiltered seawater supply was pumped through a flow-through sampling cuvette attached, via a silicone membrane, to a quadrupole mass spectrometer.  DMS was measured by detecting ions with a mass to charge ratio of 62 (m/z 62) every ~30 seconds.  The details of this system are described in Tortell et al.2.  We calibrated the MIMS DMS signal 2-3 times per week using freshly prepared standards diluted from DMS stock solutions149.  6.2.3 Process Studies and Rate Experiments A variety of experimental studies were conducted to quantify rates of key DMS/P/O production and consumption processes over the seasonal cycle.  For all rate experiments, seawater was collected from station B using a Monsoon bilge pump into a clean 20L carboy and homogenized by inverting the carboy 10 times.  Four to six replicate UVT bags were filled with 3L each, with the remaining headspace removed prior to sealing bags with gas tight clips.  The filled sampling bags were stored in the dark for <1 hour until experiments were begun and initial sub-samples were collected.  We conducted two types of rate experiments: competitive inhibition (CI) experiments and stable-isotope tracer experiments.  Competitive inhibitor experiments were used to examine gross production terms for DMS/P in the presence of analog compounds that blocked the consumption of these compounds.  Isotope tracer experiments were used to follow the production of DMS from various precursors.  After adding either competitive inhibitors or isotopically-labeled tracers to samples, the bags were mixed by inverting 10 times.  Subsamples were collected using a 60mL luerlok BD syringe through a Teflon bag port equipped with luerlok fittings.  Following this initial sampling time-point, bags were immediately transferred to an outdoor incubator maintained close to ambient sea surface temperature by a continuous flow of seawater, and covered with 1-2 layers of neutral density screening to light levels to ~ 50% of sea surface values.  Over the course of ~ 6 h (one time point every ~ 1.5 h), subsamples were filtered through a 1 µm GF/F filter into a 20 ml pre-cleaned glass bottle and analyzed using  137 the SCD for DMSP CI experiments, or purge and trap capillary inlet mass spectrometer (PT-CIMS; described below) for tracer experiments and DMS CI experiments. For competitive inhibition experiments, we used glycine betaine (Sigma Aldrich B2629) and stable isotope labeled D-6 DMS (CDN isotopes D-1509) to inhibit the consumption of DMSP and DMS, respectively.  D-6 DMS has all 6 hydrogen atoms on the two CH3 groups replaced with dueterium.  Glycine betaine is a known competitive inhibitor of DMSP uptake47, and we have found that D-6 DMS is consumed at the same rate as naturally occurring DMS (Asher et al. unpublished data), and can be used to block uptake of unlabeled DMS.  We thus used additions of 1.7µM glycine betaine according to Kiene and Gerard47, and 1.5 µM D-6 DMS as competitive inhibitors for natural DMSP and DMS uptake, respectively.  The consumption of DMS and DMSP is effectively blocked in the presence of these competitive inhibitors, enabling us to track their gross production rates.  UVT bags containing natural seawater without any additions were used as control replicates for these experiments. Gross and net DMSP production in the glycine betaine addition experiments was monitored in 20mL subsamples using the SCD.  Samples were first sparged to remove DMS, treated with NaOH, and analyzed via SCD 12-24 hours later for DMSP.  To measure net and gross DMS production, we used purge and trap capillary inlet mass spectrometry (PT-CIMS) as described by Asher et al.3.  This analytical system couples a custom-built purge and trap gas extraction system with gas chromatographic separation of various sulfur compounds and detection using Hiden Analytical quadrupole mass spectrometer.  Our PT-CIMS has a detection limit of ~0.1 nM DMS in 50mLsamples, and enables us to discriminate the isotope labeled D-6 DMS competitive inhibitor (m/z 68) from natural (unlabeled) DMS (m/z 62).  In our experiments, we used 50 mL sub-samples from control and experimental bags (i.e. with and without isotopically-labeled D6-DMS), and followed the time-course of m/z 62 and 68 over ~ 6 hours. In our DMS CI experiments, where D6-DMS is added at >200 times the concentration of natural DMS, a small secondary source of m/z 62 DMS results from ion source fragmentation of the added competitive inhibitor.  We thus corrected the measured m/z62 DMS signal for ~1% mass fragmentation from the m/z68 DMS competitive inhibitor (http://webbook.nist.gov/; Table 6.1, eq. 1). We calculated the gross and net DMS/P  138 production in CI experiments from the slope of the natural logarithm of unlabeled DMS/P concentrations over time (Table 6.1, eq. 2 and 3), using a linear regression of the concentrations during the first three time points.  In two cases when regressions had r2<0.4, or the experiment had gross DMS/P production less than net DMS production, the data were excluded from further analysis.  In total we excluded data from 2 out of 10 experiments.  Gross DMS/P production was determined from the rate of change of these compounds in samples amended with competitive inhibitors of DMS and DMSP uptake.  Net DMS/P production was determined from the rate of change of DMS and DMSP in control samples with no added inhibitors.  Inferred DMS consumption was calculated from the difference of the gross and net production rates.  Measured rate constants of DMS production and consumption were multiplied by the respective DMS/P concentrations measured in the water used for incubations to yield estimated in situ rates of DMS production and consumption (nMd-1) at station B during the time of our sampling. Isotope tracer experiments were used to quantify specific DMS production pathways (DMSP cleavage and DMSO reduction) and to measure rates of gross DMS consumption.  For these experiments, we amended four replicate bags with tracer level (i.e. < ~ 20% of ambient) additions of D-3 DMS (i.e. all 3 H atoms of one CH3 group replaced by deuterium), D-6 deuterated DMSP, and C-13 labeled DMSO to achieve final concentrations of ≤ 1nM, ≤ 0.5nM and ≤ 0.7nM.  The rate of change in labeled DMS (either D3, D6 or 13C) was measured over 4 time points using PT-CIMS.  After DMS sparging, replicate sub-samples from time points T0 and T4 were treated with 6 mL of 10N NaOH and 6mL TiCl3 to measure (via PT-CIMS) the concentrations of both labeled and natural DMSO and DMSP. A number of steps were required to obtain rate estimates from the raw tracer experiment data.  We calculated gross DMS consumption from D-3 DMS disappearance, while the rate of DMSPd cleavage and DMSOd reduction were derived from the rate of formation of D-6 and C-13 labeled DMS, respectively.  The loss rate of D-3 DMS provides an estimate for gross DMS uptake, since there is no natural formation process for this labeled species.  In previous work (see chapter 5), we have shown that the rate of labeled DMS consumption (D3 or D6) does not differ measurably from that of unlabeled DMS.  To calculate gross DMS uptake, we used a pseudo first-order equation (Table 6.1, eq. 4), where kdms_cons is the observed rate constant, t is time, and [D-3 DMS] is the concentration of the  139 added tracer.  As the tracer experiments were conducted in UV transparent bags under natural light, kdms_cons represents DMS removal from both photo-oxidation and biological DMS consumption. The C-13 labeled DMS derived from DMSOd reduction has a m/z ratio of 64, since there is a 13C label on two carbon atoms.  In our experimental system, there are two other sources of DMS with this charge to mass ratio.  The first is derived from the background pool of S34-containing DMS.  Based on the natural abundance of S34, we assume that m/z 64 DMS represents 4.3% of unlabeled DMS pool.  The second source of m/z 64 DMS results from ion source fragmentation of m/z 65 DMS from the added D3-DMS tracer.  Under the ionization conditions used in our mass spectrometer (electron impact ion source with 70 eV ionization energy), the m/z 64 DMS represents ~30% of the m/z 65 pool (http://webbook.nist.gov/).  We thus corrected the apparent DMS m/z 64 in our experiments for these two source terms (Table 6.1, eq. 5), to calculate the concentration of DMS derived specifically from the reduction of 13C-labeled DMSO. To derive gross rates of DMSO reduction, we took into account the loss rate of C-132 labeled DMS, using the DMS consumption rate constant measured with the D-3 DMS tracer (Table 6.1, eq. 6).  Gross DMSOd reduction rates were then computed from the slope of the natural logarithm of corrected m/z 64 DMS against time (Table 6.1, eq. 7). The D6-DMS derived from the cleavage of D-6 DMSPd has a m/z ratio of 68. To the best of our knowledge, D-6 DMS is the only compound with this m/z ratio in our experimental system, so DMS68was not corrected for any background signals.  The appearance of D-6 DMS over time in our experiments represents DMSPd cleavage exclusively.  Like C13-DMS and natural DMS, however, D6-DMS is consumed during these experiments and must be corrected for gross DMS consumption (Table 6.1, eq. 8).  We used the slope of the log-transformed corrected D-6 DMS concentrations over time to calculate the DMSPd cleavage rate constant (Table 6.1, eq. 9). For all tracer measurements, we assumed initial DMS concentrations of 0.1 nM in cases where the actual values were below our detection limit.  This assumption was necessary in 8% of (5 out of 60) T0 measurements.  This assumption has only a minor effect on our results.  For example, if the actual T0 concentrations were 10-fold lower than the ~0.1nM detection limit (i.e. ≤0.01nM), derived rate constants would be between 3-20%  140 higher.  As with CI experiments, DMS production rates from tracer studies were determined using a linear regression of log-transformed concentrations against time during the first three time points, and measured rate constants were multiplied by the DMS/O concentrations measured at Station B to estimate in situ rates of DMS production (nM d-1).  6.2.4 Grazing Studies We conducted two types of experiments to measure the impact of micro and meso-zooplankton grazing on DMSP cycling and DMS production.  We used dilution experiments to quantify the influence of microzooplankton grazing on natural DMS production (5 experiments between December to February), and bottle grazing experiments to examine the potential contribution of krill (E. superba) to DMS production (three experiments in February).  In the krill grazing experiments, we used isotope tracer studies to examine DMS/P cycling in detail. For dilution experiments, we followed the method of Landry and Hasset167, with a few modifications as suggested by Saló et al.157.  Briefly, seawater was sampled from ~5 m depth at dusk from Station B using Niskin bottles and immediately returned to the laboratory for subsequent processing in the 4 °C cold room.  A portion of the collected water was filtered through a 0.1 mm cartridge using acid-clean silicone tubing, and added in varying amounts to 1L acid–cleaned polycarbonate bottles.  The residual volume of the 1L bottles was filled with the unfiltered seawater collected from Station B, yielding duplicate bottles containing 100%, 75%, 50% and 25% unfiltered water without any headspace.  These replicates were spiked with nutrient additions (10mm nitrate and 0.6 mm phosphate).  To measure the natural production of chlorophyll a (Chla) and DMSP, one additional pair of bottles was filled only with unfiltered water without added nutrients.  All of the bottles were placed upright in an outdoor incubator with 2 layers of neutral density screening and flowing surface seawater.  Initial T0 levels of Chla, bacterial abundance, and sulfur compounds were sampled in the starting water used for incubations, using pre-cleaned syringes and teflon tubing.  Chla concentrations were measured using fluorometric analysis as described below (see ancillary measurements) on 200 ml GF/F filtered samples from the sampling carboy, while bacterial abundance was measured using flow cytometric analysis (see ancillary measurements).  The  141 T0 values for Chla and DMS/P in each experimental bottle were computed from the dilution fraction and the T0 value in the bulk filtered and unfiltered samples.  After 24 hours, duplicate bottles were removed from the incubator and sampled for Chla, bacterial abundance, DMSP, DMSO and DMS (T24).  The reduction in Chla (Chlagraz) was calculated as the natural logarithm of ChlaT24/ChlaT0, and the removal rate was calculated from linear regression of Chlagraz against the dilution fraction167,168. Similarly, the DMSP removed in each sample was calculated as the natural logarithm of DMSPt_T24/ DMSPt_T0157, and the removal rate (DMSPgraz_rate) was calculated from linear regression of DMSPgraz against the dilution fraction157.  The rate of DMS production due to DMSP grazed (DMSprod_rate) was calculated as the linear best fit of the change in DMS concentrations over 24 h (DMST24 - DMST0) vs. moles DMSP grazed (DMSPt-T24 - DMSPt-T0, Table 6.1, eq. 12). Data obtained from individual micro-grazing experiments were used to derive an estimate of water column DMS production (nMd-1) based on grazing rates computed over the seasonal cycle.  This estimate was calculated as a product of the rate of DMS production due to grazing (DMSprod_rate), the DMSP removal rate due to grazing (DMSPgraz_rate) and the DMSPt concentrations (Table 6.1, eq. 10).  For this analysis, we excluded results from 1 (out of 6) grazing experiment, where Chla concentrations did not show a linear dependence (r2 > 0.5) on the dilution fraction. In late February, we conducted several measurements of krill grazing rates during a period of high E. superba abundance in the waters around Palmer Station.  Experiments were conducted over a 12-hour period using Station B water dispensed into six 50L carboys.  The carboys were filled at local dusk with water from below the mixed layer (~20m depth) using a Monsoon pump.  We spiked the carboys with 1L of tracer spike solution containing D-3 DMS and D-6 DMSP to obtain final concentrations of 1.42 nM and 1.33 nM, respectively.  Carboys were then transported back to Palmer Station for further processing.  Once on station, the carboys were sampled for initial (T0) measurements of Chla, DMS/Pd and DMSPt in 50 mL sub-samples, sampled using syringes and clean teflon tubing.  DMS was analyzed via PT-CIMS, and samples were then treated with 10N NaOH and left for ~12 hours prior to DMSP analysis (PT-CIMS).  After these T0 measurements, we added 10 juvenile krill to 3 of the 6 carboys to study the impact of krill on DMSP grazing, net DMS production and rates of DMSPd cleavage.  The other 3 carboys served as experimental controls.  The krill added to  142 these carboys were obtained from net tows (700 µm mesh diameter) deployed from a zodiac equipped to locate krill using acoustic measurements.  Carboys were capped, sealed with parafilm and moved to the outdoor flow-through incubator, where they were slowly rotated every 3-4 hours to prevent the phytoplankton from settling to the bottom.  After 24 hours, the carboys were removed and sampled a second time for Chla, DMS, and DMSP. We calculated the net change in DMS, DMSP and Chla over the course of the krill grazing experiments (24h), as well as DMS production according to equation 10.  Error estimates for each rate measurement were calculated as the standard error of triplicate rates (DMS/P or Chla d-1).  Mean DMS production in these experiments was normalized to the abundance of krill (individuals m-2) in our experiments (10 krill in a 50L carboy = 0.2 krill  L-1).  These krill-specific DMS production rates were then used to derive an estimate of the depth integrated in situ DMS production from krill grazing (Table 6.1, eq. 11).  For this computation, krill-specific DMS production rates were multiplied by the krill densities (ind. m-3) derived from acoustic measurements (ind. m-2) according to Bernard et al.169 and the mean bathymetric depth of our study area (~88m; Fig. 6.1).   6.2.5 Inferring Total DMS Production/Consumption Terms from Specific Rate Measurements We estimated semi-weekly values for total DMS production and consumption rates using our rate measurements and computed sea-air fluxes (Table 6.1, eq. 12).  The following specific rates were included in the total DMS production term: 1) DMSP cleavage, 2) DMSO reduction, 3) DMS production due to microzooplankton, and 4) DMS production due to krill grazing.  The total DMS consumption term was comprised of gross DMS consumption (from both biological and photo-chemical processes), and sea-air flux (see below for details of the calculations).  Dissolved DMS consumption rates were derived from gross DMS consumption measured in tracer experiments and inferred gross DMS consumption in CI experiments.  We calculated the uncertainty for each term from the standard error of the derived rate constants and concentration measurements.  Where necessary (i.e. DMS production/ consumption rates), uncertainty was propagated using a Taylor Series expansion.   143 6.2.6 Ancillary Measurements Mixed layer depths at LTER station B were derived from a 0.125 kg m-3 density difference criteria, based on bi-weekly temperature and salinity depth profiles with a Seabird SBE 19plus Seacat Profiler.  The Palmer Station Terra Laboratory provided continuous measurements of wind speed and photosynthetically active radiation (PAR) during our study period. Krill abundance was determined every ~2-3 days from December to February using acoustic surveys169 from a standard set of transects in the LTER sampling grid.  Discrete samples for macronutrients, Chla and accessory photosynthetic pigments were also obtained semi-weekly from ~10m depth at LTER Station B.  Samples for macronutrients were filtered through a 0.2 µm membrane, frozen, and analyzed using a Bran & Luebe autoanalyzer.  200 mL samples for Chla determinations were filtered onto a glass fiber filter (GF/F with a ~0.7 µm nominal pore size), extracted in acetone for 24 hours, and measured on a fluorometer prior to and post acidification to correct for phaeopigments170.  For accessory photosynthetic pigments, 2L samples were filtered through 47mm GF/F filters and stored frozen at -80C until analysis by high performance liquid chromatography (HPLC; Goldman et al.171).  The relative abundance of diatoms, Phaeocystis and cryptophytes was determined as described by Goldman et al.171 from the abundance of three pigments: fucoxanthin, 19’hexanoyloxyfucoxanthin, and alloxanthin. Sea-air DMS flux was calculated from DMS concentrations at Station B and wind speed measurements derived from meteorological sensors at Palmer Station.  For the purposes of DMS sea-air flux calculations, we employed the Nightingale et al.70 wind speed parameterization and the Schmidt number69 to calculate gas exchange.  Due to the quadratic dependence of the piston coefficient on wind speed, we computed the piston coefficients at the resolution of wind speed measurements prior to calculating the daily mean piston coefficient.  Air-sea flux was calculated as the product of DMS concentrations measured at station B and the corresponding daily mean piston coefficient. Atmospheric DMS concentrations were assumed to be zero for the purpose of these calculations.   144 6.3 Results 6.3.1 Surface Water Hydrography and Plankton Biomass Our field campaign captured a significant portion the seasonal cycle (late spring, summer, and early fall) at Palmer Station in 2012/2013.  By November, sea ice had retreated, exposing surface waters to increased gas exchange. By December, the mixed layer shoaled from ~20m to ~8m (Fig. 6.2d), while average daily surface PAR levels increased to ~600 µE m-1 s-1, resulting in a significantly increased mixed layer mean irradiance.  In late November, we observed a massive diatom-dominated phytoplankton bloom that achieved peak Chla levels in excess of 600 mg m-2 and coincided with the drawdown of ~ 30 mM nitrate in surface waters (Fig. 6.2a, 6.2b).  This spring bloom crashed within two weeks, and nitrate levels were restored to 20 µM (Fig. 6.2a), likely due to a lateral mixing or vertical entrainment event43,165. Following the initial diatom bloom, low phytoplankton biomass persisted for ~ 2 months (Fig. 6.2a), with the assemblages containing a mixture of chlorophytes, diatoms and Phaeocystis (Fig. 6.2b).  In February, a second, deeper phytoplankton bloom developed (Fig. 6.2a), which was comprised of a mixture of diatoms and Phaeocystis (Fig. 6.2b).  Krill biomass fluctuated significantly during our sampling period, with sporadic influxes of E. Superba observed at various times between December and February.  High rates of bacterial respiration (measured with 3H-Thymidine, H. Goldman et al.171) were observed directly following the spring bloom (Fig. 6.2c), with smaller periodic increases observed through mid-February.  6.3.2 Seasonal DMS/P/O Distributions DMS concentrations, measured semi-weekly at Station B, fluctuated throughout the seasonal cycle, with an overall mean of 4.7±4.6 nM, and a range from < 0.1 nM (detection limit) to 19 nM (Fig. 6.3b).  The largest, albeit short-lived, DMS peak followed the crash of the main phytoplankton bloom in early-December.  Although neither vertical diffusivity nor lateral advection measurements were part of this study, examination of CTD data in early December shows an upwards doming of isohalines, indicative of deep-water entrainment or lateral advection172.  Although we lack data on DMS/P/O concentrations in potential entrainment source waters, we presume that they would be lower than surface values, and  145 could help explain the low DMS concentrations that followed this DMS maximum in mid-December  We also observed several other instances (e.g. December 27th and January 20th, and Jan 31st) of DMS accumulation, unrelated to changes in phytoplankton biomass.  In general, DMS measurements from MIMS showed good coherence with the semi-weekly DMS measurements from Station B, and provide additional, high frequency data (see below) through to the end of our sampling period.  The MIMS data suggest that surface DMS concentrations continued to climb through out February (after semi-weekly measurements had ceased), with a decreasing trend by early March.  DMS concentrations measured at Station B were weakly (though statistically significantly) correlated with the mixed layer depth (r = -0.43, p <0.05; Table 6.2).  In contrast, DMS concentrations at Station B did not show any significant correlations with Chla, the abundance of Phaeocystis, PAR, or UV (Table 6.2). In addition to our semi-weekly DMS measurements at station B, we followed the seasonal evolution of total and dissolved DMSP/O concentrations.  DMSPt concentrations at Station B did not exceed ~60nM for the majority of the season (mean 49.3 ± 43.4;  Fig. 6.3c), although very high concentrations (~ 150 nM) were observed during the final two sampling points in late February (Fig. 6.3c).  Note that the data collection was cut short due to an instrument malfunction.  DMSPt concentrations were closely coupled to the abundance of Phaeocystis (r = 0.85, p < 0.001; Table 6.2).  DMSPd remained < 2 nM during the early part of our sampling (until December), after which concentrations began to accumulate to values of ~ 10 nM by mid February  The large difference between DMSPt and DMSPd suggests that phytoplankton contained substantial concentrations of particulate DMSP.  Although DMSPd concentrations were not significantly correlated with DMSPt (r = 0.42, p = 0.15; Table 6.2), DMS concentrations did track DMSPd concentrations at Station B (r = 0.54, p < 0.05; Table 6.2).  DMSOt concentrations remained lower than DMSPt concentrations (≤ 30nM) during the majority of our study period, with maximum concentrations observed in mid-February (Fig. 6.3c).  During the first half our sampling season, DMSOd encompassed the bulk of the DMSOt pool, suggesting a low particulate pool.  However, during latter half of sampling season (February and March) the DMSOt pool increased significantly to ~160 nM, without a commensurate increase in the DMSOd pool.  Differences between DMSOt (43.2 ± 46.2 nM) and DMSOd (11.1 ± 15.4 nM) concentrations suggested the accumulation of a particulate  146 DMSO pool towards the end of our sampling period.  Across the full sampling season, DMSOd concentrations were well correlated with DMSOt (r = 0.78, p < 0.01; Table 6.2) and also with DMS concentrations (r = 0.75, p < 0.01; Table 6.2).  6.3.3 High Frequency DMS/P/O Measurements and Diel Cycles In addition to the regular semi-weekly sampling at Station B, we also conducted high frequency analysis of surface DMS concentrations, and diel measurements (~ every two hours) of DMSP and DMSO during two days during the mid-summer season from the SWP.  (Note that MIMS DMS data are not available during the early part of the season due to instrument problems, and discrete measurements are not available during the last part of the season due to logistical and personnel constraints).  As discussed above, results from the MIMS DMS analysis agreed well with the discrete measurements at Station B (Fig. 6.3a).    Beyond providing validation of the DMS data obtained by SCD, MIMS data provide additional information on the high frequency DMS variability that is not captured by semi-weekly discrete measurements.  As an example of this, Figure 6.4 shows apparent diel cycles in DMS concentrations during mid-February  During this one-week period, we observed increasing DMS concentrations in the morning, with maximum DMS concentrations in the afternoon (12-3pm), followed by apparent DMS consumption and decreasing surface concentrations.  The timing of maximum DMS concentrations corresponded well with the peak in photosynthetically active radiation (PAR; Fig. 6.4b).  On February26, we did not observe a daytime increase in DMS concentrations.  Rather, DMS appears to have been influenced by low PAR and high wind speeds, which led to DMS removal via sea-air flux (Fig. 6.4c). Discrete DMSPd and DMSOd measurements every 2.5 hours also revealed diel cycles at Station A (Fig. 6.5).  On December 25, both DMSPd and DMSOd showed a distinct diel cycle, with both compounds showing maximum concentrations during the mid/late afternoon.  During this sampling day, DMSPd concentrations peaked 2.5 hours earlier than DMSOd concentrations.  On January 20, DMSPd concentrations were significantly lower on January 20, and showed a much smaller diel cycle than on December 25.  In contrast, the diel cycle of DMSOd concentrations was similar for both days, albeit with an earlier increase on January 20.  We observed a statistically significant correlation between DMSOd and surface PAR on  147 both days, and between DMSPd and surface PAR on December 25 (r ≥ 0.7, p < 0.05; Fig. 6.5).  To account for the offset in peak DMSOd concentrations relative to solar noon, we used a lag of 2.5 hours for correlations with DMSOd.  On December 25th, PAR reached a maximum of ~1100 µE m-2 s-1, with moderate wind speeds (2.37±0.74 m s-1).  By comparison, maximum PAR levels were ~ 2-fold lower on January 20, while average wind speeds (4.01+1.02 m s-1) were nearly two-fold higher on this day.  The lower PAR and higher wind-speeds observed on January 20th may help explain the lower amplitude diel cycle of DMSPd concentrations and the lower correlation with PAR on that day (see discussion).  6.3.4 Rate Measurements and Process Studies We measured various DMS/P/O consumption and consumption pathways in order to understand the processes driving the observed temporal DMS/P/O variability in the WAP over a significant portion of the seasonal cycle.  In tracer experiments, we simultaneously measured rates of DMSPd cleavage, DMSOd reduction, and gross DMS consumption.  Figure 6.6 shows an example of the data obtained from tracer experiments).  Across the our full sampling season, DMSPd cleavage rates averaged 2.6 ± 4.1 d-1 (range 0-11.8 d-1), DMSOd reduction rates averaged 1.7 ± 2.2 d-1 (range 0-6.9 d-1) and gross DMS consumption averaged 3.6 ± 5.0 d-1 (range 0-13.9 d-1).  In CI experiments (see Fig. 6.7 for an example), we also measured rapid and variable rate constants of gross DMS production (mean = 2.9 ± 3.6; range 0-8.6 d -1), and net DMS change.  Rate constants of net DMS change (0.24 ± 0.72; range -1.0-1.9 d-1) were 10-fold lower, on average, than the rate of gross DMS production.  From the difference in gross DMS production and net DMS production measured in competitive inhibitor experiments, we calculated high gross DMS consumption rates (3.6 ± 1.9; range 0-5.5 d-1).    The DMS consumption rates calculated in CI experiments were in good agreement with our direct measurements from D3-DMS tracer experiments.  Compared with the highly variable rates of DMS production and consumption, we found that gross DMSP production rates (derived from Glycine Betaine experiments) remained fairly constant over the season (2.4 ± 0.23 hr-1; Table 6.3).  148  6.3.5 Grazing Experiments Dilution experiments enabled us to quantify the effect of micro-zooplankton grazing on Chla and DMS/P concentrations, and subsequent DMS production (see Fig. 6.8 for an example).  Results were tabulated from five independent micro-grazing experiments, conducted between late December and early February (Table 6.4).  These repeated experiments showed Chla grazing rates ranging from 0.1 to 0.2 d-1, with larger grazer effects on DMSP (range 0.25 – 1.7 d-1 removal rates).  As the summer progressed, and the phytoplankton community composition shifted from one dominated by either diatoms or chlorophytes to an autotrophic community composition split between Phaeocystis and diatoms (Fig. 6.2), the grazer effects on DMSP and Chla removal decreased (Table 6.4).  Indeed, the grazing effect on DMSP removal (r = 0.94, p < 0.05, n = 5) exhibited a strong negative relationship with abundance of Phaeocystis (Fig. 6.9).  For most of the experiments (four out of five), we measured a ratio of DMS production to DMSP removal less than 0.15 (Table 6.4).  In the final experiment, however, (February 9), this ratio was close to 1, suggesting that almost all of the DMSP removed by grazing was being converted to DMS.  We observed a positive correlation between DMS production associated with DMSP removal and Phaeocystis (0.91, p < 0.05, n = 5). Results from three 24-hour krill grazing experiments demonstrated that krill also significantly influenced net DMS production and net DMSPt removal.  Results from these experiments are summarized in Figure 6.10.  Relative to the control (no krill bottles), the presence of krill increased net DMS accumulation by ~140% on average, while also increasing DMSP removal by ~150% on average.  Similarly, net Chla accumulation decreased by ~90% on average in the presence of krill.  Contrary to expectation, we observed no significant change in derived rate constants for DMSPd cleavage (as inferred from changes in m/z 68 DMS) between krill and control treatments.  Together, these results suggest that the krill grazing increased DMSP removal and direct DMS production; however, the influx of dissolved DMSP due to grazing did not stimulate specific rates of algal or bacterial mediated DMSPd cleavage within the time frame of these experiments (see discussion).   149 6.3.6 Trends in DMS Production/Consumption Rate constants of DMS production derived from tracer and grazing experiments, varied considerably over the seasonal cycle (Table 6.5).  We observed low initial rates of DMSPd cleavage that ramped up rapidly following the spring bloom in early Dec to reach maximum rates >30 nMd-1 (Table 6.5), driven by high rate constants for DMSPd cleavage (Fig. 6.11).  Specific rates of DMSPd cleavage (d-1) were negatively correlated with net primary production (r = -0.84, p < 0.05; n= 5).  Similarly, DMSOd reduction rates remained below detection prior to the spring bloom.  These very high rates of DMS production from DMSPd cleavage and DMSOd reduction gave way to moderate DMS production rates from DMSPd cleavage and grazing (15.1 ± 14.6 nMd-1) for the remainder of the austral summer.  Throughout most of sampling season, DMSPd cleavage and micro-zooplankton grazing dominated total DMS production, with smaller contributions from krill grazing and two instances of DMSOd reduction in mid-December (Table 6.5).  Low in situ DMS production due to krill grazing reflects the patchy distributions of krill in the water column (Table 6.5).   Gross DMS consumption rates also reached a maximum after the spring bloom and decreased thereafter to 16.6± 14.3 nMd-1 (Table 6.5), roughly balancing DMS production from DMSPd cleavage and grazing.  Rate constants for gross DMS consumption peaked in mid-December, during a post spring-bloom period of maximum bacterial respiration (Fig. 6.11).  Across our full seasonal sample, we observed a strong linear correlation between specific rates of gross DMS consumption and bacterial respiration (r = 0.83, p < 0.01; n = 9).  Gross DMS consumption rates were >10 times higher than DMS removal due to sea-air flux (Table 6.5), suggesting that gross DMS consumption dominated DMS removal.   6.3.7 Seasonal DMS Budget In addition to examining seasonal changes in the various DMS production and consumption terms, we also computed overall seasonal means for DMSPd cleavage, DMSOd reduction, gross DMS consumption, sea-air flux, mircrozooplankton grazing, and krill grazing (Fig. 6.12).  This analysis was conducted to provide some insight into the average contribution of these various processes to DMS cycling over the seasonal cycle.  Multiple sources of DMS were required to balance the gross DMS consumption in the seasonal DMS  150 budget.  Despite the clear influence of sea-air flux on DMS concentrations during a number of short-term periods of high wind speeds (Fig. 6.4), sea-air flux contributed, on average, only a minor negative term to the DMS budget.  Average rates of DMS production over the seasonal cycle ranked as follows: 1) micro-zooplankton grazing, 2) DMSPd cleavage, 3) DMSOd reduction, and 4) krill grazing, although average DMS production from DMSPd cleavage and micro-zooplankton grazing were similar in magnitude.  Over the seasonal cycle, the mean net balance of our rate measurements was slightly positive but not significantly different from zero (Fig. 6.12).  6.4 Discussion Our results contribute to a growing database of DMS concentration measurements in the WAP region43,165.  The few available data from Palmer Station suggest that spring / summer DMS concentrations in this region average ~5nM, ranging from < 1nM to ~20nM, with several peaks over the seasonal cycle.  To date, there have been no reports of extraordinarily high DMS levels in the WAP, as observed in several Antarctic polynya systems.  Below, we discuss a number of factors that may lead to the relatively lower DMS/P concentrations in WAP.  The new observations presented here increase our understanding of the inter-annual and seasonal variability of DMS, and the related compounds DMSP and DMSO.  Our work also adds a new dimension, with higher frequency DMS/P concentration measurements, the first DMSO data, and the use of multiple experimental approaches to quantify DMS production/ consumption terms.   6.4.1 Concentrations Our observed concentrations of DMS and DMSP agree with previously published values at Palmer Station43,165, who found similar mean values and ranges.  In the absence of any previous DMSO measurements, there are no data available for comparative purposes.  Our results indicate that nearly all DMSOt remained in the dissolved phase until mid-summer, indicating that phytoplankton did not produce large quantities of intracellular DMSO, as compared to DMSP.   Moreover, DMSOd tracked DMS concentrations,  151 supporting the hypothesis that DMS is the main precursor of DMSOd in surface waters due to the photochemical and biological oxidation of DMS. Correlations between concentrations of DMS/P and ancillary variables (Table 6.2) indicate that phytoplankton species composition and water column stratification exert strong controls on DMS/P dynamics.  We observed a strong correlation between DMSPt and the abundance of Phaeocystis (Fig. 6.2); however, unlike what has been reported in the Ross and Amundsen Sea polynyas, we did not observe the accumulation of extremely high Phaeocystis biomass (i.e. > 50 mg m-2 Chla; Fig. 6.2).  As such, it is possible that the lack of massive DMS accumulation in the WAP results from the lower absolute abundance of Phaeocystis, as compared to the Ross and Amundsen Sea polynyas149.  Stratification has been suggested as a key variable influencing phytoplankton community composition in the Southern Ocean, with shallow mixed layer depths (5m - 20m) favoring diatom accumulation, and deeper mixed layer depths (20m - 50m) favoring Phaeocystis accumulations173.  The relatively shallow mixed layer depths observed in the WAP region, may thus favor diatom growth, thereby limiting DMS accumulation in surface waters.  We note, however, that high Phaeocystis biomass and DMS accumulation has been reported in stratified waters of the Amundsen Sea.  This result may reflect the physiological effects of high irradiance in well-stratified surface waters, which could subject microbial assemblages to considerable light stress, stimulating DMS/P/O productions10 and deterring bacterial consumption of DMS28.  In support of these hypotheses, we observed a correlation between DMS and mixed layer depth.  We thus suggest that phytoplankton taxonomic composition exerts a first-order control on surface water DMS/P accumulation, with mixed layer depth and irradiance exerting a secondary control by influencing plankton physiological ecology. Previous work has relied on the analysis of DMS/P/O in discrete samples, limiting our understanding of temporal variability.  Yet, high rates of DMS production/ consumption in surface waters indicate the potential for rapid changes in DMS concentrations.  Ship-based DMS surveys using MIMS have been conducted in several Antarctic regions, but these measurements cannot disentangle spatial vs. temporal components of the observed variability.  In comparison, our MIMS data from Palmer station provide an opportunity to observe clear diel cycles in DMS (Fig. 6.4), revealing the influence of PAR and wind speed on short-term fluctuations in DMS concentrations (Fig. 6.4).  Our Palmer results also  152 represent the first deployment of an automated DMSP/O system (see chapter three), enabling us to document temporal variability in DMSP/O concentrations, alongside of our MIMS DMS measurements.  As with DMS, PAR and wind (i.e. environmental forcing) appear to control DMSP/O temporal variability on short time-scales when phytoplankton biomass and species composition remain relatively constant (Fig. 6.5).  The highest DMS/P/O concentrations were observed in the afternoon (Fig. 6.4; Fig. 6.5), with high PAR driving larger amplitude cycles.  This result provides support for the role for DMS/P/O as an anti-oxidant, as previously suggested10.  Indeed, measurements of bulk phytoplankton photosynthetic efficiency (Fv/Fm), derived from active Chla fluorescence data172, demonstrated a strong down regulation of PSII electron transport capacity around mid-day, which is consistent with the induction of photo-protective mechanisms.  Our data thus suggest that increased cellular sulfur compounds could play a role in this photo-protective mechanism, by helping to limit oxidative stress causes by excess irradiance.  6.4.2 Rate Measurements We employed several experimental techniques to measure various DMS production/ consumption terms.  To the best of our knowledge, our measurements of DMSPd cleavage, DMSOd reduction, DMS production due to krill grazing, and DMS production due to microzooplankton grazing are the first of their kind for the WAP region.  Previous measurements43 of gross DMS consumption rate constants (0.71 ± 0.15 d-1) fall within the (relatively wide) range of our gross DMS consumption measurements (2.8 ± 2.2 d-1) over the seasonal cycle.  Whereas the majority of the season was characterized by relatively balanced DMS sources and sinks terms (Table 6.5), DMS production and consumption terms peaked in mid-December, following a massive diatom bloom.  These peaks appear to result from both high concentrations of algal biomass (and DMSP), and also from a stimulation of bacterial activity, likely tied to the increased availability of organic carbon.  Below, we examine controls on the different production and consumption terms over the seasonal cycle. We measured large DMS production terms from the algal/ bacterial DMSPd cleavage and DMSOd reduction.  On average, DMSPd cleavage was the second largest DMS production term (Fig. 6.12); this term consistently proved a major source for DMS (Table 6.5), indicating that DMSPd was the main precursor of DMS over the seasonal cycle.  Indeed,  153 the correlation between DMS and DMSPd supports the idea that DMSPd cleavage is a dominant DMS production term.  Consistently high specific rate constants of DMSPd cleavage drove high DMSPd cleavage rates.  In comparison, rate constants for DMSOd reduction remained below detection in the majority of tracer experiments, suggesting that DMS was the main precursor for DMSOd.  However, on one occasion (December 19), a sizable DMSOd pool (~8nM) and high DMSO reduction rate constant (7.2 d-1; comparable to DMSPd cleavage) resulted in a high rate of DMS production from DMSOd  (Table 6.5).  The low relative contribution of DMSO reduction as a source of DMS stands in contrast to recent observations showing significant rates of DMSOd reduction in sea-ice brines3.  These high rates were the product of very large DMSOd pools and rate constants of DMSOd reduction that were comparable to rate constants for DMSPd cleavage.  Although a handful of studies have documented DMSOd reduction by both phytoplankton and bacteria12,21, we speculate that the environmental conditions (e.g. in sea ice brines) or the presence of particular taxonomic groups of bacteria and algae may determine the importance of this poorly studied DMS production pathway.  It has been suggested that DMS production may be increased when algal cells have a physiological history of nutrient stress, growth is limited, and cells accumulate internal pools of sulfur intended for protein synthesis50.  In support of this hypothesis, we observed the highest rates of both DMSPd cleavage and DMSOd reduction during a period of low net primary production (NPP), following a minimum in nitrate concentrations.  More generally, we observed an inverse relationship between NPP and DMSPd cleavage (r = -0.84, p < 0.05; n= 5).  Our results suggest that grazing is an important mechanism for DMS production in the WAP.  To date, only a handful of field studies have examined how grazing influences DMS cycling156,157,174.  This work has shown that grazing can exert a strong control on DMS production in several marine environments, including the Southern Ocean.  Results from our grazing experiments indicated that micro-zooplankton and krill grazing stimulated DMS production, in part by releasing intracellular DMSP.  Due to patchy distributions of krill abundance throughout the water column, our calculations suggest that micro-zooplankton generally dominated DMSP release and subsequent DMS production.  In addition to the type of grazer, the presence of certain phytoplankton taxa may influence the rates of DMSP removal (release) and subsequent DMS production.  For example, we observed a strong  154 negative relationship between the abundance of Phaeocystis and the removal rate of DMSP (d-1) in micro-grazing experiments (Fig. 6.9), suggestive of an aversion for DMSP rich Phaeocystis and selective feeding by micro-zooplankton.  Conversely, the fraction of DMS produced from DMSP removed increased as a function of Phaeocystis abundance (r2 = 0.94, p < 0.05).  Phaeocystis cells contain compartmentalized cellular DMSP and membrane-bound DMSP-lyase enzyme, which cleaves DMSP to produce DMS16.  Intracellular DMSP released during grazing mixes with free DMSP-lyase enzyme to produce DMS in solution, independent of bacterial or algal activity.  This mechanism for direct DMS production could explain the high net DMS production without a commensurate increase in DMSPd cleavage observed in our krill grazing experiments in mid-February (Fig. 6.10). As part of our field campaign, we calculated DMS removal due to sea-air flux, and measured rates of gross DMS consumption, including biological DMS consumption and photo-oxidation.  Compared with gross DMS consumption, sea-air flux represented a minor sink for DMS (< 10%; Table 6.5), as previously demonstrated by Hermann et al (2012).  Similarly, photo-oxidation typically represents a small fraction of gross DMS consumption at Palmer Station, even in surface waters43.  For this reason, our measured rates (based on D3 DMS removal or CI experiments) should largely reflect biological consumption.  Indeed, using the relationship between the UV-A light dose and DMS removal due to photolysis observed in Toole et al.159, we estimate maximum rates of surface water DMS photo-oxidation of 2.4 ± 1.5 nM d-1, with a corresponding rate constant of 0.67 ± 0.35 d-1.  These rates represent < 20% of our measured gross DMS consumption rates, indicating a dominance of biological process in DMS consumption.  It is currently believed that heterotrophs are solely responsible for biological DMS consumption12,175.  In support of this, we found that bacterial respiration was correlated with gross DMS consumption (Fig. 6.11), which is thought to supplement DMSP consumption as a source of sulfur and energy26.  Unfortunately, we lack information on the taxonomic composition of the bacterial assemblages.  Based on previous studies, however, we would expect that the presence of particular groups (e.g. Roseobacter), would be particularly important in determining biological DMS/P consumption rates.   155 6.4.3 Conclusions and Future Outlook This study provides new insight into rapid DMS/P/O cycling in the WAP using a combination of high-resolution DMS/P/O measurements, stable isotope rate experiments, and grazing experiments.  Results from this fieldwork show that 1) DMSPt production is tied to the abundance of Phaeocystis, 2) DMSPd is the main precursor of DMS, 3) grazing, particularly by microzooplankton, is critical DMS production, 4) DMS appears to be the main precursor of DMSOd, and 5) bacterial consumption dominates DMS removal.  In addition, data from this study are consistent with an anti-oxidant role for DMS/O/P, with increase surface concentrations under conditions of high surface irradiance and shallow mixed layer depths. The Palmer Station LTER site is an important resource for studies of seasonal and inter-annual DMS/P/O variability, in WAP waters that are currently undergoing rapid ecological change.  The Station is located at the northern-most tip of the LTER sampling grid, where significant warming and changes in sea-ice dynamics have been observed over the past several decades.  Primary production and krill abundance in the northern WAP have declined, and further stratification of the water column has favored the accumulation of diatoms and cryptophytes.  New controlling factors in DMS cycling may emerge as phytoplankton bloom dynamics respond to shifts in the timing of sea-ice retreat and stratification176,177.  Based on the dependence of DMS/P production and concentrations on Phaeocystis and grazing, we expect that DMS concentrations at Palmer (and in the northern half of the LTER grid) could decline with a further shift towards diatom-dominance.  Palmer Station thus provides a unique opportunity to study the effects of climate change on the DMS cycle in polar marine waters. Looking ahead, more data are needed on the temporal variability in DMSP/O concentrations and DMS production/ consumption terms throughout the seasonal cycle.  For example, turn-over rates derived from our stable isotope tracer and competitive inhibition experiments are greater than three times higher and more variable than rate constants derived from the 2005-2006 in field campaign using a radio isotope tracer technique.  Although these differences could reflect seasonal, inter-annual, or longer-term change, our understanding of the DMS cycle at Palmer Station would benefit from an inter-comparison of stable isotope and radioisotope labeled tracer techniques.  In addition, more information is needed to  156 constrain the impact of physical processes on DMS/P/O concentrations and turnover rates.  In our study, we were unable to measure either mixing or lateral advection.  However, previous studies have suggested that dilution is an important removal mechanism for phytoplankton172 and sulfur compounds43.  Dilution due to vertical mixing or entrainment in our study area may have been particularly important between December 1- December 17172.  Recent upgrades to sampling capabilities at Palmer Station, including the acquisition of a Lidar system to measure surface current flows (O. Schofield, pers. comm.), and the potential future deployment of instrumented moorings will provide better insight into the physical processes driving DMS/P/O variability.  We believe that continued deployment of automated analytical systems for DMS/P/O concentration measurements, along with advanced process studies, will be an important addition to on-going biogeochemical studies at Palmer Station, and potentially other monitoring sites in polar marine waters.    157  Table 6.1. Equations for DMS Rate Measurements Equations: In these equations, k is the rate constant, D is the dilution factor, where 𝐷is the mean dilution factor, t is for time, the subscript t is for an individual measurement, the subscript 0 is for the initial measurement at T0, and the subscript corr is for a signal corrected for background interferences.     1.  n𝑎𝑡𝑢𝑟𝑎𝑙  𝐷𝑀𝑆 = 𝑚62  𝐷𝑀𝑆 − 0.01   ∗   𝑚68  𝐷𝑀𝑆  2.  ln [ ™? ™⌤?☣☧ ]?[ ™? ™⌤?☣☧ ]? = −𝑘 ™? _ ™?␥?????   3.  ln [ ™?? ™⌤?☣☧ ]?[ ™?? ™⌤?☣☧ ]? = −𝑘 ™? ? ™⌤ ┦✨ 〈 𝑡  4.  ln [???   ™? ]????   ™? ? = −𝑘 ™? _ ™??   5.  [𝐶13?DMS] = [𝐷𝑀𝑆™ ] − 0.3  [DMS ™ ] − 0.043  [𝐷𝑀𝑆™ ]  6.  [𝐶13?DMS] ™?? = [𝐶13?DMS] +    [? ™ ? ™? ]????™ ? ™⌤ ?  7.  ln [? ™ ?? ™? ]?   ™⌣[? ™ ?? ™? ]?   ™⌣ = 𝐾™?? _ ™???☧⠩ 𝑡  8.  D − 6  DMS ™?? = D − 6  DMS + [???   ™? ]????™? _™⌤ ?  9.  ln ???   ™? ?   ™⌣???   ™? ? = 𝐾™?␥ 𝑡  10.  𝐷𝑀𝑆   ™??? ? ™??    =   𝐷𝑀𝑆™?? _ ™ ⌤ ∗   𝐷𝑀𝑆𝑃™?? ∗    𝐷𝑀𝑆𝑃?     11.  𝐷𝑀𝑆™???    ™??   𝑊𝐴𝑃 =   𝐷𝑀𝑆 ™   ??? ∗   𝐾𝑟𝑖𝑙𝑙 ™?   ??? ∗ ???  ?  ??™ ∢   ?  ?   ™ ⌤ ?   ?   ?   ™???    ™?   12.  DMS  Budget =   𝐷𝑀𝑆™??? ? ™?? + 𝐷𝑀𝑆™??? ? ™??   𝑊𝐴𝑃 +   𝐷𝑀𝑆™?? _ ™ ⌤? + 𝐷𝑀𝑆™?? _ ™?    −𝐷𝑀𝑆™?? ?™⌤ − 𝐷𝑀𝑆™??          158 Table 6.2. Correlation Matrix for DMS/P/O and Ancillary Variables Pearson correlation coefficients between DMS/P/O concentrations at Station B and ancillary measurements over the seasonal cycle.  Significance level indicated by * for p < 0.05, ** for p < 0.01, and  *** for p < 0.001.  	   DMS	   DMSPd	   DMSPt	   DMSOd	   DMSOt	  	   	   	   	   	   	  DMSPd	   0.54*	   	   	   	   	  DMSPt	   0.34	   0.42	   	   	   	  DMSOd	   0.75**	   0.47	   0.27	   	   	  DMSOt	   0.41	   0.31	   0.66	   0.78**	   	  MLD	   -­‐0.43*	   -­‐0.65**	   -­‐0.19	   -­‐0.36	   -­‐0.23	  Chla	   0.07	   -­‐0.05	   -­‐0.11	   0.15	   -­‐0.05	  Phaeo	   0.27	   0.67**	   0.85***	   -­‐0.01	   0.39	  UV	   0.23	   -­‐0.03	   0.25	   0.35	   0.05	      159 Table 6.3. DMSP Production Measurements DMSP production measured in competitive inhibition experiments using glycine betaine. Logarithmic DMSP accumulation rates are shown for experiments that yielded reasonably linear (r2>0.4) slopes of DMSP concentrations over the experimental time course.  Error bars represent one standard error from the mean.  DATE Initial DMSPt concentration (nM) LN Gross Change DMSPt (d-1) LN Net Change DMSPt (d-1) 27-Dec 28.6 2.4 ±1.7 1.6 ± 1.9 1-Jan 34.3 2.3 ± 0.7 0.8 ± 1.4 14-Jan 90 2.0 ± 0.8 0.8 ± 0.7 28-Jan 145 2.5 ± 1.1 2.0 ± 0.6 16-Feb 158 2.6 ± 1.8 1.8 ± 2.9     160 Table 6.4. DMS Production and Removal Terms DMS production and removal terms (nM d-1) calculated from in situ measurements and experimental rates (see Table 1 for equations).  Error terms were calculated using standard error propagation of the individual rate measurements, which represent one standard error from concentration and rate measurements, respectively.  Rates where the rate constant was lower than our detection limit (< 0.5 d-1) are denoted by n.d.  Dates DMSP Cleavage (nM d-1) DMSO Reduction (nM d-1) Microzoop. Grazing (nM d-1) Krill Grazing (nM d-1) Gross DMS Consumption (nM d-1) Sea-air Flux (nM d-1) Nov-06 1.67 ± 3.3 n.d. - - 3.5 ± 4.7 0.88 ± 0.00 Nov-18 n.d n.d.  - - n.d 1.38 ± 0.74 Nov-27 - n.d - - 32.6 ± 22 0.64 ± 0.26 Dec-13 31.5 ± 13 n.d - 2.1 ± 1.8 125.8 ± 32 2.10 ± 0.73 Dec-17 4.9 ± 12 3.5±1.2 - 0.3 ± 0.2 2.9 ± 22 2.93 ± 1.01 Dec-19 37.3 ± 15 64.6±47 - - n.d 0.92 ± 0.08 Dec-20 - - 6.4 ± 3.4 0.3 ± 0.3 - 0.74 ± 0.07 Dec-27 n.d. n.d 2.6 ± 0.6 2.5 ± 2.1 - 0.75 ± 0.24 Jan-06 -  7.8 ± 4.4 0.9 ± 0.7 - 4.78 ± 1.93 Jan-14 - - - 1 ± 0.8 32.3 ± 31 9.32 ± 4.30 Jan-21 - - - 0.7 ± 0.6 23.5 ± 31 4.18 ± 1.10 Jan-26 - - 20.3 ± 12 1 ± 0.9 - 6.00 ± 2.78 Feb-09 - - 38.4 ± 16 3.3 ± 2.7 - 4.65 ± 1.23 Feb-15 6.2 ± 13 n.d - - 24.2 ± 33 7.01 ± 2.11     161 Table 6.5. Microzooplankton Grazing and DMS Production Effects of grazing on chlorophyll a (Chl a) and DMSP removal rates and DMS production measured in dilution experiments (± standard error).  Chl a and DMSP removal rates were calculated from the slope of the natural logarithm of Chl a and DMSP decrease over 24 hours versus the fraction of filtered (i.e. grazer-free) seawater.  DMS evolved from grazed DMSP is calculated from the net DMS production over time normalized to DMSP loss.   DATE Chla. Grazing Rate (d-1) DMSP Grazing Rate (d-1) mol DMS prod / mol DMSP grazed 21-Dec 0.20 ± 0.08 1.36 ± 0.15 0.09 ± 0.06 28-Dec 0.20 ± 0.06 1.68 ± 0.30 0.04 ± 0.01 7-Jan 0.21 ± 0.05 0.61 ± 0.09 0.14 ± 0.04 26-Jan 0.09 ± 0.01 0.89 ± 0.32 0.14 ± 0.01 9-Feb 0.10 ± 0.02 0.25 ± 0.02 0.96 ± 0.12     162   Figure 6.1. Map of Study Area A map of the study area, showing bathymetry and the location of and the  seawater pump intake (SWP) from which seawater was continuously sampled, and station B (Stn B), where discrete bottle samples were obtained.    163  Figure 6.2. Seasonal Changes in Ancillary Measurements Time-series of a) depth-integrated chlorophyll a (Chla) and nitrate (NO3-), b) relative abundance of diatoms, Phaeocystis and chlorophytes at 10m depth, c) bacterial and krill abundance, and c) mixed layer depth (MLD; derived from a density difference criterion, Δσt,of 0.125 kg m-3), and daily averaged photosynthetically active irradiance (PAR).  Surface ocean measurements were derived from ~ semi-weekly sampling at Station B, as described in the methods.   164   Figure 6.3. Seasonal Changes in DMS/P/O Concentrations Time course of sampling activities and DMS/P/O concentrations at Palmer Station.  Panel (a) shows the sampling dates for various processes studies and rate measurements.  Panel (b) shows DMS concentration measurements from continuous MIMS analysis of the seawater supply, and discrete measurements at station B (10 m depth).  Panels (c) and (d), show semi-weekly DMSP and DMSO total and dissolved concentrations at station B (10 m depth).  Note that axes for total and dissolved DMSP have different scales.  Error bars represent standard error from the mean.    165  Figure 6.4. High-Frequency Measurements of DMS by MIMS High frequency measurements of a) DMS (via MIMS) b) photosynthetically active radiation (PAR) and c) wind speed during a one-week period February, 2013.  Tick marks and grids indicate solar noon.     166  Figure 6.5. High Frequency Measurements of DMSP/O by OSSCAR Concentration of DMSPd (nM) and DMSOd in relation to wind speed and PAR at station A during two diel cycle studies on a) Dec-25, 2012 and a) Jan-20, 2013.     167  Figure 6.6. Example of Tracer Experiment  Examples of tracer time-course data used to derive rate constants for DMS production and consumption: a) DMS consumption b) DMSP cleavage and c) DMSO reduction.  Note the log scale on the y-axis.  Error bars represent standard error from the mean.    168  Figure 6.7. Example of DMS CI Experiment An example of a DMS competitive inhibition experiment, in which gross DMS production is measured using a competitive inhibitor for DMS consumption (open symbols) and net DMS production is measured in the absence of a competitive inhibitor (filled symbols).  Error bars represent standard error from the mean.   169   Figure 6.8. Example of Dilution Experiment Impacts of microzooplankton grazing on DMSP and DMS production.  Results of dilution experiments showing a) the net change in Chla (µg d-1) and DMSP concentrations (nMd-1); b) the resulting net DMS production (nM d-1) due to grazing (nM d-1).     170  Figure 6.9. Relationship between Phaeocystis and DMSP Removal Relationship between the abundance of Phaeocystis (µg Chla L-1) and the DMSP removal rate observed in 5 micro-grazing experiments. The taxa’s contribution of Chla serves as a measure for the abundance of Phaeocystis (Goldman et al.171), which was calculated by scaling the ancillary marker pigment 19’hex (µg 19’hex L-1) by 1.25, according to an algorithm developed for the Southern Ocean by Everitt et al. (1990) and implemented more recently by Arrigo et al. (2000).  The line represents the best-fit linear regression line (DMSPrem= -4.9 (Phaeo) + 1.7; r = 0.94, p<0.05).    171   Figure 6.10. Krill Grazing and DMS Production Derived mean values of net DMS production (change in m/z 62 DMS), and chlorophyll and total DMSP removal rates (change in DMSPt m/z 62) in three krill grazing experiments in mid-February.  Error bars represent one standard error from the mean.    172  Figure 6.11. DMS Cleavage, DMS Consumption, and Bacterial Respiration A comparison of specific rates (d-1) of DMSP cleavage and gross DMS consumption and bacterial respiration (µg C L-1 d-1) over the seasonal cycle.  Error bars represent the standard error from the mean of DM production and consumption rate terms.    173  Figure 6.12. Seasonal Summary of DMS Production and Removal Terms  Seasonal mean rates of DMS production / removal due to 1) DMSP cleavage, 2) DMSO reduction, 3) micrograzing 4) krill grazing 5) gross DMS consumption (biological DMS consumption and photo-oxidation) and 6) air-sea flux.  Error bars show one standard error from the mean.  The net balance term represents the sum of all (mean) DMS production terms (i.e. DMSP Cleavage, DMSO Reduction, micrograzing, krill grazing) minus the sum of all measured (mean) DMS loss terms (i.e. gross DMS consumption and sea-air flux).    174 7 Conclusions The goal of this thesis was to identify primary biogeochemical controls on DMS/P/O accumulation in surface waters of the Subarctic Northeast Pacific and high latitude Southern Ocean, and in Antarctic sea-ice environments.  Each chapter in this thesis addressed one or more of the following questions using field-based sampling and experimental studies: 1) How do DMS/P/O concentrations vary across various spatial/ temporal scales? 2) To what extent can biological and hydrographic parameters be used to predict DMS/P/O concentrations?  3) What biological processes control DMS production, and how do these processes respond to environmental variability?  These questions are central to our understanding of the DMS/P/O cycle in distinct marine environments, and its potential response to future climate change.   7.1.1 Major Findings and Contributions This thesis described a novel system for the sequential analysis of DMS/P/O in surface waters and presented the first field application of this method.  Using this automated DMS/P/O system in conjunction with MIMS, the results provide DMS/P/O concentration measurements at unmatched spatial and temporal resolution in the Subarctic Pacific and Southern Ocean.  This thesis also provided the first field application of a novel stable isotope tracer technique, and, to our knowledge, the first rate measurements of natural DMS/P/O cycling in Antarctic sea ice.  Application of these and other techniques has the potential to offer new insights into the cycling DMS/P/O in a variety of marine environments. High-resolution MIMS surveys in the Subarctic Northeast Pacific reveal open ocean accumulation of DMS during late summer, with considerable spatial variability, and rapid changes in DMS concentrations over diel cycles.  Strong seasonal patterns were observed in the spatial distribution of DMS along a coastal - oceanic gradient, with maximum concentrations shifting from coastal waters to offshore regions during the progression from spring to late summer.  Data obtained from six surveys across two years were used to derive empirical relationships between DMS concentrations and various hydrographic and biological variables.  The relationships were able to capture regional patterns of DMS variability across years, but failed to reproduce DMS variability at the mesoscale (1X1°).   175 Automated sequential measurements of DMS/P/O in surface waters (using the new OSSCAR system) reveal considerable spatial variability in DMSP and DMSO concentrations that is related to DMS variability.  These measurements present evidence for a relationship between phytoplankton taxonomic composition and surface water concentrations of DMS/P/O.  The application of MIMS and OSSCAR on board research vessels enables us to simultaneously measure surface water DMS, DMSP and DMSO concentrations with much higher spatial resolution than existing methods, with very little operator involvement.  These new automated methods could improve our ability to characterize surface ocean distributions of key reduced sulfur species. Application of a novel stable-isotope technique in Antarctic sea-ice allowed us to simultaneously track DMS consumption rates, and DMS production via DMSP cleavage and DMSO reduction.  Our results reveal a highly dynamic DMS/P/O cycle in Antarctic sea ice, and show, unexpectedly, that DMSO reduction can be a major contributor to DMS production.  The isotope tracer method was also applied in surface waters of the Subarctic Pacific, and West Antarctic Peninsula in order to gain insight into DMS/P/O cycling processes.  Simultaneous application of MIMS and OSSCAR was used to describe the spatial and temporal distributions of these compounds in surface waters, and to characterize the seasonal cycle of DMS/P/O concentrations and turnover rates in coastal Antarctic waters.  We also examined the importance of various bacterial and grazer-dependent processes in net DMS production.  Results from this work demonstrated that grazing and bacterial DMSPd cleavage can be dominant sources of DMS in the West Antarctic Peninsula, and that bacterial uptake controls gross DMS consumption.  Our results also illustrated an important role of surface irradiance levels in controlling DMS/P/O concentrations over diel (day-night) and seasonal cycles.  Results from the Subarctic Pacific show that rates of DMS consumption and production decreased with distance from the coast (and secondarily, with latitude), and that rate measurements can be used to predict DMS accumulation in surface waters.  7.1.2 Future Directions The analytical methods described here represent important tools for future studies, and we expect that they will facilitate research into a number of key unresolved questions  176 about the DMS/P/O cycle.  Below, we suggest some directions for continued improvements of our methods, and potential new applications for these techniques.   As discussed in chapter 3, the OSSCAR system for automated sequential analysis of DMS/P/O would benefit from improved sampling resolution (i.e. reduced analysis times).  At present, every DMS/P/O analysis cycle requires ~ 1.5 h.  This measurement frequency provides near real-time data and is likely sufficient to capture much of the variability in open ocean systems.  In highly dynamic coastal waters, however, large biogeochemical gradients are often observed over distances of a few km to tens of km106,148, thus requiring much higher measurement frequency to fully describe surface water variability.  Improved measurement frequency in the OSSCAR system could be achieved through a number of approaches that are described in chapter 3.  Among these is the use of parallel sparging chambers and traps to facilitate simultaneous conversion of DMSP and DMSO to DMS for GC analysis.  In addition, this system could be made to distinguish dissolved from particulate S pools using a gentle, in-line gravity filtration method. In addition to simultaneously measuring DMS consumption and DMS production from DMSP and DMSO as discussed in chapters 4, 5, and 6, the stable isotope tracer method could be used to trace the fate of DMS/P/O to other sulfur pools.  The fraction of DMSP incorporated into (non-DMS producing) phytoplankton cells and the DMSO yield from DMS are of particular interest.  Briefly, intracellular DMSP could be measured in phytoplankton by fixing samples, re-suspending material captured on GFF filters and measuring DMSP by NaOH hydrolysis and chla in replicate size-fractionated subsamples.  Second, by pairing the stable isotope tracer technique with the sensitive assay for DMSO using DMSO reductase enzyme, this method could be used to trace the fate of DMS to the DMSO pool, as opposed to sulfate.  Until recently, limitations on the accuracy and sensitivity of DMSO measurements have impeded the robust quantification of the DMSO yield from DMS.  Using the stable isotope tracer and DMSOr determination of DMSO methods together, however, the removal rate of DMSO and the DMS yield from DMSO could also be easily quantified.  Such an experiment could help elucidate the fate of DMSO, which remains an active area of research.  Finally, there would be a significant benefit in pairing the new stable isotope tracer method with the existing radio-isotope methods of Kiene et al.95.  For instance, whereas the radiolabeled tracer method can quantify algal vs. bacterial cleavage over a few hours, our  177 stable isotope tracer method cannot.  Collectively, these two methods can provide highly complementary information into the turnover of various reduced S species in marine waters. Overall, future use of the methodological approaches presented in this thesis will likely lead to an improved understanding of the DMS/P/O cycle in oceanic waters.  For instance, the deployment of MIMS and OSSCAR in LaGrangian studies (where individual water masses are tracked and sampled repeatedly) would improve our understanding of the temporal changes of these reduced S compounds.  In addition, a few modeling and mesocosm field studies have suggested that ocean acidification may amplify warming by lowering DMS emissions178,179, although no scientific consensus has been reached on this topic.  Simultaneous DMS production and consumption rate measurements and well resolved temporal changes in DMS/P/O concentrations in ocean acidification mesocosm experiments is yet another exciting application for MIMS, OSSCAR and the stable isotope tracer technique.  These measurements could be used to examine the sensitivity of various DMS production and removal terms to on-going ocean acidification.  Finally, the microzooplankton grazing dilution technique, utilized in chapter six of this thesis, has not been widely adopted by the oceanographic community for studies of DMS/P cycling. Further use of this method could provide information on DMS production related to grazing, thus shedding light on key differences between coastal and offshore ecosystem dynamics that lead to DMS production in marine environments.  The availability of new rate measurements in the DMS/P/O cycle180, will facilitate the development of more robust ocean models representing key processes in the marine sulfur cycle and predicting potential climate feedbacks resulting from changes in oceanic DMS emissions.    178 Bibliography 1. Asher, E. C., Merzouk, A. & Tortell, P. D. Fine-scale spatial and temporal variability of surface water dimethylsufide (DMS) concentrations and sea–air fluxes in the NE Subarctic Pacific. Mar. Chem. 126, 63–75 (2011). 2. Tortell, P. D. Dissolved gas measurements in oceanic waters made by membrane inlet mass spectrometry. Limnol. Oceanogr.-Methods 3, 24–37 (2005). 3. Asher, E. C., Dacey, J. W. H., Jarnikova, T. & Tortell, P. D. Measurement of reduced sulfur compounds in natural waters by automated sequential chemical analysis. Limnol. Oceanogr. Methods (In Press). 4. Asher, E. C., Dacey, J. W. H., Mills, M. M., Arrigo, K. R. & Tortell, P. D. High concentrations and turnover rates of DMS, DMSP and DMSO in Antarctic sea ice: DMS dynamics in Antarcic sea ice. Geophys. Res. Lett. 38, n/a–n/a (2011). 5. Dacey, J. W. H. & Stefels, J. Deuterated tracers for dynamics of the DMS system in marine waters. (2005). at <http://www.aslo.org/meetings/santiago2005/abstracts/1892.htm> 6. Bates, T. S., Lamb, B. K., Guenther, A., Dignon, J. & Stoiber, R. E. Sulfur emission to the atmosphere from natural sources. J. Atmospheric Chem. 14, 315–337 (1992). 7. Mahajan, A. S. et al. Quantifying the impacts of an updated global dimethyl sulfide climatology on cloud microphysics and aerosol radiative forcing. J. Geophys. Res. Atmospheres 120, 2524–2536 (2015). 8. Dickson, D. M. J. & Kirst, G. O. Osmotic adjustment in marine eukaryotic algae: the role of inorganic ions quarternary ammonium, tertiary sulfphonium and carbohydrate solutes in i. diatoms and rhodophyte. New Phytol. 106, 645–655 (1987). 9. Kirst, G. O. et al. Dimethylsulfoniopropionate (DMSP) in ice-algae and its possible biological role. Mar. Chem. 35, 381–388 (1991). 10. Sunda, W., Kieber, D. J., Kiene, R. P. & Huntsman, S. An antioxidant function for DMSP and DMS in marine algae. Nature 418, 317–320 (2002). 11. Kiene, R. P., Linn, L. J. & Bruton, J. A. New and important roles for DMSP in marine microbial communities. J. Sea Res. 43, 209–224 (2000). 12. Vila-Costa, M. et al. Phylogenetic identification and metabolism of marine dimethylsulfide-consuming bacteria. Environ. Microbiol. 8, 2189–2200 (2006). 13. Seymour, J. R., Simo, R., Ahmed, T. & Stocker, R. Chemoattraction to dimethylsulfoniopropionate throughout the marine microbial food web. Science 329, 342–345 (2010). 14. Nevitt, G. A. Sensory ecology on the high seas: the odor world of the procellariiform seabirds. J. Exp. Biol. 211, 1706–1713 (2008). 15. Keller, M. D., Bellows, W. K. & Guillard, R. R. L. Dimethyl sulfide production in marine-phytoplankton. Acs Symp. Ser. 393, 167–182 (1989). 16. Stefels, J. & Dijkhuizen, L. Characteristics of DMSP-lyase in Phaeocystis sp (Prymnesiophyceae). Mar. Ecol.-Prog. Ser. 131, 307–313 (1996). 17. Dacey, J. W. H. & Wakeham, S. G. Oceanic dimethylsulfide-production during zooplankton grazing on phytoplankton. Science 233, 1314–1316 (1986). 18. Malin, G., Wilson, W. H., Bratbak, G., Liss, P. S. & Mann, N. H. Elevated production of dimethylsulfide resulting from viral infection of cultures of Phaeocystis pouchetii. Limnol. Oceanogr. 43, 1389–1393 (1998).  179 19. Todd, J. D., Curson, A. R. J., Dupont, C. L., Nicholson, P. & Johnston, A. W. B. The dddP gene, encoding a novel enzyme that converts dimethylsulfoniopropionate into dimethyl sulfide, is widespread in ocean metagenomes and marine bacteria and also occurs in some Ascomycete fungi. Environ. Microbiol. 11, 1376–1385 (2009). 20. Todd, J. D. et al. DddQ, a novel, cupin-containing, dimethylsulfoniopropionate lyase in marine roseobacters and in uncultured marine bacteria. Environ. Microbiol. 13, 427–438 (2011). 21. Spiese, C. E., Kieber, D. J., Nomura, C. T. & Kiene, R. P. Reduction of dimethylsulfoxide to dimethylsulfide by marine phytoplankton. Limnol. Oceanogr. 54, 560–570 (2009). 22. Fuse, H. et al. Transformation of dimethyl sulfide and related-compounds by cultures and cell-extracts of marine phytoplankton. Biosci. Biotechnol. Biochem. 59, 1773–1775 (1995). 23. Zinder, S. H. & Brock, T. D. Production of methane and carbon dioxide from methane thiol and dimethyl sulfide by anaerobic lake-sediments. Nature 273, 226–228 (1978). 24. Hatton, A. D., Darroch, L. & Malin, G. in Oceanography and Marine Biology: an Annual Review, Vol 42 42, 29–55 (2005). 25. Levasseur, M. Impact of Arctic meltdown on the microbial cycling of sulphur. Nat. Geosci. 6, 691–700 (2013). 26. Zubkov, M. V. et al. Rapid turnover of dissolved DMS and DMSP by defined bacterioplankton communities in the stratified euphotic zone of the North Sea. Deep-Sea Res. Part Ii-Top. Stud. Oceanogr. 49, 3017–3038 (2002). 27. Toole, D. A., Kieber, D. J., Kiene, R. P., Siegel, D. A. & Nelson, N. B. Photolysis and the dimethylsulfide (DMS) summer paradox in the Sargasso Sea. Limnol. Oceanogr. 48, 1088–1100 (2003). 28. Toole, D. A., Slezak, D., Kiene, R. P., Kieber, D. J. & Siegel, D. A. Effects of solar radiation on dimethylsulfide cycling in the western Atlantic Ocean. Deep-Sea Res. Part -Oceanogr. Res. Pap. 53, 136–153 (2006). 29. Steiner, N. & Denman, K. Parameter sensitivities in a 1-D model for DMS and sulphur cycling in the upper ocean. Deep-Sea Res. Part -Oceanogr. Res. Pap. 55, 847–865 (2008). 30. Simo, R. & Dachs, J. Global ocean emission of dimethylsulfide predicted from biogeophysical data. Glob. Biogeochem. Cycles 16, (2002). 31. Vallina, S. M. & Simo, R. Strong relationship between DMS and the solar radiation dose over the global surface ocean. Science 315, 506–508 (2007). 32. Lana, A. et al. An updated climatology of surface dimethlysulfide concentrations and emission fluxes in the global ocean. Glob. Biogeochem. Cycles 25, (2011). 33. Lana, A., Simo, R., Vallina, S. M. & Dachs, J. Re-examination of global emerging patterns of ocean DMS concentration. Biogeochemistry 110, 173–182 (2012). 34. Boyd, P. W. et al. A mesoscale phytoplankton bloom in the polar Southern Ocean stimulated by iron fertilization. Nature 407, 695–702 (2000). 35. Levasseur, M. et al. DMSP and DMS dynamics during a mesoscale iron fertilization experiment in the Northeast Pacific - Part I: Temporal and vertical distributions. Deep-Sea Res. Part Ii-Top. Stud. Oceanogr. 53, 2353–2369 (2006).  180 36. Merzouk, A. et al. DMSP and DMS dynamics during a mesoscale iron fertilization experiment in the Northeast Pacific - Part II: Biological cycling. Deep-Sea Res. Part Ii-Top. Stud. Oceanogr. 53, 2370–2383 (2006). 37. Wong, C. S. et al. Temporal and spatial distribution of dimethylsulfide in the subarctic northeast Pacific Ocean: a high-nutrient-low-chlorophyll region. Tellus Ser. B-Chem. Phys. Meteorol. 57, 317–331 (2005). 38. Steiner, N. S. et al. Evaluating DMS measurements and model results in the Northeast subarctic Pacific from 1996–2010. Biogeochemistry 110, 269–285 (2012). 39. Royer, S. J. et al. Microbial dimethylsulfoniopropionate (DMSP) dynamics along a natural iron gradient in the northeast subarctic Pacific. Limnol. Oceanogr. 55, 1614–1626 (2010). 40. Tison, J. L., Brabant, F., Dumont, I. & Stefels, J. High-resolution dimethyl sulfide and dimethylsulfoniopropionate time series profiles in decaying summer first-year sea ice at Ice Station Polarstern, western Weddell Sea, Antarctica. J. Geophys. Res.-Biogeosciences 115, (2010). 41. Trevena, A. J. & Jones, G. B. Dimethylsulphide and dimethylsulphoniopropionate in Antarctic sea ice and their release during sea ice melting. Mar. Chem. 98, 210–222 (2006). 42. Sea Ice: An Introduction to its Physics, Chemistry, Biology and Geology. (Wiley-Blackwell, 2003). 43. Herrmann, M. et al. Diagnostic modeling of dimethylsulfide production in coastal water west of the Antarctic Peninsula. Cont. Shelf Res. 32, 96–109 (2012). 44. Saltzman, E. S., De Bruyn, W. J., Lawler, M. J., Marandino, C. A. & McCormick, C. A. A chemical ionization mass spectrometer for continuous underway shipboard analysis of dimethylsulfide in near-surface seawater. Ocean Sci. 5, 537–546 (2009). 45. Kiene, R. P. & Linn, L. J. The fate of dissolved dimethylsulfoniopropionate (DMSP) in seawater: Tracer studies using S-35-DMSP. Geochim. Cosmochim. Acta 64, 2797–2810 (2000). 46. Wolfe, G. V. & Kiene, R. P. RADIOISOTOPE AND CHEMICAL INHIBITOR MEASUREMENTS OF DIMETHYL SULFIDE CONSUMPTION RATES AND KINETICS IN ESTUARINE WATERS. Mar. Ecol.-Prog. Ser. 99, 261–269 (1993). 47. Kiene, R. P. & Gerard, G. Evaluation of glycine betaine as an inhibitor of dissolved dimethylsulfoniopropionate degradation in coastal waters. Mar. Ecol.-Prog. Ser. 128, 121–131 (1995). 48. Armstrong, A. Ice and algae. Nat. Geosci. 4, 819–819 (2011). 49. Simo, R. From cells to globe: approaching the dynamics of DMS(P) in the ocean at multiple scales. Can. J. Fish. Aquat. Sci. 61, 673–684 (2004). 50. Stefels, J., Steinke, M., Turner, S., Malin, G. & Belviso, S. Environmental constraints on the production and removal of the climatically active gas dimethylsulphide (DMS) and implications for ecosystem modelling. Biogeochemistry 83, 245–275 (2007). 51. Ayers, G. P. & Cainey, J. M. The CLAW hypothesis: a review of the major developments. Environ. Chem. 4, 366–374 (2007). 52. Belviso, S. et al. Comparison of global climatological maps of sea surface dimethyl sulfide. Glob. Biogeochem. Cycles 18, (2004). 53. Derevianko, G. J., Deutsch, C. & Hall, A. On the relationship between ocean DMS and solar radiation. Geophys. Res. Lett. 36, (2009).  181 54. Kettle, A. J. et al. A global database of sea surface dimethylsulfide (DMS) measurements and a procedure to predict sea surface DMS as a function of latitude, longitude, and month. Glob. Biogeochem. Cycles 13, 399–444 (1999). 55. Harrison, P. J., Boyd, P. W., Varela, D. E. & Takeda, S. Comparison of factors controlling phytoplankton productivity in the NE and NW subarctic Pacific gyres. Prog. Oceanogr. 43, 205–234 (1999). 56. Booth, B. C., Lewin, J. & Postel, J. R. Temporal variation in the structure of autotrophic and heterotrophic communities in the Subarctic Pacific. Prog. Oceanogr. 32, 57–99 (1993). 57. Martin, J. H. & Fitzwater, S. E. Iron-deficiency limits phytoplankton growth in the Northeast Pacific Subarctic. Nature 331, 341–343 (1988). 58. Gueguen, C. & Tortell, P. D. High-resolution measurement of Southern Ocean CO2 and O-2/Ar by membrane inlet mass spectrometry. Mar. Chem. 108, 184–194 (2008). 59. Tortell, P. D. & Long, M. C. Spatial and temporal variability of biogenic gases during the Southern Ocean spring bloom. Geophys. Res. Lett. 36, (2009). 60. Tortell, P. D. Small-scale heterogeneity of dissolved gas concentrations in marine continental shelf waters. Geochem. Geophys. Geosystems 6, (2005). 61. Nemcek, N., Ianson, D. & Tortell, P. D. A high-resolution survey of DMS, CO2, and O-2/Ar distributions in productive coastal waters. Glob. Biogeochem. Cycles 22, (2008). 62. Huebert, B. J. et al. Measurement of the sea-air DMS flux and transfer velocity using eddy correlation. Geophys. Res. Lett. 31, (2004). 63. Kameyama, S. et al. Equilibrator Inlet-Proton Transfer Reaction-Mass Spectrometry (EI-PTR-MS) for Sensitive, High-Resolution Measurement of Dimethyl Sulfide Dissolved in Seawater. Anal. Chem. 81, 9021–9026 (2009). 64. Marandino, C. A., De Bruyn, W. J., Miller, S. D. & Saltzman, E. S. DMS air/sea flux and gas transfer coefficients from the North Atlantic summertime coccolithophore bloom. Geophys. Res. Lett. 35, (2008). 65. Freeland, H. A short history of ocean station papa and Line P. Prog. Oceanogr. 75, 120–125 (2007). 66. Dacey, J. W. H. & Blough, N. V. Hydroxide decomposition of dimethylsulfoniopropionate to form dimethylsulfide. Geophys. Res. Lett. 14, 1246–1249 (1987). 67. Bourassa, M. A., Legler, D. M., O’Brien, J. J. & Smith, S. R. SeaWinds validation with research vessels. J. Geophys. Res.-Oceans 108, (2003). 68. Ho, D. T. et al. Measurements of air-sea gas exchange at high wind speeds in the Southern Ocean: Implications for global parameterizations (vol 33, art no L16611, 2006). Geophys. Res. Lett. 33, (2006). 69. Saltzman, E. S., King, D. B., Holmen, K. & Leck, C. Experimental-determination of the diffusion-coefficient of dimethylsulfide in water. J. Geophys. Res.-Oceans 98, 16481–16486 (1993). 70. Nightingale, P. D. et al. In situ evaluation of air-sea gas exchange parameterizations using novel conservative and volatile tracers. Glob. Biogeochem. Cycles 14, 373–387 (2000). 71. Cohen, J. Statistical Power Analysis for the Behavioral Sciences. (Routledge, 1988).  182 72. Takahashi, T. et al. Climatological mean and decadal change in surface ocean pCO(2), and net sea-air CO2 flux over the global oceans. Deep-Sea Res. Part Ii-Top. Stud. Oceanogr. 56, 554–577 (2009). 73. Bouillon, R. C. et al. The effect of mesoscale iron enrichment on the marine photochemistry of dimethylsulfide in the NE subarctic Pacific. Deep-Sea Res. Part Ii-Top. Stud. Oceanogr. 53, 2384–2397 (2006). 74. Steiner, N., Denman, K., McFarlane, N. & Solheim, L. Simulating the coupling between atmosphere-ocean processes and the planktonic ecosystem during SERIES. Deep-Sea Res. Part Ii-Top. Stud. Oceanogr. 53, 2434–2454 (2006). 75. Bouillon, R. C. & Miller, W. L. Photodegradation of dimethyl sulfide (DMS) in natural waters: Laboratory assessment of the nitrate-photolysis-induced DMS oxidation. Environ. Sci. Technol. 39, 9471–9477 (2005). 76. Liss, P. S. & Merlivat, L. in The Role of Air-Sea Exchange in Geochemical Cycling (ed. Buat-Ménard, P.) 113–127 (Springer Netherlands, 1986). at <http://www.springerlink.com/index/10.1007/978-94-009-4738-2_5> 77. Wanninkhof, R. Relationship between wind-speed and gas-exchange over the ocean. J. Geophys. Res.-Oceans 97, 7373–7382 (1992). 78. Whitney, F. A. & Freeland, H. J. Variability in upper-ocean water properties in the NE Pacific Ocean. Deep-Sea Res. Part Ii-Top. Stud. Oceanogr. 46, 2351–2370 (1999). 79. Anderson, T. R. et al. Global fields of sea surface dimethylsulfide predicted from chlorophyll, nutrients and light. J. Mar. Syst. 30, 1–20 (2001). 80. Kettle, A. J. & Andreae, M. O. Flux of dimethylsulfide from the oceans: A comparison of updated data seas and flux models. J. Geophys. Res.-Atmospheres 105, 26793–26808 (2000). 81. Chapman, E. G., Shaw, W. J., Easter, R. C., Bian, X. & Ghan, S. J. Influence of wind speed averaging on estimates of dimethylsulfide emission fluxes. J. Geophys. Res.-Atmospheres 107, (2002). 82. Harris, S. L., Varela, D. E., Whitney, F. W. & Harrison, P. J. Nutrient and phytoplankton dynamics off the west coast of Vancouver Island during the 1997/98 ENSO event. Deep-Sea Res. Part Ii-Top. Stud. Oceanogr. 56, 2487–2502 (2009). 83. Suzuki, K., Minami, C., Liu, H. B. & Saino, T. Temporal and spatial patterns of chemotaxonomic algal pigments in the subarctic Pacific and the Bering Sea during the early summer of 1999. Deep-Sea Res. Part Ii-Top. Stud. Oceanogr. 49, 5685–5704 (2002). 84. Cullen, J. T., Chong, M. & Ianson, D. British Columbian continental shelf as a source of dissolved iron to the subarctic northeast Pacific Ocean. Glob. Biogeochem. Cycles 23, (2009). 85. Archer, S. D., Ragni, M., Webster, R., Airs, R. L. & Geider, R. J. Dimethyl sulfoniopropionate and dimethyl sulfide production in response to photoinhibition in Emiliania huxleyi. Limnol. Oceanogr. 55, 1579–1589 (2010). 86. Slezak, D. & Herndl, G. J. Effects of ultraviolet and visible radiation on the cellular concentrations of dimethylsulfoniopropionate (DMSP) in Emiliania huxleyi (strain L). Mar. Ecol.-Prog. Ser. 246, 61–71 (2003). 87. Ribalet, F. et al. Unveiling a phytoplankton hotspot at a narrow boundary between coastal and offshore waters. Proc. Natl. Acad. Sci. U. S. A. 107, 16571–16576 (2010).  183 88. Gabric, A. J., Simo, R., Cropp, R. A., Hirst, A. C. & Dachs, J. Modeling estimates of the global emission of dimethylsulfide under enhanced greenhouse conditions (vol 18, art no GB2014, 2004). Glob. Biogeochem. Cycles 18, (2004). 89. Masotti, I. et al. Spatial and temporal variability of the dimethylsulfide to chlorophyll ratio in the surface ocean: an assessment based on phytoplankton group dominance determined from space. Biogeosciences 7, 3215–3237 (2010). 90. Del Valle, D. A., Kieber, D. J., Toole, D. A., Bisgrove, J. & Kiene, R. P. Dissolved DMSO production via biological and photochemical oxidation of dissolved DMS in the Ross Sea, Antarctica. Deep-Sea Res. Part -Oceanogr. Res. Pap. 56, 166–177 (2009). 91. Del Valle, D. A., Kieber, D. J., Toole, D. A., Brinkley, J. & Kiene, R. P. Biological consumption of dimethylsulfide (DMS) and its importance in DMS dynamics in the Ross Sea, Antarctica. Limnol. Oceanogr. 54, 785–798 (2009). 92. Simo, R. & Pedros-Alio, C. Role of vertical mixing in controlling the oceanic production of dimethyl sulphide. Nature 402, 396–399 (1999). 93. Simo, R., Pedros-Alio, C., Malin, G. & Grimalt, J. O. Biological turnover of DMS, DMSP and DMSO in contrasting open-sea waters. Mar. Ecol.-Prog. Ser. 203, 1–11 (2000). 94. Gali, M. & Simo, R. Occurrence and cycling of dimethylated sulfur compounds in the Arctic during summer receding of the ice edge. Mar. Chem. 122, 105–117 (2010). 95. Kiene, R. P. & Linn, L. J. Distribution and turnover of dissolved DMSP and its relationship with bacterial production and dimethylsulfide in the Gulf of Mexico. Limnol. Oceanogr. 45, 849–861 (2000). 96. Lovelock, J. E., Maggs, R. J. & Rasmussen, R. A. Atmospheric dimethyl sulphide and the natural sulphur cycle. Nature 237, 452–453 (1972). 97. Charlson, R. J., Lovelock, J. E., Andreae, M. O. & Warren, S. G. Oceanic phytoplankton, atmospheric sulfur, cloud albedo and climate. Nature 326, 655–661 (1987). 98. Vila-Costa, M. et al. Dimethylsulfoniopropionate uptake by marine phytoplankton. Science 314, 652–654 (2006). 99. Spielmeyer, A., Gebser, B. & Pohnert, G. Investigations of the Uptake of Dimethylsulfoniopropionate by Phytoplankton. ChemBioChem 12, 2276–2279 (2011). 100. Raina, J.-B. et al. DMSP biosynthesis by an animal and its role in coral thermal stress response. Nature 502, 677–680 (2013). 101. Thomas, M. A. et al. Quantification of DMS aerosol-cloud-climate interactions using the ECHAM5-HAMMOZ model in a current climate scenario. Atmospheric Chem. Phys. 10, 7425–7438 (2010). 102. Dacey, J. W. H., Howse, F. A., Michaels, A. F. & Wakeham, S. G. Temporal variability of dimethylsulfide and dimethylsulfoniopropionate in the Sargasso Sea. Deep Sea Res. Part Oceanogr. Res. Pap. 45, 2085–2104 (1998). 103. Wakeham, S. G., Howes, B. L. & Dacey, J. W. H. Dimethyl sulphide in a stratified coastal salt pond. Nature 310, 770–772 (1984). 104. Andreae, M. O. Dimethylsulfoxide in marine and freshwaters. Limnol. Oceanogr. 25, 1054–1063 (1980). 105. Park, K.-T. & Lee, K. High-frequency, accurate measurement of dimethylsulfide in surface marine environments using a microporous membrane contactor: Underway seawater DMS measurement system. Limnol. Oceanogr. Methods 6, 548–557 (2008).  184 106. Tortell, P. D., Merzouk, A., Ianson, D., Pawlowicz, R. & Yelland, D. R. Influence of regional climate forcing on surface water pCO2, ΔO2/Ar and dimethylsulfide (DMS) along the southern British Columbia coast. Cont. Shelf Res. 47, 119–132 (2012). 107. Tortell, P. D. et al. Spatial distribution of pCO2, ΔO2/Ar and dimethylsulfide (DMS) in polynya waters and the sea ice zone of the Amundsen Sea, Antarctica. Deep Sea Res. Part II Top. Stud. Oceanogr. 71-76, 77–93 (2012). 108. Simó, R., Grimalt, J. O. & Albaigés, J. Sequential Method for the Field Determination of Nanomolar Concentrations of Dimethyl Sulfoxide in Natural Waters. Anal. Chem. 68, 1493–1498 (1996). 109. Hatton, A. D., Malin, G., McEwan, A. G. & Liss, P. S. Determination of Dimethyl Sulfoxide in Aqueous Solution by an Enzyme-Linked Method. Anal. Chem. 66, 4093–4096 (1994). 110. Vila-Costa, M., Kiene, R. P. & Simo, R. Seasonal variability of the dynamics of dimethylated sulfur compounds in a coastal northwest Mediterranean site. Limnol. Oceanogr. 53, 198–211 (2008). 111. Challenger, F. & Simpson, M. I. 320. Studies on biological methylation. Part XII. A precursor of the dimethyl sulphide evolved by Polysiphonia fastigiata. dimethyl-2-carboxyethylsulphonium hydroxide and its salts. J. Chem. Soc. Resumed 1591 (1948). doi:10.1039/jr9480001591 112. Swan, H. B. et al. An interlaboratory comparison for the quantification of aqueous dimethylsulfide. Limnol. Oceanogr. Methods 12, 784–794 (2014). 113. Kropuenske, L. R. et al. Photophysiology in two major Southern Ocean phytoplankton taxa: Photoprotection in Phaeocystis antarctica and Fragilariopsis cylindrus. Limnol. Oceanogr. 54, 1176–1196 (2009). 114. Smith, C. M. The relationship between phytoplankton pigment concentrations and DMSP, DMS, and DMSO in a diatom-dominated bloom in the Ross Sea, Antarctica. (Univeristy of South Alabama, 2011). at <http://gradworks.umi.com/14/96/1496368.html> 115. Bates, T. S. et al. The cycling of sulfur in surface seawater of the northeast Pacific. J. Geophys. Res. 99, 7835 (1994). 116. Breider, T. J. et al. Impact of BrO on dimethylsulfide in the remote marine boundary layer. Geophys. Res. Lett. 37, (2010). 117. Korhonen, H., Carslaw, K. S., Spracklen, D. V., Mann, G. W. & Woodhouse, M. T. Influence of oceanic dimethyl sulfide emissions on cloud condensation nuclei concentrations and seasonality over the remote Southern Hemisphere oceans: A global model study. J. Geophys. Res.-Atmospheres 113, 16 (2008). 118. Shaw, G. E. Bio-controlled themostasis involving the sulfur cycle. Clim. Change 5, 297–303 (1983). 119. Bopp, L., Aumont, O., Belviso, S. & Monfray, P. Potential impact of climate change on marine dimethyl sulfide emissions. Tellus Ser. B-Chem. Phys. Meteorol. 55, 11–22 (2003). 120. Bopp, L. et al. Will marine dimethylsulfide emissions amplify or alleviate global warming? A model study. Can. J. Fish. Aquat. Sci. 61, 826–835 (2004). 121. Cameron-Smith, P., Elliott, S., Maltrud, M., Erickson, D. & Wingenter, O. Changes in dimethyl sulfide oceanic distribution due to climate change. Geophys. Res. Lett. 38, (2011).  185 122. Gabric, A. J., Whetton, P. H. & Cropp, R. Dimethylsulphide production in the subantarctic southern ocean under enhanced greenhouse conditions. Tellus Ser. B-Chem. Phys. Meteorol. 53, 273–287 (2001). 123. Chin, M., Rood, R. B., Lin, S. J., Muller, J. F. & Thompson, A. M. Atmospheric sulfur cycle simulated in the global model GOCART: Model description and global properties. J. Geophys. Res.-Atmospheres 105, 24671–24687 (2000). 124. Zemmelink, H. J., Houghton, L., Dacey, J. W. H., Worby, A. P. & Liss, P. S. Emission of dimethylsulfide from Weddell Sea leads. Geophys. Res. Lett. 32, (2005). 125. Zemmelink, H. J., Dacey, J. W. H., Houghton, L., Hintsa, E. J. & Liss, P. S. Dimethylsulfide emissions over the multi-year ice of the western Weddell Sea. Geophys. Res. Lett. 35, (2008). 126. Gonzalez, J. M., Kiene, R. P. & Moran, M. A. Transformation of sulfur compounds by an abundant lineage of marine bacteria in the alpha-subclass of the class Proteobacteria. Appl. Environ. Microbiol. 65, 3810–3819 (1999). 127. Holmhansen, O., Mitchell, B. G., Hewes, C. D. & Karl, D. M. Phytoplankton blooms in the vicinity of Palmer Station, Antarctica. Polar Biol. 10, 49–57 (1989). 128. Arrigo, K. R. Physical control of chlorophyll a , POC, and TPN distributions in the pack ice of the Ross Sea, Antarctica. J. Geophys. Res. 108, (2003). 129. Kiene, R. P. & Slezak, D. Low dissolved DMSP concentrations in seawater revealed by small-volume gravity filtration and dialysis sampling. Limnol. Oceanogr.-Methods 4, 80–95 (2006). 130. Griebler, C. Dimethylsulfoxide (DMSO) reduction: A new approach to determine microbial activity in freshwater sediments. J. Microbiol. Methods 29, 31–40 (1997). 131. Griebler, C. & Slezak, D. in International Association of Theoretical and Applied Limnology, Proceedings, Vol 27, Pt 4 (ed. Williams, W. D.) 27, 2492–2497 (2001). 132. Harvey, G. R. & Lang, R. F. Dimethylsulfoxide and dimethylsulfone in the marine atmosphere. Geophys. Res. Lett. 13, 49–51 (1986). 133. Gralnick, J. A., Vali, H., Lies, D. P. & Newman, D. K. Extracellular respiration of dimethyl sulfoxide by Shewanella oneidensis strain MR-1. Proc. Natl. Acad. Sci. U. S. A. 103, 4669–4674 (2006). 134. McCrindle, S. L., Kappler, U. & McEwan, A. G. in Advances in Microbial Physiology, Vol 50 50, 147–+ (2005). 135. Sciare, J., Baboukas, E., Hancy, R., Mihalopoulos, N. & Nguyen, B. C. Seasonal variation of dimethylsulfoxide in rainwater at Amsterdam island in the southern Indian Ocean: Implications on the biogenic sulfur cycle. J. Atmospheric Chem. 30, 229–240 (1998). 136. Del Valle, D. A., Kieber, D. J., Bisgrove, J. & Kiene, R. P. Light-stimulated production of dissolved DMSO by a particle-associated process in the Ross Sea, Antarctica. Limnol. Oceanogr. 52, 2456–2466 (2007). 137. Hatton, A. D. & Wilson, S. T. Particulate dimethylsulphoxide and dimethylsulphoniopropionate in phytoplankton cultures and Scottish coastal waters. Aquat. Sci. 69, 330–340 (2007). 138. Simo, R., Hatton, A. D., Malin, G. & Liss, P. S. Particulate dimethyl sulphoxide in seawater: production by microplankton. Mar. Ecol.-Prog. Ser. 167, 291–296 (1998).  186 139. Kiene, R. P. et al. Distribution and cycling of dimethylsulfide, dimethylsulfoniopropionate, and dimethylsulfoxide during spring and early summer in the Southern Ocean south of New Zealand. Aquat. Sci. 69, 305–319 (2007). 140. Quinn, P. K. & Bates, T. S. The case against climate regulation via oceanic phytoplankton sulphur emissions. Nature 480, 51–56 (2011). 141. Howard, E. C. et al. Bacterial taxa that limit sulfur flux from the ocean. Science 314, 649–652 (2006). 142. Reisch, C. R. et al. Novel pathway for assimilation of dimethylsulphoniopropionate widespread in marine bacteria. Nature 473, 208–+ (2011). 143. Levine, N. M. et al. Environmental, biochemical and genetic drivers of DMSP degradation and DMS production in the Sargasso Sea. Environ. Microbiol. 14, 1210–1223 (2012). 144. Boyd, P. W. et al. In vitro iron enrichment experiments in the NE subarctic Pacific. Mar. Ecol.-Prog. Ser. 136, 179–193 (1996). 145. Le Clainche, Y. et al. A first appraisal of prognostic ocean DMS models and prospects for their use in climate models. Glob. Biogeochem. Cycles 24, (2010). 146. Sharma, S. et al. Flux estimation of oceanic dimethyl sulfide around North America. J. Geophys. Res.-Atmospheres 104, 21327–21342 (1999). 147. Freeland, H. J., Crawford, W. R. & Thomson, R. E. Currents along the pacific coast of Canada. Atmosphere-Ocean 22, 151–172 (1984). 148. Evans, W., Hales, B., Strutton, P. G. & Ianson, D. Sea-air CO2 fluxes in the western Canadian coastal ocean. Prog. Oceanogr. 101, 78–91 (2012). 149. Tortell, P. D. et al. Spatial variability and temporal dynamics of surface water pCO(2), Delta O-2/Ar and dimethylsulfide in the Ross Sea, Antarctica. Deep-Sea Res. Part -Oceanogr. Res. Pap. 58, 241–259 (2011). 150. Barwell-Clarke, J. & Brabant, F. Institute of Ocean Sciences nutrient methods and analysis. Can. Tech. Rep. Hydrogr. Ocean Sci. 182, vi+43 (1996). 151. Wanninkhof, R. & McGillis, W. R. A cubic relationship between air-sea CO2 exchange and wind speed. Geophys. Res. Lett. 26, 1889–1892 (1999). 152. Pena, M. A. & Varela, D. E. Seasonal and interannual variability in phytoplankton and nutrient dynamics along Line P in the NE subarctic Pacific. Prog. Oceanogr. 75, 200–222 (2007). 153. Gondwe, M., Krol, M., Gieskes, W., Klaassen, W. & de Baar, H. The contribution of ocean-leaving DMS to the global atmospheric burdens of DMS, MSA, SO2, and NSS SO4=. Glob. Biogeochem. Cycles 17, (2003). 154. Wolfe, G. V. & Steinke, M. Grazing-activated production of dimethyl sulfide (DMS) by two clones of Emiliania huxleyi. Limnol. Oceanogr. 41, 1151–1160 (1996). 155. Luce, M. et al. Distribution and microbial metabolism of dimethylsulfoniopropionate and dimethylsulfide during the 2007 Arctic ice minimum. J. Geophys. Res.-Oceans 116, (2011). 156. Archer, S. D., Stelfox-Widdicombe, C. E., Malin, G. & Burkill, P. H. Is dimethyl sulphide production related to microzooplankton herbivory in the southern North Sea? J. Plankton Res. 25, 235–242 (2003). 157. Salo, V., Simo, R. & Calbet, A. Revisiting the dilution technique to quantify the role of microzooplankton in DMS(P) cycling: laboratory and field tests. J. Plankton Res. 32, 1255–1267 (2010).  187 158. Archer, S. D., Safi, K., Hall, A., Cummings, D. G. & Harvey, M. Grazing suppression of dimethylsulphoniopropionate (DMSP) accumulation in iron-fertilised, sub-Antarctic waters. Deep-Sea Res. Part Ii-Top. Stud. Oceanogr. 58, 839–850 (2011). 159. Toole, D. A. et al. High dimethylsulfide photolysis rates in nitrate-rich Antarctic waters. Geophys. Res. Lett. 31, (2004). 160. Arrigo, K. R., van Dijken, G. & Long, M. Coastal Southern Ocean: A strong anthropogenic CO2 sink. Geophys. Res. Lett. 35, (2008). 161. DiTullio, G. R. et al. Rapid and early export of Phaeocystis antarctica blooms in the Ross Sea, Antarctica. Nature 404, 595–598 (2000). 162. Wingenter, O. W. et al. Changing concentrations of CO, CH4, C5H8, CH3Br ,CH3I, and dimethyl sulfide during the southern ocean iron enrichment experiments. Proc. Natl. Acad. Sci. U. S. A. 101, 8537–8541 (2004). 163. Yang, M. et al. Lagrangian evolution of DMS during the Southern Ocean gas exchange experiment: The effects of vertical mixing and biological community shift. J. Geophys. Res.-Oceans 118, 6774–6790 (2013). 164. Stammerjohn, S. E., Martinson, D. G., Smith, R. C. & Iannuzzi, R. A. Sea ice in the western Antarctic Peninsula region: Spatio-temporal variability from ecological and climate change perspectives. Deep-Sea Res. Part Ii-Top. Stud. Oceanogr. 55, 2041–2058 (2008). 165. Berresheim, H. et al. Measurements of dimethyl sulfide, dimethyl sulfoxide, dimethyl sulfone, and aerosol ions at Palmer Station, Antarctica. J. Geophys. Res.-Atmospheres 103, 1629–1637 (1998). 166. Kiene, R. P. & Gerard, G. Determination of trace levels of dimethylsulfoxide (DMSO) in seawater and rainwater. Mar. Chem. 47, 1–12 (1994). 167. Landry, M. R. & Hassett, R. P. Estimating the grazing impact of marine micro-zooplankton. Mar. Biol. 67, 283–288 (1982). 168. Landry, M. R., Kirshtein, J. & Constantinou, J. A refined dilution technique for measuring the community grazing impact of microzooplankton, with experimental tests in the central Equatorial Pacific. Mar. Ecol. Prog. Ser. 120, 53–63 (1995). 169. Bernard, K. S., Steinberg, D. K. & Schofield, O. M. E. Summertime grazing impact of the dominant macrozooplankton off the Western Antarctic Peninsula. Deep-Sea Res. Part -Oceanogr. Res. Pap. 62, 111–122 (2012). 170. Welschmeyer, N. A. Fluorometric analysis of chlorophyll-a in the presence of chlorophyllb and pheopigments. Limnol. Oceanogr. 39, 1985–1992 (1994). 171. Goldman, J. A. L. et al. Gross and net production during the spring bloom along the Western Antarctic Peninsula. New Phytol. 205, 182–191 (2015). 172. Tortell, P. D. et al. Metabolic balance of coastal Antarctic waters revealed by autonomous p CO 2 and ΔO 2 /Ar measurements: METABOLIC BALANCE OF ANTARCTIC WATERS. Geophys. Res. Lett. 41, 6803–6810 (2014). 173. Arrigo, K. R. et al. Phytoplankton community structure and the drawdown of nutrients and CO2 in the Southern Ocean. Science 283, 365–367 (1999). 174. Kwint, R. L. J. & Kramer, K. J. M. A new sensitive tracer for the determination of zooplankton grazing activity. J. Plankton Res. 18, 1513–1518 (1996). 175. Kiene, R. P. & Bates, T. S. Biological removal of dimethyl sulphide from sea water. Nature 345, 702–705 (1990).  188 176. Vernet, M. et al. Primary production within the sea-ice zone west of the Antarctic Peninsula: I-Sea ice, summer mixed layer, and irradiance. Deep-Sea Res. Part Ii-Top. Stud. Oceanogr. 55, 2068–2085 (2008). 177. Ducklow, H. et al. West Antarctic Peninsula: An Ice-Dependent Coastal Marine Ecosystem in Transition. Oceanography 26, 190–203 (2013). 178. Riebesell, U., Gattuso, J.-P., Thingstad, T. F. & Middelburg, J. J. Preface ‘Arctic ocean acidification: pelagic ecosystem and biogeochemical responses during a mesocosm study’. Biogeosciences 10, 5619–5626 (2013). 179. Six, K. D. et al. Global warming amplified by reduced sulphur fluxes as a result of ocean acidification. Nat. Clim. Change 3, 975–978 (2013). 180. Galí, M. & Simó, R. A meta-analysis of oceanic DMS and DMSP cycling processes: Disentangling the summer paradox. Glob. Biogeochem. Cycles 29, 496–515 (2015).  


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