"Science, Faculty of"@en . "Earth, Ocean and Atmospheric Sciences, Department of"@en . "DSpace"@en . "UBCV"@en . "Nemcek, Nina"@en . "2011-03-08T00:35:57Z"@en . "2007"@en . "Master of Science - MSc"@en . "University of British Columbia"@en . "A novel technique, membrane inlet mass spectrometry (MIMS), was used to measure\r\ndimethylsulfide gas (DMS) and algal dimethylsulfoniopropionate (DMSPp) concentrations in\r\ntwo different marine ecosystems of the NE Pacific. In oceanic waters along Line P, DMS levels\r\nhad been observed to be unusually high, yet particulate DMSP levels had not been extensively\r\nmeasured. DMSPp concentrations during 3 consecutive spring cruises ranged from 0.2-63.2 nM\r\n(mean 21.5 nM , s.d. 15.0 nM ) in the upper 50 m of the water column, and varied significantly\r\nwith depth, across stations and between study years. DMSPp generally decreased with depth and\r\ndistance from the coast. DMSPp concentrations at most stations in 2003 were 2 to 3-fold higher\r\nthan in subsequent years, and were significantly correlated to the biomass of dinoflagellates\r\n(r2 = 0.46) across the survey region. Although phytoplankton biomass (chlorophyll a) also\r\ndeclined in 2004-2005, DMSPp:chl ratios did as well, indicating a physiological or taxonomic\r\nchange in the phytoplankton community.\r\nSurface DMS concentrations were measured underway along with pCO\u00E2\u0082\u0082, O\u00E2\u0082\u0082/Ar,\r\ntemperature, salinity and chlorophyll a in productive, coastal waters off British Columbia. All\r\nparameters exhibited large ranges, (pCO\u00E2\u0082\u0082, 200-747 ppm; DMS, "https://circle.library.ubc.ca/rest/handle/2429/32146?expand=metadata"@en . "M E M B R A N E INLET MASS SPECTROMETRY (MIMS): A N O V E L A P P R O A C H TO THE OCEANIC M E A S U R E M E N T OF DIMETHYLSULFIDE by NINA N E M C E K B S c , University of British Columbia, 2003 A THESIS SUBMITTED IN PARTIAL F U L F I L L M E N T OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in F A C U L T Y OF G R A D U A T E STUDIES (Oceanography) UNIVERSITY OF BRITISH C O L U M B I A March 2007 \u00C2\u00A9 Nina Nemcek, 2007 Abstract A nove l technique, membrane inlet mass spectrometry ( M I M S ) , was used to measure d imethy lsu l f ide gas ( D M S ) and a lgal d imethy lsu l foniopropionate ( D M S P p ) concentrations in two different mar ine ecosystems o f the N E Pac i f i c . In oceanic waters a long L i n e P, D M S levels had been observed to be unusua l l y h igh , yet particulate D M S P levels had not been extensively measured. D M S P p concentrations dur ing 3 consecutive spr ing cruises ranged f rom 0.2-63.2 n M (mean 21.5 n M , s.d. 15.0 n M ) in the upper 50 m o f the water c o l u m n , and var ied s igni f icant ly w i th depth, across stations and between study years. D M S P p general ly decreased w i th depth and distance f rom the coast. D M S P p concentrations at most stations i n 2003 were 2 to 3-fold higher than in subsequent years, and were s igni f icant ly correlated to the b iomass o f dinoflagel lates (r2 = 0.46) across the survey region. A l t hough phytoplankton b iomass (ch lorophy l l a) also decl ined i n 2004-2005, D M S P p x h l ratios d id as w e l l , ind icat ing a phys io log i ca l or taxonomic change i n the phytoplankton communi ty . Surface D M S concentrations were measured underway a long w i t h pCOi, Oil Ax, temperature, sa l in i ty and ch lo rophy l l a i n product ive, coastal waters o f f B r i t i sh C o l u m b i a . A l l parameters exhib i ted large ranges, (pCOi, 200-747 p p m ; D M S , < l-28 .7 n M ; ch l a, <0.1-33.2 p g L\" 1), h igh l ight ing the dynamic nature o f the region. A tight anti-correlation between pC02 and 62?Ar was observed across the region (r = 0.90), w i t h the distr ibut ions o f these gases strongly inf luenced by both b io l og i ca l (photosynthesis and respiration) and phys i ca l (upwel l ing) processes. In contrast, D M S levels w h i c h exhibi ted rap id , fine-scale f luctuations irresolvable w i th tradit ional methods, were unrelated to any single var iable. A signif icant l inear relat ionship although w i t h a different sca l ing factor to that der ived f rom open ocean data. L o w e r resolut ion was however observed between D M S and the ch lorophy l l to m i x e d layer depth ratio (r2 = 0.83), sampl ing i n this reg ion can introduce errors as large as 4 1 % o f the mean concentration for D M S , emphas iz ing the ut i l i ty o f MHVIS i n dynamic areas. In conc lus ion , M I M S proved to be a signif icant advance for D M S measurement, yet improvements need to be made for it to be a viable alternative to other methods for D M S P measurements. Table of Contents Abstract i i Table of Contents iv List of Tables v i List of Figures vu Acknowledgements ix Dedication x Co-Authorship Statement x i Chapter 1: Introduction 1 1.1 Dimethyl sulfide: Sources, Sinks and Climate Links 1 1.2 The C L A W Hypothesis: 20 Years Later 4 1.3 Membrane Inlet Mass Spectrometry 6 1.4 Thesis Objectives 7 1.5 References 8 Chapter 2: Springtime Variability in Particulate D M S P Concentrations along Line P, N E Pacific Ocean 11 2.1 Introduction 11 2.2 Materials and Methods 13 2.3 Results 19 2.4 Discussions 24 2.5 References 45 Chapter 3: High-Resolution Measurements of D M S , CO2, and 02 /Ar in Productive, Coastal Waters around Vancouver Island, Canada 49 3.1 Introduction 49 3.2 Materials and Methods 52 3.3 Results '. 58 3.4 Discussion 67 3.5 Conclusions 78 iv 3.6 References . . : 89 Chapter 4: Conc lus ions 93 4.1 Thesis Ove r v i ew 93 4.2 Eva lua t ion o f M U M S for D M S P / D M S Measurements 94 4.3 Successes and P i t fa l l s 95 4.4 Future D i rec t ions 97 4.5 References '.. '. 99 v List of Tables Table 2.1: A n c i l l a r y oceanographic data for a l l stations surveyed dur ing 2003 33 Table 2.2: Coef f i c ients o f determinat ion (r2) from linear regressions between D M S P p levels and the absolute and relat ive carbon biomass o f different phytoplankton groups for al l stations dur ing the 2003 survey 34 Table 2.3: The contr ibut ion o f D M S P p to total phytoplankton ce l l carbon in 2003 35 Table 3.1: Abso lu te and relative asymptotic interpolat ion errors a long the 6 major transects . . . .79 List of Figures Figure 2.1: Map of the NE Pacific showing the 5 major stations along Line P 36 Figure 2.2: Depth profiles of DMSPp measured at the 5 major stations along Line P during 3 consecutive spring cruises 37 Figure 2.3: Depth profiles of chlorophyll a measured at the 5 major stations along Line P during 3 consecutive spring cruises 38 Figure 2.4: Depth profiles of the DMSPp :chl ratio measured at the 5 major stations along Line P during 3 consecutive spring cruises 39 Figure 2.5: Contour plots illustrating spatial and interannual variability in springtime DMSPp and chlorophyll levels along Line P 40 Figure 2.6: Relationship between chlorophyll and DMSPp concentrations along Line P for the 3 year pooled dataset 41 Figure 2.7: The relative abundance of the different phytoplankton groups enumerated at (a) 10 m depth and (b) the chlorophyll max. at all stations surveyed in 2003 42 Figure 2.8: The contribution of the different phytoplankton groups to total carbon biomass at (a) 10 m depth and (b) the chlorophyll max. at all stations surveyed in 2003 43 Figure 2.9: 50 m depth integrated DMSPpxhl ratios plotted against the mixed layer nitrate concentration 44 Figure 3.1: Map of southwestern British Columbia, Canada showing the location of underway transects 80 Figure 3.2: Surface plots of (a) temperature (\u00C2\u00B0C), (b) salinity (psu), (c) chlorophyll a (ug L\"1), (d) pC02 (ppm), (e) 0 2 /Ar (torr ratio), and (f) DMS (nM) 81 Figure 3.3: Detailed south-north view of all variables measured along T5 82 Figure 3.4: Detailed view of all variables measured along T7 83 Figure 3.5: The correlation across all transects between (a)pCC>2 and (VAr (r = 0.90) with corresponding chl a concentrations overlaid (colourbar), and (b) chl a and (VAr (r2 = 0.19) with corresponding temperature overlaid (colourbar) 84 Figure 3.6: The correlation across all transects between chl a and DMS (r2 = 0.06) 85 vn Figure 3.7: Results o f the P C A showing the strong separation between pCC\u00C2\u00BB 2 and C V A r i n two-dimensional space and the clear part i t ioning between phys ica l (T, S) and b io log i ca l (chl a) variables 86 F igure 3.8: Average D M S concentrations plotted against C H L / M L D ratios for the lA degree grids 87 F igure 3.9: Autocor re la t ion functions for a l l parameters measured a long transect 5 (a); Average D L S for a l l parameters for the entire survey (b) 88 v i i i A c k n o w l e d g e m e n t s This project would not have been possible without the tireless efforts of Mark Buckley and Robert Stannard at Hiden Analytical in the development and troubleshooting of the MIMS. My supervisor Philippe Tortell, Celine Gueguen and Chris Payne also helped tremendously over the years with method development and repairs on \"Herbie\". The author would like to acknowledge the Captain, crew, and science personnel of the CCGS John P. Tully for their assistance during the Line P and Queen Charlotte Sound surveys. The Natural Sciences and Engineering Research Council of Canada (NSERC) and the University of British Columbia funded this project in the form of an Undergraduate Student Research Award (2003), Postgraduate Scholarship and University Fellowship (2004-2006). A sincere thank you goes to the following people for their assistance with the Line P dataset: T. Peterson for performing phytoplankton identification and enumeration in 2003, M. Robert for organizing the cruises and providing access to CTD and ancillary data, W. Richardson and J. Barwell-Clarke for nutrient analysis, L. Richier for chlorophyll analysis in 2005, and M. Arychuk and CS. Wong for method intercomparisons and fruitful discussions on DMS/DMSP trends in the NE Pacific. A special thank you goes to Darren Tuele for DMSPp sample collection in 2005 and invaluable support at sea during all cruises. I am grateful to Debby Ianson for allowing me to participate in her Queen Charlotte Sound cruise, for providing ancillary CTD, nutrient, thermosalinograph and pCOi equilibrator data, and for her guidance and motivation during the writing of Chapter 3. Dave Mackas kindly provided unpublished CTD data and Marlene Jeffries helped with Matlab figures. Finally, I must thank my labmates for their camaraderie and support, and my supervisor Philippe Tortell for conceiving the application of MIMS for DMS analysis and for sending me to sea. ix Dedication To DT for pushing me to finish and to my parents for never pushing me. x Co-Authorship Statement I w i l l be the p r imary author o f the manuscript \"fr iterannual var iab i l i t y i n springtime particulate D M S P concentrations a long a coastal to oceanic transect in the N E P a c i f i c \" wh i ch w i l l be wri t ten based on Chapter 2. I performed the D M S P p sample analysis i n a l l three years, integrated and ana lyzed the anc i l l a ry data, and w i l l wr i te the manuscr ipt . Nutr ient , ch lorophy l l and C T D data were p rov ided by M a r i e Robert o f the Institute o f Ocean Sciences. The second author, D r . T a w n y a Peterson per formed phytoplankton counts and p rov ided ce l l d imensions in 2003 f rom w h i c h I calculated ce l l b iovo lumes and the b iomass o f spec i f ic algal groups. The third author, m y supervisor D r . Ph i l i ppe Tor te l l introduced me to the MEVIS , assisted w i th sample analysis dur ing the cruise i n 2003 , and prov ided lab space and resources. H e also prov ided guidance on the project and edited the manuscript. I wrote the manuscr ipt \" A high-resolution survey o f D M S , CO2, and 02/Ar distr ibutions in product ive coastal waters\" , based on Chapter 3, w h i c h was submitted to Global Biogeochemical Cycles. I devised the sampl ing strategy, co l lected and processed D M S P and ch lorophy l l samples and operated the MEVIS dur ing the Queen Charlotte Sound survey in 2004. Post-cruise, I processed the MEVIS , PCO2 equil ibrator, and thermosal inograph data, the latter two prov ided by m y second author D r . D e b b y Ianson o f the Institute o f Ocean Sciences. Dr . Ianson prov ided a berth on her research cruise as we l l as anci l lary C T D and nutrient data. The third author D r . Ph i l i ppe Tor te l l performed the pr inc ipa l components analysis and calculated autocorrelation funct ions and interpolat ion errors for m y underway dataset. A l l other interpretations and ideas i n the d iscuss ion were m y own , however both o f m y co-authors' edits improved the f ina l manuscr ipt greatly. I cert i fy that the above statements about authorship are correct. x i 1. Introduction 1.1 Dimethylsulfide: Sources, Sinks and Climate Links M i c r o s c o p i c algae dwe l l i ng i n surface waters o f the w o r l d ' s oceans have a profound inf luence on Ear th 's c l imate through the consumpt ion and product ion o f c l imato log ica l l y active gases. One such gas, d imethy lsu l f ide ( D M S ) , is formed i n the oceans f rom the breakdown o f dimethylsul foniopropionate ( D M S P ) , a compound produced i n large quantities by many species o f phytoplankton for a variety o f metabol ic functions (see Sect ion 2.1). In 1972, James Love lock observed that D M S was ubiquitous i n marine surface waters and postulated that the f lux o f D M S f rom the oceans to the atmosphere was suff ic ient ly large to account for the \" m i s s i n g \" sulfur in global transport mode ls [Lovelock et al, 1972]. Over the past two decades, extensive oceanic D M S measurements have supported this hypothesis. Despi te a g loba l mean surface water concentration o f < 3 n M [Kettle et al, 1999], D M S is everywhere supersaturated i n the ocean result ing i n a steady f lux o f sul fur to the atmosphere. Th i s f lux represents the largest natural source o f atmospheric sul fur const i tut ing 20 % o f total g lobal emiss ions, but over 40 % o f the atmospheric sul fur burden due to the relat ively longer l i fet imes o f D M S - d e r i v e d aerosols compared to anthropogenic ones [Chin and Jacob, 1996]. It was not unt i l the late 1980's when Love l o ck and col leagues introduced the C L A W hypothesis (an acronym for the authors' names), that interest i n D M S and its potential c l imat ic effects skyrocketed [Charlson et al, 1987]. The C L A W hypothesis states that phytoplankton regulate their environment (and b y consequence the Ear th 's c l imate) through the product ion.of D M S w h i c h affects planetary albedo. A s noted earlier b y Shaw [1983], D M S vented f rom the oceans is qu i ck l y o x i d i z e d in the atmosphere to non-sea-salt sulfate (NSS-SO4) and methanesulfonic ac id ( M S A ) , species that act as c loud condensat ion nuc le i ( C C N ) , or seed 1 crystals for c l oud format ion. Charlson et al. [1987] argued that the D M S - d e r i v e d NSS -SO4 aerosols are the m a i n source o f C C N over much o f the Ear th 's surface. Thus , according to the C L A W hypothesis, h igh solar irradiance stimulates algal D M S product ion w h i c h leads to increased planetary c loud cover and a reduct ion i n the amount o f radiat ion reaching the planet 's surface. Th is in turn has a feedback effect on the phytoplankton that init iated the process through changes i n sea surface temperature and incident sunl ight, thereby c los ing the loop. However , the s ign o f this feedback was unclear at the t ime, since many o f the factors cont ro l l ing D M S product ion and its dependence on species compos i t ion were unknown . A negative feedback on D M S product ion w o u l d i m p l y a self-regulating c l imate system mit igated by the marine biota [Charlson et al, 1987]. The C L A W hypothesis st imulated the intense research efforts directed at e lucidat ing the intricacies o f the D M S cyc le that have been occurr ing over the past two decades. F r o m these efforts we have greatly improved our understanding o f the comp lex foodweb dynamics that drive the cyc l ing o f D M S P and its byproducts. A l l D M S originates from algal or particulate D M S P ( D M S P p ) . D M S P product ion varies w ide l y b y species w i t h dinof lagel lates and prymnesiophytes generally be ing major producers, and diatoms m ino r ones [Keller et al, 1989]. However , even w i th in algal groups there is s ignif icant var iab i l i ty i n ce l lu lar D M S P content [Keller et al, 1989], and external factors such as nutrient ava i lab i l i ty can increase D M S P product ion i n species generally considered to be l ow producers [Sunda et al, 2002] . Zoop lankton graz ing [Dacey and Wakeham, 1986] and v i ra l lys is [Malin et al, 1998] o f phytoplankton promote the release o f D M S P into seawater where it is rap id ly consumed by various components o f the mar ine foodweb. Heterotrophic bacter ia are the dominant sink for d issolved D M S P ( D M S P d ) , a l though recent evidence suggests autotrophic cyanobacteria and 2 diatoms are also capable o f D M S P uptake and ass imi la t ion [Vila-Costa et al, 2006] . Bacter ia possess two compet ing pathways for the metabol ic b reakdown o f D M S P . The demethylat ion pathway converts D M S P to methanethiol w h i c h is qu i ck l y incorporated into protein and bacterial b iomass, thus d iver t ing sul fur away f rom D M S [Gonzales et al, 1999]. In contrast, the cleavage pathway ut i l izes D M S P - l y a s e to convert D M S P to D M S and acrylate, a l though this is the fate o f on ly 5-10 % o f the D M S P d metabol ized i n the water c o l u m n [Kiene et al, 2000] . Thus, the product ion o f c l imato log i ca l l y important D M S constitutes on l y a sma l l fract ion o f the large f lux o f reduced sul fur in the surface ocean. Some phytoplankton species also possess the DMSP- l y a se enzyme w h i c h mixes w i th its substrate dur ing graz ing or v i ra l lysis leading to h igh D M S product ion [Malin et al, 1998]. It is thought that algae w i t h D M S P - l y a s e act iv i ty use this enzyme as an activated chemica l defence against predators, as m ic rozoop lank ton grazers are deterred by acrylate [Wolfe et al, 1997]. M o r e recently it has been shown that acrylate may also deter v i ruses, s ince h igh DMSP- l yase containing strains o f the prymnesiophyte Emiliania huxleyii appear to be immune to v i ra l attack [Evans et al, 2006] . In addi t ion, phytoplankton may use D M S P - l y a s e to trigger a power fu l antioxidant cascade, as D M S , acrylate, D M S O , and other byproducts o f D M S P are power fu l free radical scavengers [Sunda et al, 2002] . There are three major s inks for D M S in marine surface waters, (1) bacterial consumpt ion, (2) photolys is and (3) vent i la t ion to the atmosphere. The relat ive importance o f each depends on the depth interval considered and the b io log i ca l , chemica l and meteoro log ica l condit ions [Kieber et al, 1996]. Ven t i l a t i on is general ly considered a m ino r s ink because it occurs on ly at the air-sea interface, but is the dominant loss process at h igh w i n d speeds. Photo lys is converts D M S to non-volati le compounds inc lud ing D M S O and occurs over the depth range to w h i c h U V l ight can 3 penetrate [Kieber et al, 1996]. Th i s depth depends on c loud cover, geographic locat ion and the local opt ica l properties o f seawater, and can range from a few meters i n coastal waters to >75 m in opt ica l l y clear waters under h igh U V f lux [i.e. Toole et al, 2004] . In addit ion, photolysis rates are strongly inf luenced by the concentrations o f photosensit izers such as chromophor ic d isso lved organic matter ( C D O M ) [Kieber et al, 1996; Toole et al, 2004] . Bacter ia l consumpt ion o f D M S is l i ke l y the dominant s ink for D M S over most o f the surface ocean because it occurs over the greatest depth interval and over the largest range o f condit ions [Kiene et al, 1990; Kieber et al, 1996]. However , under h igh U V l ight condi t ions, photolysis can exceed b ioconsumpt ion as the dominant loss process as bacter ia i n surface waters become photoinhibi ted [Toole et al, 2004] . 1.2 T h e C L A W H y p o t h e s i s : 20 Y e a r s L a t e r There is no doubt that b iogenic sulfur aerosols have a large inf luence on global cl imate. They are the m a i n source o f C C N over m u c h o f the remote mar ine atmosphere, part icular ly in the Southern Hemisphere where there are few terrestrial sources. Moreove r , a strong coherence has been observed between seasonality i n D M S emissions and C C N numbers [Ayers and Gras, 1991], as w e l l as c loudiness [Boers et al, 1994], at least i n the Southern Hemisphere . Ice core data from Vos tok , Anta rc t i ca reveal concentrations o f NSS - S O 4 and M S A (a byproduct exc lus ive to D M S ) were s ign i f i cant ly higher dur ing g lac ia l per iods compared to interglacial ones, and were t ight ly anti-correlated to past temperature f luctuations [Legrand et al, 1991]. Thus, marine b iota inf luence c l imate v i a the oceanic sulfur cyc le , i n addi t ion to the b io log ica l carbon pump. M o d e l l i n g studies reveal that 2/3 o f the present day coo l i ng result ing f rom b io log ica l product ion i n the oceans can be attributed to D M S emiss ions , w i th on l y a 1/3 contr ibut ion f rom CO2 uptake [Watson andLiss, 1998]. M o r e recent mode ls estimate that reducing present day D M S emiss ions by ha l f w o u l d result i n a 1.6 \u00C2\u00B0 C increase in g lobal mean sea surface temperatures [Gunson et al, 2006] . A l t h o u g h it is clear that D M S emissions have a coo l i ng effect on g loba l c l imate, it is st i l l unclear what impact a changing c l imate w i l l have on future D M S emiss ions. Impending changes in solar radiat ion, surface ocean stratif ication, nutrient supply, and w i n d speeds w i l l a l l inf luence the compet ing pathways that lead to net D M S product ion, such that neither the magnitude nor the sign o f this effect is certain at present. Increases i n the extent and strength o f surface ocean stratif ication and in incident solar radiat ion predicted under g loba l wa rm ing scenarios cou ld indeed lead to increased D M S product ion for a number o f reasons. F i rs t ly , stratif ied, nutrient-depleted waters tend to favour the growth o f h igh D M S P produc ing taxa such as dinoflagel lates and prymnesiophytes [Margalef, 1978; Keller et al, 1989]. Secondly , h igh incident U V radiat ion promotes elevated D M S P product ion i n these and other species as a phys io log i ca l response to photo-oxidative stress [Sunda et al, 2002] . F ina l l y , strong U V photo inh ib i t ion o f bacterial growth i n a stratif ied water c o l u m n results i n less sulfur ass imi la t ion into b iomass and a higher proport ion o f D M S P converted to D M S [Simo and Pedros-Alio, 1999]. M o s t recent studies predict dramatic lat i tudinal var iab i l i t y in mode l l ed D M S f luxes under g lobal wa rm ing , w i th smal l to moderate increases i n net g loba l D M S emissions [Bopp et al, 2003 ; Gabric et al, 2004]. These predict ions do support the self-regulating negative feedback C L A W hypothesis, although on ly weak l y and w i t h large uncertainty. A n improved mechanist ic understanding o f the complex web that is the oceanic D M S cycle is urgently needed to reduce this uncertainty. 5 1.3 Membrane Inlet Mass Spectrometry Tradi t iona l ly , oceanic measurements o f D M S and related sul fur compounds have been performed by purge and trap gas chromatography ( P T G C ) . In this method, discrete seawater samples are sparged w i t h an inert gas to transfer the volat i le sul fur analyte into a concentrating cryogenic trap. The trap is subsequently heated to release the analyte into a gas chromatographic co lumn that separates the var ious sulphur compounds for measurement by either a chemi luminescent or photometr ic detector. A l t hough this method offers excel lent sensit iv ity (as a result o f the cryogenic t rapping step), it is labour intensive, and t ime-consuming. Furthermore, it is l imi ted to the analysis o f discrete samples, and thus offers poor spatial resolut ion capabil i t ies. In contrast, the appl icat ion o f membrane inlet mass spectrometry ( M I M S ) to the measurement o f D M S i n seawater c i rcumvents many o f these shortcomings. M I M S is not a new method; it was first introduced In the early 1960s [Hoch and Kok, 1963] and is qu i ck l y ga in ing acceptance as an alternative to gas chromatography for the rapid analysis o f volat i le organic compounds in air and water samples [Ketola et al, 1996]. MEVIS offers the advantages o f shorter analysis t ime and a m u c h larger linear, dynamic range over P T G C methods, w i t h comparable detection l imi ts [Ketola et al, 1996]. O u r adaptation o f this technique uses a gas-permeable d imethy ls i l i cone membrane as the interface between a seawater sample and the vacuum o f the mass spectrometer. The sensit iv i ty o f the system is direct ly proport ional to the surface area o f the membrane exposed to the sample. A n y gases d issolved i n the aqueous sample di f fuse through the membrane into the vacuum chamber, where they are ion ized by electron impact and separated in a quadrupole f i l ter based on their mass-to-charge (m/z) ratios. M a j o r gases (CO2, O2, A r , N2) are detected by a Faraday cup and trace gases such as D M S by a secondary electron mul t ip l i e r ( S E M ) . The system can be used w i t h a membrane inlet 6 probe for discrete appl icat ions or connected direct ly to a sh ip 's seawater intake system for . continuous, underway mon i to r ing o f d isso lved gas concentrations. B y emp loy ing the single i on moni tor ing ( S IM) mode o f the mass spectrometer software, and c yc l i ng through their respective m/z ratios, mul t ip le gases can be moni tored i n real-time, pseudo-simultaneously w i th very l itt le effort. Thus the m a i n advantages o f the M I M S system over P T G C for the analysis o f D M S are its (1) s igni f icant ly higher spatial resolut ion capabil i t ies and (2) rap id , semi-automated analysis capabil i t ies. One o f the m a i n contr ibutions o f this thesis is the development, troubleshooting and . field-testing o f a sea-going M I M S system. 1.4 Thes i s Ob jec t i v es The foremost object ive o f this work was to determine whether M I M S cou ld be successful ly appl ied to the measurement o f trace D M S and D M S P concentrations i n seawater. Once this in i t ia l goa l had been achieved, the research objectives were (1) to measure particulate D M S P concentrations i n the N E Pac i f i c a long L i ne P where such measurements d id not prev ious ly exist, (2) to examine the spatial and interannual var iab i l i t y i n particulate D M S P p levels a long L i n e P i n the context o f phytoplankton species compos i t i on and nutrient levels, (3) to use the high-resolution underway capabi l i t ies o f the M I M S to ident i fy fine-scale structure i n D M S concentrations in dynamic and product ive waters o f coastal B .C . and f ina l ly , (4) to use these high-resolution D M S data in conjunct ion w i th underway anc i l la ry parameters to examine the factors d r i v i ng the observed D M S distr ibutions. A s a result o f the comp lex cyc l ing o f D M S in seawater and its dependence on a myr iad o f b io log i ca l , meteoro log ica l , phys ica l and chemica l factors, it was hypothes ized that D M S distr ibutions w o u l d be unrelated to any single variable and w o u l d best be s imulated by algori thms incorporat ing mul t ip le components. 7 1.5 References Ayers , G.P. and J . L . Gras (1991), Seasonal relat ionship between c loud condensation nucle i and aerosol methanesulphonate i n marine air, Nature, 353, 834-835. Boers, R., G.P. Aye r s and J . L . Gras (1994), Coherence between seasonal var iat ion i n satellite-der ived c loud opt ica l depth and boundary-layer C C N concentrations at a mid-latitude Southern Hemisphere station, Tellus, Ser B, 46,123-131. Bopp , L., O. A u m o n t , S. B e l v i s o and P. M o n f r a y (2003), Potent ia l impact o f c l imate change on marine d imethy l sulphide emiss ions, Tellus, Ser B, 55(1), 11-22. Char l son , R.J., J .E . Love l o ck , M . O . Andreae, and S.G. War ren (1987), Oceanic phytoplankton, atmospheric sulfur, c loud albedo and c l imate, Nature, 326, 655-661. C h i n , M . , and D J . Jacob, (1996) Anthropogenic and natural contr ibut ions to tropospheric sulphate: a g loba l m o d e l analysis, J . Geophys. Res., 101, 18,691-18,699. Dacey, J .W .H . , and S .G. W a k e h a m (1986), Oceanic d imethy lsu l f ide : P roduc t ion dur ing zoop lankton graz ing on phytoplankton, Science, 233(4770), 1314-1316. Evans, C , G . M a l i n , W . H . W i l s o n , and P.S. L i s s (2006), Infectious titers o f Emiliania huxleyi virus 86 are reduced by exposure to m i l l i m o l a r d imethy l su lphide and acryl ic ac id , Limnol. Oceanogr. 51(5), 2468-2471. Gabr ic , A . J . , R. S i m o , R .A . C ropp , A . C . H i rs t and J . Dachs (2004), M o d e l i n g estimates o f the g lobal emiss ion o f d imethy lsu l f ide under enhanced greenhouse gas condi t ions, Global Biogeochem. Cycles, 18, G B 2 0 1 4 , do i : ;10 .1029/2003GB002183. Gonza lez , J . M . , R. P. K i e n e and M . A . M o r a n (1999), Transformat ion o f sul fur compounds by an abundant l ineage o f mar ine bacteria i n the alpha-subclass o f the class Proteobacteria, App. Environ. Microbiol. 65(9), 3810-3819. Gunson , J .R., S .A . Spa l l , T .R . Ande r son , A . Jones, I.J. Totterdel l , and M J . Woodage (2006), C l imate sensi t iv i ty to ocean d imethy lsu lphide emiss ions, Geophys. Res. Lett., 33, L 0 7 7 0 1 , do i : 10 .1029/2005GL024982. H o c h , G . and B. K o k (1963), A mass spectrometer inlet system for sampl ing gases d issolved i n l i qu id phases, Arch. Biochem. Biophys., 101, 160-170. Ke l l e r , M . D . , W . K . B e l l o w s , and R.R.L . Gu i l l a rd (1989), D i m e t h y l sul f ide product ion in marine phytoplankton, i n Biogenic sulfur in the environment, edited by E.S. Sa l tzman and W.J . Cooper , p. 167-180, A m e r i c a n Chemica l Society, Wash ington , D .C . 8 Keto la , R .A. , V . T . V i r k k i , M . Oja la , V . K o m p p a and T. Ko t i aho (1996), Compar i son o f different methods for the determinat ion o f volat i le organic compounds i n water samples, Talanta, 44, 373-382. Ket t le , A . J . et a l . (1999), A g loba l database o f sea surface d imethy lsu l f ide ( D M S ) measurements and a procedure to predict sea surface D M S as a funct ion o f latitude, longitude and month , Global Biogeochem. Cycles, 13, 399-444. K ieber , D.J., J . J iao, R.P. K i ene , and T.S . Bates (1996), Impact o f d imethy lsu l f ide photochemistry on methy l sulfur c yc l i ng i n the equatorial Pac i f i c Ocean, J. Geophys. Res., 101(C2), 3715-3722. K i ene , R.P., and T .S . Bates (1990), B i o l o g i c a l remova l o f d imethy l sul f ide f rom sea water, Nature, 345, 702-705. K i ene , R.P., L.J. L i n n , and J . A . B ru ton (2000), N e w and important roles for D M S P in marine m ic rob ia l communi t i es , J. Sea Res. 43, 209-224. Legrand, M . , C . Feniet-Saigne, E.S. Sa l tzman, C . germain, N.I. B a r k o v and V . N . Petrov (1991), Ice-core record o f oceanic emissions o f d imethy lsu lphide dur ing the last c l imate cycle, Nature, 350, 144-146. Love lock , J .E. , R.J. M a g g s and R .A . Rasmussen (1972), A tmospher i c sulphur and the natural sulphur cyc le , Nature, 237, 452-453. M a l i n , G . , W . H . W i l s o n , P.S. L i s s and G . Bratbak (1998), E levated product ion o f d imethy lsu l f ide result ing f rom v i ra l infect ion o f cultures o f Phaeocystis pouchetii, Limnol. Oceanogr. 43, 1389-1393. Margalef , R. (1978), L i fe-forms o f phytoplankton as surv iva l alternatives i n an unstable environment, Oceanol. Acta, 1, 493-509., Shaw, G.E . (1983), B io-contro l led fhermostasis i n vo l v i ng the sul fur cyc le , Clim. Change, 5, 297-303. S imo , R., and C. Pedros-A l io (1999), Ro l e o f vert ica l m i x i n g i n cont ro l l ing the oceanic product ion o f d imethy l sulphide, Nature 402, 396-399. Sunda, W. , D.J. K i ebe r , R.P. K i ene , and S. Huntsman (2002), A n ant ioxidant funct ion for D M S P and D M S i n mar ine algae, Nature, 418, 317-320. Too le , D .A . , D.J. K i ebe r , R.P. K i e n e , E . M . Wh i te , J . B i sg rove , D . A . de l V a l l e , and D. S lezak (2004), H i g h d imethy lsu l f ide photolys is rates i n nitrate-rich Antarc t i c waters, Geophys. Res. Lett., 31, L I 1307, do i :10 .1029/2004GL019863. 9 Vi la-Costa , M . , R. S imo , H . Harada, J . M . Gaso l , D. S lezak and R.P. K i e n e (2006), D imethy lsu l fon ioprop ionate uptake by marine phytoplankton, Science, 314, 652-654. Watson, A . J . and P.S. L i s s (1998), Ma r i ne b io log i ca l controls on c l imate v i a the carbon and sulfur geochemica l cyc les, Phil. Trans. Roy. Soc. Lon. Ser. B-Biol. Sci 353, (1365), 41-51. Wo l f e , G .V . , M . Steinke, and G .O . K i r s t (1997), Grazing-act ivated chemica l defense in a unice l lu lar mar ine alga, Nature, 387, 894-897. 10 2. Springtime Variability in Particulate DMSP Concentrations along Line P, NE Pacific Ocean1 2.1 Introduction For the past two decades, intense research efforts have focused on d imethy lsu l f ide ( D M S ) , a volat i le degradation product o f the algal metabolite d imethy lsu l foniopropionate ( D M S P ) . Th i s research has been largely mot ivated by the proposal that D M S may be invo lved in global c l imate regulat ion through its abi l i ty to stimulate c loud format ion and thereby affect planetary albedo [Charlson et al, 1987]. In contrast, interest i n its precursor, D M S P has ma in l y been in the rea lm o f a lgal ce l l phys io logy , w i th efforts directed largely at i l luminat ing ' the cel lular funct ion o f this compound [Malin and Kirst, 1997; Stefels and van Leeuwe, 1998; Stefels, 2000] . A l g a l D M S P product ion is species spec i f ic , w i t h prymnesiophytes and dinoflagellates be ing the m a i n producers [Keller et al, 1989]. In some species w i th in these classes, this molecu le can constitute a signif icant fract ion o f the ce l lu lar sulfur and carbon quotas [Matrai and Keller, 1994]. Y e t despite the h igh intracel lular concentrations o f D M S P found in many species, no clear consensus has been reached on its exact phys io log i ca l role. D M S P is l i ke l y a mult i faceted mo lecu le that has been suggested to funct ion as an osmoregulant, a cryoprotectant [as rev iewed by Malin and Kirst, 1997], an over f low mechan ism for excess ce l l energy [Stefels, 2000] , and an antioxidant [Sunda et al, 2002] . The enzymat ic breakdown o f D M S P into D M S and acry l ic ac id catalyzed b y algal DMSP- l y a se has also been suggested to funct ion as a graz ing deterrent [Wolfe et al, 1997] and most recently as an anti-viral defense mechanism [Evans et al, 2006] . ' A vers ion o f this chapter w i l l be submitted as 11 Nemcek , N . , T . D . Peterson, and P.D. Torte l l . Interannual var iab i l i t y i n spr ingt ime particulate D M S P concentrations a long a coastal to oceanic transect i n the N E Pac i f i c . Limnol. Oceanogr. In recent years, invest igat ions into the cyc l ing o f D M S P by the mic rob ia l food web have revealed that this compound is important not on ly to the phytoplankton species that produce it, but also to other components o f marine food webs [as rev iewed by Kiene et al, 2000] . D M S P and its degradation products ( D M S , D M S O ) constitute the largest poo l o f reduced organic sulfur in the oceans, and as such have been shown to be important growth substrates for many species o f marine bacteria [Kiene et al, 2000] . In fact, despite be ing present at a mi l l i on- fo ld lower concentration than sulfate, D M S P represents an energetical ly favourable fo rm o f sulfur and is the preferred substrate for heterotrophic bacteria [Kiene et al, 2000] . Thus , i n addit ion to its phys io log ica l role in phytoplankton, D M S P is emerging as an integral component o f the b iogeochemica l sul fur cyc le and may p lay an important role i n the mar ine ecosystem. A s more progress is made towards e luc idat ing the dynamics o f oceanic D M S P c y c l i ng , the ecologica l and b iogeochemica l roles o f this compound may turn out to be as s igni f icant as the c l imato log ica l role o f D M S . In an attempt to understand the cyc l ing o f D M S / D M S P and the factors that regulate their product ion, researchers have sought to correlate concentrations o f these compounds w i th various b io log ica l [Leek et al, 1990; Scarratt et al, 2002 ; Riseman and DiTullio, 2004] , chemica l [Turner et al, 1988; Leek et al, 1990; Curran et al, 1998; Riseman and DiTullio, 2004] and physical [Belviso et al, 1993; Simo and Pedros-Alio, 1999] variables w i th vary ing success. For example, D M S P p concentrations i n the f ie ld tend to correlate better w i t h the presence o f specif ic algal groups such as dinoflagel lates or prymnesiophytes as opposed to bu lk ch lorophy l l levels [Scarratt et al, 2002 ; Riseman and DiTullio, 2004] . In addi t ion, D M S P accumulat ions i n natural communit ies have been shown to co inc ide w i th h igh concentrations o f photoprotective pigments, suggesting a role o f l ight i n D M S P product ion [Belviso et al, 1993; Riseman and DiTullio, 12 2004] . M o r e recently, i ron ava i lab i l i ty has been suggested as an addi t ional factor in f luenc ing phytoplankton D M S P product ion as this compound may al leviate the ox idat ive stress associated w i th i ron l im i ta t ion [Sunda et al, 2002] . A s a result, algal D M S P concentrations in iron-l imited regions are expected to be re lat ive ly higher than i n other areas. Numerous measurements o f particulate D M S P ( D M S P p ) have been made in i ron-l imited, h igh nitrate, l ow ch lo rophy l l ( H N L C ) waters o f both the Southern Ocean [Meyerdierks et al, 1997; Curran et al, 1998] and the equatorial Pac i f i c [Hatton et al, 1998; Riseman and DiTullio, 2004], In both regions, the data indicate that D M S P p : c h l ratios are indeed higher i n iron-l imited vs. iron-replete areas [Curran et al, 1998; Riseman and DiTullio, 2004] . B y compar ison, almost no D M S P p data exist for the N E Pac i f i c , the third major i ron-l imited H N L C region. In fact, f rom the large quantity o f recently comp i l ed D M S / D M S P measurements that have been made throughout the wo r l d ' s oceans, it is evident that D M S P p data from the N E Pac i f i c are conspicuously absent [Kettle et al, 1999]. W e present the first extensive dataset o f particulate D M S P measurements obtained from waters o f the N E Pac i f i c , spanning both coastal and oceanic regimes. These measurements are further unique as they represent the first appl icat ion o f membrane inlet mass spectrometry (MEVIS) for discrete D M S P measurements. The distr ibut ion o f D M S P p is examined i n the context o f phytoplankton commun i t y compos i t ion and nutrient concentrations, i n an attempt to elucidate the b iogeochemica l regulators o f D M S P product ion in the N E Pac i f i c . 2.2 M a t e r i a l s a n d M e t h o d s Study area- The data presented herein were obtained dur ing three consecut ive spr ing cruises along L i ne P i n the eastern subarctic Pac i f i c Ocean onboard the CCGS John P. Tully. Sampl ing 13 was conducted between M a y 27-June 15, 2003 , June 1-18, 2004 and June 1-18, 2005 at the 5 major stations a long L i n e P, w h i c h connects Ocean Station Papa ( O S P or P26) to the southern B r i t i sh C o l u m b i a coast (F ig . 2.1). The L i ne P dataset represents one o f the longest running oceanographic t ime series and m u c h is now k n o w n about the seasonal dynamics o f phytoplankton b iomass and product ion i n this region [Harrison, 2002] . Stations P4 , P I 2 are considered \" coas t a l \" stations and are typ ica l l y characterized b y a diatom-dominated spring b l oom that decl ines w i t h the onset o f macronutrient l imi ta t ion [Boyd and Harrison, 1999]. Station P16 is i n the transit ion zone and stations P20 and P26 fa l l w i t h i n the H N L C boundary, where surface nitrate concentrations remain h igh year-round [Whitney et al, 1998]. A t these stations, phytoplankton growth is l imi ted by i ron avai labi l i ty , and there is l itt le seasonality i n both biomass and pr imary product ion [Boyd and Harrison, 1999]. A t a l l stations in this study, we assume that h igh macronutr ient concentrations i n surface waters are indicat ive o f i ron l imitat ion [Whitney et al, 1998]. DMSPp measurements- Part iculate D M S P concentrations were measured at 5-8 depths in the top 50 m o f the water c o l u m n dur ing each o f the three cruises w i t h sl ight modi f i ca t ions in the methodology f rom year to year. In a l l cases, seawater was col lected i n 10 L N i s k i n bottles deployed on a rosette sampler equipped w i th a Seabird 911+ C T D . Samples were drawn from N i s k i n bottles into r insed 250 m l polypropylene bottles and immedia te l y f i l tered. In 2003, duplicate 250 m l al iquots o f seawater were col lected at each depth and f i l tered under l ow vacuum (< 5 in . H g ) onto 25 m m GF/F filters (nominal pore s ize 0.7 u.m). The filters were then transferred to 25 m l A l l t e c h glass v ia ls equipped w i th gas tight 'mini-nert ' c losures. A 2 m l al iquot o f methanol was added to each v ia l and the filters were a l l owed to extract at -20\u00C2\u00B0 C for 14 several hours before 25 m l o f 1 N N a O H was added to the v ia ls . Sealed v ia ls were left at room temperature overnight pr io r to analysis to ensure complete sto ichiometr ic hydrolys is o f D M S P to D M S [Dacey and Blough, 1987]. Headspace i n the v ia ls was m i n i m a l , p rov id ing negl ig ib le losses o f D M S into the gaseous phase o w i n g to the h igh so lub i l i t y o f this gas ( K H = 0.48 m o l L\"1 arm\" 1) [De Bruyn et al, 1995]. Samples were analyzed onboard ship the f o l l ow ing day by M I M S as described be low. In 2004 and 2005 , a s ingle 250 m l sample was col lected at each depth and gravity f i ltered onto a 47 m m GF/F fi lter. The filters were placed into 5 m l cryov ia ls w i th 3 m l o f methanol. The vials were stored at -20\u00C2\u00B0 C unt i l analyzed in the laboratory w i th in 2-6 months. D M S P stored i n this manner is k n o w n to be stable for extended periods o f t ime (J. Dacey , pers. comm.) . P r ior to analysis, 2 m l al iquots o f the D M S P p extract in methanol were transferred into 14 m l glass serum v ia ls , topped w i t h 12 m l o f I N N a O H and sealed w i t h Tef lon-faced buty l l iners (Wheaton) and a lumin ium c r imp seals. The v ia ls were vortexed and a l lowed to react overnight at room temperature under m i n i m a l headspace. In a l l three years, D M S P p analysis was per formed 12-24 hours after base hydrolys is by measuring the D M S concentrat ion i n the l i qu id phase us ing M I M S w i t h a membrane inlet probe suppl ied by the manufacturer (H iden Ana l y t i ca l , U K ) . The probe consists o f a 1/16\" stainless steel cap i l la ry tube f itted w i t h a 0 .005 \" th ick d imethy ls i l i cone sleeve that is idea l ly suited for analyz ing smal l-volume, discrete samples. A l t hough this probe has re lat ive ly l ow sensit iv ity due to the smal l surface area o f the membrane, the reduced sensit iv i ty was not a factor as D M S P p samples were concentrated by f i l t rat ion, and the resultant D M S concentrations in the sample vials ranged f rom -50-500 n M . 15 D u r i n g analysis, the v ia ls were uncapped and the membrane probe was inserted direct ly into the l i qu id sample. The sample was stirred to ensure a constant f l o w across the membrane. D M S was measured by single i on moni tor ing o f the m/z 62 peak us ing the residual gas analysis ( R G A ) mode o f the mass spectrometer control software ( M A S s o f t , H i d e n Ana ly t i ca l ) . Exper iments w i t h b lank solut ions (2 m l methanol + 25 m l I N N a O H ) showed that no other ions were detected at m/z 62. The mass spectrometer was run w i t h an i on source emiss ion current o f 1000 u A in 2003 w h i c h was reduced to 250 p A in subsequent years f o l l o w i n g opt imizat ion o f the quadrupole mass f i l ter to a lower mass range. D M S was measured us ing a secondary electron mul t ip l ie r ( S E M ) set to a voltage o f 950 V w i th detector d w e l l and settle t imes o f 300 ms. Fo r each sample measurement, the signal intensity was a l lowed to stabi l ize and was then monitored for 2-5 minutes before subsequent samples were introduced. The s ignal dropped instantaneously f o l l o w i n g remova l o f the probe f rom the samples, ind icat ing no memory effects on the membrane. A l t h o u g h the v ia ls were uncapped, the s ignal for i nd i v idua l samples was stable for the entire durat ion o f analysis t ime, indicat ing m i n i m a l loss o f D M S to the atmosphere. A s a further test, some samples were left st irr ing, uncapped, w i t h the probe i n place for up to 30 minutes w i th no s igni f icant decl ine i n the D M S signal (data not shown) . The data stream f rom the M I M S was exported to a spreadsheet and the mean m/z 62 s ignal intensity for each sample was determined by averaging over a two-minute interval dur ing the stable signal plateau. In a l l three years, a ca l ibrat ion curve was generated for each set o f samples us ing fresh D M S P standards consist ing o f k n o w n al iquots o f sterile D M S P so lut ion (Research P lus Inc.) prepared in methanol and I N N a O H . Standards were prepared at the same t ime, i n the same v ia ls , and in the same vo lumes o f methanol and sod ium hydroxide as the samples to prevent matr ix effects on the 16 membrane, to ensure equivalent hydrolys is t imes, and to compensate for any smal l losses o f D M S to the headspace. Ancillary oceanographic measurements- Samples for ch lo rophy l l a (chl a) analysis were drawn from the same N i s k i n bottles as the D M S P p samples and concentrations were determined us ing a f luorometr ic method f o l l o w i n g filtration o f seawater samples onto 25 m m GF/F filters and extraction o f p igments i n 9 0 % acetone for 24 hours [Parsons et al., 1984]. Seawater nutrient concentrations were determined us ing a ship-board auto-analyzer wh i l e temperature and sal inity data used to calculate potential density (O\"T) were obtained f rom the C T D . The lower boundary o f the m i x e d layer was def ined as the depth at w h i c h the value o f o r changed by 0.02 from that o f the surface. In 2 0 0 3 , 1 4 C pr imary product iv i ty and ca lc i f i ca t ion rate measurements were made at the ch lorophy l l a m a x i m u m at each station. Fo r these determinations, 200 m l o f seawater were col lected into acid-washed polycarbonate bottles and sp iked w i t h 20-50 u C , H14CC>3 (50 m C i mmol \" 1 ) . Bott les were incubated i n an onboard P lex ig lass f low-through incubator at 30 % surface irradiance and in situ temperature for 24 hours. F o l l o w i n g incubat ion, samples were harvested onto 25 m m GF/F filters, careful ly r insed w i th 0.2 p m filtered seawater to wash away unf ixed H 1 4CC*3, and immedia te ly frozen i n sc int i l la t ion v ia ls at - 2 0 \u00C2\u00B0 C . U p o n return to the laboratory, filters were ac id i f i ed w i t h 1 m l 50 % phosphor ic ac id , capped, and p laced on a shaker table overnight. Inorganic calcite col lected on the GF/F was l iberated as CO2 b y this ac id treatment. Th i s CO2 was trapped i n phenethylamine (base)-soaked fi lters (13 m m GF/D ) stuck to the caps o f the v ia ls , wh i l e organic carbon was left behind on the pr imary GF/F . The base-soaked filters were placed in fresh v ia ls and capped, and both sets o f v ia ls ( compr is ing the inorganic and 17 organic 1 4 C fraction) were counted on a sc int i l la t ion counter after the addi t ion o f 10 m l o f sc int i l la t ion cockta i l (Scint isafe, F isher Sc ient i f ic ) . Four bottle replicates were run at each station and both product iv i ty and ca lc i f i ca t ion measurements were corrected for 1 4 C uptake in dark bottles. In 2003 , samples for phytoplankton enumeration and t axonomy were col lected f rom two depths at each station into 250 m l glass amber bottles and f i xed w i t h hexamethylene-tetramine-buffered fo rma l in to a f ina l concentrat ion o f 0.4 % . Ident i f icat ion and counts were performed on 40 or 100 m l subsamples us ing inverted mic roscopy w i t h U t e rmoh l sett l ing chambers fo l l ow ing a 24-hour settl ing per iod. C e l l vo lumes for ind iv idua l phytoplankton taxa were estimated us ing ce l l d imensions and basic geometric formulas (spheres, prolate spheres, cy l inders, etc.). C e l l carbon quotas were determined f rom vo lume estimates us ing the formulas o f Montagues et al. [1994] for f lagellates: C = 0.109 * V 0 9 9 1 (1) and Strathmann [1967] for d iatoms: log C = -0.314 + 0.712 * log V (2) where C is carbon content per ce l l i n pg , and V is ce l l vo lume in p m 3 18 2.3 Results Application of MIMS to DMSPp analysis- MEVIS proved to be an effect ive new tool for measuring particulate D M S P concentrations i n seawater. S ince D M S P p was measured direct ly f rom the l i qu id phase, analysis t ime was greatly reduced compared to that typ ica l o f purge and trap gas chromatography ( P T G C ) , w h i c h requires sample sparging. In 2003 when MEVIS was first field-tested, i nd i v idua l D M S P samples took an average o f 5 minutes to analyze, a l l ow ing a complete depth prof i l e w i t h standards to be run in duplicate i n under 2 hours. A l t hough the method was s imple and rap id , it produced h igh qual i ty data. The ca l ibrat ion curves were linear over a 40-fold concentrat ion range, and sample repl icat ion was good (mean s.d. o f duplicates = 9.5 % ) . A s a result, on l y s ingle samples were col lected at each depth i n subsequent years. In 2004, a swi tch to a thinner membrane y ie lded faster response t imes, reduc ing the analysis t ime for each sample to 2 minutes. The reduct ion i n sensit iv i ty (compared to the large membrane cuvette) result ing from the smaller surface area o f the membrane inlet probe d id not h inder our ab i l i ty to accurately measure D M S P p concentrations in any samples. A l t hough detection l imi ts were not exp l i c i t l y determined, 15 n M samples were easi ly measurable relative to b lanks. T a k i n g into account sample concentration by f i l t rat ion, y ie lded a detection l imi t o f <1 n M in situ D M S P p . Th is was we l l be low the major i ty o f concentrations encountered dur ing the surveys. Due to the smal l size o f the membrane probe, it is poss ib le to s igni f icant ly reduce the vo lume o f the sample vials used (i.e. f rom 14 m l to 3 ml ) , w h i c h w o u l d reduce the vo lume o f the in i t ia l seawater sample required and m in im ize potential filtration artefacts [as in Kiene and Slezak, 2006] . 19 Variability in DMSPp concentration - Particulate D M S P concentrations i n the upper 50 m o f the water co lumn a long L i n e P showed considerable var iab i l i ty w i t h depth, across stations, and between study years (F ig . 2.2). D u e to inclement weather, samples were not col lected at P I 6 in 2005. D M S P p levels ranged f rom a l ow o f 0.2 n M at 50 m depth at P 4 i n 2004 to a h igh o f 63.2 n M at 10 m depth at P I 2 i n 2003 , w i t h an overa l l mean concentrat ion for the 3 year survey o f 21.5 n M (n = 94, s.d. = 15.0 n M ) . In general, the coastal stations (P4, P12) had higher m a x i m u m D M S P p concentrations than the offshore stations (P16-P26), i n conjunct ion w i th higher levels o f ch lorophy l l a (F igsf 2.2, 2.3). The one notable except ion was P12 in 2005 when both the D M S P p and ch lo rophy l l concentrations were un i fo rmly low. C h l o r o p h y l l levels a long L i ne P were usual ly less than 1 p g L\"1, except for at P 4 i n 2004, where a prominent subsurface ch lorophy l l m a x i m u m o f 2.2 p g L\"1 was encountered (F ig. 2.3). B o t h the D M S P p and ch lo rophy l l a depth prof i les at P 4 and P 1 2 were characterized by subsurface m a x i m a and more vert ica l var iab i l i ty than found offshore (F igs. 2.2, 2.3). A t these stations, the D M S P p m a x i m u m occurred near the base o f the rather sha l low m i x e d layer (<20 m), wh i l e the ch lo rophy l l a m a x i m u m general ly occurred just be l ow this depth. In contrast, stations P16-P26 had deeper m i x e d layers (30-40 m) and re lat ive ly un i fo rm depth prof i les w i th very l itt le vert ica l structure i n either D M S P p or ch lo rophy l l a concentrations i n the upper 50 m o f the water c o l u m n (Figs. 2.2, 2.3). The progressive loss o f ver t ica l structure i n the prof i les mov ing offshore f r om P 4 to P26 co inc ided w i th the progressive deepening o f the m i x e d layer and l i ke l y also to a deepening o f the euphotic zone due to decreased phytoplankton biomass (as estimated by ch lo rophy l l a). Var i ab i l i t y i n D M S P p concentrations a long L i n e P between survey years was h igh (F ig. 2.2). Th is var iab i l i ty was most pronounced at P4 and P I 2 , and appeared to be dr iven at least i n 20 part by h igh var iab i l i t y i n ch lo rophy l l a levels (F ig. 2.3). The lone except ion was station P20 wh i ch had relat ively constant D M S P p and ch lorophy l l a concentrations i n a l l three years. In general, D M S P p concentrations across the survey region were highest i n 2003 and decreased s igni f icant ly i n the f o l l o w i n g years by i n many cases more than 50 % (F ig . 2.2). The mean D M S P p concentrations across a l l stations i n 2004 (14.6 n M ) and 2005 (16.6 n M ) were less than ha l f the mean concentrat ion in 2003 (33.6 n M ) . A l t hough this decrease i n D M S P p levels appeared to occur i n conjunct ion w i t h a decrease i n ch lo rophy l l a concentrations (F ig. 2.3), D M S P p x h l ratios also dropped in subsequent years, ind icat ing that the decl ine i n D M S P p levels was due to more than just a drop in overa l l phytoplankton b iomass (F ig . 2.4). D M S P p x h l ratios along L i ne P ranged f rom 0.67 n m o l ug\" 1 at 50 m at P4 i n 2004 to 103 n m o l u.g\"' at 20 m at P26 in 2003 (F ig. 2.4). The mean ratio across a l l stations i n 2003 (61.1 n m o l pg\" 1 ) was almost twice the mean ratio observed i n 2004. (33.6 n m o l pg\" 1 ) and 2005 (37.0 n m o l p.g\"'). Contour plots i l lustrat ing the interannual and spatial var iab i l i ty o f both D M S P p and ch lo rophy l l a a long L ine P are presented i n F igure 2.5. DMSPp in relation to other variables- W h i l e D M S P p concentrations var ied to some extent w i th total phytoplankton b iomass , other factors inc lud ing the phys io log ica l status and species compos i t ion o f the phytoplankton communi ty also l i ke l y affected the observed distr ibutions. A l o n g L i n e P, the depth prof i les o f D M S P p general ly resembled those o f ch lo rophy l l at ind iv idua l stations, but overa l l there was on ly a weak l inear correlat ion between the two variables for the 3-year poo led dataset (r2 = 0.20, n = 94, p<0 .0001 ; F i g . 2.6). T w o signif icant outliers characterized by h igh ch lorophy l l concentrations confounded this relat ionship (P4, 20 and 25 m depth i n 2004) . W i t h these outliers removed the pos i t i ve l inear correlat ion between 21 D M S P p and ch lo rophy l l a was s igni f icant ly strengthened (r2 = 0.45, n - 92,/?<0.0001), although a good deal o f scatter remained. To examine other factors potent ia l ly in f luenc ing D M S P p var iabi l i ty , anci l lary measurements i nc lud ing pr imary product iv i ty and ca lc i f i ca t ion rates, as w e l l as detailed phytoplankton species counts were made dur ing the 2003 survey. F o r interannual compar ison purposes, on l y D M S P p data col lected at the 5 major time-series stations a long L i ne P are presented, however , measurements were also obtained at addi t ional surrounding stations each year. In 2003 , D M S P p concentrations and support ing parameters were also measured at stations A 3 , A 4 , and A 6 (see F i g . 2.1 for station locations) and are inc luded i n the f o l l o w i n g analyses to increase the sample s ize (Table 2.1). Deta i led phytoplankton commun i t y compos i t ion data were col lected at 10 m depth in the m i x e d layer and at the ch lo rophy l l m a x i m u m at a l l stations dur ing 2003 . Va r i ous sma l l , unident i f ied flagellates were numer ica l l y dominant at the more coastal stations (P4, P I 2 , P I 6 , A 6 ; F i g . 2.7). A l t h o u g h these flagellates were st i l l abundant offshore (P20, P26 , A 3 , A 4 ) , smal l diatoms and prymnesiophytes increased in abundance at these oceanic stations (F ig. 2.7). D M S P p concentrations were not l inear ly correlated to the abundance o f any phytoplankton group. However , when phytoplankton abundance was converted to carbon b iomass us ing the appropriate convers ion factors (see methods), the relative contr ibut ion o f each group to the total communi ty changed dramat ica l ly (F ig . 2.8). Despi te contr ibut ing less than 10 % to total ce l l abundance, dinoflagel lates dominated carbon biomass at P 4 and P I 2 , and made a signif icant contr ibut ion at the other stations (F ig . 2.8). A s ignif icant, pos i t ive relat ionship was observed between D M S P p concentrations and the relative carbon biomass o f dinoflagel lates across the survey region ( r 2 = 0.46, n = \6,pO.05, Table 2.2). Sma l l , pennate diatoms also contributed 22 signi f icant ly to carbon b iomass at most stations part icular ly at the ch lo rophy l l max. at stations P20 , A 4 and A 6 (F ig . 2.8b). However , d iatom biomass was lowest at P 4 , P12 and A 3 , the 3 stations w i th the highest D M S P p concentrations result ing i n a s igni f icant inverse relationship between diatom biomass and D M S P p levels across al l stations (r2 = 0.33, n = 16 ,p<0.05, Table 2.2). G i v e n the estimates o f phytoplankton carbon b iomass , the percent ce l l carbon as D M S P in each sample was calculated us ing a C :S ratio o f 5:1 for D M S P (Table 2.3). P r imary product i v i t y rates measured at the ch lo rophy l l a m a x i m u m in 2003 were ~3 times higher at P4 than at any other station and thus appeared to be unrelated to D M S P p levels (Table 2.1). In contrast, ca lc i f i ca t ion rates expressed as a percentage o f pr imary product iv i ty rates were highest i n offshore H N L C waters (P20, P26 , A 3 and A 4 ; Table 2.1), i n accordance w i th an increased proport ion o f prymnesiophytes (Figs. 2.7b, 2.8b). There was a signif icant l inear relat ionship between relative ca lc i f i ca t ion rates and the relat ive carbon biomass o f prymnesiophytes (o f w h i c h coccol i thophores are a prominent subgroup) across the survey region (r2 = 0.69, n = S,p = 0 .01, data not shown). However , neither absolute nor relative ca lc i f i cat ion rates were correlated to D M S P p or D M S P p x h l levels. The effect o f var iable nutrient concentrations on the cross-transect and inter-cruise var iabi l i ty in D M S P p levels was examined. Depth integrated D M S P p x h l ratios are plotted against the m i x e d layer nitrate concentrat ion at each station i n F igure 2.9. A l t h o u g h there was some indicat ion that h igher D M S P p x h l ratios were associated w i t h h igh nitrate, l ow i ron waters in 2003, this trend d id not ho ld i n subsequent years (F ig. 2.9). Furthermore, var iab i l i ty in D M S P p x h l ratios between study years appeared unrelated to surface nitrate levels. 23 2.4 Discussion The stations o f the N E Pac i f i c connected by L i n e P boast one o f the longest and most comprehensive open ocean t ime series, and m u c h oceanographic data has been col lected here over the past 50 years, part icu lar ly at Ocean Station Papa (P26) [see rev iew Harrison, 2002] . Despite a l l the in format ion amassed on the dynamics o f phytoplankton product iv i ty and biomass [Booth et al, 1993; Boyd and Harrison, 1999], there have been v i r tua l l y no D M S P measurements made i n this reg ion [see Kettle et al, 1999]. P r io r to the recent publ icat ion o f D M S P p measurements obtained dur ing the S E R I E S iron-fert i l izat ion exper iment conducted near Station Papa [Levasseur et al, 2006] , the on ly D M S P data for the entire N E Pac i f i c consisted o f a single depth prof i l e taken o f f the coast o f Wash ington state [Bates et al, 1994]. Th is is surpris ing g iven the numerous measurements o f surface water D M S concentrations [ Watanabe et al, 1995; Aranami et al, 2 0 0 1 ; Wong et al, 2005] that have been made i n the area. A recently publ ished 6-year t ime series o f D M S concentrations a long L i n e P [Wong et al, 2005] conf i rms earlier observations that this reg ion hosts some o f the highest spring/summer D M S levels i n the wo r l d [Kettle etal,1999], and may thus be an important source o f this gas to the atmosphere. S ince the product ion o f D M S ul t imately depends on the algal product ion o f D M S P , measurements o f this poo l constitute the first step towards understanding the factors leading to elevated D M S levels i n the N E Pac i f i c . The data presented herein compr ise the first extensive set o f D M S P p measurements made i n this region. These data reveal that concentrations o f D M S P p exhibit s ignif icant inter-cruise and spatial var iabi l i ty , both w i t h depth and across the survey area. Spatial Variability in DMSPp levels- H i g h spatial var iab i l i ty i n D M S P p levels was observed in each survey year w i t h a 2-3 fo ld range i n m a x i m u m concentrations a long L ine P. In general, 24 D M S P p concentrations decreased offshore, though D M S P p x h l ratios showed no consistent trend between coastal and H N L C waters. A t stations P 4 and P 1 2 where prominent depth m a x i m a in both ch lorophy l l a and D M S P p were observed (Figs. 2.2, 2.3), D M S P p m a x i m a were always shal lower than ch lo rophy l l a m a x i m a . Th i s is a c o m m o n l y reported phenomenon [Turner et al, 1988; Belviso et al, 1993; Dacey et al, 1998], and is l i ke l y due to the oppos ing effects o f nutrients and l ight on the synthesis o f ch lorophy l l and D M S P . In surface waters, condit ions that promote oxidat ive stress such as h igh l ight [Sunda et al, 2002 ; Slezak and Herndl, 2003] and low nutrients [Bucciarelli and Sunda, 2003] may promote increased algal product ion o f D M S P . In contrast, deeper i n the water c o l u m n phytoplankton photoaccl imate i n response to higher nutrient and l o w l ight condi t ions by increasing their intracel lular ch lo rophy l l a concentrations [Falkowski andLaRoche, 1991]. A l t h o u g h it has been suggested that depth differences in D M S P p and ch lo rophy l l a m a x i m a may s imp ly reflect changes i n species compos i t ion w i th depth, [Dacey et al, 1998], we d id not observe any dramatic shifts in phytoplankton communi ty compos i t ion between surface and deep waters (Figs. 2.7, 2.8). In fact, the b iomass o f prominent DMSP-producers such as dinoflagel lates [Keller et al, 1999], was higher at the ch lorophy l l max. than at 10 m depth i n the m i x e d layer suggesting a phys io log ica l dr iver for the depth structure o f D M S P p levels (F ig . 2.8). DMSP in relation to other variables- Consistent w i th the results o f many previous studies [Belviso et al, 1993; Townsend and Keller, 1996; Dacey et al, 1998], on l y a weak correlation between particulate D M S P and ch lo rophy l l a concentrations was observed over the 3-year survey per iod a long L i n e P (F ig . 2.6). S ince signif icant D M S P product ion is conf ined to a few classes o f phytoplankton [Keller et al, 1989], D M S P p concentrations should be more strongly 25 correlated to the presence o f certain taxonomic groups than to bu l k ch lo rophy l l concentrations, as has been observed prev ious l y [Scarratt et al, 2002] . In 2003 , o n l y a weak l inear correlat ion was observed between D M S P p levels and bu lk ch lo rophy l l at the 8 stations surveyed (r2 = 0.32, n = 46 , p< 0.001). Instead, D M S P p was more strongly correlated to the relative carbon biomass o f dinoflagel lates (r2 = 0.46, n = 16, \u00C2\u00A3><0.05, Table 2.2), a group k n o w n for h igh D M S P quotas [Keller et al, 1989]. Interestingly, D M S P p concentrations also showed a s ignif icant, inverse relationship to the relat ive carbon biomass o f diatoms (r2 = 0.33, n = 16,/?<0.05, Table 2.2), wh i ch general ly produce l itt le D M S P [Keller et al, 1989]. In contrast, there was no l inear correlation between D M S P p concentrations and the b iomass o f prymnesiophytes (Table 2.2), another group o f prominent DMSP-producers , despite the fact that this group increased in abundance in F f N L C waters i n conjunct ion w i th elevated D M S P p : c h l ratios and higher relative ca lc i f icat ion rates (Table 2.1). Th i s is not entirely surpr is ing as i n addi t ion to taxonomic effects, D M S P product ion is in f luenced by phys icochemica l factors i n c l ud ing nutrient avai labi l i ty [Stefels and van Leeuwe, 1998; Sunda et al, 2002 ; Bucciarelli and Sunda, 2003] and both v is ib le [Belviso et al, 1993; Stefels and van Leeuwe, 1998] and U V l ight [Slezak and Herndl, 2003] . These factors vary w ide l y in the marine environment and may obfuscate attempts to l ink the algal D M S P poo l w i t h any speci f ic phytoplankton group. Contribution of DMSP to carbon biomass- Based on measured D M S P p concentrations and the estimates o f carbon b iomass determined f rom phytoplankton ce l l counts, the relative contr ibut ion o f D M S P to total autotrophic ce l l carbon was determined for each sample i n 2003 (Table 2.3). Apar t f rom at stations P 4 and P I 2, the estimates o f percent ce l l carbon as D M S P i n this study were much higher (mean 20.3 \u00C2\u00B115 %) than those reported i n prev ious f ie ld studies (1-10 % ) , 26 despite the fact that the corresponding D M S P p x h l ratios fe l l w i t h i n the range o f those in the literature [see rev iew b y Kiene et al, 2000; Simo et al, 2002] . A l t h o u g h many o f these previous studies d id not estimate carbon b iomass direct ly and s imp l y assumed a C x h l ratio o f 50 g g\" 1 [i.e. Simo et al, 2002] , it is poss ib le that some o f the C x h l ratios calculated herein are underestimates (i.e. 6.5 g g\"' at 10 m P 1 6 , Table 2.3). Carbon-to-chlorophyl l ratios can range from 20-160 g g\" 1 and vary w i th depth, latitude, season [Taylor et al, 1997], and p rox im i t y to the coast [Chang et al, 2003] . W i t h on ly a few exceptions (P4, P12) , C x h l ratios i n this study were a l l less than 30 g g\" 1 , values more typical o f nutrient-replete, l ight- l imited deep waters [Taylor et al, 1997], rather than iron-l imited surface waters. The C x h l ratio estimated for the i ron-l imited station P26 (20-30 g g\"1) is about ha l f the 50 g g\" 1 spring/summer average determined prev ious ly for this station [Booth et al, 1993]. Stations P I 6 and A 6 where the lowest C x h l ratios were measured were dominated by smal l unidenti f iable flagellates (F ig . 2.7), whose exact numbers, d imens ions and thus carbon content were prone to larger errors. In addi t ion, preserved cel ls have a tendency to shrink, leading to reduced vo lume estimates and hence underestimated carbon quotas. Furthermore w i th the microscopic method used, it was not possible to enumerate cyanobacter ia such as Synechococcus spp. wh i ch are k n o w n to be abundant at the oceanic stations a long L i n e P [Booth et al, 1993], and l i ke l y made a s igni f icant contr ibut ion to total autotrophic carbon b iomass. Th i s uncertainty in the b iomass estimates may also have confounded attempts to relate D M S P levels to the carbon biomass o f i nd i v idua l phytoplankton groups across the survey region. A p p l y i n g a constant C x h l ratio o f 50 g g\" 1 to a l l stations y ie lds a percent contr ibut ion o f D M S P to ce l l carbon o f 2.5-18.8 % (mean 8.4 % ) , more in l ine w i th other oceanic f ie ld studies [Kiene et al, 2000 ; Simo et al, 2002] . 27 Inter-cruise Variability in DMSPp along Line P- Va r i ab i l i t y i n spr ingt ime D M S P p concentrations a long L i n e P between survey years was h igh , and was most pronounced at the coastal stations P4 and P12 (Figs. 2.2, 2.5). Stations P16 and P 2 0 showed m u c h less var iabi l i ty i n D M S P p levels, wh i l e at P26 a dramatic (>3-fold) decl ine i n D M S P p levels occurred after 2003, w i th concentrations rema in ing re lat ive ly constant in the f o l l o w i n g two years (Figs. 2.2, 2.5). Ac ross the survey reg ion , D M S P p concentrations (and D M S P p :chl rations) were general ly highest in 2003. The surveys i n a l l 3 years occurred dur ing almost the same three-week per iod, however, the t im ing o f the spr ing b l o o m at the coastal stations may have di f fered leading to the h igh var iab i l i ty i n both D M S P p (F ig . 2.2) and ch lorophy l l a levels (F ig . 2.3) observed at P4 and P I 2. A s a result o f the re lat ive ly sha l low winter m i x e d layer, tight coup l ing between autotrophs and grazers, and persistent i ron-l imitat ion, stations P20-P26 exhib i t m u c h less seasonality in phytoplankton b iomass [Boyd and Harrison, 1999], such that the t im ing o f the cruises wou ld have less impact on the observed var iabi l i ty . Due to a lack o f knowledge o f the seasonal var iab i l i ty i n D M S P p concentrations in this region, it is d i f f i cu l t to speculate whether the inter-cruise differences represent true inter-annual var iabi l i ty or different phases o f the seasonal cycle. Me thodo log i ca l differences in D M S P p analysis also exist between study years, as samples were analyzed at sea i n 2003 , and several months after sample co l lec t ion i n 2004 and 2005. However , this is l i k e l y not a signif icant factor in determining inter-cruise var iabi l i ty , since samples were always measured relative to fresh standards and D M S P p is k n o w n to be stable i n methanol for months (J. Dacey , pers. comm.) . Dif ferences i n sea surface temperature greater than 2 \u00C2\u00B0 C were observed between study years at P26 , and these a long w i t h changes i n nutrient supply and ch lo rophy l l b iomass support the changes in D M S P p levels, arguing against a methodolog ica l cause o f var iabi l i ty . 28 The differences i n D M S P p x h l ratios at ind iv idua l stations between study years (F ig. 2.4), coupled w i th the weak overa l l correlat ion between D M S P p and ch lo rophy l l a (F ig . 2.6) point to a taxonomic or phys io log i ca l dr iver o f D M S P p var iabi l i ty . In 2003 , D M S P - r i c h dinoflagellates made a signif icant contr ibut ion to total autotrophic carbon at a l l stations (F ig . 2.8). A l though phytoplankton commun i t y compos i t ion data were not col lected i n subsequent years, it is possible that a decl ine i n dinoflagel lates cou ld account for the drop in D M S P p levels and D M S P p x h l ratios after 2003. Dramat i c shifts in the biomass o f dinoflagel lates and prymnesiophytes (up to 2 orders o f magnitude) have been shown to occur i n consecutive years at P26 [Wong et al, 2006] . Nutr ient condi t ions can also inf luence D M S P p var iabi l i ty . S ince D M S P and its byproducts have been shown to funct ion as power fu l antioxidants [Sunda et al, 2002] , algal D M S P product ion is expected to increase under ox idat ive stressors such as i ron-l imitat ion. In 2003, there was some evidence to this effect as D M S P p x h l ratios at i nd i v idua l stations increased in conjunct ion w i t h h igher surface nitrate levels, an indicator o f i ron-l imitat ion i n this region (F ig. 2.9). Ev idence for a pos i t ive relat ionship between D M S P p x h l ratios and nitrate concentrations also comes from H N L C waters o f the Southern Ocean [Curran et al, 1998; Jones et al, 1998]. In this latter study, the highest D M S P p concentrations co inc ided w i th l ow ch lorophy l l levels and h igh nitrate waters [Curran et al, 1998; Jones et al, 1988]. The authors d id not direct ly relate this observat ion to i ron concentrations, but i ron ava i lab i l i ty is k n o w n to l imi t phytoplankton growth i n these waters [Curran et al, 1998]. Perhaps the strongest f ie ld evidence for the effect o f i ron on D M S P product ion comes from the i ron-l imited waters o f the equatorial Pac i f i c where D M S P p x h l ratios were 2-6 higher i n i ron depleted offshore waters than in iron-rich coastal waters [Riseman and DiTullio, 2004] . 29 The pos i t ive trend observed between D M S P p x h l ratios and nitrate levels i n 2003 d id not occur i n subsequent years (F ig . 2.9). In 2004, a large decl ine i n the D M S P p x h l inventory occurred at station P 2 6 despite a relat ively constant nitrate concentrat ion. Interestingly in 2005, the spring surface nitrate concentrat ion at P26 p lummeted to about h a l f the usual 14 u M value. In a l l 3 years, the prev ious winter ' s surface nitrate concentrat ion at P 2 6 was a constant ~14 p M , indicat ing that s igni f icant nitrate d rawdown had occurred at this station i n 2005, poss ib ly as a result o f an i ron in ject ion. A l t h o u g h both ch lorophy l l (F ig . 2.3) and D M S P p levels (F ig. 2.2) showed a sl ight increase oyer 2004 levels i n conjunct ion w i t h this d rawdown , D M S P p x h l ratios remained largely unchanged f rom the previous year (F igs. 2.4, 2.9). W i thout taxonomic data for 2004 and 2005 it is d i f f i cu l t to determine whether the interannual var iab i l i ty in D M S P levels a long L i n e P was due to changes i n phytoplankton communi ty compos i t ion or phys io logy. E ven w i th this data it is d i f f i cu l t to tease apart these two factors in the f ie ld since changes i n nutrient concentrations tend to be l i nked to f lor ist ic shifts in the phytoplankton communi ty . Th i s was the case i n the equatorial Pac i f i c where an increase in D M S P p x h l ratios under i ron-l imitat ion was also associated w i t h a shift f rom diatoms to cryptophytes and prymnesiophytes [Riseman and DiTullio, 2004] . S m a l l cel ls such as flagellates that are prominent DMSP-produce r s [Keller et al., 1989] tend to dominate when nutrient concentrations are low. Whether this f lor ist ic shift occurs prec ise ly because these cel ls have a competit ive advantage under l ow nutrient condit ions not on ly because o f their greater surface area to vo lume ratio, but also because o f their D M S P product ion capabi l i t ies , has yet to be shown. 30 DMSPp variability in relation to DMS concentrations- The or ig ina l impetus for this work was to determine i f the unusua l ly h igh spring/summer D M S levels characteristic o f the H N L C region around Station Papa [Kettle et al., 1999; Wong et al, 2005] were associated w i t h h igh D M S P p concentrations. D M S levels up to 25 n M have been reported dur ing the spring/summer months at stations a long L i n e P [Wong et al, 2005] . These values are except ional for open ocean areas where D M S rarely exceeds 5 n M [Kettle et al, 1999], and are more typ ica l o f product ive, coastal regions. Despite the unusua l ly h igh D M S levels at Station Papa, the D M S P p poo l had prev ious ly not been measured i n this region. G i v e n that D M S P p is the ult imate source o f a l l D M S , its measurement prov ides important insight into the factors d r i v ing h igh D M S concentrations. Unfortunate ly , D M S concentrations were uncharacter ist ical ly l o w in June dur ing our 3-year survey per iod, as determined by M I M S in 2003 [Tortell, 2005] and independently by P T G C ( C S . W o n g , unpubl ished data). A t the oceanic stations beyond P 1 2 , D M S concentrations were less than 5 n M i n a l l three years ( C S . W o n g , unpubl ished data). Th i s trend is consistent w i th the general decl ine i n spr ing and summer D M S levels that occurred a long L i n e P f o l l ow ing the 1998-1999 transit ion f rom an E l N i n o to a L a N i n a event [Wong et al, 2005] . The >10-fold decrease in D M S levels at P26 between 1998 and 1999 occurred i n conjunct ion w i th a dramatic drop i n the b iomass o f dinoflagel lates and prymnesiophytes [Wong et al, 2006] , and thus presumably, D M S P p . The Wong et al. [2005] t ime series on l y presents data to 2 0 0 1 ; interestingly, dur ing the S E R I E S i ron fert i l izat ion experiment conducted near Station Papa in Ju ly o f 2002, D M S levels were once again h igh at >15 n M in the contro l waters outside the iron-enriched patch [Levasseur et al, 2006] . Perhaps more s igni f icant ly , corresponding D M S P p concentrations were between 98-130 n M and prymnesiophytes were very abundant [Levasseur et al, 2006] . A s noted by Levasseur et al. [2006], both the D M S and D M S P p concentrations 31 measured pr ior to S E R I E S were 2-6 times higher than those measured at the onset o f previous i ron fert i l izat ion experiments i n H N L C waters o f the equatorial Pac i f i c and Southern Ocean. Moreover , D M S and D M S P p levels i n Ju ly 2002 were at least 2-3 fo ld higher than any measured at the oceanic stations dur ing the June 2003-2005 surveys. Thus , the h igh D M S levels observed prev ious ly i n the N E Pac i f i c H N L C region appear to occur i n conjunct ion w i th h igh D M S P p concentrations. In the absence o f suff ic ient evidence l i n k i n g D M S P p var iab i l i t y to variations in nutrient supply dur ing the present study, we conclude that changes i n D M S P p concentrations are l i ke l y due to taxonomic shifts i n the phytoplankton communi ty , and dr ive the var iab i l i ty in D M S levels observed i n recent years i n this H N L C region. 32 Table 2.1: A n c i l l a r y oceanographic data for a l l stations surveyed dur ing 2003 . M i x e d layer depth ( M L D ) is def ined as a change in sigma-t (density) o f 0.02 from that o f the surface. P r imary product iv i ty (PP) and ca lc i f i ca t ion rates (Calc.) were determined at the ch l a max. at each station. D M S P p x h l ratios are 50 m depth integrated values. S T A T I O N M L D ^ o ' \" ] P P C a l c - Calc ./PP D M S P p x h l (m) (MM) (ug C L-' d-') (ug C V1 d'1) (%) (umol rug1 ni2) P4 10 0.1 31.8 0.42 1.33 58.4 \u00C2\u00B13.8 P12 25 3.3 11.6 0.29 2.48 62.9 \u00C2\u00B1 1.5 P16 40 6.1 7.3 0.01 0.16 55.4 \u00C2\u00B1 1.7 P 2 0 40 8.0 8.3 0.35 4.25 56.8 \u00C2\u00B1 1.1 P26 35 13.9 9.3 0.57 6.15 79.2 \u00C2\u00B12.4 A 3 <10 15.4 9.5 0.26 2.69 1 1 8 . 2 + 1 . 6 A 4 <10 9.9 7.5 0.31 4.12 81.8 \u00C2\u00B1 1.8 A 6 <10 3.0 10.5 0.12. 1.16 49.5 \u00C2\u00B1 0.9 Table 2.2: Coef f i c ients o f determinat ion (r2) f rom l inear regressions between D M S P p levels and the absolute and relat ive carbon b iomass o f different phytoplankton groups for a l l stations dur ing the 2003 survey; n = 16, * denotes statist ical ly s ignif icant re lat ionships w i thp < 0.05, jdenotes inverse relat ionship. Group DMSPp vs. C L\"1 DMSPp vs. % C contribution misc. flagellates 0.13 0.016 Prymnesiophytes 4 . 1 x l 0 \" 6 9 . 3 x l 0 \" 3 Dinoflagellates 0.33* 0 .46* Prasinophytes 0.070 0.044 Diatoms 8 . 0 x l 0 \" 3 0 .33* t Cryptophytes 0.12 0.048 34 Table 2.3: The contr ibut ion o f D M S P p to total phytoplankton ce l l carbon in 2003 . Phytop lankton b iovo lume was determined from mic roscop i c c e l l counts and the appropriate geometric formulas at two depths at each station. B i o v o l u m e was converted to carbon b iomass us ing the formulas o f Strathmann [1967] for diatoms and Montagues et al. [1994] for flagellates. Station depth chl a C biomass (HSL\"1) DMSPp C:chl ratio DMSPp:chl % cell C as (m) (nM) ( us\"1) (nmol jig\"1) DMSP P4 10 0.66 65.8 60.3 99.1 90.8 5.5 20 0.90 88.5 59.7 98.5 66.4 4.0 P12 10 0.76 21.3 63.2 27.9 63.2 7.9 30 0.96 22.6 56.8 23.6 56.8 6.2 P16 10 0.44 2.8 26.6 6.53 60.5 55.6 40 0.62 10.8 29.2 17.5 47.5 16.3 P20 10 0.38 9.8 22.6 26.2 60.1 13.8 50 0.43 31.2 21.2 72.1 48.9 4.1 P26 10 0.42 12.6 37.3 30.1 88.8 17.7 50 0.58 14.9 27.0 20.5 46.5 13.6 A3 10 0.39 8.2 59.1 21.1 152 43.1 25 0.51 11.3 56.7 22.1 111 30.2 A4 10 0.37 7.9 37.0 21.5 101 28.1 40 0.44 5.8 12.5 13.3 28.7 12.9 A6 10 0.59 7.1 40.2 12.0 68.0 34.0 25 0.62 6.3 33.5 10.1 54.0 32.1 Figure 2.1: Map of the N E Pacific showing the 5 major stations along Line P. Also shown are supplemental stations where measurements were made in different years (A3 A 4 A 6 in 2003-W3, W6, W8 in 2005). 36 Figure 2.2: D e p t h profi les o f D M S P p measured at the 5 major stations a long L i n e P dur ing 3 consecutive spring cruises. Er ror bars i n 2003 represent standard deviations o f dupl icate samples. F i g u r e 2 .3 : Dep th profi les o f ch lo rophy l l a measured at the 5 major stations a long L i n e P dur ing 3 consecutive spr ing cruises. Note the dif ferent scale for station P4. F i g u r e 2.4: Dep th profi les o f the D M S P p x h l ratio measured at the 5 major stations a long L i n e P dur ing 3 consecutive spring cruises. N o t e that the ratios were general ly highest in 2003. Figure 2.5: Contour plots illustrating spatial and interannual variability in springtime D M S P p and chlorophyll levels along Line P. Figure 2.6: Re la t ionship between ch lo rophy l l and D M S P p concentrations a long L i ne P for the 3 year poo led dataset (r2 = 0.20, n = 94, /?<0.0001). L inear regression shown is w i th the two outliers i n brackets exc luded (r2 = 0.45, n = 92,/?<0.0001). 41 100 80 -) \"35 O 60 H \"TO o o 40 20 0 P4 P12 P16 P20 P26 A3 A4 A6 \"o5 O 6 0 ro o P20 P26 Station V7777Z\ Diatoms f V \ ^ l Dinoflagellates mm Prymnesiophytes i i Misc. flagellates P W J Prasinophytes i i i i i n i Cryptophytes I444444I Microcystis sp. Figure 2.7: The relat ive abundance (percent o f total cel ls) o f the different phytoplankton groups enumerated at (a) 10 m depth and (b) the ch lorophy l l max. at a l l stations surveyed i n 2003. See Table 2.3 for depths o f ch lo rophy l l max . 42 c o ro O ro o K-<4\u00E2\u0080\u0094 o P4 . P12 P16 P20 P26 A3 A4 A6 P4 P12 P16 P20 P26 A3 A4 A6 Station Diatoms CV\X] Dinoflagellates Rasaaa Prymnesiophytes ' ' Misc. flagellates i w v i Prasinophytes rrrrmi Cryptophytes 11111 [ i Microcystis sp. F i g u r e 2.8: The contribution of the different phytoplankton groups to total carbon biomass at (a) 10 m depth and (b) the chlorophyll max. at all stations surveyed in 2003. See Table 2.3 for depths of chlorophyll max. 43 140 120 T 100 H E \"o E \u00C2\u00B1 Z: o cL Q_ c/3 80 P4 60 -t. 40 \u00E2\u0080\u00A2 2003 - \u00E2\u0080\u00A2 2004 o 2005 20 P4 i W8 P4 P12 \u00E2\u0080\u00A2 A6 \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 P12 o P12 P16 \u00E2\u0080\u00A2 W6 o P20 \u00E2\u0080\u00A2 P20 P20 P26 W3 P16 A4 \u00E2\u0080\u00A2 6 8 10 MLD N0 3 \" (uM) \u00E2\u0080\u0094 i \u00E2\u0080\u0094 12 A3 \u00E2\u0080\u00A2 P26 \u00E2\u0080\u00A2 P26 \u00E2\u0080\u0094i\u00E2\u0080\u0094 14 16 Figure 2.9: 50 m depth integrated D M S P p x h l ratios plotted against the m i x e d layer nitrate concentration. In 2003 , stations A 3 , A 4 and A 6 are inc luded ; i n 2005 , stations W 3 , W 6 and W 8 are inc luded (see F i g . 2.1 for locations). No te that i n 2003 , elevated D M S P p x h l ratios were found i n h igh nitrate waters, but not i n subsequent years. 44 2.5 References: Axanami , K., S. Watanabe, S. Tsunoga i , M . Hayash i , K . Furuya , and T. Nagata (2001), B iogeochemica l var iat ion i n d imethy lsu l f ide , phytoplankton pigments and heterotrophic bacterial product ion i n the Subarctic No r th Pac i f i c dur ing summer, J. Oceanogr. 57, 315-322. Bates, T.S. , R.P. K i e n e , G . V . W o l f e , P .A. Ma t ra i , F.P. Chavez , K .R . B u c k , B .W. B lomqu is t , and R.L. C u h e l (1994), The cyc l ing o f sulfur i n surface seawater o f the northeast P a c i f i c , / . Geophys. Res. 99, 7835-7843. Be l v i so , S., P. Buat-Menard , J-P. Putaud, B .C . Nguyen , H . Claustre , and J . Neveux (1993), S ize d istr ibut ion o f d imethylsul foniopropionate ( D M S P ) in areas o f the tropical northeastern A t l an t i c Ocean and the Mediterranean Sea, Mar. Chem. 44, 55-71. Booth , B.C. , J . L e w i n , and J . R. Poste l (1993), Tempora l var iat ion i n the structure o f autotrophic and heterotrophic communit ies i n the subarctic Pac i f i c , Prog. Oceanogr. 32, 57-99. B o y d , P., and P.J. Ha r r i son (1999), Phytoplankton dynamics i n the N E subarctic Pac i f i c , Deep-Sea Res. II. 46, 2405-2432. Bucc ia re l l i , E., and W . G . Sunda (2003), Influence o f C O 2 , nitrate, phosphate, and sil icate l imi ta t ion on intracel lu lar d imethylsul foniopropionate i n batch cultures o f the coastal d ia tom Thalassiosirapseudonana, Limnol. Oceanogr. 48, 2256-2265. Chang , J . , F. Sh iah, G . G o n g , and K.P . Ch i ang (2003), Cross-shel f var ia t ion i n carbon-to-ch lorophy l l a ratios i n the East C h i n a Sea, summer 1998, Deep Sea Res. II, 50, 1237-1247. Char l son , R.J., J .E . Love l o ck , M . O . Andreae, and S.G. War ren (1987), Oceanic phytoplankton, atmospheric sulfur, c l oud albedo and cl imate, Nature, 326, 655-661. Curran, M .A . J . , G .B . Jones, and H . Bu r ton (1998), Spatial d is t r ibut ion o f d imethy lsu l f ide and d imethy lsu l foniopropionate i n the Austra las ian sector o f the Southern Ocean, J. Geophys. Res. 103, 16677-16689. Dacey, J .W .H . , and N . V . B l o u g h (1987), Hyd rox ide decompos i t ion o f D M S P to fo rm D M S , Geophys. Res. Lett. 14, 1246-1249. Dacey, J .W .H . , F A . H o w s e , A . F . M i chae l s , and S.G. W a k e h a m (1998), Tempora l var iabi l i ty o f d imethy lsu l f ide and dimethylsul foniopropionate i n the Sargasso Sea, Deep-Sea Res. I 45, 2085-2104. 45 D e B ruyn , W.J. , E. Swartz , J . H . H u , J . A . Shorter, P. Dav idov i t s , D.R. Worsnop , M.S . Zahniser , and C E . K o l b (1995), Henry's law solubi l i t ies and Setchenow coeff ic ients for b iogenic reduced sul fur species obtained f rom gas-l iquid uptake measurements, J. Geophys. Res. 100, 7245-7251. Evans, C , G . M a l i n , W . H . W i l s o n , and P.S. L i s s (2006), Infectious titers o f Emiliania huxleyi virus 86 are reduced by exposure to m i l l im o l a r d imethy l sulphide and acryl ic ac id, Limnol. Oceanogr. 51(5), 2468-2471. F a l k o w s k i , P .G. and J . L a R o c h e (1991), A c c l i m a t i o n to spectral i rradiance in algae, J. Phycol, 27, 8-14. Harr i son , P.J. (2002), Stat ion Papa t ime series: insights into ecosystem dynamics , J. Oceanogr. 58, 259-264. Hatton, A . D . , S . M . Turner, G . M a l i n , and P.S. L i s s (1998), D ime thy l su lphox ide and other b iogenic sulphur compounds in the Galapagos P l ume , Deep-Sea Res. II45, 1043-1053. Jones, G.B. , M . A . J . Cur ran , H .B . Swan, R . M . Greene, F.B. G r i f f i ths , and L .A . C lementson, (1998), Inf luence o f different water masses and b io log i ca l act iv i ty on dimethylsul f ide and d imethy lsu l foniopropionate i n the subantarctic zone o f the Southern Ocean dur ing A C E l,J. Geophys. Res. 103 (D13), 16691-16701. Ke l l e r , M . D . , W . K . B e l l o w s , and R.R.L . Gu i l l a rd (1989), D i m e t h y l sul f ide product ion in marine phytoplankton, i n Biogenic sulfur in the environment, edited by E.S. Sa l tzman and W.J . Cooper , p. 167-180, A m e r i c a n Chemica l Society, Wash ington , D .C . Kett le , A . J . et a l . (1999), A g loba l database o f sea surface d imethy lsu l f ide ( D M S ) measurements and a procedure to predict sea surface D M S as a funct ion o f latitude, longitude and month , Global Biogeochem. Cycles, 13, 399-444. K i ene , R.P., L.J. L i n n , and J . A . B ru ton (2000), N e w and important roles for D M S P in marine m ic rob ia l communi t i es , J. Sea Res. 43, 209-224. K i ene , R.P., and D. S lezak (2006), L o w d isso lved D M S P concentrations i n seawater revealed by smal l-volume gravity f i l t rat ion and dialysis sampl ing , Limnol. Oceanogr.: Methods, 4, 80-95. Leek, C , U . Larsson, L.E. Bagander, S. Johansson, and S. Ha jdu (1990), D ime thy l sulf ide in the Ba l t i c Sea: annual var iab i l i ty i n relat ion to b io log i ca l act iv i ty , J. Geophys. Res. 95, 3353-3363. 46 Levasseur, M., M.G. Scarratt, S. Michaud, A. Merzouk, CS. Wong, M. Arychuk, W. Richardson, R.B. Rivkin, M. Hale, E. Wong, A. Marchetti, and H. Kiyosawa (2006), DMSP and DMS dynamics during a mesoscale iron fertilization experiment in the Northeast Pacific- Part I: temporal and vertical distributions, Deep-Sea Res. II, 53, 2353-2369. Malin, G., and G.O. Kirst (1997), Algal production of dimethyl sulfide and its atmospheric role, J. Phycol. 33, 889-896. Matrai, P.A., and M.D. Keller (1994), Total organic sulfur and dimethylsulfoniopropionate in marine phytoplankton: intracellular variations, Mar. Biol. 119, 61-68. Meyerdierks, D., B. Bolt, and G.O. Kirst (1997), Spatial and vertical distribution of dimethylsulfoniopropionate (DMSP) during spring in the Atlantic sector of the Southern Ocean, Deep-Sea Res. II44, 283-297. Montagnes, DJ.S, JA. Berges, PJ. Harrison, and F.J.R. Taylor, (1994), Estimating carbon, nitrogen, protein and chlorophyll-a from volume in marine phytoplankton, Limnol. Oceanogr. 39(5), 1044-1060. Parsons, T. R., Y. Maita, and C M . Lalli (1984), A manual of chemical and biological methods for seawater analysis, Pergamon Press, New York, N. Y. Riseman, S.F. and G.R. DiTullio (2004), Particulate dimethylsulfoniopropionate and dimethylsulfoxide in relation to iron availability and algal community structure in the Peru Upwelling System, Can. J. Fish. Aquat. Sci. 61 (5), 721-735. Scarratt, M.G., M. Levasseur, S. Michaud, G. Cantin, M. Gosselin, and S.J. de Mora (2002), Influence of phytoplankton taxonomic profile on the distribution of dimethylsulfide and dimethylsulfoniopropionate in the northwest Atlantic, Mar. Ecol. Prog. Ser. 244, 49-61. Slezak, D. and G. J. Herndl (2003), Effects of ultraviolet and visible radiation on the cellular concentrations of dimethylsulfoniopropionate (DMSP) in Emiliania huxleyii (Strain L), Mar. Ecol. Prog. Ser. 246, 61-71. Stefels, J. and M.A. van Leeuwe (1998), Effects of iron and light stress on the biochemical composition of Antarctic Phaeocystis spp. (Piymnesiophyceae). I. Intracellular DMSP concentrations, J. Phycol. 34, 486-495. Stefels, J. (2000), Physiological aspects of the production and conversion of DMSP in marine algae and higher plants, J. Sea Res. 43, 183-197. Simo, R., and C. Pedros-Alio (1999), Role of vertical mixing in controlling the oceanic production of dimethyl sulphide, Nature 402, 396-399. 47 S imo , R., S .D. A rche r , C . Pedros-A l io , L. G i l p i n , and C .E . S te l fox-Widd icombe (2002), Coup led dynamics o f d imethy lsu l foniopropionate and d imethy lsu l f ide c y c l i ng and the mic rob ia l food web i n surface waters o f the No r th At lan t i c , Limnol. Oceanogr., 47, 53-61. Strathmann, R.R. (1967), Es t imat ing the organic carbon content o f phytoplankton from ce l l vo lume or p lasma vo lume , Limnol. Oceanogr.,12, 411-418. Sunda, W. , D J . K i ebe r , R.P. K i e n e , and S. Huntsman (2002), A n ant ioxidant funct ion for D M S P and D M S i n mar ine algae, Nature, 418, 317-320. Taylor , A . H . , R J . geider, and F J . H . G i lber t (1997), Seasonal and lat i tudinal dependencies o f phytoplankton carbon-to-chlorophyl l a ratios: Results o f a mode l l i ng study, Mar. Ecol. Prog. Ser 152, 51-66. Tor te l l , P .D. (2005), D i s s o l v e d gas measurements i n oceanic waters made by membrane inlet mass spectrometry, Limnol. Oceanogr. Methods, 3, 24-31. Townsend, D.W., and M . D . K e l l e r (1996), D imethy l su l f ide ( D M S ) and d imethy lsu l foniopropionate ( D M S P ) i n relat ion to phytoplankton i n the G u l f o f Ma ine , Mar. Ecol. Prog. Ser. 137, 229-241. Turner, S . M . , G . M a l i n , P.S. L i s s , D.S. Harbour , and P . M . H o l l i g a n (1988), The seasonal var iat ion o f d imethy lsu l f ide and d imethy lsu l foniopropionate concentrations in nearshore waters, Limnol. Oceanogr. 33, 364-375. Watanabe, S., H . Y a m a m o t o , and S. Tsunogai (1995), D i m e t h y l sul f ide w ide l y vary ing in surface water o f the eastern No r th Pac i f i c , Mar. Chem. 51, 253-259. Whi tney, F A . , C S . W o n g , and P.W. B o y d (1998), Interannual var iab i l i t y i n nitrate supply to surface waters o f the Northeast Pac i f i c Ocean, M a r . Ecol. Prog. Ser. 170, 15-23. Wo l f e , G .V . , M . Steinke, and G .O . K i r s t (1997), Grazing-act ivated chemica l defense in a unice l lu lar mar ine alga, Nature, 387, 894-897. W o n g , C.S., S.E. W o n g , W A . R i chardson , G .E . Smi th , M . D . A r y c h u c k , and J .S. Page (2005), Tempora l and spatial d istr ibut ion o f d imethy lsu l f ide i n the subarctic northeast Pac i f i c Ocean: a high-nutrient-low ch lorophy l l region, Tellus 57B, 317-331. Wong , C.S. , S.E. W o n g , A . Pena, and M . Levasseur (2006), C l i m a t i c effect on D M S producers i n the N E sub-Arct ic Pac i f i c : E N S O on the upper ocean, Tellus 58B, 319-326. 48 3. High-Resolution Measurements of DMS, C 0 2 and 0 2 / A r in Productive Coastal Waters around Vancouver Island, Canada2 3.1 Introduction Rising concern about global climate change, driven mainly by high anthropogenic CO2 emissions [Denman et al, 1996], has led to increased efforts to quantify the ocean's role as a source or sink of climatologically-active gases. These include greenhouse gases such as CO2 that lead to a rise in global temperatures, as well as gases such as dimethylsulfide (DMS), which can potentially cool Earth's climate through the formation of cloud condensation nuclei that promote cloud formation and scatter incoming radiation [Charlson et al, 1987]. Decades of oceanographic gas surveys have culminated in the synthesis of thousands of measurements of /?C0 2 and DMS into monthly global climatologies for both gases [Takahashi etal, 2002 and Kettle et al, 1999, respectively]. Although the number of measurements continues to steadily increase, data from continental shelf waters remain sparse. As a result, these regions suffer from low spatial and temporal resolution and are thus poorly represented in global gas climatologies [Takahashi et al, 2002; Kettle et al, 1999]. This is a significant limitation since coastal regions, despite their small areal extent, play a disproportionately large role in air-sea gas exchange due to their high productivity and dynamic physics. Coastal waters are particularly large sources of trace biogenic gases such as DMS. Neglecting to include these areas in global climatologies may thus impart significant errors on DMS flux estimates. Moreover, because DMS has a short atmospheric lifetime [Chin and Jacob, 1996], it is especially important to identify its local sources, particularly in coastal areas where biogenic sulfur sources affect the relative importance of anthropogenic ones [Jones et al, 2001]. 2 A version of this chapter has been submitted for publication. Nemcek, N . , D. Ianson, and P.D. Tortell. A high-resolution survey of DMS, C 0 2 , and 0 2 /Ar distributions in productive coastal waters. Global Biogeochem. Cycles 49 C o m p o u n d i n g the p rob lem o f l ow measurement resolut ion is our poor understanding o f the under ly ing processes d r i v i ng the observed gas distr ibut ions. F o r C O 2 , the relative importance o f the so lub i l i t y pump versus the b io log i ca l pump i n oceanic carbon c yc l i ng needs to be evaluated [Volk andHoffert, 1985]. A l t hough b io log i ca l processes such as photosynthesis, respiration and ca lc i f i ca t ion are be l ieved to dominate the seasonal var iat ions i n pCOi over most o f the surface ocean [Takahashi et al, 2002] , at h igh latitudes temperature changes l i ke l y p lay a larger role i n d r i v ing air-sea C O 2 f luxes [Murata and Takizawa, 2003] . In our study region, strong b io log i ca l C O 2 d r awdown occurs in summer wh i l e out-gassing dominates in winter, yet net annual C O 2 f luxes are s t i l l poor l y constrained [Ianson and Alien, 2002] . The D M S cyc le is more complex than that o f C 0 2 and we are st i l l far f rom a mechanist ic understanding o f the factors d r i v ing D M S product ion [see Simo, 2004, and references therein]. A l t hough it is clear that D M S originates f rom the algal metabol i te d imethy lsu l foniopropionate ( D M S P ) , there are species-specif ic differences i n D M S P product ion [Keller et al, 1989], wh i ch i n turn are affected by the nutrient status o f the cel ls [Sunda et al, 2002 ; Bucciarelli and Sunda, 2003]. Moreover , the process by w h i c h D M S P is released f rom cel ls and converted to volat i le D M S involves the entire p lankton commun i t y ( f rom viruses to grazers) [Simo, 2001] , wh i ch i tse l f is p ro found ly affected by the phys icochemica l environment [Simo and Pedros-Alio, 1999]. A s a result, attempts to relate D M S levels to total ch lo rophy l l , phytoplankton species compos i t ion , or even D M S P concentrations have proven largely unsuccessful [Leek et al, 1990; Kettle et al, 1999]. Recent efforts have focussed on f ind ing empi r i ca l relat ionships between D M S levels and combinat ions o f relevant b io log i ca l , phys ica l , and chemica l variables that w o u l d obviate the need for a fu l l mechanist ic understanding o f the processes invo l ved . Th is approach has led to the construct ion o f a lgori thms a imed at predic t ing D M S concentrations f rom remotely 50 sensed data such as ocean co lour , and wel l-constrained, or easi ly measurable environmental parameters [i.e. Anderson et al, 2 0 0 1 ; Simo andDachs, 2002 ; Belviso et al., 2004a]. The algori thm o f Simo and Dachs [2002] (hereafter referred to as S&D2002) is part icular ly appeal ing as it has shown success at s imulat ing oceanic D M S based so le ly on the ratio between SeaW iFS surface ch lo rophy l l and a c l imato log ica l m i x e d layer depth. There are currently no comparable algorithms for use i n coastal waters as most are either based on open-ocean c l imatologies [Anderson et al., 2 0 0 1 ; Simo and Dachs, 2002] , and/or have d i f f i cu l t y s imulat ing D M S levels in diatom-dominated coastal waters [Belviso et al, 2004a] . U l t imate l y , more in format ion is needed on the spatial and seasonal distr ibutions o f oceanic gases i n conjunct ion w i th process studies a imed at e luc idat ing the factors dr i v ing the observed distr ibut ions. Th i s goa l has been hampered by l imi tat ions o f current analyt ical methods wh i ch often lack the sampl ing resolut ion to capture fine-scale var iab i l i t y and measure ind iv idua l gases i n iso lat ion. Insuff ic ient sampl ing resolut ion is part icu lar ly prob lemat ic for D M S wh i ch is at best measured a few t imes an hour w i th purge-and-trap gas chromatography. A s emphasized by Simo and Dachs [2002] , \" i t is not enough to quant i fy the annual D M S emiss ion f lux from the g lobal ocean; rather it is necessary to resolve its spatial and tempora l va r iab i l i t y . . . w h i c h exceeds our capabi l i ty for sampl ing and measur ing i n the f i e l d . \" Recent work has demonstrated that membrane inlet mass spectrometry ( M I M S ) provides the capabi l i ty to resolve spatial var iab i l i ty i n D M S and other gases [Tortell, 2005a, 2005b]. Th is method is idea l ly suited for underway surveys since a suite o f both major and trace gases can be measured s imul taneous ly at a rate o f twice per minute, p rov id ing a dramatic increase i n spatial resolut ion over current methods, part icular ly for D M S . Here in , we present the first dedicated appl icat ion o f this method to dynamic , product ive coastal waters o f f B r i t i sh C o l u m b i a , Canada. 51 The emphasis o f the present study was to examine the covar iance o f the gas and hydrographic data to relate observed D M S levels to other environmental var iables and to test the appl icabi l i ty o f the predict ive S&D2002 D M S a lgor i thm for use i n coastal regions. O u r f indings suggest that D M S concentrations i n h i gh l y product ive areas can be related to the C H L / M L D ratio suggested by these authors, a l though b y a different scal ing factor. Our results also emphasize the ut i l i ty o f multi-gas measurements and the need to sample gases at appropriate spatial scales, wh i ch are especial ly short i n coastal waters. 3.2 Materials and Methods Study Area and Sampling- D i s s o l v e d gas, temperature, sa l in i ty and ch lo rophy l l a measurements were made underway a long several transects o f f the west coast o f B r i t i sh C o l u m b i a , Canada between A u g . 11-19, 2004 onboard the CCGSJohn P. Tully (see F i g . 3.1 for transect locations). The cruise track crossed many dynamic oceanographic features i nc lud ing nearshore straits inf luenced by strong t ida l m i x i n g (e.g. T l a , T 8 , F i g . 3.1), as w e l l as open she l f areas that are sites o f upwe l l i ng and the format ion o f both cyc lon ic and ant icyc lonic eddies. The northern shel f transects ( T l b - 5 , F i g . 3.1) were made i n Queen Charlotte ( Q C ) Sound , an area that serves as the source region for ant icyc lonic (downwel l ing) eddies that carry coastal waters r i ch i n i ron offshore to the G u l f o f A l a s k a [Johnson et al., 2005] . D u r i n g summer, th is region is thought to experience re laxat ion from strong winter downwe l l i ng , w i th very l itt le actual upwe l l ing . Transect 5 crosses from this re laxat ion region into an area where sporadic summer upwe l l i ng is expected o f f Vancouve r Island. Nea r the southern end o f Vancouve r Island, transects 6-7 enter the Juan de F u c a eddy, a persistent, loca l ized summert ime feature. T h i s cyc lon i c (upwel l ing) eddy supports a h igh l y product ive , diatom-dominated commun i t y [Marchetti et al, 2004]. 52 Gas Measurements- Membrane inlet mass spectrometry ( M I M S ) was used to measure d issolved gases underway ( D M S , O2, A r , and C O 2 ) as recently descr ibed i n detai l i n Tortell [2005a]. B r ie f l y , seawater f rom the sh ip 's underway intake system (4.5 m depth) was pumped through polypropylene tubing into a sampl ing cuvette connected to the mass spectrometer. F l o w rates through the cuvette were contro l led w i th a gear pump and were kept constant at - 2 0 0 m l m in \" 1 . A gas permeable d imethy ls i l i cone membrane inside the cuvette acted as the interface between the water sample and the vacuum o f the mass spectrometer. A f t e r d i f fus ion through the membrane, gases were measured i n the vacuum chamber by s ingle i on mon i to r ing (S IM) o f the signal intensities at the relevant m/z (mass to charge) ratios (32, 40 , 44 and 62 for O2, A r , C O 2 , and D M S , respectively) . W e use a H i d e n Ana l y t i ca l H A L 2 0 tr iple f i l ter quadrupole mass spectrometer w i th an electron impact ion source set at a 500 p A emiss ion current. A l l gases were measured at a frequency o f twice per minute, or approximate ly every 160 m at the typical vessel cru is ing speed o f - 1 0 knots. The D M S signal output f rom the M I M S was cal ibrated us ing standards prepared by hydrolysis o f sterile D M S P stock solutions (Research P lus Inc.) i n 1 N N a O H . A l iquo ts o f the l iqu id D M S standard were added to 500 m l vo lumes o f DMS- f r ee seawater (>1000m depth) and were always d i luted 10 5 -10 6 -fo ld to keep the p H o f the f ina l D M S standard constant, and prevent matr ix effects on the membrane. Standards were analyzed on the M I M S by rec irculat ing the l iqu id through the sampl ing cuvette us ing a gear pump connected to a manua l sampl ing valve. The detection l im i t based on a 3:1 signal-to-noise ratio o f the b lanks was 1 n M . O n the last day o f sampl ing we exper ienced a power fai lure w h i c h caused the instrument to shut down. Th is resulted i n a ma l func t ion i n the secondary electron mul t ip l i e r ( S E M ) , the detector used to measure D M S . Thus , D M S data were not avai lable a long T8 (F ig . 3.2f). 53 Gas standards were not avai lable for C O 2 , O2 or A r . Howeve r , independent measurements o f pC02 obtained f rom an underway ER.pC0 2 equi l ibrator ( L I -COR LI-6262) were used to cal ibrate the MEVIS C O 2 s ignal . E a ch day o f sampl ing was cal ibrated separately to compensate for shifts i n the m/z 44 basel ine o f the mass spectrometer. Coef f ic ients o f determination for the pC02 ca l ibrat ion were as fo l lows : r = 0.98 for T I , r2 = 0.60 for T 2 , T 3 , and T4 , r2 = 0.87 for T 5 , r = 0.96 for T 6 , and r2 == 0.93 for T 7 and T 8 . The poor correlat ion between the MEVIS C O 2 s ignal and the equi l ibrator data for T2-4 is expected because the range o f p C 0 2 values encountered a long these transects (304-335 ppm) was sma l l , such that any noise in the MEVIS C O 2 s ignal and any offsets in the t ime stamps o f the two instruments were ampl i f ied . Howeve r , despite the poor ca l ibrat ion curve for these transects, the average absolute difference between corresponding measurements f rom the two instruments was ~5 p p m (data not shown). The oxygen and argon data were not cal ibrated; however the O2 s ignal was normal ized to A r to y ie ld a b io l og i ca l l y relevant parameter representing the balance between photosynthesis and respiration. Th i s is poss ib le because oxygen and argon have s imi la r solubi l i t ies in seawater and argon concentrations are unaffected by b io log i ca l processes [Craig and Hayward, 1987]. Thus, the ( V A r ratio is a strict measure o f b io log i ca l oxygen w i t h phys ica l processes that affect gas concentrations, such as bubble injections and temperature and sal in i ty changes, removed. DMSP measurements- Pe r iod i ca l l y dur ing the underway gas analysis, discrete samples were col lected for particulate D M S P ( D M S P p ) and d isso lved D M S P ( D M S P d ) analysis f rom the underway intake l ine. These consisted o f 250 m l al iquots o f seawater w h i c h were immediate ly gravity f i l tered onto 47 m m GF/F fi lters to separate the particulate and d isso lved D M S P fractions. The filters conta in ing D M S P p were placed i n 5 m l cryov ia ls to w h i c h 3 m l o f methanol 54 were added. These samples were stored at - 2 0 \u00C2\u00B0 C unt i l analyzed b y M I M S in the laboratory several months later (see be low) . D M S P stored i n methanol is k n o w n to be stable for extended periods o f t ime (J. Dacey , pers. comm.) . The filtrates conta in ing D M S P d were transferred to 250 m l ground-glass stoppered bottles, ac id i f ied w i t h 500 p i o f 1 2 N HC1 to prevent m i c rob i a l degradation o f D M S P d , and bubbled w i th air for 30 minutes to remove D M S . A 5 m l a l iquot o f 1 0 N N a O H was then added to the samples to hydro lyze the D M S P to D M S . The bottles were stoppered and left to react overnight under m i n i m a l headspace. P r io r to analysis on the f o l l o w i n g day, 3.6 m l o f 1 2 N HC1 were added to the samples to lower the p H to 9.5, the tolerable range for the sampl ing membrane. Samples were then pumped into the cuvette v i a the manua l samp l ing va lve and analyzed on the M I M S by single i on moni tor ing o f the m/z 62 s ignal w i t h d w e l l and settle t imes o f 300 ms. Ca l ib ra t ion standards were made by adding al iquots o f sterile D M S P to 250 m l vo lumes o f D M S / D M S P - f r e e deep seawater and were treated the same as the samples. In the laboratory, D M S P p samples were analyzed o n the M I M S us ing a membrane inlet probe spec i f i ca l l y designed for sma l l vo lume, discrete samples. Th i s probe consists o f a 1/16\" stainless steel cap i l l a ry w i t h sma l l holes at one end fitted w i t h a 0 .005 \" th ick d imethy ls i l i cone sleeve. P r io r to analysis, 2 m l al iquots o f the D M S P extract i n methanol were placed into 14ml serum via ls to w h i c h 12 m l o f 1 N N a O H was added, complete ly f i l l i n g the v ia ls . The vials were immediate ly sealed w i t h gas-tight a luminum cr imp seals w i th Tef lon-faced buty l rubber liners and left to react overnight to ensure complete convers ion o f D M S P to D M S . D M S P concentrations were measured as D M S in the l iqu id phase by M I M S by insert ing the membrane inlet probe direct ly into the v ia ls . Standards were prepared by adding al iquots o f sterile D M S P to 2 m l o f methanol and hyd ro l yz ing overnight w i t h 12 m l o f 1 N N a O H as above. 55 W e are aware that our D M S P data may be subject to recently ident i f ied f i l trat ion artefacts [Kiene and Slezak, 2006] . A l t h o u g h we used gravity f i l t rat ion, our d isso lved D M S P numbers may be overestimated and our particulate D M S P underestimated as a result o f the relat ively large vo lumes o f seawater f i l tered [Kiene and Slezak, 2006] . Howeve r , s ince the two fractions came f rom the same water sample, total D M S P concentrations ( D M S P d + D M S P p ) should be accurate. Fo r the purpose o f this study we are more interested i n the covar iance o f the D M S P and D M S data, rather than absolute D M S P values, and we use D M S P as an anc i l la ry parameter to investigate differences i n D M S levels between sites. The magnitude o f the f i l t rat ion artefacts is l i ke l y dependent on the health and species compos i t ion o f phytoplankton communi t ies , and may be smaller i n coastal upwe l l i ng systems than i n some o f the environments studied by Kiene and Slezak [2006]. Hydrographic Measurements- Surface temperature, sa l in i ty and ch lo rophy l l a f luorescence data were col lected i n con junct ion w i th the d issolved gas measurements us ing a SeaB i rd SBE-25 C T D cont inuously sampl ing from the same seawater intake system as the M I M S . D u r i n g parts o f transects 1 and 3, the SBE-25 was not logg ing data due to loss o f battery power , thus gaps exist in the sal inity and ch lo rophy l l data a long these transects (F ig . 3.2b-c). Fortunately, the temperature data were avai lable for these areas from the underway pCOi equi l ibrator. The fluorescence s ignal was cal ibrated w i t h discrete ch lorophy l l a measurements {r2= 0.97, n = 18) determined f luorometr ica l l y f o l l ow ing filtration o f seawater onto GF/F filters and extraction in 9 0 % acetone for 24 hours [Parsons et al., 1984]. A t a number o f locat ions a long the cruise track, C T D prof i les were also obtained us ing a SeaB i rd S B E 911+ C T D attached to a rosette sampler. 56 Data Binning Procedures- In order to test the S&D2002 a lgor i thm, we b inned a l l the underway D M S and ch lo rophy l l a data onto lA x lA degree surface grids and calculated an average value for each parameter for each square. F o l l o w i n g the b inn ing procedure, the data were reduced -100-fo ld to 29 ind i v idua l squares each consist ing o f a single C H L and D M S value. A corresponding m i x e d layer depth ( M L D ) was assigned to each data pair. In most cases, M L D s were determined f rom C T D density prof i les and were def ined as a 0.125 crt change f rom the surface value in accordance w i t h the cr i ter ia o f the a lgor i thm. In Q C Sound ( T l b - 5 ; F i g . 3.1) where we had good C T D coverage, l inear interpolat ion was used to assign an M L D value to each gr id based on neighbour ing C T D prof i les . D u e to the narrow range o f M L D s i n this reg ion (6-12 m) , any error introduced b y this interpolat ion was smal l . C T D data were not avai lable dur ing our cruise i n the v i c in i t y o f T l a in Q C Strait, or for T6 and T 7 o f f B a r k l e y Sound. Strong t idal forc ing and turbulent f l ows , rather than w i n d or buoyancy f luxes contro l M L D s at the head o f Q C Strait ( T l a ) , such that the upper m i x e d layer is consistently deep (40-80 m, P. C u m m i n s , pers. comm.) . W e used hor izonta l gradients in surface temperature and pCC>2 to ident i fy these t ida l ly m i x e d regions, assuming that higher p C C h and colder temperatures were indicat ive o f deeper m i x i n g . Th i s estimate y ie lded a hor izonta l gradient i n M L D s decreasing f r om 80 m at the head o f Q C Strait to 50 m towards the mouth. F o r T 6 and T7 o f f Ba rk l ey Sound , M L D s were estimated based on 26 C T D prof i les taken i n the same area three weeks later (D. M a c k a s , unpubl ished data). M i x e d layer depths dur ing the latter survey ranged f rom 8-24 m, and fe l l w i t h i n the 9 5 % C.I. o f average summert ime M L D s determined f rom several decades worth o f C T D data col lected in that area [Thomson and Fine, 2003] . G i v e n that w inds were weak to moderate pr ior to, and fo l l ow ing our occupat ion, M L D s estimated for T6 and T 7 f rom the latter survey are l i ke l y accurate. Once a m i x e d layer depth was assigned to 57 each Vi x Vi degree square, we calculated the expected D M S concentrat ion us ing the C H L / M L D ratio and the appropriate equation from S&D2002 (see results). Spatial and Statistical Analysis- W e examined the spatial var iab i l i t y o f the gas and hydrographic data us ing two different statistical approaches. Lagged autocorrelat ion funct ions were computed to estimate decorrelat ion length scales ( D L S ) for the var ious parameters as described by Murphy et al. [2001]. These funct ions compute the correlat ion o f point measurements at steadily increasing sampl ing intervals, or lags. A s the sampl ing interval increases, the probabi l i ty that two points separated by that interval are related approaches zero. The D L S thus gives an indicat ion o f the spatial length scale at w h i c h measurements become independent o f each other. To estimate the errors that cou ld result f rom undersampl ing h igh ly var iable surface data, we calculated interpolation errors f o l l o w i n g Hales and Takahashi [2004]. Essent ia l ly , this approach calculates the average error obtained by resampl ing h igh resolut ion data sets w i t h increas ingly lower resolut ion. P r inc ip le components analysis ( P C A ) was also used to examine the covariance o f the gas distr ibutions and the hydrographic measurements [Shaw, 2003] . 3.3 Results Surface Gas and Hydrographic Distributions- The surface distr ibut ions o f temperature, sal inity and ch lo rophy l l a are shown alongside pC02, 02/Ar and D M S data from the M I M S in Figures 3.2a-f, respectively. A l l parameters exhibi ted large ranges h igh l igh t ing the dynamic nature o f the study region. Surface temperatures ranged f rom 10.0-18.6 \u00C2\u00B0 C (avg. 16.2 \u00C2\u00B0 C ; F i g . 3.2a), sal inity ranged from 24.2- 32.3 psu (avg. 31.4; F i g . 3.2b), wh i l e ch lo rophy l l a concentrations ranged f rom <0.1-33.2 p g L\"1 (avg. 2.6 p g L\"1; F i g . 3.2c). 58 Gas distr ibut ions over the study region were also h igh l y var iable. The partial pressure o f C O 2 dur ing this survey ranged f rom undersaturated values as l ow as 201 p p m to h igh ly supersaturated at 747 p p m w i th an average o f 362 ppm. The surface maps show spatial associations between the d istr ibut ion o f pCOi and that o f several phys i ca l and b io log i ca l variables (F ig . 3.2a-d). A l t h o u g h C O 2 concentrations were be low atmospheric equ i l ib r ium values over most o f the study reg ion, they were strongly undersaturated in areas that had h igh ch lorophy l l concentrat ions, indicat ive o f h igh phytoplankton b iomass , and presumably a strong b io log ica l C O 2 s ink (F ig . 3.2c-d). These lowpCOi areas are evident i n F i g . 3.2d a long T l a-b i n Q C Strait, and a long T 6 and T 7 in what is l i ke l y the JdF eddy. In contrast, regions o f h igh / ? C 0 2 occurred i n JdF Strait (T8) and at the head o f Q C Strait ( T l a ) , areas in f luenced by strong t idal m i x i n g w h i c h br ings deep waters enriched in respired C O 2 to the surface. A l o n g T l a in Q C Strait we observed a dramatic transit ion f rom supersaturated to undersaturated condit ions characterized by an almost 500 p p m drop inpCOi levels (747 p p m to 255 ppm) i n the span o f 26 minutes or a distance o f 8 k m (F ig . 3.2d). Smal le r regions o f C O 2 supersaturation also occurred to a lesser extent when the cruise track crossed the shel f break i n areas o f l oca l i zed upwe l l i ng (e.g. T 5 , T 7 ; F i g . 3.2d). The phys i ca l l y induced changes inpCOi are corroborated by the corresponding low temperatures i n these regions w h i c h reflect deep waters m i x i n g to the surface (F ig . 3.2a). The average value o fpCC \u00C2\u00BB 2 (362 ppm) indicates that the survey area on the who le was near, or s l ight ly be low equ i l i b r ium w i t h respect to the atmosphere, but w i th loca l areas o f large d isequi l ibr ia . C V A r ratios expressed as a ratio o f the m/z 32/40 ion currents f rom the M I M S ranged f rom 7.9-23.5 w i t h an average o f 15.1 (F ig. 3.2e). These data are uncal ibrated and thus strictly qualitat ive; however , they do prov ide an indicat ion o f the degree o f b io log i ca l oxygen saturation 59 [Craig and Hayward, 1987]. Prev ious laboratory and f ie ld studies [i.e. Tortell, 2005b] us ing air-equil ibrated water samples have shown that C V A r ratios o f 12-15 represent an oxygen saturation o f 1 0 0 % over a large range o f temperature and sal in i ty condi t ions. The range in the measured saturation ratio represents changes i n ion source performance over mu l t ip le cruises and vary ing operating condi t ions, as opposed to temperature and sal in i ty effects on the ratio per se. Thus , although we cannot p inpoint an exact value for 100 % O2 saturation, C V A r values greater than 15 characterize supersaturated waters, wh i l e values less than 12 represent undersaturation. H i g h C V A r ratios i n our survey reg ion (>16) cp inc ided w i t h areas o f elevated ch lo rophy l l concentrations and l o w pC02 ind icat ing apparent b io log i ca l l y induced oxygen supersaturation (T I , T 5 , T 7 ; F i g . 3.2c-e). A large area o f oxygen undersaturation occurred i n JdF Strait (T8) concurrent w i t h the elevated pC02 levels o f this t ida l ly m i x e d zone (F ig . 3.2d-e). D imethy l su l f ide concentrations ranged from undetectable (<1 n M ) to 28.7 n M w i th an average o f 5.8 n M (F ig . 3.2f). Areas o f h igh D M S levels were conf ined to QC Sound and corresponded to regions o f moderate ch lorophy l l a levels (F igs. 3.2c, 3.2f). In contrast to the other gases, D M S concentrations cou ld not be related i n a general sense to the phys ica l or b io log ica l environment. Concentrat ions o f this gas were l ow i n regions o f both very h igh and low ch lorophy l l b iomass and in both w a r m and co ld waters. D i s so l v ed , particulate and total D M S P levels were unrelated to the observed D M S concentrations w i th coeff ic ients o f determination (r2) for al l parameters <0.02 (data not shown). D M S P p concentrations ranged f rom 8.7-167 n M (mean 45 n M ) w i t h the highest levels occurr ing i n areas o f h igh b iomass o f f Ba rk l ey Sound (T7) and in QC Strait ( T l a ) , where D M S levels were relat ively l ow (<10 n M ; F i g . 3.2f). In contrast, the region o f h igh D M S i n QC Sound co inc ided w i th re lat ive ly l ow D M S P p concentrations (30-60 40 n M ) . D i s so l ved D M S P concentrations ranged from 5.4-135 n M (mean 27 n M ) w i th most values fa l l i ng toward the lower end o f the spectrum. A l t hough the maps o f near-surface gas distr ibutions shown i n F i g . 3.2 prov ide a general overv iew o f the b io log i ca l and hydrographic properties o f the samp l ing area, they obscure much o f the fine-scale structure revealed by the h igh resolut ion M I M S data; F igures 3.3 and 3.4 show expanded v i ews o f two transects o f f the west coast o f Vancouve r Is land (T5 and T7 ) that h ighl ight the covar iance o f the gas and hydrographic data in greater detai l . F igure 3.3 shows an 8 hour transect (T5) where the highest D M S concentrations were encountered. Interesting phys ica l features are apparent i n this high-resolution v iew, inc lud ing a reg ion that appears to be inf luenced by loca l i zed upwe l l i ng just south o f 50.8 \u00C2\u00B0 N . Th i s area is i l lustrated by a large and sudden drop in temperature, a sl ight increase i n sal inity, a dramatic spike i n pCOi and an associated decrease i n the 02/Ar ratio. A n anomalous ly h igh leve l o f D M S P d (135 n M ) was also associated w i t h this strong temperature front (F ig 3.3a, 3.3c). M o v i n g northward from this upwe l l i ng zone, both D M S concentrations and ch lo rophy l l levels increased ( in conjunct ion w i t h increasing temperatures) unt i l the two variables became uncoup led at 51.0 \u00C2\u00B0 N (F ig. 3.3a). F rom this point , D M S concentrations cont inued to increase despite a drop i n ch lo rophy l l levels and relat ively constant D M S P p concentrations. Another sharp temperature front was encountered at the northernmost sect ion o f the transect, where a sudden increase i n temperature was associated w i th a rapid, large drop in D M S concentrations. Th is greater than 20 n M decl ine in D M S levels occurred i n under 20 minutes, over a short distance (~5.5 km) , but was represented by ~40 ind iv idua l data points (F ig . 3.3a). Th i s type o f resolut ion cou ld not have been achieved w i th other sampl ing techniques. The high-resolut ion data re inforce the observation-that D M S seems to vary independently o f the other parameters, but changes dramat ica l ly at frontal regions where ' 61 two water masses converge. In contrast, pC02 and (VAr showed a striking anti-correlation along the sampling transect, reflecting the coupling of these gases through photosynthesis and respiration (r2 = 0.89,/KO.OOOl, Fig 3.3b). Figure 3.4 shows a second, 3 hour cross-shelf transect (T7) which captured the peak in chlorophyll levels encountered during our survey. In contrast to T5 (Fig. 3.3), the high levels of chlorophyll along T7 were associated with relatively low and constant DMS concentrations despite the presence of a large gradient in DMSPp levels (45-167 nM, Fig. 3.4a). Similar to the observations of T5, however, was the poor correlation between DMS and other measured biological and physical variables. The most striking feature of this transect is the strong correlation between chlorophyll concentrations and the CVAr ratio (r2= 0.86,p<0.001, Fig. 3.4a-b). This is particularly remarkable considering that the measurements came from independent instruments measuring different parameters: chlorophyll a, a measure of phytoplankton biomass, and the CVAr ratio, a proxy for net community productivity [Kaiser et al., 2005]. The CVAr and pC02 distributions were once again anti-correlated, although not as strongly as along transect 5 (r2 = 0.64,p< 0.0001, Fig. 3.4b). Covariance of Gas and Hydrographic Data- Figure 3.5 illustrates the correlation between pairs of variables for the entire dataset. The strong anti-correlation between pC02 and CVAr that was obvious for T5 and T7 (Figs. 3.3b, 3.4b) is also apparent for the pooled dataset (r2 = 0.90, p<0.0001; Fig. 3.5a). Overlaid on this scatterplot is the chlorophyll concentration. The highest chlorophyll is associated with the highest 02/Ar values and the lowest pC02 values, while low chlorophyll occurs over a larger range of 02/Ar andpC02 values (Fig. 3.5a). A trend is also evident between 02/Ar and chlorophyll although there is considerably more scatter around this 62 relationship (r2 = 0.19,/?<0.0001; F i g . 3.5b). F r o m the temperature data over la id on this plot it is clear that the ma in deviat ions f rom l inearity occur at co ld surface temperatures that represent water masses f rom the t ida l l y m i x e d Juan de F u c a and Q C Straits. These upwe l l ed waters br ing w i th them the l ow C V A r signatures o f deep water that has been subject to oxygen loss due to respiration. Da ta f rom these areas thus appear as a negative anomaly on the O2/A1 vs. ch lorophy l l p lot (F ig . 3.5b). E x c l u s i o n o f a l l the data at temperatures be low 13.0 \u00C2\u00B0 C yields a much tighter pos i t ive relat ionship between ( V A r and ch lo rophy l l levels i n these \" a g e d \" surface waters (r2 = 0.73, p<0.0001; see discussion) . D M S is plotted against ch lo rophy l l concentration in F igure 3.6. A s i l lustrated i n F igures 3.2-3.4 and evident f rom this p lot , D M S concentrations were not correlated to ch lo rophy l l levels (r2 = 0.06; F i g . 3.6), nor to any other single variable. A l though the data i n F i g . 3.6 appear to fa l l into clusters, these clusters d id not co inc ide w i th geographical locat ion, nor were they related to temperature, sal in i ty , /7CO2 or C V A r levels. W e used pr inc ipa l components analysis ( P C A ) to examine the under ly ing associations between the variables i n our mul t id imens iona l data set. Th i s method is part icular ly useful when many variables are inter-correlated w i th each other as is the case for our gas and hydrographic measurements. U s e o f this technique y ie lded two predict ive factors ( l inear combinat ions o f the or ig ina l variables) w h i c h expla ined 7 4 % o f the total variance represented by the s ix under ly ing parameters. The results o f the analysis revealed several clear patterns in the covariance o f the data set. The most evident o f these was the dist inct separation o f /7CO2 and 02/Ar i n two-dimensional space, indicat ive o f the strong anti-correlation between these variables along al l sampl ing transects (F ig . 3.7). The second noticeable feature o f the P C A was the separation between b io log i ca l var iables (i.e. ch lorophy l l ) and phys ica l ones (temperature and salinity). The locat ion o f _ p C 0 2 and 02/Ar on the plot suggests that these parameters were inf luenced almost 63 equal ly b y phys ica l and b io log i ca l d r i v ing forces. In contrast, D M S clustered most t ight ly w i th ch lorophy l l ind icat ive o f its b io log i ca l or ig in . A Test of a Predictive DMS Algorithm- N o t surpr is ingly, none o f the correlat ive analyses we appl ied were capable o f exp la in ing the distr ibut ion o f D M S i n our survey. O f a l l the variables, ch lorophy l l c lustered most t ight ly w i t h D M S in the P C A (F ig . 3.7), and co-varied w i th D M S along parts o f some transects (F ig . 3.3a), but was a poor predictor o f D M S i n general (F ig. 3.6). W e also observed areas where D M S changed sharply w i th temperature and sal in i ty across fronts (F ig . 3.3a, 3.3c), suggest ing a direct, or indirect, phys ica l in f luence such as m i x e d layer depth, on D M S concentrations. Va r i ab i l i t y i n the M L D leads to var iable nutrient and l ight levels, wh i ch inf luence both phytop lankton and bacterial growth rates and species compos i t ions , and thereby D M S P product ion and its subsequent breakdown [Simo and Pedros-Alio, 1999]. Furthermore, surface D M S levels may be related to the thickness o f the m i x e d layer b y a s imple d i lu t ion mode l , such that D M S levels are h igh where M L D s are sha l low and v ice versa [Aranami and Tsunogai, 2004] . Simo and Dachs [2002] had successful ly used the C H L / M L D ratio as a combined b io log ica l/phys ica l predictor var iable for D M S i n their a lgor i thm. Cap i t a l i z ing on our abundant underway measurements o f both ch lorophy l l and D M S , we were able to test the appl icabi l i ty o f this predict ive a lgor i thm in the coastal waters o f our survey. F o l l o w i n g the b i nn ing procedure (see methods), we appl ied the appropriate equations to calculate the predicted D M S concentration based on the magnitude o f the C H L / M L D ratio. In a l l cases, this ratio was greater than 0.02 and we thus used equation 2 o f S&D2002 that l inearly relates D M S to the C H L / M L D ratio according to: 64 D M S = 55.8 * ( C H L / M L D ) + 0.6. (1) In al l but two cases, the S&D2002 equation overestimated the D M S concentration by a factor o f at least 2, result ing i n a poor fit to the data (F ig. 3.8, dashed l ine). However , when the actual D M S concentrations were plotted against their corresponding C H L / M L D ratios, we observed a good l inear fit between the two variables (r2 = 0.83, n = 27,/><0.0001; F i g . 3.8), after exc lud ing two s ignif icant outl iers f rom the regression. There was s t i l l no correlat ion between binned D M S and ch lo rophy l l levels, indicat ing that the l inear trend was not the result o f the b inn ing and averaging process itself. The relat ionship between D M S and C H L / M L D exists despite the fact that many o f the ch lo rophy l l values exceeded the 15 p g L\"1 cutof f o f the or ig ina l S&D2002 a lgor i thm, and m i x e d layer depths for some o f the points were estimated us ing data f rom other cruises (open symbols in F i g . 3.8; see methods). W h e n on l y data f rom Q C Sound were used (where M L D s were measured expl ic i t l y ) , the coeff ic ient o f determinat ion for the l inear relat ionship improved to r2 - 0.96 without m u c h effect on the slope (c losed symbols in F i g . 3.8). The resultant fit to our data is : Th is slope is less than ha l f that o f the or ig ina l S&D2002 formulat ion. Our results suggest that the C H L / M L D ratio may be a useful p roxy for s imulat ing D M S concentrations even in product ive coastal areas, a l though w i t h a s igni f icant ly different slope. D M S - 21.0 * ( C H L / M L D ) - 0 . 1 . (2) 65 Spatial Analyses- In addi t ion to examin ing the covariance o f our measured parameters, we capita l ized on the h igh resolut ion nature o f the dataset to quant i fy the length scales o f var iab i l i ty o f the gas and hydrographic data. W e computed lagged autocorrelat ion funct ions ( A C F ) a long ind iv idua l transects f rom w h i c h the decorrelat ion length scales ( D L S ) were def ined as the first zero crossing o f the funct ion. A d iagram o f the A C F for the s ix measured variables a long T5 is shown in F igure 3.9a. A l o n g this transect, D M S had the shortest D L S o f 10.5 k m indicat ing that its distr ibut ion var ied over the shortest distance. In contrast, the phys ica l parameters, temperature and sal inity showed longer length scales o f var iab i l i ty w i th D L S o f 23-28 k m . The str ik ing feature o f this f igure is the strong s imi lar i ty between the funct ions o f ch lo rophy l l a, pCOz and 02/Ar. A l l three had roughly the same shape and a D L S o f - 1 7 k m , fa l l i ng between that o f D M S and the phys ica l parameters. Th i s trend illustrates the tight coup l i ng between these three parameters w i t h changes i n phytoplankton biomass l i ke l y d r i v i ng the var iab i l i t y in the J9CO2 and 0 2 / A r distr ibutions. The D L S a long the six major transects analyzed ( T l and T3 were exc luded due to gaps i n the hydrographic data, see methods) ranged f rom 2.5-32.2 k m w i th mean values for the six measured parameters ranging from 7-14 k m (F ig . 3.9b). A l t h o u g h the gases appeared to vary on shorter spatial scales than the hydrographic data, the differences were not statistically signif icant (see discussion) . A n addit ional analysis was used to estimate the errors that cou ld result f rom lower frequency sampl ing . F o l l o w i n g the work o f Hales and Takahashi [2004], we resampled our h igh frequency data at ever coarser resolut ion, and calculated the average error result ing f rom linear interpolations between the coarsely sampled data. A s the samp l ing frequency decreases, the interpolat ion error increases to an asymptotic value. B e yond this point , coarser sampl ing has 66 l itt le to no effect on the magnitude o f the interpolat ion error [see Hales and Takahashi, 2004, F i g . 19]. Table 3.1 summar izes the asymptotic interpolat ion errors calculated for each o f the six parameters a long the s ix major transects surveyed. Fronta l regions ev ident ly had signif icant effects on the magnitude o f the asymptotic interpolat ion error for a l l parameters. T 5 crossed a loca l upwe l l i ng zone and was characterized by two sharp temperature fronts (F ig . 3.2a, F i g . 3.3c). Consequent ly , a long this transect the interpolat ion errors for most parameters were large, part icular ly for D M S w h i c h changed dramat ica l ly at frontal zones (F ig . 3.3a). The asymptotic interpolat ion error for D M S in this case was almost 100 % o f the 8.6 n M m e a n concentration a long this transect (Table 3.1). In contrast, T 7 and T 8 were character ized by large, sharp gradients in ch lo rophy l l concentrations (Figs. 3.2c, 3.4a). These large changes i n phytoplankton biomass drove h igh var iab i l i t y i n the pC02 and 02/Ar levels w h i c h coup led w i th the m i x i n g o f deep waters to the surface a long other parts o f these transects resulted i n large ranges i n the PCO2 (~450 p p m a long T8 ) , and 02/Ar measurements. A s a result, the absolute errors for ch lorophy l l , pC02 and 02/Ar were largest a long transects 7 and 8 (Table 3.1). 3.4 D i s cuss i on Spatial Scales of Variability- B i ogeochemica l var iab i l i ty i n this coastal zone was observed both in the large range o f values measured and in the short distances over w h i c h gas and surface water hydrography var ied. B o t h the autocorrelation functions and the interpolat ion error analyses showed that, on average, the major i ty o f the var iab i l i ty i n the gas, temperature, sal inity and ch lorophy l l distr ibut ions occurred on spatial scales o f less than 20 k m (F ig . 3.9b, Table 3.1), consistent w i t h the expected Rossby radius (the correlat ion length for phys i ca l properties) in this 67 region. A l t hough the mean D L S seemed to be shorter for the gases (<10 km) , than for the hydrographic parameters (12-14 km) , the differences were not stat ist ical ly s ignif icant (F ig. 3.9b) ow ing to large var iab i l i t y i n the D L S . However , m u c h o f this var iab i l i t y appears to result f rom an apparent artefact o f the analysis i n w h i c h the D L S for any g iven parameter increases w i th increasing transect length. Thus , part o f the var iab i l i t y i n the mean D L S for each parameter is due to var iab i l i ty i n the transect length, mak ing compar ison o f D L S for a g iven parameter d i f f icu l t between transects and even between studies [i.e. Murphy et al, 2 0 0 1 ; Hales and Takahashi, 2004 ; Tortell, 2005b] . Nonetheless, differences i n D L S between parameters along any single transect are mean ingfu l ( F ig . 3.9a), and the overa l l mean values are probably representative o f inherent differences between these parameters (F ig . 3.9b). Thus it appears that the gases do vary on shorter length scales than temperature, sa l in i ty and ch lo rophy l l , w i th D M S poss ib ly exh ib i t ing the shortest D L S as suggested by prev ious data [Tortell, 2005b] . M o r e robust statistical approaches may be needed to quantify the relevant spatial scales o f var iab i l i ty o f our dataset, but the results c lear ly indicate that gases exhibi t var iab i l i t y over distances that are not suff ic ient ly resolved i n many field studies. Asympto t i c interpolat ion errors were calculated to quant i fy the magnitude o f the errors result ing f rom l o w resolut ion sampl ing. Our calculat ions show that for D M S and ch lorophy l l in particular, insuff ic ient sampl ing resolut ion can introduce errors approaching 100 % o f the mean concentration. Large errors (-20 % ) are also associated w i th pCOi and 0 2 / A r measurements (Table 3.1). A l t h o u g h the relative error for pCOi and 0 2 / A r is smal ler than that for D M S or ch lorophy l l a, a 46 p p m average absolute error i n the mean p C 0 2 concentrat ion w o u l d be h igh ly signif icant i f accurate estimates o f regional air-sea f luxes were required, and cou ld even change the direct ion o f the f lux . The same is true i n the case o f 0 2 /Ar , where a unit change in the ratio 68 can represent a 5-10 % change i n the saturation state o f O2. Th i s w o u l d have large impl icat ions for estimates o f net commun i t y product iv i ty from ( V A r ratios [as i n Kaiser et al, 2005]. W h i l e the asymptot ic interpolat ion errors represent m a x i m u m uncertainty associated w i th l ow resolut ion sampl ing , e.g. hydrostations w i th separations o f 20-60 k m , s ignif icant errors can also occur dur ing underway sampl ing at insuff ic ient resolut ion. A more meaningfu l appl icat ion o f the analysis is to calculate actual sampl ing errors for underway measurements w i th a given sampl ing frequency. W e calculated interpolat ion errors for our high-resolut ion pC02 and D M S measurements resampled at the frequency o f our pC02 equi l ibrator (5 min . ) , and a typical underway D M S sampl ing frequency o f 30 minutes. Fo r the pC02 equi l ibrator, the interpolation errors a long ind i v idua l transects ranged from 1.4-8.3 p p m , equivalent to a m a x i m u m error o f 1.5 % relative to the mean. Th i s re lat ively smal l error w o u l d not be part icu lar ly s ignif icant for most b iogeochemica l studies, and it thus appears that the sampl ing resolut ion o f the pC02 equil ibrator is suff ic ient to capture nearly a l l o f the under ly ing var iabi l i ty . In contrast, a 30 minute sampl ing reg ime for D M S introduced errors o f between 0.7-2.8 n M , or as h igh as 41 % o f the mean a long a g iven transect. Th is error i n the mean D M S concentrat ion translates to an equivalent error i n the D M S f lux , an estimate already prone to large errors due to the uncertainty in the gas transfer coeff ic ient [Nightingale et al, 2000] . pCC>2 and 02/Ar Distributions- The distr ibutions oi~pC02 and 0 2 /Ar dur ing this survey exhibited considerable patchiness w i t h regions o f strong undersaturation and supersaturation in close p rox im i t y to each other (e.g. 500 p p m change in/7CO2 over 8 k m , T l a , F i g . 3.2d). The range o f /7CO2 concentrations encountered (201-747 ppm) was s imi lar to other summert ime values reported for the southwest coast o f Vancouver Island [Ianson et al, 2003] , and the Oregon 69 coastal upwe l l i ng system [Hales et al, 2005] . In a l l three surveys, regions o f intense C O 2 supersaturation were conf ined to narrow nearshore strips w i th the major i ty o f the outer shel f areas undersaturated as a result o f b io log i ca l d rawdown. The large regions o f C O 2 undersaturation found a long the she l f (T6, T7) and in Q C Sound (T lb-5) l i ke l y compensated for the loca l l y intense C O 2 sources ident i f ied i n the straits ( T l a , T8 ) . W e suggest that this region is st i l l an overa l l C O 2 s ink dur ing the t ime o f the survey, a l though s ign i f i cant ly smal ler due to the consistent t idal sources. It is important to recognize that these t idal ly-inf luenced regions o f persistently h igh C O 2 must have a large impact on net annual carbon budgets, whereas loca l ized upwe l l ing a long the she l f leads to transiently h igh C O 2 concentrations that are usual ly qu i ck l y drawn d o w n by b i o l og i c a l product ion. A l t h o u g h there is s t i l l considerable debate as to whether coastal upwe l l i ng zones are net sources or s inks o f C 0 2 over an annual cyc le [Ianson and Allen, 2002 ; Hales et al, 2005 ] , Hales et al. [2005] suggest that C O 2 uptake i n the upwe l l i ng margins a long the west coast o f N o r t h A m e r i c a is equivalent to 5 0 % o f the entire summert ime, oceanic Nor th Pac i f i c C O 2 s ink. These f indings c lear ly demonstrate the disproport ionately large inf luence o f coastal margins i n the oceanic carbon cycle and the need to incorporate these areas into global C O 2 c l imato logies . There is m u c h current interest i n ident i fy ing the relative importance o f b io log ica l processes versus phys i ca l ones i n d r i v ing the distr ibut ion (and hence air-sea f lux) o f C O 2 [Sarmiento et al, 2000] . Our results demonstrate that var iab i l i ty i n surface pCOi was largely inf luenced by the oppos ing b io log i ca l processes o f photosynthesis and respirat ion i n this product ive, coastal marg in . Th i s is evident f rom the tight anti-correlation between pCOi and O2/AX (a str ict ly b i o l og i ca l parameter) a long a l l sampl ing transects even i n the finest resolut ion (Figs. 3.3b, 3.4b, and 3.5a), and the occurrence o f l o w pC02 levels i n regions o f h igh biomass 70 along certain transects (e.g. T i b , T 7 , F i g . 3.2c-d). However , it was the phys ics o f this region (upwel l ing, t idal m ix i ng ) that drove the large-scale pC02 d istr ibut ions and accounted for the large range o f pC02 levels b y p rov id ing a mechan ism for deep waters enriched i n reminera l ized CO2 to reach the surface. A tmospher i c exchange also affects the distr ibut ion o f pC02 i n surface waters although its effects are not as pronounced as those o f m i x i n g and b io log i ca l d rawdown. Th i s is due in part to the s low rate o f CO2 exchange w h i c h occurs on timescales o f days to weeks and is about 10 times s lower than the rate o f O2 exchange [Broecker and Peng, 1982]. These differentia] gas exchange rates may exp la in the divergent slopes o f the pC02 vs. 02/Ar relat ionship on the left side o f the plot i n F i g . 3.5a. Here , /?C02 concentrations o f - 2 0 0 p p m are associated w i th two very different levels o f ch lo rophy l l and 02/Ar ratios. A t the top o f the p lot , the highest ch lorophy l l levels (red area, F i g . 3.5a) correspond to the highest 02IAx ratios and lowest pC02 values measured, yet lower d o w n on the y-axis, the same pC02 levels are associated w i th much lower ch lo rophy l l and 02/Ar levels (F ig . 3.5a). Thus it appears that the O2 concentrations in these latter waters were able to re-equilibrate to the surrounding b iomass m u c h faster than CO2 levels. Due to the longer \" h i s t o r y \" o f the pC02 signature, a strong anti-correlation between pC02 levels and ch lo rophy l l concentrations was on ly evident i n areas where b iomass was h igh , presumably where product iv i ty rates were at their peak (F ig . 3.4a-b). The 02/Ar ratio has been used as a p roxy for the net commun i t y product ion o f a water mass integrated over a t ime scale that depends on the gas transfer ve loc i ty and the depth o f the m ixed layer [Kaiser et al, 2005] . W e observed a strong l inear relat ionship between 02/Ar and ch lorophy l l levels in surface waters indicat ing increasing pr imary product iv i ty w i th increasing phytoplankton b iomass (F ig . 3.5b). However , several factors compl ica te the relat ionship between 71 chlorophyll and 02/Ar. On the one hand, deep mixing brings up cold subsurface waters which carry the low 02 /Ar values characteristic of oxygen loss due to respiration. This creates an initial oxygen deficit in surface waters which persists despite the growth of phytoplankton and the addition of photosynthetically derived oxygen. This scenario may explain the negative anomaly in the 02/Ar vs. chlorophyll relationship (blue regions in Fig . 3.5b), and highlights the difficulty of estimating net community productivity from 02/Ar ratios in regions with significant upwelling or vertical mixing [Kaiser et al, 2005]. On the other hand, high 02 /Ar ratios may persist in the mixed layer despite the presence of low surface chlorophyll levels (red regions in Fig. 3.5b). These waters may have warmed and stratified to the extent o f cutting off the supply o f nutrients from below the mixed layer, thus either creating a subsurface chlorophyll maximum, or causing the phytoplankton to be removed from the surface before the O2/AJ productivity signature could be reset by atmospheric exchange. Thus, knowledge of the relevant timescales of both gas exchange and phytoplankton turnover rates is critical when inferring production rates from gas distributions. Factors Driving DMS Distributions- The range of D M S concentrations encountered during our survey was large and variable, from undetectable (<1 nM) to almost 30 n M . The upper end of this range is at least an order of magnitude higher than current estimates o f the global mean D M S concentration [Belviso et al, 2004b], and comparable to data obtained in various other coastal regions [Leek etal, 1990; Townsend and Keller, 1996; Locarnini et al, 1998; Tortell, 2005b]. However, our measurements provide much higher spatial resolution than previous surveys and thus offer new insight into the small-scale patchiness of D M S distributions in coastal waters (Fig.3.2f). 72 W e compared our D M S data to measurements taken in the same general area (coastal B r i t i sh C o l u m b i a between 48-55\u00C2\u00B0 N ) , dur ing the same week, the prev ious year by automated purge-and-trap gas chromatography sampl ing at a frequency o f once every 30 minutes (J.E. Johnson, unpubl i shed data avai lable at http://saRa.pmel.noaa.gov/dms/'). The range o f D M S concentrations encountered i n 2003 (0.7-17.6 n M ) was smal ler than that observed dur ing our 2004 survey (<1- 28.7 n M ) , but the mean D M S concentrations were remarkably s imi lar (5.3 n M vs. our value o f 5.8 n M ) . It is possible that the range o f concentrations dur ing 2003 was lower because Johnson's survey may have been unable to resolve the true var iab i l i t y o f the D M S distr ibutions (as indicated b y our 30 m i n . interpolat ion error analysis). A s observed prev ious ly [Hales and Takahashi, 2004 ; Tortell, 2005b] , we note that the measured range o f a g iven parameter is m u c h more prone to sampl ing errors than the regional mean. It is thus not surpris ing that the overa l l mean o f Johnson 's survey was s imi lar to that o f the present survey, despite a much smal ler range o f reported D M S concentrations. However , many environmental factors could account for the interannual var iab i l i ty i n D M S concentrations. Var ia t ions i n w i n d ve loc i ty could be responsible for the difference i n m a x i m u m D M S levels between the two years. W inds were very l ight dur ing our entire cruise (rarely exceeding 5 m s\"1), compared to those reported in 2003 (max 13 m s \" 1 , mean 5 . 1 m s\"1). A s a result, a larger gas transfer ve loc i ty i n 2003 cou ld have created a larger D M S loss term i n surface waters, act ing against the accumulat ion o f h igh D M S concentrations i n the m i x e d layer. It should be noted, however , that vent i lat ion to the atmosphere is on l y one s ink for D M S and often on ly a m ino r one [Kiene and Bates, 1990]. Photolys is [Kieber et al, 1996; Toole et al, 2004] and bacterial consumpt ion [Kiene and Bates, 1990] can be m u c h larger s inks for D M S , and these a long w i th under ly ing differences in phytoplankton species compos i t ion and b iomass, zoop lankton graz ing rates, nutrient supply and 73 l ight intensity cou ld a l l account for interannual var iab i l i ty i n D M S concentrations. Despite these potential differences, we observed a str ik ing s imi lar i ty i n the spatial d is t r ibut ion o f D M S between the two years. B o t h the 2003 survey and our 2004 survey reported peak D M S levels in the region north o f Vancouve r Island between 51-52 \u00C2\u00B0 N . The area o f summer relaxat ion over the broad shel f i n Q C Sound may thus be a region o f persistently h igh D M S dur ing August . The h igh D M S concentrations (>10 n M ) observed i n Q C Sound dur ing our survey were associated w i t h moderate levels o f ch lorophy l l ( T i b , T 4 , T 5 ; F igs . 3.2c, 3.2f). In contrast, D M S concentrations were l o w in areas where ch lo rophy l l levels exceeded 15 p g L\"1 ( T l a , T 7 ; F igs . 3.2c, 3.2f). S ince D M S P product ion by phytoplankton is k n o w n to be h igh l y species-specific [Keller et al, 1989], var iab i l i t y in phytoplankton commun i t y compos i t ion may account for differences in D M S levels between regions. Th is taxonomic effect is potent ia l ly confounded, however, b y d i f fe r ing env i ronmenta l condit ions across our sampl ing region. S ince D M S P and its byproducts are power fu l ant ioxidants, cel ls increase their D M S P (and D M S ) product ion when exposed to ox idat ive stressors such as l ow nutrients and h igh U V l ight [Sunda et al, 2002] . These condit ions general ly occur i n later stage b looms when surface waters have stratif ied, cutting o f f the supply o f nutrients f rom be low the thermocl ine and expos ing cel ls to higher levels o f U V light. Strat i f ied waters tend to favour flagellate groups such as prymnesiophytes that have adapted to l i v i n g under l ow nutrient, h igh l ight condit ions [Margalef, 1978] and are perhaps non-coinc identa l ly the same groups that are prominent DMSP-produce rs [Keller et al, 1989]. Furthermore, nutrient l imi ta t ion can induce D M S P product ion i n groups such as diatoms [Sunda et al, 2002 ; Bucciarelli and Sunda, 2003] , that have tradi t ional ly been considered low D M S P -producers [Keller et al, 1989]. The h igh D M S concentrations i n Q C Sound occurred where m ixed layer depths were less than 12 m and surface nitrate was depleted, suggesting an oxidat ive 74 stress effect on D M S P / D M S product ion. In contrast, the waters i n QC Strait and in the JdF eddy ( T l a , T7) were rece iv ing a steady nutrient supply through loca l i zed upwe l l i ng or deep m i x i n g result ing i n h igh (or measurable) surface nitrate and h igh phytoplankton b iomass , w i th little D M S product ion. The occurrence o f l o w D M S levels i n recently upwe l l ed waters has been observed prev ious ly [Belviso et ai, 2003] . However , since changes i n phytoplankton species compos i t ion general ly occur i n conjunct ion w i th changing envi ronmenta l condi t ions, it is d i f f icu l t to d is t inguish between species compos i t ion effects and nutrient/light effects in determining D M S levels. A s i n prev ious studies [Locarnini et al., 1998; Tortell, 2005b] , D M S concentrations dur ing this survey changed dramat ica l ly at fronts, regions where abrupt gradients in nutrient concentrations, l ight regimes, product iv i ty and p lankton commun i t y compos i t ion are common . This trend reflects the comp lex interplay o f physics and b i o l ogy that characterizes the oceanic D M S cycle [see Simo, 2004] . The anomalous ly h igh leve l o f D M S P d (135 n M ) measured o f f the northwest coast o f Vancouve r Is land (F ig . 3.3a) also occurred at a sharp temperature front. The except ional ly h igh abundance o f zoop lankton encountered at this site (R. E l-Sabaawi , unpubl ished data) may have been responsible for massive grazer-mediated release o f D M S P [Dacey and Wakeham, 1986]. Towards Global DMS Prediction- Intense efforts have been a imed at understanding the factors cont ro l l ing D M S product ion i n the oceans, and quant i fy ing its atmospheric f l ux i n order to evaluate the feas ib i l i ty o f the hypothesized bio logica l ly-mediated homeostasis ( C L A W hypothesis) [Charlson et al., 1987]. It has proved d i f f i cu l t to project future oceanic D M S emissions i n a changing c l imate due to an incomplete mechanist ic understanding o f the D M S 75 cycle and uncertainties i n the g loba l and seasonal distr ibutions o f this gas. These uncertainties coupled w i th uncertainties i n the gas transfer coeff ic ient [Nightingale et al, 2000] hamper the abi l i ty o f atmospheric sul fur mode ls to evaluate the modu la t ing effect o f this gas on g lobal cl imate. T w o complementary approaches are needed to better constrain the g lobal distr ibut ion o f D M S . Accurate , spat ia l ly resolved g lobal D M S measurements w i t h good seasonal coverage are the first step, and M I M S can greatly facil itate this endeavour. E v e n w i th an automated M I M S system, it w o u l d s t i l l be unfeasible to map the entire ocean at suff ic ient temporal resolut ion. Hence, the second approach invo lves deve loping predict ive a lgor i thms that simulate D M S levels based on w e l l constrained b iogeochemica l parameters. A recent compar i son o f several such algorithms has shown that they have various strengths and weaknesses and di f fer in their abi l i ty to accurately repl icate D M S levels both seasonally and reg iona l ly [Belviso et al, 2004b]. W e chose to evaluate the a lgor i thm o f Simo and Dachs [2002] w i t h our coastal dataset because it is one o f the few without a modeled term that relies on two c o m m o n l y measured parameters ( ch lorophy l l and M L D ) . A s noted above, the m i x e d layer depth encompasses a number o f factors such as nutrient ava i labi l i ty , l ight ava i lab i l i ty and phytoplankton succession, (which appeared to exp la in many o f the regional differences i n observed D M S levels), into a single variable. A l t h o u g h the or ig ina l Simo and Dachs [2002] a lgor i thm fa i led to re-create the D M S levels found dur ing our survey, w i th a different slope the C H L / M L D ratio was a good predictor o f surface D M S concentrations (r2 = 0 .83; F i g . 3.8). Th i s is part icular ly impressive consider ing the complete lack o f a relat ionship between D M S and ch lo rophy l l alone (r2 - 0.06; F i g . 3.6). 76 Our slope (21.0) is less than ha l f that o f the or ig ina l a lgor i thm (55.8) w h i c h may result from the higher relative proport ion o f diatoms (with lower D M S P content per unit ch lorophyl l ) [Keller et al, 1989] i n coastal regions compared to oceanic waters. O u r coastal data encompassed a m u c h larger range o f C H L / M L D ratios (0.03-2.26) than or ig ina l l y used to formulate the a lgor i thm (<0.20), and we observed a good fit to the D M S data up to a C H L / M L D ratio o f 1.0. The two prominent outl iers that were exc luded from the regression had C H L / M L D ratios greater than 1.0 and came from the region o f h igh b iomass o f f B a rk l e y Sound. A l t hough we d id not exp l i c i t l y measure M L D s outside the area o f Q C Sound , we chose to incorporate the whole survey reg ion into the a lgor i thm to expand the ranges o f M L D , D M S , and ch lorophy l l concentrations used. Thus we re l ied on previous knowledge o f the area, and data f rom other cruises to estimate M L D s outside o f Q C Sound (open symbols , F i g . 3.8). A s a result, there is potential error i n the M L D s for the two outl iers from Ba rk l e y Sound . Howeve r , based on the measured D M S and ch lo rophy l l concentrations, M L D s for the two outl iers w o u l d have had to have been unreasonably deep (75-100 m) i n order to fit the curve. It is poss ib le that the linear relationship s imp l y does not ho ld beyond C H L / M L D ratios greater than 1.0 w h i c h wou ld characterize rare regions o f very h igh biomass and re lat ive ly sha l low m i x e d layers. The area o f f Ba rk ley Sound where these h igh C H L / M L D ratios were observed was l i ke l y part o f the Juan de Fuca eddy, where exceed ing ly h igh phytoplankton b iomass is sustained throughout the summer through the constant in ject ion o f nutrients into the eddy core [Marchetti et al, 2004] . The S&D2002 a lgor i thm uses both a l inear and a negative logar i thmic equation to mode l D M S concentrations based on the magnitude o f the C H L / M L D ratio. O u r results indicate a l inear equation works i n the C H L / M L D range o f 0.02-1.0 but not beyond. E i ther no relat ionship exists between D M S and C H L / M L D beyond C H L / M L D > 1 or a different equation applies. M o r e data 77 in product ive regions where these h igh C H L / M L D ratios are l i ke l y to be found is needed to test this poss ib i l i ty . 3.5 C o n c l u s i o n s Our f indings extend the ut i l i ty o f the C H L / M L D ratio as a predictor o f D M S levels into h igh ly product ive coastal waters. App l i c a t i on o f s imi la r analyses i n var ious coastal waters is needed to test the general appl icab i l i t y o f our der ived a lgor i thm. Th i s w o u l d a id i n determining whether the relat ionship is speci f ic to this part icular reg ion, or season, or more broadly appl icable to dist inct coastal systems (i.e. temperate). O u r results also emphasize the ut i l i ty o f membrane inlet mass spectrometry for reso lv ing spatial var iab i l i t y i n gas distr ibut ions, and the need for high-frequency samp l ing i n coastal waters i f the a im is to accurately quant i fy f luxes o f both major and trace gases i n these important regions. The appl icat ion o f M I M S (or other comparable analyt ical tools) is thus l i ke l y to s igni f icant ly increase our understanding o f oceanic gas distr ibutions over the c o m i n g decade. A s more in format ion becomes avai lable on the concentration and var iab i l i t y o f gases i n dynamic coastal regions, new insight may be gained into the b io log i ca l and phys i ca l controls on gas distr ibutions. In the case o f the D M S cycle, a better mechanist ic understanding is required o f the under ly ing product ion and consumpt ion processes, and this may be obtained by coup l ing focused process studies w i t h real-time underway surveys. U l t imate ly , this in format ion w i l l a id i n the development o f better predict ive algorithms wh i ch are needed to understand the potent ia l impact o f D M S emissions on future c l imate, and conversely, the impact o f c l imate change on the oceanic cycle o f D M S . 78 Table 3.1: Abso lu te and relative asymptotic interpolat ion errors a long the 6 major transects Temp. (\u00C2\u00B0C) Sal. (psu) Chl a (ug L\"1) 02/Ar (torr ratio) />C02(ppm) DMS (nM) Transect *abs. error frel. error (%) abs. error rel. error (%) abs. error rel. error (%) abs. error rel. error (%) abs. error rel. error (%) abs. error rel. error (%) 2 0.45 2.7 0.06 0.19 0.15 22 0.15 1.0 12 3.7 1.25 48 4 0.28 1.6 0.17 0.54 0.80 65 0.28 1.8 8 2.5 6.0 90 5 2.25 15 0.23 0.71 3.50 97 1.30 8.2 45 13.9 8.0 93 6 0.18 1.1 0.15 0.47 0.80 97 0.70 4.3 30 9.9 1.5 52 7 0.90 6.1 0.15 0.47 7.50 62 2.00 11.5 55 18.0 1.0 24 8 0.35 3.1 0.10 0.32 2.75 55 1.75 UJ 125 23.4 \u00E2\u0080\u0094 ~ mean 0.74 5.0 0.14 0.45 2.6 66 1.0 6.9 46 12 3.6 61 * absolute asymptot ic interpolat ion error represents the m a x i m u m error i n the mean ( in the respective units o f each parameter), obta ined when the actual data a long each transect is re-sampled at ever coarser reso lut ion. The asymptotic error was reached at an average samp l ing distance o f 34 k m . The characteristic error length scale (analogous to the D L S ) is computed as 2/3 the distance to the asymptote (-20 km) [Hales and Takahashi, 2004] . t relative asymptotic interpolat ion error is the percentage o f the mean value a long each transect that the absolute error represents 52\u00C2\u00B0N 51 \u00C2\u00B0N Queen \ 2 ^ Charlotte Sound A 50\u00C2\u00B0N British Columbia yQueen Charlotte Strait l a 49\u00C2\u00B0N Vancouver Island 48\u00C2\u00B0N 132\u00C2\u00B0W 130\u00C2\u00B0W Strait of Juan de FUG J . i 128\u00C2\u00B0W 126\u00C2\u00B0W 124\u00C2\u00B0W 122\u00C2\u00B0W Figure 3.1: Map o f southwestern British Columbia, Canada showing the location of underway transects. J 80 52 N 49 N o 48 N 0 51 N o 49 N 10 12 14 16 18 20 10 15 20 25 F i g u r e 3.2: Surface plots o f (a) temperature ( \u00C2\u00B0C ) , (b) sa l in i ty (psu), (c) ch lo rophy l l a (u.g L\" 1), (d) pC02 (ppm), (e) 0 2 /Ar (torr ratio), and (f) D M S (nM) . See F i g . 3.1 for transect labels. See methods for explanat ion o f gaps i n sa l in i ty , ch lo rophy l l and D M S plots. 81 1 O L 0 r 450 - 400 - 350 (ppm) - 300 CM O o Q . - 250 - 200 r 18 Latitude (degrees N) F i g u r e 3.3: Deta i led south-north v i ew o f a l l variables measured a long T5 (see F i g . 3.1 for location); (a) D M S ( _ ) , ch l a (\u00E2\u0080\u0094), D M S P d (\u00E2\u0080\u00A2), D M S P p ( A ) ; (b) 0 2 / A r ( _ ) , pC02 (\u00E2\u0080\u0094); (c) sal inity ( _ ) , and temperature (\u00E2\u0080\u0094). No te the h igh D M S P d that co inc ides w i th the sharp temperature front south o f 50.8 \u00C2\u00B0 N . Lo ca l i z ed upwe l l i ng is evident just south o f 50.8 \u00C2\u00B0 N . 82 0 10 20 30 40 50 60 Distance (km) Figure 3.4: Detailed view of all variables measured along T7 (see Fig . 3.1 for location); (a) D M S ( _ ) , chl a (\u00E2\u0080\u0094), D M S P p . ( A ) ; (b) 0 2 /Ar ( - ) , p C 0 2 (\u00E2\u0080\u0094); (c) salinity ( _ ) , and temperature (\u00E2\u0080\u0094). Note the strong correspondence of the chl a and 0 2 /Ar traces and the mirror images of the pC02 and 0 2 /Ar data. D M S P d data were not available for this transect due to the malfunction of the S E M detector near the end of the cruise. 83 pC02 [ppm] chl a [ug/L] Figure 3.5: The correlation across all transects between (a) pC02 and 0 2 / A r (r2= 0.90) with corresponding chl a concentrations overlaid (colourbar), and (b) chl a and 0 2 / A r (r2 = 0.19) with corresponding temperature overlaid (colourbar). Deviations from linearity in (b) mainly occur at cold surface temperature representing deep-mixed water masses, exclusion of these data from the regression improves the correlation to r2 = 0.73. Chl a (jug L\"1) F i g u r e 3.6: The correlat ion across a l l transects between ch l a and D M S (r2 = 0.06). Note that h igh D M S was general ly associated w i th l ow to moderate ch l a levels (<10 ug L\"1) wh i le regions w i th h igh ch l a had re lat ive ly l o w D M S concentrations. 85 1.0 Temp .\u00E2\u0080\u00A2j _o J \u00E2\u0080\u0094 1 \u00E2\u0080\u0094 i \u00E2\u0080\u0094 1 \u00E2\u0080\u0094 \u00E2\u0080\u00A2 \u00E2\u0080\u0094 1 \u00E2\u0080\u0094 1 \u00E2\u0080\u0094 i \u00E2\u0080\u0094 1 \u00E2\u0080\u0094 1 \u00E2\u0080\u0094 1 \u00E2\u0080\u0094 1 \u00E2\u0080\u0094 1 \u00E2\u0080\u0094 1 \u00E2\u0080\u0094 1 \u00E2\u0080\u0094 1 \u00E2\u0080\u0094 1 \u00E2\u0080\u0094 1 \u00E2\u0080\u0094 -1.0 -0.5 0.0 0.5 1.0 Factor 1 Figure 3.7: Results of the PCA showing the strong separation between pC02 and 02/Ar in two-dimensional space and the clear partitioning between physical (T, S) and biological (chl a) variables. 86 CHL/MLD (mg m\"4) F i g u r e 3 . 8 : Average D M S concentrations plotted against C H L / M L D ratios for the !/4 degree grids. The dashed l ine represents the predicted D M S concentrat ion based on equation 2 o f S&D2002 ( D M S = 55.8 * ( C H L / M L D ) + 0.6). The so l id l ine is the l inear regression o f the actual data ( D M S = 21.0 * ( C H L / M L D ) - 0.1, r = 0.83, n = 27,/?<0.0001); Y error bars are standard errors o f the D M S mean. F i l l e d symbols represent data points w i t h measured M L D s , open symbols represent data po ints w i t h estimated M L D s (see methods). The two outl iers on the far right were exc luded from the regression and represent data from transect 7. 87 0 5 10 15 20 Decorrelation Length Scale (Km) F i g u r e 3.9: Autocor re la t ion functions for a l l parameters measured a long transect 5 (a); the D L S is the first zero cross ing o f the funct ion ; average D L S for a l l parameters for the entire survey (b). Note that D M S appears to have the shortest length scale o f var iab i l i ty , a l though the results are not statistical ly s igni f icant (see discussion) . 88 3.6 References: Anderson , T.R. , S .A . Spa l l , A . Y o o l , P. C i p o l l i n i , P .G. Cha l lenor , and M.J .R. Fasham (2001), G l o b a l f ie lds o f sea surface d imethy lsu l f ide predicted f rom ch lo rophy l l , nutrients and l ight, J. Mar. Syst., 30, 1-20. A r anam i , K., and S. Tsunoga i (2004), Seasonal and regional compar ison o f oceanic and atmospheric d imethy lsu l f ide i n the northern No r th Pac i f i c : D i l u t i o n effects on its concentrat ion dur ing winter , J. Geophys. Res., 109, D 1 2 3 0 3 , do i :10.1029/2003JD004288. Be l v i so , S., A . Sc iandra , C . Cop in-Montegut (2003), Mesosca l e features o f surface water D M S P and D M S concentrations i n the At lant i c Ocean o f f M o r o c c o and in the Mediterranean Sea, Deep-Sea Res. I, 50, 543-555. Be l v i so , S., C . M o u l i n , L. B o p p , and J . Stefels (2004a), Assessment o f a g loba l c l imato logy o f oceanic d imethy lsu l f ide ( D M S ) concentrations based on S e a W i F S imagery (1998-2001), Can. J. Fish. Aquat. Sci., 61, 804-816. Be l v i so , S., L. B o p p , C . M o u l i n , J . C . Orr , T. R. Anderson , O. A u m o n t , S. C h u , S. E l l io t t , M . E. Ma l t r ud , and R. S imo (2004b), Compar i son o f g loba l c l ima to log i ca l maps o f sea surface d imethy l sul f ide, Global Biogeochem. Cycles, 18, G B 3 0 1 3 , do i :10 .1029/2003GB002193. Broecker , W.S. , and T . H . Peng (1982), Tracers in the Sea, L amont Doher ty G e o l . Observ. Press, Pal isades, N . Y . Bucc ia re l l i , E., and W . G . Sunda (2003), Influence o f C O 2 , nitrate, phosphate, and si l icate l imi ta t ion o n intracel lu lar d imethy lsu l foniopropionate i n batch cultures o f the coastal d ia tom Thalassiosirapseudonana, Limnol. Oceanogr. 48, 2256-2265. Char lson, R.J., J .E . Love l o ck , M . O . Andreae, and S.G. War ren (1987), Ocean ic phytoplankton, atmospheric sulfur, c l oud albedo and c l imate, Nature, 326, 655-661. C h i n , M . , and D.J. Jacob, (1996) Anthropogen ic and natural contr ibut ions to tropospheric sulphate: a g loba l m o d e l analysis, / Geophys. Res., 101, 18,691-18,699. C ra ig , H . , and T . H a y w a r d (1987), O x y g e n supersaturation i n the ocean: b io log i ca l versus phys ica l contr ibut ions, Science, 235, 199-202. Dacey, J .W .H . , and S .G. W a k e h a m (1986), Oceanic d imethy lsu l f ide : P roduct ion dur ing zoop lankton graz ing on phytoplankton, Science, 233(4770), 1314-1316. Denman , K., E. H o f m a n n , and H . Marchant (1996), M a r i n e b iot ic responses to environmental change and feedbacks to c l imate, i n Climate Change 1995IPCC, pp. 487-516, Cambr idge U n i v . Press, N e w Y o r k . 89 Hales , B., and T. Takahash i (2004), High-resolut ion b iogeochemica l invest igat ion o f the Ross Sea, Antarc t i ca , dur ing the A E S O P S (U.S. J G O F S ) program, Global Biogeochem. Cycles, 18, G B 3 0 0 6 , do i : 10 .1029/2003GB002165. Hales , B., T . Takahash i , and L. Bandstra (2005), A tmospher i c C O 2 uptake by a coastal upwe l l i ng system, Global Biogeochem. Cycles, 19, G B 1 0 0 9 , do i : 10 .1029/2004GB002295. Ianson, D., and S.E. A l l e n (2002), A two-dimensional n i t rogen and carbon f lux mode l i n a coastal upwe l l i ng reg ion, Global Biogeochem. Cycles, 16(1), 1011, do i : 10 .1029/2001GB001451. Ianson, D., S.E. A l l e n , S.L. Har r i s , K.J. Or ians, D.E. Va re l a , and C.S . W o n g (2003), The inorganic carbon system in the coastal upwe l l i ng reg ion west o f Vancouve r Island, Canada, Deep-Sea Res. I, 50, 1023-1042. Johnson, W .K . , L .A . M i l l e r , N .E . Sutherland, and C.S. W o n g (2005), Iron transport by mesoscale H a i d a eddies i n the G u l f o f A l a s k a , Deep-Sea Res. II, 52 (7-8), 933-953. Jones, A . , D .L . Roberts , M. J . Woodage , and C E . Johnson, (2001) Indirect sulphate aerosol forc ing i n a c l imate m o d e l w i th interactive sulphur cyc le , J. Geophys. Res., 106, 20,293-20,310. Ka iser , J . , M . K . Reuer , B. Barnett, and M . L . Bender (2005), M a r i n e product i v i t y estimates f rom continuous C V A r ratio measurements by membrane inlet mass spectrometry, Geophys. Res. Lett., 32, L I 9 6 0 5 , do i :10 .1029/2005GL023459. Ke l l e r , M . D . , W . K . B e l l o w s , and R.R.L . Gu i l l a rd (1989), D i m e t h y l sul f ide product ion in marine phytoplankton, i n Biogenic sulfur in the environment, edited by E.S. Sa l tzman and W.J . Cooper , p. 167-180, A m e r i c a n Chemica l Society, Wash ington , D . C Kett le , A . J . , M . O . et a l . (1999), A globa l database o f sea surface d imethy lsu l f ide ( D M S ) measurements and a procedure to predict sea surface D M S as a funct ion o f latitude, longitude and month , Global Biogeochem. Cycles, 13, 399-444. K ieber , D.J., J . J iao, R.P. K i e n e , and T.S . Bates (1996), Impact o f d imethy lsu l f ide photochemistry on methy l sulfur cyc l ing i n the equatorial Pac i f i c Ocean , J. Geophys. Res., 101(C2), 3715-3722. K i ene , R.P., and T.S . Bates (1990), B i o l og i c a l remova l o f d imethy l sul f ide f rom sea water, Nature, 345, 702-705. K i ene , R.P., and D. S lezak (2006), L o w dissolved D M S P concentrations i n seawater revealed by smal l-volume gravity f i l t rat ion and dialysis sampl ing , Limnol. Oceanogr.: Methods, 4, 80-95. 90 Leek, C , U . La rsson , L.E. Bagander, S. Johansson, and S. Ha jdu (1990), D i m e t h y l sulf ide in the Ba l t i c Sea: annual var iab i l i t y i n relat ion to b io log i ca l act iv i ty, J. Geophys. Res., 95, 3353-3363. Loca rn in i , S.J.P., S . M . Turner , and P.S. L i s s (1998), The distr ibut ion o f d imethy lsu l f ide , D M S , and d imethy lsu l fon iopropionate , D M S P i n waters o f f the Western Coast o f Ireland, Cont. Shelf Res., 18, 1455-1473. Marchet t i , A . , V . L . Trainer, and P.J. Har r i son (2004), Env i ronmenta l condit ions and phytoplankton dynamics associated w i th Pseudo-nitzschia abundance and domoic acid in the Juan de F u c a eddy, Mar. Ecol. Prog. Ser., 281, 1-12. Marga lef , R. (1978), L i fe-forms o f phytoplankton as surv iva l alternatives i n an unstable environment, Oceanol. Acta., 1, 493-509. Murata , A . , and T. T a k i z a w a (2003), Summert ime C O 2 s inks i n she l f and slope waters o f the western A r c t i c Ocean , Cont. Shelf Res., 23, 753-776. Mu rphy , P.P., Y . N o j i r i , D .E . Har r i son , and N . K . L a r k i n (2001), Scales o f spatial var iabi l i ty for surface ocean p C 0 2 i n the G u l f o f A l a s k a and B e r i n g Sea: Towards a sampl ing strategy, Geophys. Res. Lett, 28, 1047-1050 Night inga le , P.D., P.S. L i s s , and P. Schlosser (2000), Measurements o f air-sea gas transfer dur ing an open ocean algal b l o o m , Geophys. Res. Lett. 27, (14), 2117-2120. Parsons, T. R., Y . M a i t a , and C M . L a l l i (1984), A manual of chemical and biological methods for seawater analysis, Pergamon Press, N e w Y o r k , N . Y . Sarmiento, J .L . , P. Mon f r a y , E. Ma ier-Re imer , O. A u m o n t , R.J. Murnane , and J .C . Orr (2000), Sea-air C 0 2 fluxes and carbon transport: A compar ison o f three ocean general c i rculat ion models , Global Biogeochem. Cycles, 14(4), 1267-1281. Shaw, P.J.A (2003), Multivariate statistics for the environmental sciences, O x f o r d U n i v . Press, N e w Y o r k , N . Y . S imo , R. (2001), P roduc t ion o f atmospheric sulphur by oceanic p lankton : b iogeochemica l , eco logica l and evolut ionary l inks , Trends in Ecol. & Evol, 16(6), 287-294. S imo , R. (2004), F r o m cel ls to globe: approaching the dynamics o f D M S ( P ) i n the ocean at mul t ip le scales, Can. J. Fish. Aquat. Sci., 61, 673-684. S imo , R., and C . Pedros-A l io (1999), Ro l e o f vert ica l m i x i n g i n cont ro l l ing the oceanic product ion o f d imethy l sulphide, Nature, 402, 396-399. 91 S imo , R. and J. Dachs (2002), G l o b a l ocean emiss ion o f d imethy lsu l f ide predicted f rom b iogeophys ica l data, Global Biogeochem. Cycles, 16(4), 1078, do i : 10 .1029/2001GB001829. Sunda, W. , D.J. K i ebe r , R.P. K i e n e , and S. Huntsman (2002), A n ant ioxidant funct ion for D M S P and D M S in mar ine algae, Nature, 418, 317-320. Takahashi , T , S .C. Sutherland, C . Sweeney, A . Po i sson , N . M e t z l , B. T i l b rook , N . Bates, R. Wann inkho f , R.A. Fe l l y , C . Sabine, J. O la fsson , and Y. N o j i r i (2002), G l o b a l sea-air CO2 f lux based on c l imato log ica l surface ocean pCC>2, and seasonal b io log i ca l and temperature effects, Deep-Sea Res. II, 49(9-10), 1601-1622. Thomson , R.E., and I.V. F ine (2003), Es t imat ing m i x e d layer depth f rom oceanic prof i le data, J. Atmos. Oceanic Technol., 20(2), 319-329. Too le , D .A . , D.J. K i ebe r , R.P. K i ene , E . M . Wh i te , J. B i sgrove , D . A . de l V a l l e , and D. S lezak (2004), H i g h d imethy lsu l f ide photolys is rates i n nitrate-rich Antarc t i c waters, Geophys. Res. Lett., 31, L I 1307, do i :10 .1029/2004GL019863 . Tor te l l , P .D. (2005a), D i s s o l v e d gas measurements i n oceanic waters made by membrane inlet mass spectrometry, Limnol.Oceanogr. Methods, 3, 24-37 . Torte l l , P .D. (2005b), Small-scale heterogeneity o f d isso lved gas concentrations in marine continental she l f waters, Geochem. Geophys. Geosyst., 6, Q 1 1 M 0 4 , do i : 10 .1029/2005GC000953 . Townsend, D.W. , and M . D . K e l l e r (1996), D imethy l su l f ide ( D M S ) and d imethy lsu l fon iopropionate ( D M S P ) in relat ion to phytop lankton i n the G u l f o f Ma ine , Mar. Ecol. Prog. Ser., 137, 229-241. V o l k , T. , and M.I. Hof fer t , (1985) Ocean carbon pumps: analysis o f relat ive strengths and eff ic iencies i n ocean-driven atmospheric pC02 changes, i n The carbon cycle and atmospheric C02: Natural variations, archean to present, Geophys. Monogr. Ser., edited by E.D. Sundquist and W.S . Broecker , pp. 99-110, A G U , Wash ington , D .C . 92 Chapter 4: Conclusions 4.1 Thesis Overview The importance o f constra in ing g loba l D M S fluxes and unrave l l ing the complex oceanic sulfur cycle is becoming increas ingly clear as we face a changing g loba l c l imate. B iogen i c sulfur emissions from the wo r l d ' s oceans have a large impact on the Ear th 's c l imate and may help to counteract the effects o f increasing greenhouse gas emissions [Watson and Liss, 1998; Gunson et al, 2006] . However , to evaluate the magnitude o f this impact we need (1) a better mechanist ic understanding o f the D M S cyc le , and (2) more advanced methodo logy to facil itate measurement o f D M S in the oceans. The preceding two chapters that formed the body o f this w o r k tackled these two goals. Each focused on a different appl icat ion o f membrane inlet mass spectrometry to the study o f biogenic sulfur compounds i n the oceans. These chapters introduced new methodology for measuring both d isso lved and particulate D M S P concentrations as w e l l as volat i le D M S , and prov ided new insights into the distr ibutions o f these important compounds . The focus o f Chapter 2 was on discrete, vert ica l measurements o f springtime D M S P p i n oceanic waters a long L i ne P over a three year t ime per iod. Th i s survey attempted to relate D M S P p concentrations to the compos i t ion o f the phytoplankton commun i t y and offered insight into the vert ical var iab i l i ty o f this compound on an interannual t imescale i n an ocean bas in where such measurements d id not prev iously exist. In contrast, Chapter 3 focused on cont inuous, real-time, surface measurements o f D M S in coastal surface waters around Queen Charlotte Sound over a week-long per iod in summer. Th is study examined the high-resolution co-variance o f mu l t ip le parameters in relation to D M S distr ibutions i n a h igh l y dynamic and product ive region. 93 4.2 Evaluation of MIMS for DMSP/DMS Measurements In i t ia l ly , our results from the L i n e P and Queen Charlotte Sound surveys indicated that M I M S was a p romi s i ng new technique for measur ing oceanic D M S P p (and D M S P d ) concentrations. Measurement prec is ion was good, ca l ibrat ion curves were l inear over a large range and analysis t imes were short (see Sect ion 2.3). Howeve r , some recently publ ished data as w e l l as some o f our o w n f indings indicate that there are potent ia l prob lems w i t h this method. F i rst ly , we discovered that pro longed exposure o f the d imethy ls i l i cone membrane to strong base (such as that used to hydro lyze D M S P samples) leads to a breakdown o f the membrane and subsequent s i l i cone contaminat ion o f the mass spectrometer ion source. Th i s results in a drastic loss o f sensit iv i ty w h i c h can on l y be rect i f ied by cost ly replacements o f the entire i on source. Th is p rob lem does not affect the qual i ty o f the data, but leads to frequent, impract ica l and expensive repairs. D i s so l v ed D M S P samples are also typ ica l ly measured at h igh p H . W e attempted to circumvent the above p rob lem for D M S P d analysis dur ing our Queen Charlotte Sound survey by lower ing the p H o f the seawater sample to 9.5 fo l l ow ing the 24 hour hydrolys is per iod (see Sect ion 3.2). Th i s reduced p H pro longed the l i fe o f the membrane wi thout af fect ing the D M S in the sample. Howeve r , this method d i d not produce re l iable results when tested w i th extracted D M S P p samples w h i c h lacked the buf fe r ing capacity o f seawater. E v e n when exact addit ions o f acid and base were made, s l ight var iab i l i ty in the p H between samples affected the permeabi l i ty o f the membrane w h i c h i n turn led to var iab i l i ty i n the D M S signals for a g iven concentration. Thus, D M S P p samples had to be measured at h igh p H us ing the small-area probe w i th the risk o f damaging the system. 94 The second recently ident i f ied p rob lem that makes M I M S not idea l for d isso lved and particulate D M S P analysis is the re lat ively large sample vo lumes required. In the case o f D M S P p , we fdtered 250 m l o f seawater to compensate for the reduced sensit iv i ty o f the small-area membrane inlet probe used. In hindsight, this vo lume cou ld have been reduced to 50 m l i f a 3 m l v ia l had been used dur ing analysis without affect ing the ~1 n M detection l imi t o f in situ D M S P p (see Sect ion 2.3). Howeve r , l ow ambient D M S P d concentrations need to be measured on the more sensit ive large-area membrane cuvette that requires a re lat ive ly large vo lume o f recirculat ing sample (>200 ml ) . It has on l y recently come to l ight that f i l ter ing large vo lumes o f water for D M S P d analysis (even by gravity) leads to elevated D M S P d levels due to ruptured phytoplankton ce l ls [Kiene and Slezak, 2006] . These f i l t rat ion artifacts can be quite signif icant and increase w i t h the v o l u m e o f water f i l tered [Kiene and Slezak, 2006] . It is n o w recommended that on ly the first 3.5 m l o f filtrate co l lected by gravity f i l t rat ion be used to determine D M S P d concentrations, wh i l e D M S P p concentrations are best calculated f rom total D M S P ( D M S P t = D M S P d + D M S P p ) determined f rom whole seawater samples [Kiene and Slezak, 2006]. A l t hough the sensi t iv i ty o f the M I M S is suff ic ient to measure ambient D M S P t concentrations in most regions o f the wo r l d ' s oceans, an unconcentrated 3.5 m l D M S P d sample cou ld not be measured w i th either the sma l l or large-area membrane inlets w i t h the current M I M S conf igurat ion g iven that the g loba l average D M S P d concentrat ion is estimated at < 3 n M [Kiene and Slezak, 2006] . 4.3 Successes a n d P i t f a l l s D u r i n g the L i n e P surveys we were able to achieve our o r ig ina l goa l o f adapting M I M S for oceanic D M S P p measurements, but the data may have underest imated the true D M S P p 95 concentrations i n this reg ion because o f the f i l trat ion method used. Howeve r , the fi ltration-induced artifact is l i ke l y not as severe for D M S P p , w h i c h is present at 10-100-fold higher concentrations than D M S P d [Kiene and Slezak, 2006] . Furthermore, the major i ty o f investigators have col lected D M S P p samples i n m u c h the same way i n the past, w i t h many us ing more damaging vacuum filtration [i.e. Ledyard and Dacey, 1996; Matrai and Vernet, 1997] and others f i l ter ing up to 1 L vo lumes o f seawater [Dacey et al, 1998]. Therefore, a l though the true D M S P p concentrations i n the N E Pac i f i c may indeed be s l ight ly higher than we measured, our data are st i l l useful for compar i son w i t h other studies i n other regions o f the wo r l d ' s oceans. In 2003, we used l ow vacuum f i l t rat ion, whereas i n subsequent years we swi tched to the more gentle method o f gravity filtration. W e w o u l d thus expect that 2003 D M S P p concentrations were potential ly more underestimated than i n later years. Th i s impl ies that the true dec l ine i n D M S P p levels a long L ine P between 2003 and 2004 may be even larger than our results suggest (see Sect ion 2.3). Perhaps the biggest l imi ta t ion o f the L i n e P D M S P p survey l ies i n the phytoplankton data. The mic roscop i c method used to count ind iv idua l cel ls appeared to underestimate the true autotrophic b iomass as evident from the resultant l ow C x h l ratios and the h igh estimates o f cel lular D M S P quotas (see Sect ion 2.4). Th is l i ke l y hampered attempts to relate D M S P p levels to the biomass o f spec i f ic phytoplankton groups i n 2003 . The biggest shortfa l l o f the 3-year dataset however, is the lack o f support ing phytoplankton data for 2004 and 2005. A s a result, we were unable to d is t inguish between the roles o f phytoplankton taxonomy and phys io logy in dr i v ing cross-station or interannual D M S P p var iabi l i ty . Nonetheless, our study prov ided the first D M S P p measurements made in this important and wel l-studied H N L C reg ion that prov ide a reference point for future measurements i n this area. 96 The Queen Charlotte Sound survey on the other hand was a resounding success and we were able to meet or exceed a l l o f our research goals. W e were able to show that D M S levels exhibi ted rapid, large f luctuations over spatial scales that cou ld not be resolved w i th tradit ional G C methods. M o r e important ly , we were able to conv inc ing l y relate surface D M S levels to the ratio o f ch lorophy l l /mixed layer depth, something that had prev ious ly not been achieved in h igh product iv i ty , coastal regions. Furthermore, we obtained interesting results on the strong co-variance o f ch lo rophy l l concentrat ions, 0 2/Ar levels and P C O 2 levels. These data highl ighted the power o f M I M S for measur ing mul t ip le gases at high-resolution s imultaneously , and thereby prov id ing valuable anc i l la ry in format ion useful for interpreting D M S distr ibut ions. Th is dataset cou ld , however, have been improved b y cal ibrat ing the O2/AJ s ignal to give more meaningfu l % ' O2 saturation values. Th i s was not possible dur ing this cruise due to t ime and man-power constraints. In the future, more frequent sampl ing o f total D M S P w o u l d be useful to examine its co-variance w i t h D M S . Samp l ing for total D M S P w o u l d require less effort than separating the dissolved and particulate fract ions, and w o u l d bypass some o f the issues associated w i th these measurements as ment ioned above. 4.4 F u t u r e D i r e c t i o n s W e have shown that M I M S has capabi l i t ies that far exceed many o f those o f P T G C for the measurement o f oceanic D M S concentrations, and have thus m o v e d closer to accompl ish ing the second goal out l ined above: the development o f advanced methodologies that s imp l i f y oceanic D M S measurements. The one area where M I M S currently fal ls short is i n sensitivity. Due to a lack o f a concentrat ing step for D M S , the current detection l im i t o f - 1 n M is not suff icient to accurately measure this gas i n many oceanic regions, part icu lar ly dur ing the winter 97 months. Sens i t iv i ty can be improved i n the future by increasing the surface area o f the membrane; however , this w o u l d l i ke l y necessitate the employment o f a water trap to counteract the negative impact o f an increase i n water vapour i n the vacuum o f the mass spectrometer. A s noted above, the M I M S system in its present conf igurat ion is not usefu l for d isso lved D M S P measurements, and thus w i l l not replace P T G C for smal l vo lume , h igh sensit iv i ty applications. In order to predict the impact o f D M S emissions on future c l imate and the impact o f a changing c l imate on the sul fur cyc le , we need to achieve the pr imary goal as stated above: a mechanist ic understanding o f the under ly ing processes that dr ive the product ion and consumpt ion o f D M S P , D M S and related compounds. Recent progress i n this area has been steady, w i th 3 5\"S tracer studies mapp ing the fate o f D M S P ass imi la t ion and destruction pathways [Kiene and Linn, 2000 ; Vila et al, 2004] and p rov id ing estimates o f sul fur cyc l ing rates [Toole et al, 2004] . Future appl icat ions o f M I M S should a im to couple its cont inuous sampl ing capabil it ies w i th detai led process studies. Fo r example, rates o f m i c rob i a l D M S / D M S P decay cou ld be moni tored i n real-time without the need for radioisotopes or destructive, interval sampl ing. Furthermore, cap i ta l iz ing on the real-time nature o f the M I M S output i n the field w o u l d a l low for immediate , targeted sampl ing of anci l lary parameters at D M S \" f ron t s \" such as those observed dur ing the Queen Charlotte Sound survey (see Sect ion 3.3). Such studies w i l l further enhance our understanding o f the intricacies o f the mar ine sul fur cyc le and lead us closer to p rov ing or refut ing the C L A W hypothesis. 98 4.5 References Dacey, J .W .H . , F .A . H o w s e , A . F . M i chae l s , and S.G. W a k e h a m (1998), Tempora l var iabi l i ty o f d imethy lsu l f ide and dimethylsul foniopropionate i n the Sargasso Sea, Deep-Sea Res. I 45, 2085-2104. Gunson , J .R. , S .A . Spa l l , T .R . Anderson , A . Jones, I.J. Totterdel l , and M. J . Woodage (2006), C l imate sensi t iv i ty to ocean d imethy lsu lphide emiss ions, Geophys. Res. Lett., 33, L 0 7 7 0 1 , d o i : 1 0 . 1 0 2 9 / 2 0 0 5 G L 0 2 4 9 8 2 . K i ene , R.P., and L.J. L i n n (2000), D is t r ibut ion and turnover o f d i sso lved D M S P and its relat ionship w i t h bacter ial product ion i n the G u l f o f M e x i c o , Limnol. Oceanogr., 45, 849-861. K i ene , R.P. and D. S lezak (2006), L o w d isso lved D M S P concentrations i n seawater revealed by smal l-volume gravity f i l t rat ion and dia lys is sampl ing , Limnol. Oceanogr.: Methods, 4, 80-95. Ledyard, K . M . and J . W . H . Dacey (1996), M i c r o b i a l c yc l i ng o f D M S P and D M S in coastal and o l igotrophic seawater, Limnol. Oceanogr., 41, 33-40. Mat ra i , P. A . and M . Vernet (1997), Dynamics o f the vernal b l o o m in the margina l ice zone o f the Barents Sea: d imethy lsu l f ide and d imethy lsu l foniopropionate budgets, J. Geophys. Res., 102, 22965-22979. Too le , D .A . , D.J. K i ebe r , R.P. K i ene , E . M . Wh i te , J . B i sg rove , D . A . de l V a l l e , and D. S lezak (2004), H i g h d imethy lsu l f ide photolys is rates i n nitrate-rich Antarc t i c waters, Geophys. Res. Lett., 31, L I 1307, do i : 10 .1029/2004GL019863. V i l a , M . , R. S i m o , R.P. K i e n e , J . P inhass i , J . A . Gonza lez , M . A . M o r a n , and C . Pedros-Al io (2004), U s e o f microautoradiography combined w i t h f luorescence i n situ hybr id izat ion to determine d imethy lsu l foniopropionate incorporat ion by mar ine bacter ioplankton taxa, Appl. Environ. Microbiol., 70 (8), 4648-4657. Watson, A . J . and P.S. L i s s (1998), M a r i n e b io log ica l controls on c l imate v i a the carbon and sul fur geochemica l cyc les , Phil. Trans. Roy. Soc. Lon. Ser. B-Biol. Sci 353, (1365), 41-51. 99 "@en . "Thesis/Dissertation"@en . "10.14288/1.0053229"@en . "eng"@en . "Oceanography"@en . "Vancouver : University of British Columbia Library"@en . "University of British Columbia"@en . "For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use."@en . "Graduate"@en . "Membrane inlet mass spectrometry (M(MS) : a novel approach to the oceanic measurement of dimethylsulfide"@en . "Text"@en . "http://hdl.handle.net/2429/32146"@en .