Chemosphere 78 (2010) 1063–1070

Contents lists available at ScienceDirect

Chemosphere

journal homepage: www.elsevier .com/locate /chemosphere

Vertical variability of seawater DMS in the South Pacific Ocean and its implicationfor atmospheric and surface seawater DMS

Gangwoong Lee a, Jooyoung Park a, Yuwoon Jang a, Meehye Lee b,*, Kyung-Ryul Kim c, Jae-Ryoung Oh d,Dongseon Kim d, Hi-Il Yi d, Tong-Yup Kim e

a Department of Environmental Science, Hankuk University of Foreign Studies, Yongin, South Koreab Department of Earth and Environmental Sciences, Korea University, Seoul, South Koreac School of Earth and Environmental Sciences, Seoul National University, Seoul, South Koread Korea Ocean Research & Development Institute, Ansan, South Koreae Korea Polar Research Institute, Songdo, South Korea

a r t i c l e i n f o a b s t r a c t

Article history:Received 29 April 2009Received in revised form 20 October 2009Accepted 23 October 2009Available online 15 January 2010

Keywords:Dimethyl sulfideDissolved DMSPParticulate DMSPSouth PacificSea-to-air fluxMixed layer depth

0045-6535/$ - see front matter � 2009 Published bydoi:10.1016/j.chemosphere.2009.10.054

* Corresponding author. Tel.: +82 2 3290 3178; faxE-mail address: meehye@korea.ac.kr (M. Lee).

Shipboard measurements of atmospheric dimethylsulfide (DMS) and sea surface water DMS were per-formed aboard the R/V Onnuri across the South Pacific from Santiago, Chile to Fiji in February 2000.Hydrographic profiles of DMS, dissolved dimethylsulfoniopropionate (DMSPd), and particulate DMSPp

in the upper 200 m were obtained at 16 stations along the track. Atmospheric and sea surface waterDMS concentrations ranged from 3 to 442 pptv and from 0.1 to 19.9 nM, respectively; the mean valuesof 61 pptv and 2.1 nM, respectively, were comparable to those from previous studies in the South Pacific.The South Pacific Gyre was distinguished by longitudinal-vertical distributions of DMS, DMSPd, andDMSPp, which was thought to be associated with the characteristic modification of biological activitiesthat occurs mainly due to significant change in water temperature. The averaged DMS maximumappeared at 40 m depth, whereas DMSPp and DMSPd maxima coincided with that of dissolved oxygencontent at 60–80 m. The sea-to-air fluxes of DMS were estimated to be 0.4–11.3 lmol d�1 m�2

(mean = 2.8 lmol d�1 m�2). A fairly good correlation between atmospheric DMS and sea-to-air DMS fluxindicated that atmospheric DMS concentration was more sensitive to change in physical parameters thanits photochemical removal process or surface seawater DMS concentrations.

� 2009 Published by Elsevier Ltd.

1. Introduction

Dimethylsulfide (DMS), the most dominant sulfur speciesthroughout the ocean, is formed by enzymatic cleavage of dim-ethylsulfoniopropionate (DMSP), which is produced by a varietyof phytoplankton species. Most DMSP is consumed by bacteriaand only a fraction is used to produce DMS (Kiene, 1996). AlthoughDMS is removed by bacterial consumption (Kiene and Bates, 1990),sea surface layers are always supersaturated with it, which impliesa net flux of DMS to the atmosphere (Huebert et al., 2004). As a re-sult, approximately 1% of the DMSP produced in sea water is trans-ported to the air in the form of DMS through sea-to-air flux (Bateset al., 1994; Simó and Pedrós-Alió, 1999).

After being released into the atmosphere, DMS is readily oxi-dized to non-sea-salt sulfate (nss-SO2�

4 ) and methane sulfonate(MSA) in the atmospheric boundary layer. Atmospheric DMS ismainly oxidized by OH during the day and nitrate radical (NO3)at night. The oxidation of atmospheric DMS seems to contribute

Elsevier Ltd.

: +82 2 3290 3189.

largely to the formation of aerosol containing nss-SO2�4 in the mar-

ine troposphere. According to the CLAW hypothesis (Charlsonet al., 1987), sulfate and MSA produced from oceanic DMS affectthe Earth’s radiation balance through the formation of cloud con-densation nuclei (CCN), thereby altering cloud properties. Theoverall effect of these couplings on climate is negative feedback,meaning that it tends to stabilize the climate. In recent studies,DMS was positively correlated with atmospheric CCN (Vallinaet al., 2006, 2007) and solar radiation (Toole and Siegel, 2004; Val-lina and Simó, 2007) over most of the global ocean, which supportsthe DMS–climate feedback loop for open-ocean environments.

Researchers estimate that the gaseous DMS flux from the oceanto the atmosphere lies between 23 and 35 Tg S yr�1 (Kettle and An-dreae, 2000; Simó and Dachs, 2002; Kloster et al., 2006). The oce-anic DMS flux compromises �30% of global sulfur sources (IPCC,2001) and its contribution to global nss-SO2�

4 is 27%, both of whichare similar in magnitude (Kloster et al., 2006). The mean annualcontribution of DMS to the climate-relevant nss-SO2�

4 column bur-den is the greatest (43%) in the relatively pristine Southern Hemi-sphere, where a lower oxidative capacity of the atmosphere, alarger sea-to-air transfer of DMS, and a larger surface area of

Fig. 1. Ship track and stations (closed circles) for vertical seawater sampling.Numbers above and below each station indicate the station number and date inFebruary 2000, respectively.

1064 G. Lee et al. / Chemosphere 78 (2010) 1063–1070

emission lead to an elevated atmospheric DMS burden (Gondweet al., 2003). Therefore, the vast area of the South Pacific is thekey region in which to test the validity of the CLAW hypothesis.

DMS flux from the ocean has been estimated through theparameterization of wind fields and the maps of DMS concentra-tions in the global ocean (Liss and Merlivat, 1986; Wanninkhof,1992; Kettle et al., 1999). Although Kettle et al. (1999) compileda seawater DMS and DMSP database of more than 15 000 measure-ments over the global ocean, the temporal and spatial coverage ofDMS is still poor. To obtain a global view of DMS distribution bytime, determining oceanic DMS concentration has to be ap-proached using various empirical parameterizations of field obser-vation datasets, such as chlorophyll a distribution (Anderson et al.,2001), climatological mixed layer depth (Simó and Dachs, 2002),and SeaWiFS ocean color measurements (Belviso et al., 2006).

Kettle et al. (1999) found no significant correlations betweenDMS and other oceanographic parameters and no simple algorithmto create temporal fields of sea surface DMS concentrations basedon these parameters. Thus, to reduce the great uncertainty inher-ent in estimates of DMS flux, more measurements with greatertemporal and spatial resolution are necessary. This is particularlytrue for the South Pacific, where measurements of DMS and DMSPare still very sparse. DMS in the South Pacific has been studiedmost extensively by Bates and his group (Bates and Quinn, 1997;Bates et al., 1998; Bates, 2004). Most of their research, however,has been concentrated on the equatorial Pacific, which is a regionthat exhibits relatively high DMS emissions throughout the year(Bates and Quinn, 1997). Unlike the equatorial Pacific, the centralSouth Pacific (20–50�S) should have large seasonal and spatial vari-ations of DMS levels due to distinct seasonality and latitudinalvariations in sea surface temperature. The central South Pacifichas an area of 30 � 106 km2 and covers about 8% of all oceansand seas worldwide, yet only four sets of latitudinal transit dataare available for this region in the Global Surface Seawater DMSDatabase (Bates, 2004). Convincing evidence also exists for the sea-sonality of DMSP and DMS concentrations and DMS flux in theSouthern Ocean (Simó and Dachs, 2002; Vallina et al., 2007).

Kettle et al. (1999) reported that having DMS measurements isnot enough to explain the global DMS distribution, particularly inthe South Pacific and Indian Ocean. To evaluate the role of DMSin climate change at regional to global scales also requires mea-surements of atmospheric and sea water DMS concentrations,quantification of its sea-to-air flux, and identification of the factorsthat control them. In this experiment, we concurrently measuredsea water DMS, dissolved DMSP, and particulate DMSP at variousdepths, mainly within the thermocline, and the atmospheric DMSalong the ship track from Chile to Fiji. We then characterized thebehaviors of atmospheric DMS and its sea-to-air flux with regardto various factors such as sea surface DMS and DMSP, momentumflux, and mixed layer depth MLD. The results of this study will beuseful to evaluate global ecosystem models for DMS productionand to accurately determine the global DMS budget.

2. Methods

As a part of the Southern Pacific Ocean Dynamic Studies, mea-surements of atmospheric DMS and surface water DMS were madeon board the research vessel (R/V) Onnuri, which left Punta Arenas,Chile on February 5, 2000 and arrived at Fiji on March 4, 2000. Thestudy area lies between 20�S and 50�S and runs from the equatorialPacific to the Southern Ocean, in which scarce sets of DMS andDMSP measurements are available. Fig. 1 shows the ship trackand the stations at which hydrocast samples were taken. On Febru-ary 20 and 21, hydrocast sampling was cancelled in order to keepthe ship on schedule.

For vertical profiles of DMS and DMSP, sea water samples werecollected using 11 Niskin bottles from 3 m below the surface downto 200 m at 16 stations. During each hydrocast, conductivity, tem-perature, and depth were continuously determined with a CTDalong with dissolved oxygen content. Water samples were takenfrom 3 m below surface down to 200 m at each station: seven sam-ples between 3 m and 100 m, and four samples between 100 m and200 m at a 25 m interval. The hydrocasts were conducted at dawnso that bacterial and phytoplankton populations would be mini-mally affected upon exposure to ultraviolet radiation (Kiene andLinn, 2000). Upon retrieval of the bottles, sea water was gentlydrawn from each Niskin bottle into a 130 mL DO bottle through ty-gon tubing; the new bottle was overflowed with sea water 2–3times so that no air would be trapped inside. At each hydrocast sta-tion, surface water was collected using a bucket for the surfaceDMS measurement. Along the cruise track, surface seawater alsowas sampled 4–6 times a day for DMS analysis using the continu-ous seawater pumping system on the ship.

To measure sea water DMS, an aliquot of 30 mL was withdrawnfrom the DO bottle with a 50 mL syringe. The water-filled syringethen was connected to an air-tight filter holder equipped with a47 mm Whatman GF/F glass fiber filter. By applying gentle pres-sure to syringe, the water sample was filtered into a 100 mL gasstripper bottle. Next, the filtrate was purged with high purity he-lium at 100 mL min�1 for 20 min and the purged air was passedthrough a Nafion dryer (Perma-Pure, Inc., USA) to remove watervapor. Finally, DMS was captured in a carbosieve 300 adsorptiontrap (Supelco, USA). The dissolved DMSP (DMSPd) in the purged fil-trate and the particulate DMSP (DMSPp) on the filter were con-verted to DMS using a strong alkaline solution for 1–2 h and thenmeasured as DMS (Kim and Andreae, 1987). A common practiceis to leave the particulate DMSP samples in the alkaline solutionat least 12 h. As a result, our DMSPp values were likely underesti-mated due to incomplete hydrolysis of DMSPp. DMS and DMSP insea water were measured right after the water samples were col-lected from the upper depths. Some samples from lower depths,however, could not be analyzed immediately and had to wait amaximum of 4 h; during this time they remained in the dark at

0100200300400500

pptv

0

4

8

12

16

nM

6-Feb 11-Feb 16-Feb 21-Feb 26-Feb 2-Mar

0

4

8

12

m/s

ec

0

0.4

0.8

1.2

N/m

2

Atmospheric DMS

Momentum Flux

Wind Speed

Sea surface DMS

(a)

(b)

(c)

(d)

Fig. 2. Variations of (a) atmospheric DMS, (b) surface water DMS, (c) momentumflux, and (d) wind speeds during the course of the experiment.

G. Lee et al. / Chemosphere 78 (2010) 1063–1070 1065

room temperature. Traps containing seawater DMS were stored ina freezer at �70 �C until gas chromatography (GC) analysis couldbe conducted.

Air DMS was collected in adsorption traps (carbosieve 300)every 3–6 h (4–6 times a day) at the upper bridge deck of the ship.Air was passed through a Nafion dryer filled with silica gel and a KItrap at �100 mL min�1 (1 atm, 25 �C) ahead of an adsorption tubeto remove water vapor and oxidants, respectively. AtmosphericDMS was analyzed immediately after being collected.

The DMS in a trap was desorbed by rapid heating to 300 �Cusing a thermal desorption unit (Supelco, USA) for 3 min and in-jected directly into a HP 6890 GC equipped with a 30-in. packedsuper Q column (Alltech, USA) and a sulfur chemiluminescencedetector (Siever, USA). The flow rate of carrier gas (He) was set at30 mL min�1. The oven temperature was programmed to initiateat 60 �C for 3 min, then the temperature was raised to 200 �C at arate of 15 �C min�1, held there for 5 min. Detection limits of atmo-spheric and seawater DMS were 1.5 pptv and 0.01 nM, respec-tively. DMS concentrations were calibrated using a permeationdevice (13 ng min�1 at 30 �C, VICI, USA). The analytical precisionsof atmospheric and seawater DMS measurements were 12% and13%, respectively, which accounted for uncertainties associatedwith variance of daily DMS standards, air flow, and water volumeover the course of the experiment.

Meteorological data, such as wind speed, wind direction, airtemperature, and pressure, were obtained from the shipboardautomated weather system (Vaisala, Finland), which was designedfor accurate measurement of true wind direction and speed. How-ever, the data for the first 2 weeks of the experiment were lost be-cause they were overwritten daily under the automatic savingmode. After February 16, meteorological data were manuallysaved. Modeled wind speeds and momentum fluxes were derivedfrom the FNL archive produced by the National Centers for Envi-ronmental Prediction (NCEP). The FNL model data used in thisstudy were archived in the Air Resources Laboratory of the Na-tional Oceanic and Atmospheric Administration (NOAA, http://www.arl.noaa.gov/ready). While the FNL wind speed and directionand those observed on the ship were similar in their pattern of var-iation, their one-to-one correlation was poor.

3. Results and discussion

3.1. Distribution of atmospheric and surface seawater DMS

Fig. 2 shows the variations of atmospheric and seawater DMS,momentum fluxes, and wind speeds over the course of the study.The atmospheric DMS concentrations ranged from 3 to 442 pptvwith a mean of 61 pptv. For the first week of the experiment, theatmospheric DMS concentrations remained below 50 pptv. Mostof the high atmospheric DMS concentrations (>100 pptv) occurredbetween February 14 and 24 and between stations 4 (37�590S,101�580W) and 13 (28�510S, 153�330W), which is the subtropicalregion of the South Pacific. Although atmospheric DMS over theocean originates from the ocean, we did not find a significant cor-relation between atmospheric and surface seawater DMS concen-trations in this study. Although sea surface and atmospheric DMSconcentrations were raised over the same period, their variationsclearly differed from each other. Watanabe et al. (1995) observeda strong positive correlation between the spatial distributions ofsurface seawater and air DMS over the temperate North Pacific.However, other previous studies conducted in various marine envi-ronments failed to find similar correlations due to complexities insea-air transfer mechanisms of DMS (Berresheim et al., 1991;Church et al., 1991) and removal processes of atmospheric DMS(Kieber et al., 1996).

In this study, meteorological parameters affecting sea-to-airflux played a more pronounced role in controlling atmosphericDMS than did surface water DMS concentrations. The period of ele-vated atmospheric DMS concentrations, especially between Febru-ary 17 and 24, was characterized by higher wind speed (>5 m s�1).Although wind data were lost and the overall correlation was notavailable for the first 10 d of the experiment, atmospheric DMSconcentrations were closely related to the momentum fluxes.However, the poor correlations between observed wind and FNLdata hinder further discussion about the relationship betweenatmospheric DMS and the calculated momentum flux. Nonethe-less, it is evident that the intimate coupling of sea-to-air flux withwind speed and momentum flux played a significant role in deter-mining atmospheric DMS concentrations over the study region.

The surface seawater DMS concentrations varied considerablyfrom 0.1 to 19.9 nM (mean = 2.1 nM) and were elevated concur-rently with atmospheric DMS at stations 4–13 in the subtropicalregion. Our results indicate that in summer, seawater DMS concen-trations of the subtropical region in the South Pacific clearly dif-fered from those of tropical and temperate regions. These spatialdifferences in surface seawater DMS will be examined in detail ina later section.

Our measurements of atmospheric and surface seawater DMSare in reasonable agreement with DMS concentrations from previ-ous studies conducted in the Pacific Ocean. For atmospheric DMS,our mean was close to that obtained by Nguyen et al. (1983) duringthe spring, but their measurements (40–139 pptv) were less scat-tered than ours (3–442 pptv); our values also fall within the rangesobserved at a coastal site (36�160S) in New Zealand (Wylie and deMora, 1996) and in the North Pacific (Watanabe et al., 1995).

Over the South Pacific Subtropical Gyre, the levels of atmo-spheric and seawater DMS are highest during February and March(�3–4 nM), based on the global database of measurements (Kettle

1066 G. Lee et al. / Chemosphere 78 (2010) 1063–1070

et al., 1999; Kettle and Andreae, 2000) and calculations based onsatellite observations of primary production (Longhurst et al.,1995). For surface seawater DMS, our mean of 2.1 nM lies nearthe lower end of all measurements throughout the Pacific Ocean.Excluding temperate and tropical regions, however, the mean con-centration of 2.9 nM for the subtropics is comparable to that fromthe compiled database. It is noteworthy that the overall mean of2.1 nM in our study was very similar to that calculated (2.4 nM)with an algorithm using mean MLD (Simó and Dachs, 2002), whichwill be discussed later. These results show that the measured DMSconcentrations in this study were in accordance with those esti-mated from the global database (Kettle et al., 1999; Simó andDachs, 2002).

3.2. Vertical distribution of water DMS

The hydrographic profiles of DMS, DMSPd, and DMSPp were ob-tained at 16 stations along the cruise track, and we used these datato construct longitude–depth distributions (Fig. 3). In general, DMSmaxima occurred 20–40 m below the sea surface. Such sub-surfacemaxima are commonly observed in the open ocean (Dacey et al.,1998; Scarratt et al., 2002; Wong et al., 2005). Below the maxima,DMS concentrations decreased exponentially and almost reached

0.03

0.20.

090.

080.

07

0.06

0.050.04

0.4

0.3

0.7 0.

70.6

0.6 0.5

0.5 0.

4

0.4

0.7

0.6

0.5

0.4

0.60.5

0.4

0.9

0.9

0.80.7

0.3

0.8

0.30.2

0.4

0.90.8

0.07

0.06

0.050.04

0.03

0.9

0.9

0.02

0.9

0.8

3

2

2

0.02

4

0.09

0.08

0.07

3

0.07

0.09

0.08

0.07

0.06

0.05

0.04

0.03

0.02

0.2

87654

0.6

0.5

3

23

0.079

0.07

0.90.8

0.70.6 0.

5

0.90.80.7

0.60.5

0.40.3

0.2

4

32

0.2

0.09

0.08

0.07

0.06

0.050.04

0.03

0.02

0.3

5

760.02

0.2

0.09

0.08

0.07

0.060.05

4

4

0.04

0.03

0.4

0.3

0.22

0.2

0.2

0.20.090.08

0.2

0.80.70.6

0.5

0.1

1

1

1

1

1

0.1

0.1

1

1 0.1

0.10.1

1

Longitude

Dep

th M

20

40

60

80

100

120

140

160

180

200

0.060.050.04

0.30.2

0.090.080.07

0.9 0.80.7

0.60.50.4

3 45

2

0.030.02

0.90.80.70.60.5

20.9

65

3

4

45

0.9

5

0.09

0.08

0.07

20.90.80.70.60.50.4

6

5

7

40.60.50.4

0

0

0

0.0000

0.000

0.1

1

1

1 1

0.1

1

0

1

Lo

-160 -150 -140 -130 -120 -110 -100 -90

-160 -150 -140 -13

Dep

th M

20

40

60

80

100

120

140

160

180

200

(a) (

(c)

Fig. 3. Contoured vertical distributions of (a) DMS,

the detection limit at 200 m. A water mass with high concentra-tions of DMS was distinguishable between 105�W (38�S) and150�W (30�S); across these boundaries, the abundance of phyto-plankton species and productivity significantly changed as well.In the region between 30�S and 40�S, chlorophytes are the domi-nant species during summer. This area lies in the transition zonefrom diatom-dominated Antarctica to the cyanobacteria-domi-nated southern central Gyre (Gregg et al., 2003). As we did not ana-lyze phytoplankton species, we could not confirm that thedistinctively high DMS concentrations to the north of 38�S wasdue mainly to the change from diatoms to chlorophytes. However,the abrupt change in DMS distribution likely was related with bio-logical activity, which depends greatly on phytoplankton species.We also detected a discernable change in the depths of DMS max-ima between two regions: at higher latitudes, the DMS maximumoccurred �20 m, whereas it was deeper in the central Gyre(�40 m). The variation in the depth of DMS maxima suggests spa-tial differences in biological activity.

In the transition zone between 105�W and 150�W where rela-tively high sub-surface DMS concentrations occurred, DMSPp con-centrations also were enhanced. In contrast, DMSPd concentrationswere consistently low in this region. If we consider the change inthe DMSPd/DMS ratio is indicative of bacterial conversion of DMSPd

0.06

0.050.04

0.3

0.2

0.090.080.070.06

0.90.80.70.60.50.4

2

4

3

3

2

2

0.050.04

0.2

4

0.9

0.9

0.80.70.60.

28

765

0.5

8

0.2

0.8

0.9

0.8

0.090.080.070.060.050.04

0.030.02

5

43

3

0.9

0.02

5

3

2

0.9

2

2

0.30.2

0.4

0.3

0.2

3

0.90.80.70.60.5 0.4

0.4

0.2

2

0.3

0.2

0.090.080.07

0.06

0.05

0.04 0.03 0.0

2

0.90.80.70.60.50.4

0.50.4

0.2

0.090.2

0.060.050.040.03

0.40.30.5

0.4

0.03

0.4

0.80.70.60.50.4

2

0.9

0.80.70.6

43

7 65

433

0.40.3

0.25

6

43

56

0.20.60.50.4

0.3

0.09

0.9

0.8

0.7

0.6

2

0.9

20.4

0.1

1

1

1

1

0.1

1

1

1

0.1

1 1

0.1

10.1

0.1

0.1

1

1

Longitude

Dep

th M

20

40

60

80

100

120

140

160

180

200

2

43

4

0.60.50.4.3

.2

.2

09.08.07.06.05

0.04

9.8.7.60.5

0.2

60.2

9876

0.2

0.90.8

0.7

0.6

6

5

4

5

0.3

0.2

0.09

0.08

0.07

0.06

0.05

0.04

0.03

3

65

7

0.70.6

0.5

0.4

0.90.8

0.70.6

0.8

0.09 0.2

0.90.8

87

6 5 43

2

0.3

98765432

0.50.4

9

87

65432

0.9

0.09

9

0.90.80.70.60.50.4

0.30.2

9

876

0.2

0.60.50.4

0.9 0.8

3

1

.1

10

1 1

0.1

10

0.1

1

1

110

0.1

10

1

10

10

ngitude

-160 -150 -140 -130 -120 -110 -100 -90

0 -120 -110 -100 -90

b)

(b) dissolved DMSP, and (c) particulate DMSP.

G. Lee et al. / Chemosphere 78 (2010) 1063–1070 1067

to DMS (Jones et al., 1998), the bacterial conversion of DMSPd

would be greatest over the transition region. While the levels ofDMS and DMSPp within the water column rapidly decreased mov-ing north from 30�S (west of 150�W), DMSPd increased noticeablyin this subtropical oligotrophic region; this indicates that the rateof bacterial conversion of DMSPd to DMS or the direct release ofDMS from phytoplankton likely decreased. Evidently, the changes

DMSDissolved DMSPParticulate DMSP

0 1 2 3 4 5Concentration (nM)

200

160

120

80

40

0

Dep

th (m

)

8 12 16 20 24Temperature oC

200

160

120

80

40

0

Dep

th

24 24 25 25 26 26 27Sigma-θ

5 5 6 6 6Oxygen mg/l

34.0 34.2 34.4 34.6 34.8 35.0Salinity %o

TemperatureSalinitySigma- θOxygen

(a)

(b)

Fig. 4. Vertical profiles of (a) DMS and dissolved and particulate DMSP concentra-tions (averaged for all stations) and (b) density (sigma h), temperature, salinity, andoxygen contents at station 8.

in biological activity, including phytoplankton composition andbacterial activities, played a crucial role in the distribution ofDMS over the research area. Further investigation is needed to clar-ify the mechanisms responsible for this finding and their relativeimportance to DMS and DMSP distributions.

In Fig. 4a, vertical profiles at all 16 stations were averaged forDMS, DMSPd, and DMSPp. The maximum depth appeared at 40 mfor DMS and 60–80 m for DMSPp and DMSPd, which is within thedepth range of the Chl a maximum reported for the South Pacific(Maritorena et al., 2002; Grob et al., 2007). As shown in typical pro-files of density, temperature, salinity, and oxygen contents(Fig. 4b), the surface mixed layer was well developed throughoutthe study region, even though the depth of the mixed layer variedslightly from station to station (mainly due to meteorological andoceanographic conditions). The depth for maximum dissolved oxy-gen coincided with the depth at which the maximum concentra-tions of DMSPp and DMSPd were found. Considering thatdissolved oxygen content is an indication of biological productionwithin the euphotic zone, the conversion of DMSPd to DMS wouldbe higher within the mixed layer. Indeed, the variations in DMSand DMSPd were out of phase between 20 m and 60 m (Fig. 4a).

3.3. DMS flux into the air

We estimated the sea-to-air fluxes of DMS using the followinggas exchange model for the 12 stations for which wind speedand water temperature data were available:

FluxDMS ¼ klðCl � Cg=HÞ ffi kl � Cl

where kl is the gas transfer velocity; Cl and Cg are the concentrationsof gas in seawater and the atmosphere, respectively; and H is thesolubility of the gas in seawater. Because seawater is supersatu-rated relative to atmospheric DMS, the flux of DMS can be ex-pressed as the product of surface water concentrations and thetransfer velocity of DMS in the interface of sea and air. Variousmethods have been employed to parameterize kl using wind speedand physio-chemical properties of surface water (Liss and Merlivat,1986; Wanninkhof, 1992; Nightingale et al., 2000). Here, we usedthe parameterization schemes of Liss and Merlivat (1986) (hereinaf-ter LM86) and Wanninkhof (1992) (hereinafter W92) that havebeen most widely used in calculations of gas transfer velocity. Intheir schemes, the gas transfer velocity is regarded as being depen-dent on wind speed at 10 m height (u10) and the Schmidt number(ScDMS), which is expressed as a function of temperature (Saltzmanet al., 1993):

ScDMS ¼ 2674:0� 147:12t þ 3:726t2 � 0:038t3

where t is water temperature (�C).In LM86, the gas transfer velocity is defined differently accord-

ing to wind speed:

Kl;DMS ¼ 0:17u10ð660=ScDMSÞ2=3½cm h�1�u10 6 3:6 ðsmooth regimeÞ

Kl;DMS ¼ ð2:85u10 � 9:65Þð660=ScDMSÞ1=2½cm h�1�3:6 < u10 6 13 ðrough regimeÞ

Kl;DMS ¼ ð5:9u10 � 49:3Þð660=ScDMSÞ1=2½cm h�1�u10 P 13 ðwave-breaking regimeÞ

In comparison, the gas transfer velocity given by W92 is simplerthan that of LM86:

Kl;DMS ¼ 0:31u210ð660=ScDMSÞ1½cm h�1�

Using these equations, we calculated the DMS fluxes for eachstation and the results are given in Table 1. The sea-to-air fluxesof DMS by LM86 ranged from 0.4 to 11.3, with a mean of

1068 G. Lee et al. / Chemosphere 78 (2010) 1063–1070

2.8 mol d�1 m�2. In comparison, DMS fluxes by W92 were greaterby almost a factor of two. Recent studies reported that the mostdefensible gas exchange velocities fall between those of LM86and W92 (Nightingale et al., 2000; Huebert et al., 2004), so themean flux of DMS in this study would be between 2.8 mol d�1 m�2

(LM86) and 4.5 mol d�1 m�2 (W92). Using the LM86 method, Ket-

Table 1Calculated DMS fluxes using observed and model-derived winds.

Stations Surface waterDMS (nM)

Watertemperature (�C)

Wind speed(m s�1)

DMS fluxes(lmol d�1 m�2)

W92 LM86

1 0.8 15.8 7.0 2.1 1.43 4.2 16.5 3.5 3.0 0.46 1.0 20.1 3.2 0.6 0.17 1.2 21.4 5.7 2.5 1.69 1.4 22 7.8 5.4 3.510 3.1 22 9.4 18.0 11.311 3.0 23.7 6.8 9.6 6.512 0.7 24.4 4.0 0.8 0.313 1.3 26 5.0 2.2 1.314 0.6 26.5 6.3 1.8 1.215 1.2 27.3 6.2 3.3 2.216 1.3 27.2 6.9 4.7 3.2

Means 2.1 21.6 6.0 4.5 2.8

× ×

(a)

(c)

Fig. 5. Correlations between atmospheric DMS and (a) observed wind speeds and (b) DBoxes and bars represent 25–75 percentiles and 5–95 percentiles, respectively. Lines andsamples were collected from midnight to 6 am.

tle and Andreae (2000) estimated the mean DMS flux as 4–5 mol d�1 m�2 at latitudes from 30�S to 40�S using global climato-logical wind fields and compiled surface seawater DMS. Theirhigher flux value can be attributed in part to their higher concen-trations of surface DMS. Our data included measurements fromsome tropical and temperate regions where water DMS levels wererelatively low. If we limited our analysis to the same latitudinalzone studied by Kettle and Andreae (stations 6–13), the mean fluxof DMS would increase to 3.5 mol d�1 m�2. Considering the uncer-tainties in two studies, we found no significant difference.

Although wind is the most important physical factor controllingDMS flux, field studies rarely report that atmospheric DMS concen-trations were dependent on wind speeds (Huebert et al., 2004). Inour study, we also found a reasonable quadratic correlation be-tween atmospheric DMS and wind speed (Fig. 5a). Andreae et al.(1995) and Ayers et al. (1995) reported a similar relationship. Be-cause the surface DMS concentrations were determined along theship track as well as at hydrocast stations, we calculated sea-to-air DMS fluxes for all available surface seawater DMS measure-ments using the LM86 method. Sea water temperatures were notmeasured during transit and interpolated from those obtained nearhydrocast stations. Fig. 5b illustrates that air DMS was well corre-lated with DMS flux, which is of great consequence considering thepoor correlation between surface seawater and atmospheric DMSconcentrations. Some systematic deviations occur in Fig. 5b and

− −

× − −

(b)

(d)

MS fluxes, and diurnal variations of (a) atmospheric and (b) surface seawater DMS.diamonds inside the box denote median and mean, respectively. No surface water

1 2 3 4 5 6 7 8 9 10 111213141516

Station number

0

2

4

6

8

10

Surfa

ce W

ater

DM

S n

M

0

10

20

30

40

50

MLD

m

DMSMLD

0 2 4 6 8 10Surface Water DMS nM

50

40

30

20

10

0

MLD

m

R2 = 0.443834

Y = 9.819 * ln(X) + 23.830

(b)(a)

Fig. 6. (a) The spatial variations of surface DMS and mixed layer depth (MLD) and (b) the relationship between surface DMS concentrations and MLD. The best fitted line wasexpressed as a logarithmic equation. MLD was defined as the depth at which water density increased by 0.03 kg m�3 or temperature dropped by 0.1�C.

G. Lee et al. / Chemosphere 78 (2010) 1063–1070 1069

the atmospheric concentration is a direct result of the competitionbetween seawater source and atmospheric sinks. However, thiscorrelation implies that levels of atmospheric DMS were more sub-ject to change in physical parameters (such as wind speed) thansurface seawater DMS concentration. Watanabe et al. (1995)pointed that with fast removal of atmospheric DMS, a proportionalcorrelation between atmospheric DMS and flux should be possible.

In the open ocean, the most important process of removal ofatmospheric DMS is its oxidation by OH radicals (Chen et al.,2000). Because OH radicals are produced by sunlight, the removalof atmospheric DMS is most efficient during the day. All atmo-spheric DMS measurements throughout the cruise were plottedagainst local times in Fig. 5c. In our experiment, diurnal variationsof atmospheric DMS showed a general tendency toward low con-centrations during the day. Most studies that have reported a typ-ical pattern of atmospheric DMS variation with minima in theafternoon were conducted at island sites or in areas in which oce-anic and atmospheric conditions were relatively homogeneous andpersistent (Ayers et al., 1995; Bandy et al., 1996; Sciare et al.,2000). In fact, Cooper and Saltzman (1991) reported that diurnalvariations were not evident in open oceans due to large-scalechanges in seawater and atmospheric conditions, which effectivelymasked any effects due to oxidation processes. We found no signif-icant diurnal variation in seawater DMS (Fig. 5d), mainly becauseits short-term variability was coupled not only with the variabilityof solar radiation, but also with other physical parameters such aswind speed and mixing (Toole and Siegel, 2004).

3.4. MLD vs. surface seawater DMS

Simó and Dachs (2002) constructed the following straightfor-ward but meaningful algorithm to estimate surface seawaterDMS concentrations [nM] using MLD [m] and Chl a [mg m�3] bycombining more than 2000 DMS data points with concurrent sur-face Chl a measurements and climatological MLDs:

DMS ¼ �LnðMLDÞ þ 5:7 for Chl a=MLD < 0:02DMS ¼ 55:8ðChl a=MLDÞ þ 0:6 for Chl a=MLD 6 0:02

ð1Þ

Although we did not measure Chl a in our study, it is not likelythat Chl a concentrations were higher than 1 mg m�3 based on re-sults reported by Gregg and Casey (2004). Thus, ratios of Chl a/MLDin this experiment would be smaller than 0.02 with the observedmaximum MLD of 45 m. Therefore, it is reasonable to assume thatour data fall into the first regime, where MLD is the sole factor thatdetermines the surface seawater DMS concentration. Using the

equation above, our mean MLD of 28 m resulted in 2.4 nM ofDMS, which was just 10% higher than the overall observed meanof 2.1 nM for surface seawater DMS. Thus, Simó and Dachs(2002) empirical equation produced a reasonable mean valuewhen applied to large areas of the South Pacific.

The logarithmic equation above suggests a negative correlationbetween surface seawater DMS concentration and MLD. In thisstudy, however, surface seawater DMS and MLD for each stationwere positively correlated (Fig. 6). This disagreement can be re-solved by considering the vertical distribution of seawater DMSand the physical properties of the water column (Fig. 4). Whilethe mean depth of water DMS maxima was 40 m, most MLDs wereshallower than 40 m (mean = 28 m). Under such conditions, DMSconcentrations in the surface water would be enhanced by a dee-per MLD, which would enable the surface water to be mixed withwater enriched in DMS below the MLD. Because Simó and Dachs(2002) equation was derived from global measurements of datacombined with climatological MLD, cautions should be exercisedwhen applying the MLD algorithm to estimate water DMS varia-tions at a certain region for a short period of time.

4. Conclusions

We successfully measured atmospheric DMS and seawaterDMS, DMSPd, and DMSPp along an east–west transect cruise inthe South Pacific during February 2000. The mean concentrationsof 61 pptv and 2.11 nM for atmospheric and surface seawaterDMS, respectively, agreed well with those from previous studiesover similar latitudes and times of the year. Our results also agreedwell with water DMS concentrations estimated from the globaldatabase over the South Pacific in summertime. The South PacificGyre was identified by characteristic distributions of DMS, DMSPd,and DMSPp along latitude and depth. The distinctive variations ofthese species were closely coupled with characteristic modificationin geophysical and biological parameters. Averaged DMS maximaof whole study appeared at 40 m, which was deeper than the meanMLD of 28 m. The DMSPp and DMSPd maxima occurred at 60–80 m,where dissolved oxygen content also was at a maximum. The sea-to-air fluxes of DMS varied from 0.4 to 11.3, with a mean of2.8 mol d�1 m�2. For the region of the South Pacific Gyre (30–40�S), the mean flux was 3.5 lmol d�1 m�2, which was lower thanthat estimated by Kettle et al. (1999) (4–5 lmol d�1 m�2) using thesame parameterization with global climatological wind fields andcompiled surface seawater DMS. Neither clear correlation betweenatmospheric DMS and surface seawater DMS nor significant

1070 G. Lee et al. / Chemosphere 78 (2010) 1063–1070

diurnal variations of atmospheric DMS were observed in our study.Instead, we found a reasonable correlation between atmosphericDMS and sea-to-air DMS flux. Finally, Simó and Dachs’ (2002) esti-mated sea surface DMS concentration calculated using mixed layerdepth agreed well with the observed mean in our study. Because ofinherent differences in methodology, however, extreme careshould be exercised when estimating short-term variations ofDMS using only MLD.

Acknowledgements

This work was funded by the Korea Meteorological Administra-tion Research and Development Program under Grant CATER 2007-3204.

References

Anderson, T.R., Spall, S.A., Yool, A., Cipollini, P., Challenor, P.G., Fasham, M.J.R., 2001.Global fields of sea surface dimethylsulfide predicted from chlorophyll,nutrients and light. J. Mar. Syst. 30, 1–20.

Andreae, M.O., Elbert, W., de Mora, S.J., 1995. Biogenic sulfur emissions and aerosolsover the tropical South Atlantic 3. Atmospheric dimethylsulfide, aerosols andcloud condensation nuclei. J. Geophys. Res. 100, 11335–11356.

Ayers, G.P., Gillett, R.W., Ivey, J.P., Schäfer, B., Gabric, A., 1995. Short-term variabilityin marine atmospheric dimethylsulfide concentration. Geophys. Res. Lett. 22(18), 2513–2516.

Bandy, A.R., Thornton, D.C., Blomquist, B.W., Chen, S., Wade, T.P., Ianni, J.C., Mitchell,G.M., Nadler, W., 1996. Chemistry of dimethyl sulfide in the equatorial Pacificatmosphere. Geophys. Res. Lett. 23, 741–744.

Bates, T.S., 2004. Global surface seawater dimethylsulfide (DMS) database. In: PaperPresented at 8th International Global and Atmospheric Chemistry Conference,Christchurch, New Zealand.

Bates, T.S., Quinn, P.K., 1997. Dimethylsulfide (DMS) in the equatorial Pacific Ocean(1982 to 1996): evidence of a climate feedback. Geophys. Res. Lett. 24, 861–864.

Bates, T.S., Kiene, R.P., Wolfe, G.V., Matrai, P.A., Chavez, F.P., Buck, K.R., Blomquist,B.W., Cuhel, R.L., 1994. The cycling of sulfur in surface seawater of the NortheastPacific. J. Geophys. Res. 99, 7835–7843.

Bates, T.S., Kapustin, V.N., Quinn, P.K., Covert, D.S., Coffman, D.J., Mari, C., Durkee,P.A., DeBruyn, W., Saltzman, E., 1998. Processes controlling the distribution ofaerosol panicles in the lower marine boundary layer during the First AerosolCharacterization Experiment (ACE-I). J. Geophys. Res. 103, 16369–16383.

Belviso, S., Thouaeau, G., Schmidt, S., Reigstad, M., Wassmann, P., Arashkevich, E.,Stefels, J., 2006. Significance of vertical flux as a sink for surface water DMSP andas a source for the sediment surface in coastal zones of northern Europe. Estuar.Coast. Shelf Sci. 68, 473–488.

Berresheim, H., Andreae, M.O., Iverson, R.L., Li, S.M., 1991. Seasonal variations ofdimethylsulfide emissions and atmospheric sulfur and nitrogen species over thewestern North Atlantic Ocean. Tellus B 43, 353–372.

Charlson, R.J., Lovelock, J.E., Andreae, M.O., Warren, S.G., 1987. Oceanicphytoplankton, atmospheric sulfur, cloud albedo and climate. Nature 295,655–661.

Chen, G., Davis, D., Kasibhatla, P., Bandy, A., Thornton, D., Huebert, B., Clarke, A.,2000. A study of DMS oxidation in the tropics: comparison of Christmas Islandfield observations of DMS, SO2, and DMSO with model simulations. J. Atmos.Chem. 37, 137–160.

Church, T.M., Tramontano, J.M., Whelpdale, D.M., Andreae, M.O., Galloway, J.N.,Keene, W.C., Knap, A.H., Tokos, J., 1991. Atmospheric and precipitationchemistry over the North Atlantic Ocean: shipboard results, April–May 1984.J. Geophys. Res. 96 (D10), 18705–18726.

Cooper, D.J., Saltzman, E.S., 1991. Measurements of atmospheric dimethyl sulfideand carbon disulfide in the western Atlantic boundary layer. J. Atmos. Chem. 12,153–168.

Dacey, J.W.H., Howse, F.A., Michaels, A.F., Wakeham, S.G., 1998. Temporal variabilityof dimethylsulfide and dimethylsulfoniopropionate in the Sargasso sea. Deep-Sea Res. I 45, 2085–2104.

Gondwe, M., Krol, M., Gieskes, W., Klaassen, W., de Baar, H., 2003. The contributionof ocean-leaving DMS to the global atmospheric burdens of DMS, MSA, SO2, andNSS SO2�

4 . Global Biogeochem. Cycles 17 (2), 1056. doi:10.1029/2002GB001937.Gregg, W.W., Casey, N.W., 2004. Global and regional evaluation of the SeaWiFS

chlorophyll data set. Remote Sens. Environ. 93, 463–479.Gregg, W.W., Ginoux, P., Schopf, P.S., Casey, N.W., 2003. Phytoplankton and iron:

validation of a global three-dimensional ocean biogeochemical model. Deep-Sea Res. II 50, 3143–3169.

Grob, C., Ulloa, O., Li, W.K.W., Alarcón, G., Fukasawa, M., Watanabe, S., 2007.Picoplankton abundance and biomass across the eastern South Pacific Oceanalong latitude 32.5�S. Mar. Ecol. Prog. Ser. 332, 53–62.

Huebert, B.J., Blomquist, B.W., Hare, J.E., Fairall, C.W., Johnson, J.E., Bates, T.S., 2004.Measurement of the sea–air DMS flux and transfer velocity using eddycorrelation. Geophys. Res. Lett. 31, L23113. doi:10.1029/2004GL021567.

Jones, G.B., Curran, M.A.J., Swan, H.B., Greene, R.M., Griffiths, F.B., Clementson, L.A.,1998. Influence of different water masses and biological activity ondimethylsulphide and dimethylsulphoniopropionate in the subantarctic zoneof the Southern Ocean during ACE. J. Geophys. Res. 103 (D13), 16691–16701.

Kettle, A.J., Andreae, M.O., 2000. Flux of dimethylsulfide from the oceans: acomparison of updated data sets and flux models. J. Geophys. Res. 105, 26793–26808.

Kettle, A.J. et al., 1999. A global database of sea surface dimethylsulfide (DMS)measurements and a procedure to predict sea surface DMS as a function oflatitude, longitude, and month. Global Biogeochem. Cycles 13 (2), 399–444.

Kieber, D.J., Jiao, J., Kiene, R.P., Bates, T.S., 1996. Impact of dimethylsulfidephotochemistry on methyl sulfur cycling in the equatorial Pacific Ocean. J.Geophys. Res. 101, 3715–3722.

Kiene, R.P., 1996. Production of methanethiol from dimethylsulfoniopropionate inmarine surface waters. Mar. Chem. 54, 69–83.

Kiene, R.P., Bates, T.S., 1990. Biological removal of dimethyl sulphide from seawater. Nature 345, 702–705.

Kiene, R.P., Linn, L.J., 2000. Distribution and turnover of dissolved DMSP and itsrelationship with bacterial production and dimethylsulfide in the Gulf ofMexico. Limn. Oceanogr. 45, 849–861.

Kim, K.-H., Andreae, M.O., 1987. Determination of carbon disulfide in natural watersby adsorbent preconcentration and gas chromatography with flamephotometric detection. Anal. Chem. 59, 2670–2673.

Kloster, S., Feichter, J., Maier-Reimer, E., Six, K.D., Stier, P., Wetzel, P., 2006. DMScycle in the marine ocean-atmosphere system – a global model study.Biogeoscience 3, 29–51.

Liss, P.S., Merlivat, L., 1986. Air–sea exchange rates: introduction and synthesis. In:Buat-Menard, P. (Ed.), The Role of Air–Sea Exchange in Geochemical Cycling. D.Reidel Publishing, Dordrecht, Holland, pp. 113–127.

Longhurst, A., Sathyendranath, S., Platt, T., Caverhill, C., 1995. An estimate of globalprimary production in the ocean from satellite radiometer data. J. Plankton Res.17, 1245–1271.

Maritorena, S., Siegel, D., Peterson, A., 2002. Optimization of a semi-analytical oceancolor model for global-scale applications. Appl. Opt. 41, 2705–2714.

Nguyen, B.C., Bonsang, B., Gaudry, A., 1983. The role of the ocean in the globalatmospheric sulfur cycle. J. Geophys. Res. 88, 10903–10914.

Nightingale, P.D., Malin, G.C., Law, S., Watson, A.J., Liss, P.S., Liddicoat, M.I., Boutin, J.,Upstill-Goddard, R.C., 2000. In situ evaluation of air–sea gas exchangeparameterizations using novel conservative and volatile tracers. GlobalBiogeochem. Cycles 14 (1), 373–388.

Saltzman, E.S., King, D.B., Holmen, K., Leck, C., 1993. Experimental determination ofthe diffusion coefficient of dimethylsulfide in water. J. Geophys. Res. 98, 16481–16486.

Scarratt, M., Levasseur, M., Michaud, S., Cantin, G., Gosselin, M., de Mora, M.J., 2002.Influence of phytoplankton taxonomic profile on the distribution ofdimethylsulfide and dimethylsulfoniopropionate in the northwest Atlantic.Mar. Ecol. Prog. Ser. 244, 49–61.

Sciare, J., Mihalopoulos, N., Dentener, F.J., 2000. Interannual variability ofatmospheric dimethylsulfide in the southern Indian Ocean. J. Geophys. Res.105 (D21), 26369–26378.

Simó, R., Dachs, J., 2002. Global ocean emission of dimethylsulfide predicted frombiogeophysical data. Global Biogeochem. Cycles 16 (4), 1018. doi:10.1029/2001GB 001829.

Simó, R., Pedrós-Alió, C., 1999. Role of vertical mixing in controlling the oceanicproduction of dimethyl sulphide. Nature 402, 396–399.

Toole, D.A., Siegel, D.A., 2004. Light-driven cycling of dimethylsulfide (DMS) in theSargasso Sea: closing the loop. Geophys. Res. Lett. 31, L09308. doi:10.1029/2004GL019581.

Vallina, S.M., Simó, R., 2007. Strong relationship between DMS and the solarradiation dose over the global surface ocean. Science 315, 506–508.

Vallina, S.M., Simó, R., Gassó, S., 2006. What controls CCN seasonality in theSouthern Ocean? A statistical analysis based on satellite-derived chlorophylland CCN and model-estimated OH radical and rainfall. Global Biogeochem.Cycles 20, GB1014. doi:10.1029/2005GB002597.

Vallina, S.M., Simó, R., Gassó, S., de Boyer-Montégut, C., del Río, E., Jurado, E., Dachs,J., 2007. Analysis of a potential ‘‘solar radiation dose–dimethylsulfide–cloudcondensation nuclei” link from globally mapped seasonal correlations. GlobalBiogeochem. Cycles 21, 2004. doi:10.1029/2006GB002787.

Wanninkhof, R., 1992. Relationship between wind speed and gas exchange over theocean. J. Geophys. Res. 97, 7373–7382.

Watanabe, S., Yamamoto, H., Tsunogai, S., 1995. Relation between theconcentrations of DMS in surface seawater and air in the temperate NorthPacific region. J. Atmos. Chem. 22, 271–283.

Wong, C.S., Wong, S.E., Richardson, W.A., Smith, G.E., Arychuk, M.D., Page, J.S., 2005.Temporal and spatial distribution of dimethylsulfide in the subarctic northeastPacific Ocean: a high-nutrient-low-chlorophyll region. Tellus B 57, 317–331.

Wylie, D.J., de Mora, S.J., 1996. Atmospheric dimethylsulfide and sulfur species inaerosol and rainwater at a coastal site in New Zealand. J. Geophys. Res. 101(D15), 21041–21049.