CDOM Sources and Photobleaching Control Quantum Yields for

Subscriber access provided by ATHABASCA UNIV

Article

CDOM sources and photobleaching control quantum yields for oceanic DMS photolysis Martí Galí, David John Kieber, Cristina Romera-Castillo, Joanna D. Kinsey, Emmanuel Devred, Gonzalo Luis Pérez, George R. Westby, Cèlia Marrasé, Marcel Babin, Maurice Levasseur, Carlos M. Duarte, Susana Agusti, and Rafel Simo Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.6b04278 • Publication Date (Web): 14 Nov 2016 Downloaded from http://pubs.acs.org on November 22, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Environmental Science & Technology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 40

Environmental Science & Technology

CDOM sources and photobleaching control quantum yields for oceanic DMS photolysis

Martí Galí,1* David J. Kieber,2 Cristina Romera-Castillo,3,4 Joanna D. Kinsey,2,5 Emmanuel Devred,1,6 Gonzalo L. Pérez,7 George R. Westby,2 Cèlia Marrasé,8 Marcel Babin,1 Maurice Levasseur,1 Carlos M. Duarte,9 Susana Agustí9 and Rafel Simó8

*Corresponding author: [email protected] Address: Takuvik Joint International Laboratory (Université Laval – CNRS). Biology Department, Université Laval. 1045 Avenue de la Médecine G1V 0A6 Québec (QC) Canada. Phone: 418-656-2131 #7740. Fax: 418-656-2339

1. Takuvik Joint International Laboratory (Université Laval – CNRS). Biology Department, Université Laval. 1045 Avenue de la Médecine G1V 0A6 Québec (QC) Canada 2. Department of Chemistry, College of Environmental Science and Forestry, State University of New York, 1 Forestry Drive, Syracuse, New York 13210 3. Rosenstiel School of Marine and Atmospheric Science, RSMAS/OCE, University of Miami, Miami, FL 33149

ACS Paragon Plus Environment

1


Page 2 of 40

4. Department of Limnology and Bio-Oceanography, Center of Ecology, University of Vienna, 1090 Vienna, Austria 5. Department of Marine, Earth and Atmospheric Sciences, North Carolina State University, 2800 Faucette Drive, Raleigh, NC 27695 6. Fisheries and Oceans Canada, Bedford Institute of Oceanography, Dartmouth, NS B2Y 4A2, Canada. 7. Instituto INIBIOMA (CRUB Comahue, CONICET), Quintral 1250, 8400 S.C. de Bariloche, Rio Negro, Argentina 8. Department of Marine Biology and Oceanography, Institut de Ciències del Mar (CSIC). Passeig Marítim de la Barceloneta 37-49, 08003 Barcelona, Catalonia, Spain 9. King Abdullah University of Science and Technology (KAUST), Red Sea Research Center (RSRC), Thuwal, 23955-6900, Saudi Arabia


2

Page 3 of 40


1

Abstract

2

Photolysis is a major removal pathway for the biogenic gas dimethylsulfide (DMS) in the

3

oceans’ surface. Here we tested the hypothesis that apparent quantum yields (AQY) for DMS

4

photolysis varied according to the quantity and quality of its photosensitizers, chiefly

5

chromophoric dissolved organic matter (CDOM) and nitrate. AQY compiled from the literature

6

and unpublished studies ranged across three orders of magnitude at the 330 nm reference

7

wavelength. The smallest AQY(330) were observed in coastal waters receiving major riverine

8

inputs of terrestrial CDOM (0.06 to 0.5 m3 (mol quanta)-1). In open-ocean waters, AQY(330)

9

generally ranged between 1 and 10 m3 (mol quanta)-1. The largest AQY(330), up to 34 m3 (mol

10

quanta)-1), were seen in the Southern Ocean potentially associated with upwelling. Despite the

11

large AQY variability, daily photolysis rate constants at the sea surface spanned a smaller range

12

(0.04-3.7 d-1), mainly because of the inverse relationship between CDOM absorption and AQY.

13

Comparison of AQY(330) with CDOM spectral signatures suggests there is an interplay between

14

CDOM origin (terrestrial versus marine) and photobleaching that controls variations in AQYs,

15

with a secondary role for nitrate. Our results can be used for regional or large-scale assessment of

16

DMS photolysis rates in future studies.


3


Page 4 of 40

17

1. Introduction

18

Dimethylsulfide (DMS) is primarily produced in the sunlit layer of the ocean from microbial of

transformations

20

synthesized by phytoplankton, macroalgae and corals.3 The sea-air flux of DMS, estimated at 20-

21

28 Tg S y-1 (ref.

22

DMS can affect climate through its effects on aerosol and cloud formation,8–10 and DMS plays an

23

important ecological role as an infochemical.11,12 However, the sea-air DMS flux represents only

24

a small fraction (generally 5—15%) of the DMS produced in the upper mixed layer (UML) of

25

the ocean.13 The remainder is removed from the water column through biological14 and

26

photochemical15,16 processes. In a shallow upper mixed layer (UML) and under high-irradiance

27

conditions typical of summer, photolysis can become the main removal mechanism for

28

DMS,13,17–19 controlling its sea-surface concentration and flux into the atmosphere.13

4,5

dimethylsulfoniopropionate

(DMSP),1,2

19

a

multifunctional

osmolyte

), is large compared to many other marine organic volatiles.6,7 Ocean-emitted

29 30

DMS does not absorb photons at the solar wavelengths that reach the Earth’s surface. Rather, it

31

is oxidized by reactive species generated upon light absorption by optically-active substances in

32

seawater known as photosensitizers.15 It is generally assumed that chromophoric dissolved

33

organic matter (CDOM) and nitrate20 are the main DMS photosensitizers in seawater. Although

34

several studies have reported on the kinetics and the spectral dependence of DMS photolysis in

35

different marine and estuarine environments, the factors that control the observed variability in

36

DMS photolysis rates, and the relative importance of CDOM and nitrate as photosensitizers,

37

remain controversial.21,22 Even less is known about the molecular-scale mechanisms of DMS

38

photolysis, the specific oxidants involved (e.g., the hydroxyl radical or other radical species20,23)


4

Page 5 of 40


39

and their temperature dependence.24 This limits our ability to model DMS photolysis across

40

contrasting marine environments and to predict its future trends under global-change scenarios.

41 42

DMS photolysis follows pseudo-first order kinetics16,23 at ambient seawater DMS

43

concentrations.16 Thus, instantaneous photolysis rates (d[DMS]/dt) can be calculated as the

44

product of the pseudo first-order rate constant (kp) and DMS concentration:

45 46

d[DMS]/dt = kp [DMS]

(1)

47 48

In turn, kp can be modeled as the product of three spectral variables integrated across the relevant

49

wavelength range (Supporting Information (SI) Figure S1):

50 51

kp = ∫ Ed(λ) aCDOM(λ) AQY(λ) dλ

(2)

52 53

where Ed(λ) is the downwelling irradiance (mol quanta m-2 s-1), aCDOM(λ) is the CDOM

54

absorption coefficient (m-1), and AQY(λ) is the apparent quantum yield for DMS photolysis [m3

55

(mol quanta)-1].

56 57

An AQY is used here instead of a true quantum yield to define the efficiency for DMS

58

photolysis because the precursors giving rise to DMS photolysis in seawater are not known and

59

therefore the absorption term includes all components of CDOM, including contributions of

60

light-absorbing substances (chromophores) that do not participate in DMS photolysis. Thus,

61

wavelength-dependent AQY are minimum estimates of the true wavelength-dependent QYs. In


5


Page 6 of 40

62

general, AQYs decrease exponentially with increasing wavelength21,24–26 and become

63

insignificant at wavelengths longer than 400 nm, as is the case for DMS photolysis. The

64

photolysis of DMS is driven by ultraviolet (UV) radiation between 300-400 nm in seawater, with

65

a peak near 330 nm.21,24–26

66 67

Generally, CDOM is the main absorber of UV radiation and thus controls UV penetration in

68

oceanic waters.27 CDOM absorption coefficients typically decrease sharply from river-influenced

69

coastal waters to offshore waters, reflecting the balance between photobleaching of terrestrially-

70

derived CDOM, autochthonous CDOM production28,29 and mixing between the oceanic and

71

coastal end-members.27 Owing to its characteristic optical signature, CDOM can be monitored

72

using ocean color remote sensing,27 enabling large-scale photochemistry assessments. Satellite-

73

based estimates of photochemical rates have been made for dissolved organic carbon (DOC)

74

photomineralization30 and for carbon monoxide photoproduction31 in seawater. Yet, to our

75

knowledge this approach has not been applied to DMS photolysis, likely because a synthesis

76

study of DMS photolysis AQYs was missing. Here we analyze the variability in DMS photolysis

77

rate constants and AQYs in a comprehensive global dataset, gaining insights into the

78

environmental factors affecting photolysis. This should improve our ability to remotely sense and

79

model DMS photolysis in surface waters globally.

80 81

2. Methods

82

2.1 Database

83

We assembled a global database of DMS photolysis measurements in the upper mixed layer of

84

the ocean (UML), including both published and previously unpublished data (Figure 1 and SI


6

Page 7 of 40


85

Table S1). Individual studies were labeled with the name of the lead author followed by the

86

publication year, or with the name of the authors followed by the study area in the case of

87

unpublished studies. Some studies were divided into two or three regions according to the

88

authors' criteria. We did not apply any correction to the data reported in the original papers, and

89

the reader is referred to Table S2 for methodological details of each study. The data were

90

assigned to three different subsets (D1-D3) according to the type of measurements performed and

91

the increasing uncertainty of AQY estimation:

92

D1 (n = 50): Studies specifically designed to measure AQY (see section 2.4).

93

D2 (n = 78): Studies where kp was determined in at least one spectral treatment along with

94

aCDOM spectra and spectral UV downwelling irradiances (Ed). This allowed AQY to be

95

determined at the reference wavelength (330 nm) using spectral optimization methods after

96

making a few simplifying assumptions (see section 2.5).

97

D3 (n = 18): kp was determined, but aCDOM and/or UV irradiances did not have sufficient

98

spectral resolution to determine AQY(330) with the methodology used for the D2 subset. Instead,

99

AQY(330) for this dataset were estimated from a fitted equation based on data from the D1 and

100

D2 subsets (see section 2.6).

101 102

Samples were further classified into three biogeochemical domains: river-influenced coastal

103

waters that include the Mackenzie Estuary and Shelf ("Taalba13MES" subset), plus the

104

Mississippi plume ("Kinsey_PLU") and its transitional waters towards the Gulf of Mexico at a

105

distance less than 50 km from the coast ("Kinsey_INT"); other coastal/shelf areas that include

106

samples collected less than 50 km from the coast over a water column shallower than 200 m and

107

excluding river-influenced waters; and the open ocean that includes the remaining samples.


7


Page 8 of 40

108 109

When available, ancillary variables were reported including temperature in situ and during the

110

incubation, salinity, nitrate concentration (nitrate + nitrite in some studies), chlorophyll a, and

111

dissolved organic carbon concentration (DOC). Missing temperature, salinity and nitrate data

112

were filled with climatological data from the World Ocean Atlas32–34 (only for open-ocean

113

samples), and missing bottom depths were obtained from the General Bathymetric Chart of the

114

Oceans (www.gebco.net/). Available CDOM spectra were used to compute (i) the spectral slope

115

ratio SR, defined as S275_295/S350_400 where S is the spectral slope of ln(aCDOM) for the spectral

116

intervals 275—295 nm and 350—400 nm;35 (ii) the absorbance ratio between 254 and 365 nm,

117

aCDOM(254)/aCDOM(365);36 and (iii) the specific ultraviolet absorption (often referred to as

118

SUVA)37 defined as aCDOM(254)/DOC. SR and aCDOM(254)/aCDOM(365) are proxies for the mean

119

CDOM molecular weight and degree of photobleaching, whereas aCDOM(254)/DOC is a proxy for

120

aromaticity.

121 122

Our analysis includes only the results from filtered seawater incubations, precluding any

123

assessment of particle-associated DMS photolysis. Measurement of particle-associated DMS

124

photolysis is challenging20 due to concurrent DMS photoproduction by marine particles,18,38 and

125

might be important in detritus- and sediment-laden coastal waters.20

126 127

All statistical analyses were done using Matlab® 2013b, except for the statistics shown in

128

Table 1 that were calculated using R 3.2.039 (multcompView package40). Significance thresholds

129

and reported confidence intervals correspond to α = 0.05. Throughout the text, r refers to

130

Pearson's linear correlation coefficient and rS to Spearman's rank correlation coefficient.


8

Page 9 of 40


131 132

2.2 Radiative transfer model

133

The downwelling spectral irradiance just below the sea surface, Ed,0-(λ), was computed using

134

the Santa Barbara DISORT Atmospheric Radiative Transfer (SBDART)41 model as implemented

135

by Bélanger et al. (ref. 42). In this configuration, SBDART outputs the spectral irradiance (µmol

136

quanta m-2 s-1) at a given location every 3 h at a spectral resolution of 5 nm between 290 and 700

137

nm using climatological satellite-observed atmospheric properties at each location and date.

138

Further details on SBDART implementation are given in SI section 1.

139 140

2.3 kp determination and scaling to daily irradiance

141

The pseudo-first order rate constant can be determined by evaluating DMS loss over time

142

under controlled irradiance using the integrated form of eq 1:

143 144

kp,OBS = ln([DMS]0/[DMS])/t

(3)

145 146

where kp,OBS is the observed photolysis rate constant, [DMS]o is the initial DMS concentration

147

and [DMS] is the concentration at irradiation time t.

148

expressed in several different units. To allow for a comparison, all the kp,OBS were normalized to

149

the daily climatological irradiance at their corresponding sampling location using the equation:

In the compiled studies, kp,OBS was

150 151

kp = kp,OBS Ed,day / Ed,INCUB

(4)

152


9


Page 10 of 40

153

where Ed,INCUB is the reported incubation irradiance and Ed,day is the daily climatological

154

irradiance calculated with SBDART, integrated within the same radiation band that was reported

155

in each study (see SI section 2). Note that normalization to the particular band reported in each

156

study would have been nearly impossible without the use of a spectrally resolved radiative

157

transfer model. When kp,OBS was reported in photon dose-1 units, kp was converted to d-1 units

158

assuming reciprocity.43 Henceforth, kp in our study is expressed in d-1 and represents the pseudo-

159

first order photolysis rate constant just below the sea surface for a given location and date and the

160

corresponding 24 h climatological irradiance.

161 162

2.4 AQY(λ) determination in the D1 subset

163

Studies belonging to the D1 subset were designed to determine AQY(λ) using a

164

monochromatic24,25 or polychromatic21,26 approach, as described elsewhere. Results obtained

165

from both approaches indicate that the AQY spectrum for DMS photolysis can be modeled using

166

a decreasing exponential function:

167 168

AQY(λ) = AQYref exp [SAQY (λ–λref)]

(5)

169 170

where AQYref is the AQY at the reference wavelength, λref, and SAQY is the AQY spectral slope.

171

Here we set the reference wavelength at 330 nm, corresponding to the peak in the photolysis

172

response,21,24–26 and converted AQY(λ) reported at different wavelengths to AQY(330) using eq

173

5 when needed. The unit for the DMS AQY is m3 (mol quanta)-1. This is equivalent to 10-6 mol

174

DMS photolyzed (mol quanta)-1 normalized to 1 nmol L-1 DMS, as used in other studies.24

175


10

Page 11 of 40


176

2.5 AQY(330) determination in the D2 subset

177

Studies designated as D2 were not designed to determine AQY(λ). Yet, since aCDOM and

178

irradiance spectra between 300 and 400 nm were accurately known, we were able to use a

179

nonlinear optimization method to determine the AQY spectrum that satisfied the observed kp,

180

according to eq 2 and 5 (see SI for methodological details). This approach is similar to that

181

employed for polychromatic D1 incubations except that, when the experimental kp was

182

determined under a single spectral treatment, it required making assumptions regarding the value

183

of SAQY. That is, since SAQY could not be estimated through optimization, an SAQY value had to

184

be chosen a priori to estimate AQY(330) through optimization. The sensitivity of optimized

185

AQY(330) to the chosen SAQY was estimated by calculating AQY(330) for thirteen SAQY values

186

ranging between 0.0200 and 0.0700 nm-1 (see section 3.2). This test showed that, on average,

187

estimated AQY(330) varied by ±36% (relative standard deviation about the mean) between the

188

smallest and the largest SAQY values tested. We took the mean of the thirteen AQY(330) and

189

assumed that this coefficient of variation represented the maximal error associated to AQY(330)

190

in the D2 subset. The uncertainty was also computed over a narrower spectral-slope range,

191

between 0.034 and 0.058 nm-1, which encompassed the open-ocean AQY spectra. This yielded a

192

correspondingly smaller uncertainty (< ±10%) in AQY(330).

193 194

Occasionally, kp was determined under two to four spectral treatments in samples classified in

195

the D2 subset38. This allowed for the determination of both AQY(330) and SAQY following the

196

same polychromatic modeling approach that was used for the D1 subset, but with larger

197

uncertainty due to the smaller number of spectral treatments applied.

198


11


Page 12 of 40

199

2.6 AQY(330) determination in the D3 subset

200

The photolysis of DMS just below the sea surface generally peaks around 330 nm21,24–26 (SI

201

Figure S1), suggesting that kp should be roughly proportional to the product Ed(330) aCDOM(330)

202

AQY(330). Here we took advantage of this spectral response to derive an empirical method to

203

estimate AQY(330). In the first step, a standardized daily photolysis rate constant, k*p, was

204

calculated by scaling kp to an arbitrary standard daily irradiance (Ed,daySTA). For the D1 subset,

205

k*p was calculated through forward application of eq 2 when the in situ aCDOM(λ) was available.

206

For D2 and D3, it was estimated as:

207 208

k*p = kp Ed,daySTA / Ed,dayINCUB

(6)

209 210

where Ed,daySTA is the output of SBDART at 45°N and 30°W for June 21 (summer solstice). For

211

D2 and D3, eq 6 was applied using a broadband irradiance value obtained by integrating the

212

Ed,daySTA spectrum across the same radiation band that was monitored during each incubation

213

(Ed,dayINCUB), as done in section 2.3.

214 215 216

In the second step, k*p was used to plot AQY(330) from the D1 and D2 subsets against k*p/aCDOM(330) to obtain the following regression equation (Figure 2A):

217 218

log10AQY(330) = (–0.21±0.01) + (1.00±0.01)log10[k*p/aCDOM(330)]

(7)

219 220

with r2 = 0.99, n = 111 (type II major axis regression). The highly significant relationship

221

confirmed that the broadband photolysis of DMS in seawater exposed to sea-surface solar


12

Page 13 of 40


222

radiation can be approximated by the peak response wavelength at 330 nm. Furthermore, this

223

procedure allowed us to include in our analysis AQY(330) data from the D3 subset (or, more

224

generally, any samples where only aCDOM(330), kp and the broadband irradiance were known).

225 226

3. Results and discussion

227

3.1 kp and AQY(330) variability

228

Our dataset covers a wide environmental range spanning the Arctic through temperate and

229

tropical oceans to the Antarctic Seas, and from estuarine and coastal areas to the open ocean

230

(Figure 1). Since most measurements were obtained between the spring and fall equinoxes, the

231

daily UV irradiance typically varied within a relatively narrow range (Figure 1B). By contrast,

232

aCDOM, nitrate and AQY(330) varied by two to three orders of magnitude both across latitudes

233

and from oceanic to coastal waters (Figure 1C-E), suggesting that their role in controlling kp

234

variability was more important than that of irradiance. Since the patterns displayed by aCDOM and

235

nitrate have been described elsewhere, below we examine the variability of the daily DMS

236

photolysis rate constants at the sea surface (kp) and the corresponding AQY(330).

237 238

In the ensemble of studies, kp varied between 0.044 and 3.66 d-1, with a median of 0.55 d-1, and

239

displayed regular oceanographic patterns (Table 1, Figure 1F). Typically, kp ranged between 0.2

240

and 0.6 d-1 in low-latitude (35°S to 35°N) open ocean environments16,24,44,45 and in temperate46 to

241

subpolar21,25,47 northern hemisphere oceans, with slightly lower kp at low latitudes. A larger

242

scatter was observed in northern polar seas, with an overall median of 0.28 d-1 but with higher kp

243

in the Greenland Sea19 compared to the Beaufort Sea and the northern Baffin Bay.26 Higher kp of

244

ca. 1 d-1 were observed in waters nearby the Antarctic Peninsula and in the Ross Sea; yet, the


13


Page 14 of 40

245

highest kp values (median 2.2 d-1) were consistently observed in the Southern Ocean22,48

246

particularly at latitudes between 50° and 70°S in the Indian and Pacific sectors.

247 248

In the open ocean, the 50% central values (interquartile range) of AQY(330) were between 1.3

249

and 10.5 m3 (mol quanta)-1, with a median of 3.1 m3 (mol quanta)-1. In tropical and subtropical

250

oceans, as well as in northern temperate to subpolar seas, the median AQY(330) was 1.7 m3 (mol

251

quanta)-1. A larger scatter was observed at latitudes above 65°N, with values up to 8 m3 (mol

252

quanta)-1 in the Greenland Sea but less than 1 m3 (mol quanta)-1 in the northern Baffin Bay.

253

Moderate to high values were observed south of 65°S, with a median of 6.4 m3 (mol quanta)-1.

254

As with kp, the highest AQY(330) were seen in Sub-Antarctic waters with a maximum of 34 m3

255

(mol quanta)-1. Previously, the highest AQY(330) reported in the literature was 2.5 m3 (mol

256

quanta)-1 in the Sargasso Sea.24 Therefore, our study extends the upper range of DMS photolysis

257

AQY by an order of magnitude, and indicates that AQY(330) between 3 and 30 m3 (mol quanta)-

258

1

are not exceptional. The maximal AQY(330) occurred within a narrow range in salinity and

259

temperature characteristic of the Antarctic Polar Front,49 suggesting a link with upwelling of

260

deep, potentially old50 water masses (SI Figure S2).

261 262

Coastal waters influenced by the Mackenzie26 and the Mississippi river outflows differed from

263

the general patterns in kp and AQY(330) seen in the open ocean, adding another dimension

264

(inshore-offshore) to the latitudinal gradients. A closer examination of these coastal transects

265

shows that kp increased by four- to tenfold toward the mouth of these major rivers compared to

266

nearby oceanic waters (e.g., compare "Kinsey_PLU" to "Kinsey_INT" and "Kinsey_GOM", and

267

"Taalba13MES" to "Taalba13CB" in Figure 1). This trend corresponded to a 40 to 70-fold


14

Page 15 of 40


268

increase in aCDOM(330) and a ca. fivefold decrease in AQY(330) with decreasing salinity, such

269

that AQY(330) was usually less than 0.2 m3 (mol quanta)-1 close to river mouths (Figure 2B

270

inset). Pooling together these two river-influenced coastal datasets, the linear correlation

271

coefficient between AQY(330) and salinity for salinities less than 31 was r = 0.78 (p = 3·10-6, n =

272

25). For comparison, the correlation between aCDOM(330) and salinity was r = –0.95 for the same

273

pooled datasets. Because of the opposite changes in aCDOM(330) and AQY(330), kp in river-

274

influenced waters were not significantly different, overall, from those in other coastal areas

275

(Table 1). In samples from coastal and continental shelf areas with limited or no riverine

276

influence, such as those from the coastal NW Mediterranean,38,45,51 NW Atlantic18 and

277

Antarctica,52 kp and AQY(330) were similar or slightly higher than those in temperate to subpolar

278

oceanic waters (excluding the Sub-Antarctic region).

279 280

3.2 AQY and aCDOM magnitude and spectral shape

281

When the full range of AQY(330) and aCDOM(330) was examined (Figure 2B), a highly

282

significant inverse relationship was found between log10AQY(330) and log10aCDOM(330),

283

according to the regression equation:

284 285

log10AQY(330)] = (–0.32±0.11) – (1.08±0.13)log10aCDOM(330)

(8)

286 287

with r2 = 0.70, p < 10-16, n = 111. The D3 subset was not included in the regression, since it was

288

derived from another regression involving aCDOM(330). At aCDOM(330) less than 0.4 m-1

289

characteristic of oceanic waters, the AQY(330) vs. aCDOM(330) relationship diverged into two

290

distinct trends, roughly defined by high- and low-latitude samples. This divergence was more


15


Page 16 of 40

291

marked at low aCDOM(330), with AQY(330) ranging between 1 and 34 m3 (mol quanta)-1 at

292

aCDOM(330) of 0.1 m-1.

293 294

AQY(330) from different study areas were grouped according to the value of spectroscopic

295

indicators of CDOM origin, molecular weight and photobleaching (Figure 3). AQY(330)

296

generally increased with SR and the aCDOM(254)/aCDOM(365) ratio and decreased with the

297

aCDOM(254)/DOC ratio. Low AQY(330) characteristic of river outflow areas coincided with the

298

lowest spectral slope ratios (SR < 2, Figure 3A), low to intermediate values of the

299

aCDOM(254)/aCDOM(365) ratio (3 to 15, Figure 3B) and intermediate to high aCDOM(254)/DOC

300

ratios (generally 0.02 to 0.07 m2 mmol-1, Figure 3C). Furthermore, AQY(330) in river-influenced

301

samples showed a monotonic increase with SR (rS = 0.75, p = 10-4, n = 21) and the aCDOM(254)/

302

aCDOM(365) ratio (rS = 0.74, p = 7 10-5), and a decrease with aCDOM(254)/DOC (rS = 0.72, p =

303

0.01). Overall, this suggests that the offshore increase in AQY(330) is related to photobleaching

304

of terrestrial CDOM, accompanied by a decrease in the mean CDOM molecular weight35 and

305

CDOM aromaticity.37

306 307

In oceanic waters, the relationship between AQY(330) and SR branched at SR > 2, setting apart

308

high-AQY Sub-Antarctic waters from other oceanic regions (Figure 3A). Samples from diverse

309

subtropical to subpolar seas clustered around an SR of 2, whereas samples with SR between 3.5

310

and 6.3 and AQY(330) smaller than 4 m3 (mol quanta)-1 originated mainly from the open

311

Mediterranean Sea and the subtropical gyres, with maximal SR in the Sargasso Sea (July-

312

August). Comparison between AQY(330) and the aCDOM(254)/aCDOM(365) ratio revealed a

313

similar branching pattern, with separation into two groups at aCDOM(254)/aCDOM(365) of ca. 10.


16

Page 17 of 40


314

The grouping of samples was, however, slightly different from that produced by SR. In this case,

315

the samples from the Indian Ocean subtropical gyre in late austral summer showed the highest

316

aCDOM(254)/aCDOM(365) ratio with values above 30. High SR and aCDOM(254)/aCDOM(365) have

317

both been interpreted as the signature of intense CDOM photobleaching at the sea surface, which

318

is overall consistent with the observed trends. The high SR (with a maximum of 5 at 60°S) seen in

319

high-AQY Sub-Antarctic samples in the austral spring is more intriguing, and perhaps related to

320

CDOM transformations other than photobleaching.53

321 322

Despite

the

general

positive

covariation

between

AQY(330)

and

either

SR

or

323

aCDOM(254)/aCDOM(365), both indices showed a local minimum in samples from the Greenland

324

Sea and the Antarctic Peninsula with AQY(330) between 1 and 10 m3 (mol quanta)-1. This

325

pattern might reflect the effect of higher local biological productivity on the CDOM pool, since

326

these samples exhibited a higher median chlorophyll concentration (0.69 µg L-1) than the other

327

oceanic samples (ca. 0.20 µg L-1). Yet, this interpretation remains speculative.

328 329

The inverse relationship between AQY and aCDOM at the 330 nm reference wavelength

330

extended across the spectral domain. The median AQY spectral slope (SAQY) in the Mackenzie

331

estuary and shelf (0.0275 nm-1, n = 14) was significantly smaller (p = 0.0002, Kruskal-Wallis

332

test) than in oceanic samples (0.0462 nm-1, n = 36). In fact, log10AQY(330) and SAQY displayed a

333

significant positive linear correlation in the D1 subset studies (r = 0.54, p = 2·10-4, n = 47; SI

334

Figure S3). This suggests that the AQY spectrum becomes steeper and thus shortwave-shifted as

335

it increases in magnitude.

336


17


Page 18 of 40

337

As described in Section 2.5, we estimated SAQY in some D2 samples that could then be

338

compared to D1 samples, and a good agreement between both subsets was found (SI section 3).

339

Interestingly, SAQY in samples from the Sub-Antarctic and Antarctic Peninsula region was

340

significantly higher than in the other oceanic samples (p = 2·10-5, Kruskal-Wallis test; pooled D1

341

and D2 subsets). The median of Sub-Antarctic and Antarctic Peninsula samples was 0.0535 nm-1

342

(n = 23), compared to 0.0389 nm-1 in the rest of oceanic samples (n = 29). These results indicate

343

that high AQY and steep AQY spectral slopes are characteristic of Antarctic waters and set them

344

apart from other oceanic regions including Arctic waters.

345 346

3.3 AQY and nitrate

347

The role of nitrate as a DMS photosensitizer was previously examined in two studies. In the

348

Northeast Pacific, AQY(330) increased with nitrate concentration at a rate of 0.15 m3 (mol

349

quanta)-1 µM-1 (n = 7; "Bouillon2004" dataset).21 In the Southern Ocean, kp increased linearly

350

with added nitrate which corresponded to a slope between 0.17 and 0.19 m3 (mol quanta)-1 µM-1

351

("Toole2004" dataset,22 SI section 4). In our dataset, we further assessed the role of nitrate by

352

regressing the residuals left after fitting AQY(330) to aCDOM(330) (eq 8) against nitrate

353

concentration (Figure S4A). This gave a regression slope of 0.17 ± 0.04 m3 (mol quanta)-1 µM-1

354

(r2 = 0.41, p < 10-12, n = 111), similar to the nitrate dependence calculated from the two previous

355

regional studies. The slope was not significantly different if river-influenced samples were

356

excluded.

357 358

Fig. 3D displays the relationship between AQY(330) and nitrate. Despite the visual scatter, a

359

highly significant correlation existed between the two variables (r = 0.55, p < 10-10; rS = 0.64, p

CDOM Sources and Photobleaching Control Quantum Yields for

Recommend Documents