Predicting Reactive Intermediate Quantum Yields ... - ACS Publications

Subscriber access provided by UB + Fachbibliothek Chemie | (FU-Bibliothekssystem)

Article

Assessing dissolved organic matter photo-reactivity in a subtropical wetland ecosystem: Correlations between optical properties, antioxidant capacity, and the photochemical formation of reactive intermediates Garrett McKay, Wenxi Huang, Cristina Romera-Castillo, Jenna Crouch, Fernando L. Rosario-Ortiz, and Rudolf Jaffe Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.6b06372 • Publication Date (Web): 09 Apr 2017 Downloaded from http://pubs.acs.org on April 10, 2017

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 41

Environmental Science & Technology

1

Predicting Reactive Intermediate Quantum Yields from

2

Dissolved Organic Matter Photolysis using Optical Properties

3

and Antioxidant Capacity

4 5 6 7 8 9 10 11 12

Garrett Mckay1, Wenxi Huang2, Cristina Romera-Castillo3,#, Jenna Crouch1, Fernando L. Rosario-Ortiz1 and Rudolf Jaffé2,3*

13 14 15

3) Southeast Environmental Research Center, Florida International University, Miami, Fl, USA

16

*Corresponding author: Rudolf Jaffé; Phone: 305.348.2456; Fax: 305.348.4096;

17

[email protected]

18 19 20 21 22

1) Department of Environmental Engineering, University of Colorado, Boulder, CO, USA. 2) Department of Chemistry and Biochemistry, Florida International University, Miami, Fl, USA

# Present address: Department of Limnology and Bio-Oceanography, University of Vienna, 1090 Vienna, Austria

23 24

ABSTRACT

25

The antioxidant capacity and formation of photochemically produced reactive

26

intermediates (RI) was studied for water samples collected from the Florida Everglades

27

with different spatial (marsh versus estuarine) and temporal (wet versus dry season)

28

characteristics. Measured RI included triplet excited states of dissolved organic matter 1

ACS Paragon Plus Environment


29

(3DOM*), singlet oxygen (1O2), and the hydroxyl radical (•OH). Single and multiple

30

linear regression modeling were performed using a broad range of extrinsic (to predict RI

31

formation rates, RRI) and intrinsic (to predict RI quantum yields, ΦRI) parameters.

32

Multiple linear regression models consistently led to better predictions of RRI and ΦRI for

33

our data set, but poor prediction of ΦRI for a previously published data set 1, probably

34

because the predictors are intercorrelated (Pearson’s r > 0.5). Single linear regression

35

models were built with data compiled from previously published studies (n≈120) in

36

which E2:E3, S, and ΦRI values were measured, which revealed a high degree of

37

similarity between RI-optical property relationships across DOM samples of diverse

38

sources. This study reveals that •OH formation is, in general, decoupled from 3DOM* and

39

1

40

ΦRI for 1O2 and 3DOM* correlated negatively with antioxidant activity (a surrogate for

41

electron donating capacity) for the collected samples, which is consistent with

42

intramolecular oxidation of DOM moieties by 3DOM*.

O2 formation, providing supporting evidence that 3DOM* is not a •OH precursor. Finally,

43

2


Page 2 of 41

Page 3 of 41


44

INTRODUCTION

45

Dissolved organic matter (DOM) plays an important biogeochemical role in

46

aquatic ecosystems 2. The structure of DOM is highly dependent on its source (e.g.,

47

terrestrial vs. aquatic) and history (e.g. microbial, photochemical processing)

48

that absorbs ultraviolet and visible light is classified as chromophoric dissolved organic

49

matter (CDOM)

50

waters and is a source and sink of reactive intermediates (RI)8-10, which likewise

51

influences several biogeochemical processes such as carbon mineralization, pathogen

52

inactivation, and trace metal speciation and cycling

53

has shown that light absorption by DOM can lead to photobleaching, production of

54

carbon monoxide and carbon dioxide, and formation of lower molecular weight organic

55

compounds 2,5,13,14.

6,7

3-5

. DOM

. This fraction determines the depth of the photic zone in natural

1,11-13

. In addition, previous research

56

Absorption of light by DOM in aquatic environments (λ > 290 nm) produces RI

57

including hydrogen peroxide (H2O2), singlet oxygen (1O2), excited triplet states (3DOM*),

58

superoxide radical anion (O2−•), and the hydroxyl radical (•OH)15-21. The formation

59

pathways, measurement, and importance of these RI has been discussed in detail in

60

several recent reviews22-25. The redox properties of DOM have also recently received

61

significant interest, as this may be a controlling factor of DOM’s ability to scavenge free

62

radicals, interact with metals, and inhibit photosensitized contaminant degradation

63

As such, electron accepting capacity (EAC) and electron donating capacity (EDC) have

64

been quantified for freshwater and soil humic substances using direct and mediated 3


26-30

.


29,31

65

electrochemical methods.

66

for EDC), have been reported for DOM of various origins 32.

Page 4 of 41

Similarly, the antioxidant capacities of DOM (a surrogate

67

Although it has been known for some time that RI are formed from DOM

68

photolysis, there is still limited information on how DOM quantity and quality impact RI

69

formation rates (RRI) and apparent quantum yields (ΦRI), respectively. In terms of DOM

70

quantity, Peterson et al. (2012) demonstrated that DOC concentration and the absorption

71

coefficient at 300 nm (a300) were good predictors of 1O2 steady state concentrations

72

([1O2]ss) for a notably large dataset collected from Lake Superior (USA) and inflowing

73

rivers

74

rates and steady state concentrations from surface water samples from the Florida

75

Everglades, and also observed that RI formation rates positively correlated with light

76

absorption, specifically the absorption coefficient at 254 nm (a254)

77

samples enriched in terrestrial DOM in contrast to microbial DOM Similar results

78

observed by Haag and Hoigné (1986),35 strongly suggest that the rate of RI formation is a

79

function of the amount of chromophoric dissolved organic matter (CDOM) (eq. 1).

33

. More recently, Timko et al. (2014) measured 3DOM*, 1O2, and •OH formation

34

, particularly for

80

RRI = ∑ Ra,λ Φ RI ,λ = [CDOM]∑ ka,λ Φ RI ,λ all λ all λ

81

In equation 1, Ra-DOM,λ and ka-DOM,λ are the rate of light absorption and the specific rate of

82

light absorption by DOM, respectively. Note that there is a distinction here between

83

DOM and CDOM in that CDOM represents a fraction of the total DOM concentration.

84

Although it is not feasible to determine the CDOM concentration exactly, this value is

85

often strongly correlated to DOM concentration. 4


(1)

Page 5 of 41


86

In terms of DOM quality, several reports have examined links between RI quantum

87

yields (ΦRI) and DOM optical and physicochemical properties

88

consistently shown that ΦRI (3DOM*, 1O2, and more recently •OH) positively correlate

89

with both spectral slope (S) and absorbance ratios (e.g., Abs250/Abs365 = E2:E3) 1,22,36,40-42

90

and fluorescence quantum yields42-44, all of which are surrogates for molecular weight

91

42,45

92

regression

93

these relationships, the most commonly explored being the relationship between Φ1O2 and

94

E2:E3. These optical properties vary with DOM molecular weight and inverse

95

correlations between ΦRI and DOM molecular weight have been consistently

96

demonstrated.39,42,45,46

5,35-39

. It has been

. Most research has focused on S or E2:E3 to model RI quantum yields using linear 1,10,36,37

. These models have similar results (i.e. regression coefficients) for

97

Besides S and E2:E3, few intrinsic variables have been explored in regression

98

analysis of RI data. Timko et al., reported a strong coupling between RI formation rates

99

and the abundance of terrestrial humic substances for a similar set of Everglades DOM

100

samples using excitation emission matrix (EEM) fluorescence measurements and parallel

101

factor analysis (PARAFAC) 34. These authors also reported linear relationships between

102

RI formation rates and steady state concentrations and DOM optical properties, but for a

103

rather small dataset. To our knowledge, most regression modeling of RI data thus far

104

have used a single regressor as opposed to a multivariate approach. An issue that further

105

hinders complete understanding of this topic is that many reports do not always provide

106

complete datasets in tabular form, which would facilitate broader modeling applications.

5



107

In this study, we examine the relationship between photochemical production of

108

RI and DOM physicochemical properties for whole water samples (n = 21) from the

109

Florida Everglades [Everglades National Park (ENP), Florida, USA]. The samples were

110

comprised of DOM of diverse sources and environments (see below). A diverse set of

111

DOM physicochemical properties were explored in linear regression analysis with

112

measured RRI and ΦRI. Additional regression analyses were done incorporating data from

113

previous studies in an attempt to compare and consolidate existing models. In an effort to

114

better predict ΦRI based on DOM chemical properties, antioxidant capacity was measured

115

for each sample using a recently developed assay

116

steadily growing knowledge base of the relationships between DOM optical properties

117

and RI production, this study provides one of the first analyses of relationships between

118

ΦRI and DOM antioxidant capacity.

32

. As well as contributing to the

119

120

EXPERIMENTAL METHODS

121

Chemicals. Furfuryl alcohol (FFA, 97+%, TCI America), phenol (99+%, Sigma Aldrich),

122

2,4,6-trimethylphenol (TMP, 99+% Alfa Aesar), benzene (99+%, Alfa Aesar), methanol

123

(HPLC grade, VWR), and phosphoric acid (85%, JT Baker) were used as received.

124

Varian Bond Elut PPL cartridges were purchased from Agilent Technologies. Suwannee

125

River NOM (1R101N) was purchased from the International Humic Substances Society,

126

dissolved at either ~5 or ~10 mgC L-1 in 10 mM pH 7.2 phosphate buffer, and filtered

127

through 0.7 µm muffled glass fiber filters (GF/F) prior to use. 6


Page 6 of 41

Page 7 of 41


128

Sampling. As the largest subtropical wetland in the United States, the Everglades not

129

only exhibits diverse vegetation, but is unique in its low concentrations of iron, phosphate,

130

nitrate, and nitrite 47, and exhibits high variability in DOM character 48,49

131

Two major drainages were sampled in ENP (Figure 1): Shark River Slough (SRS)

132

and the Taylor Slough (TS), which drain through mangrove estuaries into the Florida

133

Shelf and Florida Bay (FB) respectively, creating a salinity gradient along SRS and TS

134

transects

135

(between June and November) and dry seasons (December to May) with about 80% of

136

total annual precipitation occurring in the wet season. In the dry season, freshwater

137

inflows are reduced and estuarine residence time and salinity intrusions increase. These

138

altering processes are responsible for nutrient availability and water quality51 which

139

includes changes in DOM sources and residence times, with potential effects on its

140

reactivity. The Shark River Slough estuary is strongly connected to the Gulf of Mexico

141

and receives significant tidal influence 52.

50

. The seasonal climate variations in the Everglades represent clear wet

142

Samples were collected from long-term monitored FCE-LTER sites located in the

143

Shark River Slough (SR2, SR4, and SR6), Taylor Slough (TS2, TS3, and TS7) and

144

Florida Bay (FB21) in September, October, and November of 2014. Samples were

145

collected as surface water (maximum depth of 1 ft) in pre-rinsed (trace metal grade nitric

146

acid, base, then deionized water) polyethylene bottles and transported back to the lab on

147

ice. Samples were then filtered through muffled 0.7 µm filters (Gelman GF/F) and stored

148

in polyethylene bottles at 4 °C.

7



Page 8 of 41

149

DOM composition varies on spatial scales in the Everglades, in part due to source

150

material changes 48,49. Upstream SRS and TS, both freshwater marshes, are dominated by

151

sawgrass (Cladium), spikerush (Eleocharis) and periphyton. TS sites dry more frequently

152

and for a longer duration (short hydroperiod) compared to SRS sites (long hydroperiod)

153

and are thus characterized as marl soils rather than peat. Biomass in downstream TS and

154

SRS sites is dominated by mangroves

155

trees, lower rates of litterfall and root production

156

seagrass. Overall, these vegetation and hydrological patterns result in highly variable

157

DOM character on spatial and temporal scales.

158

Antioxidant activity measurement. DOM antioxidant activity was assessed by reacting

159

2,2-diphenyl-1-picrylhydrazyl (DPPH•) with DOM in methanol solution (eqs 2,3) 32,57,58.

53,54

, however, TS sites have significantly smaller 55,56

. FB biomass is dominated by

160

DPPH• + antioxidant → DPPH—H + A•

(2)

161

DPPH• + R• → DPPH—R

(3)

162

DPPH• has a characteristic absorbance at 515 nm that is reduced by reaction with an

163

antioxidant (AH) or a radical species (R•). This assay has been described in detail

164

previously

165

was carried out using PPL cartridges (methanol as eluent). The resulting methanolic

166

extracts were mixed with DPPH• reagent (70.9 µM), shaken, and left to react in the dark

167

at room temperature for 15 minutes. The loss in absorbance at 515 nm was measured

168

using a Varian Cary 50 spectrophotometer. The antioxidant activity or redox potential is

169

expressed as (eq 4)

57,58

and recently adapted to DOM

32

. Briefly, solid phase extraction (SPE)

8


Page 9 of 41


170

 A −A  A× A =  1− A C  ×100 AB  

171

where AA is the absorbance of the antioxidant sample at the end of the reaction time, AC

172

is the absorbance of the sample blank at the beginning of the reaction time (t=0), and AB

173

is the absorbance of the methanol blank at the beginning of the reaction time (t=0). The

174

antioxidant activity of a sample is reported as the A×A50 value, which is the antioxidant

175

activity for a 50 ppm methanol extracted sample based on the calibration curve (see 32). A

176

greater A×A50 value corresponds to a greater DOM antioxidant capacity.

(4)

177

It is important to note at this point that these values were measured on PPL

178

extracts, whereas ΦRI values were measured on whole water samples. Control

179

experiments performed to assess the validity of this approach showed that the comparison

180

is valid within the scope of this work (see Text S2 for detailed discussion).

181

DOC and optical property measurements. DOC concentrations were measured after

182

filtration (0.7 µm GF/F) with a Shimadzu TOC-5000 analyzer using high temperature

183

combustion. DOC concentrations of the PPL extracts were determined by drying an

184

aliquot (200 µL) of each methanol DOM extract with N2 gas and freeze drying under

185

vacuum for 2 hr to insure complete dryness of the samples. Once dried, all the samples

186

were re-dissolved with 10 mL Milli-Q water for DOC measurement.

187

DOM absorbance was measured at both labs participating in this study.

188

Absorbance and fluorescence spectra were measured at FIU for whole water samples

189

using an Aqualog spectrofluorometer (Horiba) with a 1 cm quartz cuvette. The excitation 9



Page 10 of 41

190

wavelength was scanned from 240 nm to 621 nm with an increment of 3 nm. The

191

emission wavelength was scanned from 241 nm to 622 nm in 1.5 nm steps. UV-Vis

192

spectra were collected from 240 nm to 621 nm with an increment of 3 nm simultaneously.

193

The optical parameters used in this study that derived from these measurements include

194

HIX (humification index)

195

Slope Ratio (SR) 5. Absorbance spectra at CU Boulder were measured in triplicate using a

196

Cary 100-Bio in a 1 cm quartz cuvette, scanned from 800 to 200 nm, and baseline

197

corrected to deionized water. The optical parameters used in this study that derived from

198

these measurements included absorption coefficients (Naperian, m-1) at 254 and 300 nm

199

(a254 and a300), E2:E3 (Abs250/Abs365), spectral slope (S300-600), and specific ultraviolet

200

absorbance at 254 nm (SUVA254) and 300 nm (SUVA300). S300-600 was calculated by

201

nonlinear regression over 300 to 600 nm using a single exponential model with λ = 350

202

nm as the reference wavelength. S is used to denote spectral slope in general and (when

203

provided) subscripts denote the wavelength range over which the regression was

204

performed.

205

Photochemical methods and regression modeling. The methods used here to measure

206

RI formation rates and calculate ΦRI values have been discussed in detail previously and

207

more information is provided in the Supplemental Information

208

effort was made to distinguish between free •OH and low energy hydroxylators in •OH

209

measurements64,65. Linear regression modeling was performed using the Statistics

210

Toolbox in Matlab (R2015a). Model terms and prediction intervals were evaluated at

211

2 α=5%. The 95% prediction interval (PI) and R2 predicted ( Rpred ) values were calculated

59-61

; absorption coefficient at 255 nm [a255 (cm⁻¹)], and the

10


38,62,63

. We note that no

Page 11 of 41


212

2 to assess uncertainty in predicting future observations. Rpred was chosen as a model

213

diagnostic over R2 because it penalizes for model over-fitting (especially important for

214

multiple linear regression). The Kolmogrov-Smirnov test was used to assess residual

215

distribution (studentized residuals were used). Normal probability plots of studentized

216

residuals and plots of studentized residuals versus effects were further evaluated for

217

normalcy.

218

RESULTS AND DISCUSSION

219

General observations and cluster analysis. Selected water quality data (DOC values,

220

absorption coefficients, and salinities) and RI data (formation rates and quantum yields)

221

are provided as supplementary information. Samples could generally be divided into

222

three groups based DOC concentrations and absorption coefficients. Low absorbing

223

waters with lower DOC concentrations were from the seagrass-dominated, marine,

224

Florida Bay site (FB21). Freshwater marsh samples (TS2, TS3, SRS2) and the upper

225

estuary samples (SRS4) comprised the second group, having intermediate to high

226

absorption coefficients and DOC values. SRS4 is at the interface of freshwatwer marshes

227

and estuarine, mangrove-influenced waters (Figure 1) and thus this sample’s DOM

228

character is a function of season (i.e. freshwater marsh and estuarine influenced in wet

229

and dry season, respectively). The third group consisted of two estuarine samples (SRS6

230

and TS7), which are associated with mangrove forests and significant tidal action (SRS6)

231

and saltwater intrusion (TS7 during dry season).

11



Page 12 of 41

232

To further identify relationships based on optical character and reactivity between

233

samples, hierarchical cluster analysis was performed using ΦRI values, antioxidant

234

activity (A×A50) values, and optical properties (E2:E3, a255, HIX). The results from these

235

analyses are shown in Figure S1, and as previously reported

236

was evident, likely driven by DOM source based on the sites’ soil type and plant cover

237

characteristics. The cluster clearly separated the dataset (using monthly samples; Figure

238

S1a) into sub-environment types as follows: Taylor River mangrove estuary site (TS7;

239

long inundation), Shark River mangrove estuary site (SRS6; tide influenced), Shark River

240

mangrove ecotone (SRS4), Taylor River freshwater marsh sites (TS2, TS3; short

241

hydroperiod), Shark River freshwater marsh site (SRS2; long hydroperiod) and marine

242

the marine Florida Bay site (FB21; seagrass dominated). Cluster analysis using averaged

243

monthly data was also performed (Figure S1b) with similar results, with the samples

244

grouping into similar sub-environments such as Florida Bay, fresh water marshes, and

245

mangrove-influenced sites. This suggests that Everglades DOM can be categorized based

246

on a combination of compositional features (optical properties) and photoreactivity (RI

247

data). A notable exception in the general cluster distribution however, was that the

248

November TS7 sample fell out from the general mangrove group (Figure S1a). The

249

salinity value for this sample was 25.3 practical salinity units (PSU), which is about three

250

times higher than the other two months at this site (average = 9.6 PSU) indicating that the

251

TS7 site was strongly influenced by Florida Bay water intrusions at that time. November

252

is the beginning of dry season in the Everglades, when reduced freshwater discharge at

253

this site has previously shown to lead to a shift from estuary DOM to marine DOM

12


48,49

, a clear spatial cluster

49

,

Page 13 of 41


254

explaining why the TS7 November sample had a low a₂₅₅ and DOC concentration, more

255

similar to Florida Bay samples.

256

These observations are consistent with previous results, in which the fringe

257

mangrove environment in ENP has been reported to have higher DOC concentrations and

258

highly aromatic DOM 49,66, leading to higher absorption coefficients for TS7 as compared

259

to TS2 and TS3 which are overlaying marl soils. Mangrove-influenced sites (SRS6 and

260

TS7) had consistently greater SUVA254 values than sites further upstream, whose DOM

261

input was dominated by peat soils and vegetative leaching, including significant

262

periphyton-derived inputs

263

greatest antioxidant capacity, likely a result of an increased proportion of polyphenols

264

such as tannins

265

analysis (including RI data) and previously reported clusters 48,49, (based on molecular or

266

optical properties only) for DOM in diverse Everglades sub-environments, suggesting

267

that this sample set is well suited to investigate the relationships between DOM source

268

and reactivity as reflected through RI production and antioxidant activity.

269

RI Formation

270

Regression models between RI formation rates and DOM quantity. RRI and ΦRI are

271

shown in the SI and fall within the range of previously measured values for each species

272

23

273

potential predictors. Significant positive correlations were observed between all RRI and

274

2 these predictors, although Rpred varied greatly (0.1 to 0.91). The positive correlations

32

49

. In addition, mangrove-influenced samples exhibited the

. It is interesting to see such a clear agreement between this cluster

. Figure S3 shows regression models for RRI with DOC concentration, a254, and a300 as

13



Page 14 of 41

275

observed between RRI and DOC concentration, a254, and a300, all of which describe the

276

quantity of DOM or CDOM, supports the hypothesis that DOM is the photosensitizer

277

producing these species. The linear relationship between ROH and DOC concentration

278

suggests that DOM (and not nitrite or nitrate) is the source of •OH in this system,

279

particularly when considering that nitrate and nitrite concentrations are known be low in

280

the Everglades

281

(Laboratory for Environmental and Geological Studies, CU Boulder). All samples were

282

below the detection limit for nitrite (0.01 mgNO2- L-1) and the highest nitrate concentration

283

was 3.5 mgNO3- L-1 (~0.056 mM). These levels are too low to be a competitive source of

284

•

285

2 degree of correlation with all regressors was observed for RTMP (0.1< Rpred

Predicting Reactive Intermediate Quantum Yields ... - ACS Publications

Recommend Documents