Drivers of pH Variability in Coastal Ecosystems - ACS Publications

Mar 20, 2019 - Lewis, Buchanan, Jhaveri, Sullivan, Kelly, Das, Taub, and Keoleian. 0 (0),. Abstract: A large portion of life cycle transportation impa...
0 downloads 0 Views 1MB Size
Subscriber access provided by UNIVERSITY OF ADELAIDE LIBRARIES

Critical Review

Drivers of pH variability in coastal ecosystems Jacob Carstensen, and Carlos M. Duarte Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.8b03655 • Publication Date (Web): 20 Mar 2019 Downloaded from http://pubs.acs.org on March 23, 2019

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 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 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.

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

Critical Review

Drivers of pH variability in coastal ecosystems

1 2 3

Jacob Carstensena,* and Carlos M. Duarteb,c

4 5

Aarhus University, Department of Bioscience, Frederiksborgvej 399, DK-4000 Roskilde,

6

a

7

Denmark

8

b

9

(RSRC), Thuwal, 23955-6900, Saudi Arabia

King Abdullah University of Science and Technology (KAUST), Red Sea Research Center Arctic Research Centre, Department of Bioscience, Aarhus University, C.F. Møllers Allé 8, DK-

10

c

11

8000 Århus C, Denmark

12 13 14

*

Corresponding author: [email protected]

15 16 17

1 ACS Paragon Plus Environment

Environmental Science & Technology

Page 2 of 41

Critical Review 18

ABSTRACT

19 20

A synthesis of long-term changes in pH of coastal ecosystems shows that, in contrast to the

21

uniform trends of open-ocean acidification (-0.0004 to -0.0026 pH units yr-1) driven by increased

22

atmospheric CO2, coastal ecosystems display a much broader range of trends (-0.023 to 0.023

23

pH units yr-1) and are as likely to show long-term increase as decline in pH. The majority of the

24

83 investigated coastal ecosystems displayed non-linear trends, with seasonal and interannual

25

variations exceeding 1 pH unit for some sites. The high pH variability of coastal ecosystems is

26

primarily driven by inputs from land. These include freshwater inputs that typically dilute the

27

alkalinity of seawater thereby resulting in reduced buffering, nutrients enhancing productivity

28

and pH, as well as organic matter supporting excess respiration driving acidification. For some

29

coastal ecosystems, upwelling of nutrient-rich and corrosive water may also contribute to

30

variability in pH. Metabolic control of pH was the main factor governing variability for the majority

31

of coastal sites, displaying larger variations in coastal ecosystems with low alkalinity buffering.

32

pH variability was particularly pronounced in coastal ecosystems with strong decoupling of

33

production and respiration processes, seasonally or through stratification. Our analysis

34

demonstrate that coastal pH can be managed by controlling inputs of nutrients, organic matter,

35

and alkalinity. In well-mixed coastal waters, increasing productivity can improve resistance to

2 ACS Paragon Plus Environment

Page 3 of 41

Environmental Science & Technology

Critical Review 36

ocean acidification, whereas increasing productivity enhances acidification in bottom waters of

37

stratified coastal ecosystems. Environmental management should consider the balance

38

between the negative consequences of eutrophication versus those of acidification, to maintain

39

biodiversity and ecosystem services of our coastal ecosystems.

40

3 ACS Paragon Plus Environment

Environmental Science & Technology

Page 4 of 41

Critical Review 41

INTRODUCTION There is growing concern that ocean acidification by anthropogenic CO2, clearly reflected

42 43

in long-term trends toward a decline by 0.02 pH units decade-1 in open-ocean waters1–4, may

44

significantly impact marine calcifiers, such as corals and mollusks1,5–7. This has led to increased

45

attention to coastal ecosystems, where many of the vulnerable calcifiers inhabit. Coastal

46

ecosystems have been shown to display complex and diverse patterns of pH change over time8–

47

12,

48

open-ocean surface waters1–4. Indeed, changes in pH in coastal ecosystems appear

49

idiosyncratic, displaying a diversity of patterns11,13.

50

in contrast to the more monotonic, uniform decrease in pH in response to increased CO2 in

Efforts to understand the regulation of pH, including anthropogenic impacts, on coastal

51

waters have revealed a complex set of interacting processes driving pH changes8,11,12,14,15. Initial

52

attempts to understand variability in coastal pH extended concepts and models developed for

53

the open ocean, dominated by the expected ocean acidification trends derived from

54

anthropogenic CO2 emissions and mixing between the marine and freshwater end-members

55

controlling carbonate chemistry in coastal ecosystems16. However, examination of the control of

56

coastal pH has identified the involvement of a variety of additional drivers, including watershed

57

and biological processes, which support a variety of metabolically intense communities8–12,14,15.

4 ACS Paragon Plus Environment

Page 5 of 41

Environmental Science & Technology

Critical Review 58

In particular, coastal eutrophication has been shown to exert a strong control on patterns of pH

59

change9,10,17–19 consistent with the strong nutrient control of biological activity in coastal

60

ecosystems20–23. Because the timing and strength of management of nutrient inputs to coastal

61

ecosystems differ across the world24, we expect coastal ecosystems to exhibit a diversity of

62

patterns of pH variability. However, this expectation derives largely from reports documenting

63

pH changes in individual coastal ecosystems, which have not allowed the detection of patterns

64

of changes in pH, either in terms of the repertoire of seasonal and long-term trends exhibited, or

65

the relationships with potential drivers, such as inputs of nutrients, levels of chlorophyll a,

66

salinity or alkalinity buffering. We argue here that a critical review of the patterns of pH change

67

in coastal ecosystems provides an improved opportunity to resolve the contributions of different

68

factors, including the role of eutrophication processes, in driving seasonal and long-term

69

changes in pH. In turn, understanding drivers of change in coastal pH may allow interventions,

70

at the watershed scale, to reach desirable target pHs25, thereby helping protect coastal

71

ecosystems.

72

Here we critically review patterns of long-term pH change in coastal ecosystems

73

across the world, based on a synthesis of available data (ranging from 6 to 67 years), to (1)

74

characterize the patterns and rates of change at seasonal and interannual time scales, (2)

5 ACS Paragon Plus Environment

Environmental Science & Technology

Page 6 of 41

Critical Review 75

assess the role of mixing between freshwater and marine waters that characterize these

76

ecosystems, and (3) quantify the effect of eutrophication and changes in phytoplankton

77

abundance on seasonal and decadal patterns of change in pH in coastal ecosystems.

78

6 ACS Paragon Plus Environment

Page 7 of 41

Environmental Science & Technology

Critical Review 79

VARIATIONS IN COASTAL pH

80

Records of long-term changes in pH in surface coastal waters (0-10 m) are available from public

81

databases from a number of sources (Table 1), representing 83 different coastal ecosystems in

82

11 regions of the world (Fig. S1), and a total number of 147,042, 186,956, and 148,731 samples

83

for pH, salinity, and Chla, respectively (Table 1). These datasets allow the characterization of

84

the long-term changes (>10 year with the exception of two French sites) in pH, along with

85

relevant ecosystem properties, such as salinity, temperature, chlorophyll a (Chla), total nitrogen

86

(TN) and total phosphorus (TP). Total alkalinity (TA) is an important property affecting pH,

87

allowing coastal ecosystems to be classified as weakly or strongly buffered, depending on

88

whether the TA for the freshwater entering the ecosystems, which was reported for 47 of the

89

coastal ecosystems for which time series are available, was below or above 1.2 mmol kg-1 ,

90

(~half ocean TA), respectively, while the degree of buffering was estimated from the literature

91

for the remaining coastal ecosystems (see Table S1 for details on classification).

92

The pH in surface waters of coastal ecosystems can be characterized through their overall

93

mean levels, as well as seasonal and interannual variations. However, comparing pH across

94

coastal ecosystem can be cumbersome due to differences in procedures and pH scales used.

95

Coastal and estuarine monitoring programs generally use glass electrodes for measuring pH

7 ACS Paragon Plus Environment

Environmental Science & Technology

Page 8 of 41

Critical Review 96

that are calibrated on a series of low ionic strength buffer solutions available from the US

97

National Bureau of Standards (NBS). The NBS scale was employed by all monitoring authorities

98

in the long-term data sets available, except for France where the seawater scale (SWS) was

99

employed (1997-2016; Table 1). Laboratory measurements of pH cannot be readily compared,

100

as temperature and hence pH may have shifted from in-situ conditions. Although the pH scale of

101

older monitoring data cannot always be retrieved from data sources or literature, they were

102

assumed reported on the NBS scale, i.e. maintaining the same analysis and reporting scale

103

over time. In order to characterize pH trends across ecosystems, pH was converted to the total

104

scale, using salinity-dependent conversion of pH on the NBS scale26 and assuming pH on SWS

105

and total scales to be equal. pH measured on the NBS scale expresses an activity as opposed

106

to total, free and SWS scales representing the hydrogen ion concentration. Since hydrogen ion

107

activity varies with salinity, comparison of pH levels in relation to changes in atmospheric CO2

108

and metabolic processes across coastal ecosystems covering a broad span in salinity is more

109

meaningful on the total scale. Although potentiometric measurements of pH are inherently less

110

precise (±0.1; ref. 11), the uncertainty introduced by this measurement error on mean values is

111

small relative to seasonal and interannual variations, as mean values were typically calculated

112

from tens or hundreds of observations.

8 ACS Paragon Plus Environment

Page 9 of 41

Environmental Science & Technology

Critical Review 113

Variations in the water quality variables were partitioned into spatial variation between stations

114

within coastal ecosystem, seasonal variation using monthly means and trends using annual

115

means with a general linear model (GLM)27. This decomposition approach is based on a

116

parametric decomposition of spatial and temporal variations, which is particularly useful for

117

irregularly sampled monitoring data. Chla, TN and TP were log-transformed prior to this

118

decomposition because variations typically scaled with the means, whereas pH, salinity and

119

temperature were not transformed. This implies that sources of variations were additive for

120

salinity and temperature, and multiplicative for Chla, TN, and TP as well as pH that represents a

121

log-scale. Means for the log-transform of Chla, TN, and TP were subsequently back-

122

transformed as geometric means using the exponential function. For comparative purposes,

123

long-term records of pH in oceanic waters were also retrieved from existing databases, the

124

Hawaii Ocean Time-series (HOT; http://hahana.soest.hawaii.edu/hot/) and the Bermuda Atlantic

125

Time Series (BATS; http://bats.bios.edu/), with the corresponding long-term slopes of

126

decreasing pH and their SE (Table S2). Ocean acidification (OA) is subsequently characterized

127

by the mean slope of these time series of -0.0018 yr-1 (Table S2).

128

9 ACS Paragon Plus Environment

Environmental Science & Technology

Page 10 of 41

Critical Review 129

Interannual variations in pH were further investigated using a Generalized Additive Model

130

(GAM) containing a parametric component (linear trend with slope β) and a non-parametric

131

component (spline S()) to describe deviations from the ocean acidification (OA) by subtraction of

132

0.0018 yr-1. The pH trends of the 83 ecosystems were partitioned into four categories: (1) no

133

deviation from OA (β and S() are both non-significant), (2) linear trend with slope different from

134

OA (β is significant and S() is non-significant), (3) non-linear trend with slope equal to OA (β is

135

non-significant and S() is significant), and (4) non-linear trend with slope different from OA (β

136

and S() are both significant). Furthermore, we tested if the magnitude of trends, given by the

137

absolute value of β, as well as seasonal and interannual ranges in pH were different for weakly

138

versus strongly buffered systems by analysis of variance.

139 140

pH time series from 83 coastal ecosystems across the world (Fig. 1; Fig. S3) demonstrated a

141

broad span of trends with slopes ranging from -0.023 to 0.023 yr-1, i.e. trends up to one order of

142

magnitude larger than ocean acidification (OA) and 46 sites (~55%) experiencing declining pH

143

(Fig. 2). Coastal systems with weak buffering (Table S1) did not show larger trends than

144

stronger buffered systems (F1,81=0.51; p=0.4793). In fact, only 16 sites exhibited changes over

145

time similar to the OA trend (Table 2), whereas most sites (1) experienced trends deviating

10 ACS Paragon Plus Environment

Page 11 of 41

Environmental Science & Technology

Critical Review 146

significantly from OA by different slopes (13 sites), (2) multiannual to decadal departures from a

147

linear trend similar to OA (20 sites), or (3) multiannual to decadal departures from a linear trend

148

different from OA (34 sites). This diversity of trend patterns was not specific to any particular

149

region but was observed across all regions. Coastal pH was decreasing significantly faster than

150

OA for 21 sites, whereas changes over time in pH were significantly larger than OA for 26 sites

151

(Table 2).

152 153

The coastal systems displayed large seasonal and interannual variations in pH, typically up to 1

154

pH unit, although seasonal and interannual variability could be as high as 1.4 and 1.6,

155

respectively. The predominant drivers of pH variations were investigated at different scales:

156

spatial, seasonal, and interannual. Drivers of seasonal and interannual variation were

157

investigated by calculating the range among monthly and annual means for each ecosystem

158

and examining their relationships to salinity and Chla seasonal and interannual ranges. We

159

used range as a more informative statistic of variability although it increases with the number of

160

observations, but this bias was small given the lengths of time series (Fig. S2). We found that

161

temporal variations in coastal pH far exceeded those observed in open ocean environments

11 ACS Paragon Plus Environment

Environmental Science & Technology

Page 12 of 41

Critical Review 162

(Table S2, Fig. 2) and were more closely coupled to variability of chlorophyll a (Chla) than

163

salinity (Fig. 3).

164

Largest seasonal ranges in pH were found at high-latitude sites in the Baltic Sea and

165

Danish Straits, whereas pH in low- and mid-latitude sites typically varied by less than 0.4 over

166

the year. Nevertheless, at the low- and mid-latitude sites pH seasonal variability was still

167

positively correlated with variability in Chla (Fig. 3B; r=0.48; p=0.0002 for log-transform of Chla).

168

Similar pH associations with salinity were not observed in general, although pH seasonal

169

variability in WA estuaries was significantly linked to salinity variations (Fig. 3A; r=0.79;

170

p=0.0350). Seasonal ranges in pH were significantly higher for coastal systems with weak

171

buffering (F1,81=10.39; p=0.0018), although one high-buffered coastal ecosystem, the

172

permanently stratified Mariager Fjord in the Danish Straits, deviated from this general pattern

173

with ~1 pH unit seasonal range despite high total alkalinity (TA).

174

Interannual variability exceeded 1 pH unit for two sites in the Baltic Sea and one site in

175

Neuse River Estuary. Variations in pH increased with salinity variability (Fig. 3C), but the

176

relationship to Chla was even stronger (Fig. 3D). However, coastal systems with large

177

interannual variability in salinity also had large variability in Chla (r=0.56; p1.2 meq kg-1

203

(strong buffering) had modest variations in pH across the salinity gradient, demonstrating that

204

pH variability due to mixing would be less than 0.3. Seasonal variability in temperature among

205

sites (temperature ranges from 6.3 to 24 °C) could almost account for similar pH variability over

206

the season, although for some coastal systems (Baltic Sea, Danish Straits, and Hong Kong; Fig.

207

S4A) amplifying salinity-induced pH variability, while balancing the effect of salinity for other

208

systems (French Atlantic coast, North Sea, Texas estuaries, and WA estuaries; Fig. S4B).

209

Low pH mean values were almost exclusively found for coastal systems with weak

210

buffering (Fig. 4B). Coastal systems within regions having similar end-members (e.g. sites along

211

the mainstem of Chesapeake Bay) underlined that mixing overall controls the mean pH level.

212

However, the salinity gradients were less pronounced for regions with high alkalinity buffering.

14 ACS Paragon Plus Environment

Page 15 of 41

Environmental Science & Technology

Critical Review 213

Importantly, mean levels of Chla did not explain variation in pH among coastal systems (Fig.

214

4D).

215

Our simulations also showed that imbalance between production and respiration could

216

change pH substantially, particularly in low-salinity coastal systems with low alkalinity buffering

217

(Fig. 4C). Whereas the effect of changing net DIC uptake was minor for well-buffered systems

218

at salinities above ~10 and low-buffered systems at salinities above ~20, drastic changes in pH

219

could be observed for small DIC changes in systems with low salinity and low buffering.

220

Although the position and magnitude of the sharp increase in pH varied with the TA and pH

221

values employed for the end-members, the simulated changes in pH could clearly encapsulate

222

those observed on both seasonal and interannual scales (Fig. 3).

223 224

INSIGHTS INTO THE PATTERNS AND DRIVERS OF LONG-TERM CHANGE IN pH IN

225

COASTAL ECOSYSTEMS

226

The analysis of long-term changes in pH across 83 coastal ecosystems shows that

227

these ecosystems present a much broader range (~ 20 times larger) of rates of change in pH

228

than the open ocean does. In contrast to oceanic time-series, which show long-term trends of

229

decline in pH ranging from -0.0004 to -0.0026 pH units yr-1 (Table S2, Fig. 2), long-term

15 ACS Paragon Plus Environment

Environmental Science & Technology

Page 16 of 41

Critical Review 230

changes in pH in coastal systems range from -0.023 to 0.023 yr-1, with 65% of the ecosystems

231

showing no monotonic trends, and 45% of the ecosystems studied showing positive trends, in

232

contrast of the negative trends expected from CO2 emissions. The modal trajectory, i.e. that

233

displayed by the largest number of coastal ecosystems, was one of broad oscillations around a

234

long-term trend in pH departing from that expected from open-ocean ocean acidification.

235

Provided the broad range of behaviors, we submit that, unlike the case for the open

236

ocean, an average rate of pH change across coastal ecosystems is not representative of the

237

expected behavior of ecosystems encompassed by the data set assembled here. Indeed, there

238

is considerably variability in pH trends not only among coastal ecosystems in widely distributed

239

regions but also within regions. Yet, the pH trajectories of coastal ecosystems are not entirely

240

idiosyncratic, as some general patterns emerged from our analyses that help develop

241

expectations for other coastal ecosystems not included here, and informs about controlling

242

mechanisms and intervention options. Indeed, understanding the drivers of variability in coastal

243

pH is of fundamental importance, as these ecosystems are inhabited by many of the calcifiers

244

(e.g. mollusks, crustaceans, echinoderms, corals) that supply fundamental services30, and are

245

vulnerable to ocean acidification1,6.

16 ACS Paragon Plus Environment

Page 17 of 41

Environmental Science & Technology

Critical Review 246

A previous conceptual model of pH regulation in coastal ecosystems considered three

247

components: (1) an oceanic end-member, in which variability is largely affected by air-sea CO2

248

exchange characterized by a monotonic decline by about 0.02 pH unit decade-1 due to ocean

249

acidification, (2) a freshwater end-member, often, but not always, characterized by low pH and

250

CO2 supersaturation31–34, and (3) biological metabolism, either raising pH for autotrophic

251

ecosystems (production > respiration) or decreasing pH for heterotrophic ones (production