Subscriber access provided by UNIV OF NEBRASKA - LINCOLN
Article
Flux of total and methyl mercury to the northern Gulf of Mexico from U.S. estuaries Clifton Buck, Chad R. Hammerschmidt, Katlin Bowman, Gary A. Gill, and William M Landing Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.5b03538 • Publication Date (Web): 27 Oct 2015 Downloaded from http://pubs.acs.org on October 29, 2015
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 27
Environmental Science & Technology
1
Flux of total and methyl mercury to the northern Gulf of Mexico
2
from U.S. estuaries
3 4
Clifton S. Buck,†* Chad R. Hammerschmidt,‡ Katlin L. Bowman,‡ Gary A. Gill, § and William M.
5
Landingǁ
6 7
†Department of Marine Science, Skidaway Institute of Oceanography, University of Georgia, 10 Ocean
8
Science Circle, Savannah, Georgia 31411, United States
9
‡
Department of Earth & Environmental Sciences, Wright State University, 3640 Colonel Glenn Hwy.,
10
Dayton, Ohio 45435, United States
11
§Marine Science Laboratory, Pacific Northwest National Laboratory, Sequim, Washington 98382, United
12
States
13
ǁDepartment of Earth, Ocean, and Atmospheric Science, Florida State University, Tallahassee, Florida
14
32306, United States
15
*
Corresponding Author: email:
[email protected]; phone: 912-598-2418; fax: 912-598-2310
16 17
Abstract: To better understand the source of elevated methylmercury (MeHg) concentrations in Gulf of
18
Mexico (GOM) fish, we quantified fluxes of total Hg and MeHg from 11 rivers in the southeastern United
19
States, including the 10 largest rivers discharging to the GOM. Filtered water and suspended particles
20
were collected across estuarine salinity gradients in Spring and Fall 2012 to estimate fluxes from rivers to
21
estuaries and from estuaries to coastal waters. Fluxes of total Hg and MeHg from rivers to estuaries varied
22
as much as 100-fold among rivers. The Mississippi River accounted for 59% of the total Hg flux and 49%
23
of the fluvial MeHg flux into GOM estuaries. While some estuaries were sources of Hg, the combined
24
estimated fluxes of total Hg (5200 mol y-1) and MeHg (120 mol y-1) from the estuaries to the GOM were
25
less than those from rivers to estuaries, suggesting an overall estuarine sink. Fluxes of total Hg from the 1 ACS Paragon Plus Environment
Environmental Science & Technology
26
estuaries to coastal waters of the northern GOM are approximately an order of magnitude less than
27
atmospheric deposition rates. However, fluxes from rivers are significant sources of MeHg to estuaries
28
and coastal regions of the northern GOM.
29 30
2 ACS Paragon Plus Environment
Page 2 of 27
Page 3 of 27
31
Environmental Science & Technology
INTRODUCTION
32
Mercury (Hg) contamination of coastal marine ecosystems poses a potential health risk to humans.
33
Inorganic forms of Hg can be transformed to methylmercury (MeHg) in coastal waters and sediments1,2 as
34
well as in estuaries.3,4,5,6 Mercury contamination of the coastal ocean is concerning because consumption
35
of marine fish is the primary vector for human exposure to MeHg,7 and the majority of commercial
36
fisheries are in coastal waters.8 Methylmercury bioaccumulates in marine organisms and biomagnifies in
37
food webs often resulting in high concentrations in commercially valuable predatory fish (e.g., grouper,
38
mackerel, swordfish).9 Consumption of contaminated fish has resulted in as much as 30% of the U.S. Gulf
39
coast population having blood MeHg concentrations that increase risk of neurodevelopmental problems in
40
children.10
41
Fluvial inputs are often the principal source of Hg to estuaries,11,12,13,14 with much of the riverine Hg
42
derived from atmospheric deposition to the watershed.13 The relative significance of river discharge
43
sources versus direct atmospheric deposition decreases as the surface area of the receiving waters
44
increases, such that rivers are estimated to account for less than 10% of Hg delivered to the remote
45
continental margin and open ocean.2,15,16,17 Riverine inputs of Hg to the global ocean are estimated to be
46
20–60% that of atmospheric deposition with 90% of the river-derived Hg confined to sediments on
47
continental margins.18,19,20 However, very limited data exist for riverine Hg fluxes to the global ocean,18,21
48
especially the net flux through estuaries to the coastal zone including around the Gulf of Mexico
49
(GOM).22
50
We investigated the flux of total Hg (HgT) and MeHg from 11 U.S. rivers and estuaries to the
51
northern Gulf of Mexico (nGOM). The nGOM is an ideal study region because the estuaries vary with
52
regard to flow and morphology and include the Mississippi and Atchafalaya River systems, which are
53
estimated to deliver as much as 75% of the HgT load to the nGOM.23 While the nGOM coast receives the
54
highest rates of atmospheric Hg wet deposition in the U.S. (≥18 µg m-2 HgT),24 the riverine inputs are
55
hypothesized to dominate relative to atmospheric inputs in this area because of high rates of freshwater
56
discharge. The GOM is also an important seafood fishery, and by extension a potential source of MeHg 3 ACS Paragon Plus Environment
Environmental Science & Technology
57
exposure, accounting for 15% of U.S. marine commercial and about 40% of the marine recreational fish
58
catch in 2008.10
59
We sought to (1) quantify the riverine fluxes of HgT and MeHg to estuaries around the nGOM, (2)
60
determine whether, and to what extent, estuarine processes (e.g., scavenging, Hg methylation) alter
61
riverine fluxes to the Gulf, and (3) quantify net fluxes of HgT and MeHg from estuaries to coastal waters
62
of the nGOM. The results of this synoptic survey present a comprehensive set of Hg concentrations and
63
estimated fluxes from the largest estuarine systems in the nGOM. Our sampling approach was intended to
64
maximize spatial coverage, in lieu of temporal resolution, during the expected wet and dry seasons of
65
2012.
66 67
METHODS AND MATERIALS
68
Water sampling. Filtered water and suspended particles were sampled from 11 estuaries in the nGOM in
69
spring (May–June) and fall (September–October) of 2012 (Figure 1). Excluding the Escambia River,
70
these are the ten largest American rivers discharging to the Gulf. Surface water was sampled by peristaltic
71
pump from the freshwater end-member through eight additional locations along the salinity gradient
72
down the main channel to the seawater end-member of each estuary. Sampling was focused on the oligo-
73
and mesohaline reaches of each estuary to observe mixing behavior in the areas expected to be most
74
influenced by changes in salinity. Target salinities for sampling were about 0, 3, 6, 9, 12, 15, 20, 25 and
75
35. Duplicate samples were collected at salinities of about 0 and 12. Salinity was measured with a YSI
76
multi-parameter water quality sonde that was calibrated with standards traceable to the U.S. National
77
Institute of Standards and Technology (NIST). While this sampling approach disregards shorter time scale
78
fluctuations in Hg concentrations, the long-term water discharge records allow for the estimation of flux
79
to the coastal ocean with reasonable certainty.
80
An aluminum pole was used to extend acid-cleaned Teflon tubing into surface water (~1 m depth)
81
upstream of the boat, which was either anchored or motored slowly into the current. The Teflon tubing
82
was connected to acid-cleaned C-Flex tubing (Masterflex) in the pump roller, and water was pushed 4 ACS Paragon Plus Environment
Page 4 of 27
Page 5 of 27
Environmental Science & Technology
83
through Teflon tubing to a series of Teflon “tee” fittings that divided the flow between in-line glass fiber
84
filters (GF/F, for suspended particles) and 0.4 µm capsule filters (for filtered water). The GF/F filters (47
85
mm, 0.7 µm, Whatman 1825-047) were cleaned prior to use by heating for four hours at 550 ºC, and the
86
capsule filters (AcroPak-200, polyethersulfone membrane, Pall 12941) were cleaned with 0.1 M HCl.25
87
Samples for analysis of filtered HgT and MeHg were collected in pre-cleaned glass and polycarbonate
88
bottles, respectively.26 Samples were stored in the dark on ice and shipped within 24 hours of sampling to
89
Pacific Northwest National Laboratory (PNNL) and Wright State University (WSU), where they were
90
either analyzed immediately (filtered MeHg) or preserved by freezing (filters) or acidifying (filtered HgT).
91
Prior to and between sampling expeditions, all components of the pumping and filtering system were
92
cleaned in a class-1000 clean laboratory by first rinsing with ultrapure water (UHP; >18.2 MΩ-cm),
93
followed by immersion in 0.5 M HCl (sub-boiling quartz distilled) and rinsing with UHP water. All
94
sampling equipment was double bagged in plastic before leaving the clean lab.
95 96
Mercury Analyses. Filtered (< 0.4 µm) HgT in 0.5 L samples of water was measured at PNNL following
97
U.S. EPA Method 1631E. Briefly, this method uses cold-vapor atomic fluorescence spectrometry
98
(CVAFS) to measure Hg that has been oxidized to Hg(II) with BrCl and then reduced to Hg0. Elemental
99
Hg is purged from the sample and concentrated on Au, from which it is desorbed and quantified. The
100
detection limit for filtered HgT was 0.5 pM and the relative percent difference (RPD) of duplicate samples
101
ranged between 5 and 7%.
102
Filtered MeHg and particulate MeHg and HgT were extracted and quantified at WSU. Methylmercury
103
in filtered water was determined from an acidified 2 L sample (0.2 M H2SO4; >12 h), then neutralized
104
with KOH, buffered with acetate, and derivatized with sodium tetraethyl borate before purging,
105
concentrating on Tenax, and quantification by gas-chromatographic CVAFS.27,28 We suspect that nearly
106
all of the measured MeHg was as monomethyl mercury, as opposed to dimethyl mercury, although the
107
two species were not determined independently. Glass fiber filters containing particles filtered from 0.5 to
108
3.8 L of water were digested with 4.57 M HNO3 for 12 h in a water bath (60°C) prior to MeHg analysis 5 ACS Paragon Plus Environment
Environmental Science & Technology
109
by gas-chromatographic CVAFS).29 An aliquot of each digestate was oxidized with BrCl solution for >12
110
h prior to determination of HgT by dual-Au CVAFS.30,31,32 Method detection limits were 0.030 pM for
111
filtered MeHg, 0.002 pM for particulate MeHg and 0.02 pM for particulate HgT. Reproducibility between
112
duplicate samples averaged 15 ± 9 RPD for filtered MeHg, 13 ± 9 RPD for particulate MeHg, and 9 ± 8
113
RPD for particulate HgT. The following discussion will refer to the sum concentrations of the filtered plus
114
particulate fractions for both MeHg (ΣMeHg) and HgT (ΣHgT); however, an additional manuscript will
115
describe mercury partitioning as well as the role of other chemical parameters including total suspended
116
solids. The cumulative analytical uncertainty for the summed dissolved and particulate fractions was
117
calculated by standard propagation of errors techniques.
118 119
Flux Model. We estimated the flux of ΣMeHg and ΣHgT from rivers to estuaries (Friv) and from estuaries
120
to the coastal ocean (Fsea) based on measured Hg concentrations and river discharges. Flux estimates were
121
calculated for each season as well as for the year 2012. Seasonal Friv were estimated as the product of the
122
observed concentration (C0) at the zero salinity river end-member (or lowest salinity sample) and river
123
discharge (Driv; equation 1). In each river, the mean daily discharge (m3 d−1), over a 90 day period
124
bracketing each sampling event, was used to estimate daily Hg flux (mol d-1) from each system during the
125
spring and fall. River discharges were obtained from USGS gauges closest to the river/estuary confluence
126
(www.waterdata.usgs.gov/nwis) and the water flux was assumed to be constant downstream of the gauge
127
(Supplemental Table S1). The average daily discharge for 2012 was calculated from USGS discharge data
128
and used to calculate the estimated annual flux (mol yr-1) by multiplication with the mean of the two
129
seasonal C0 values. Additionally, long-term average discharge was determined from at least 10 years of
130
USGS flow measurements (Supplemental Table S1).
131
The Hg flux from the estuary to the coastal ocean (Fsea) was estimated from the product of Driv and
132
the modeled effective riverine end-member concentration (C*; equation 2). Seasonal and annual Hg Fsea
133
estimates were produced following the same procedure as for Friv. The C* concentration was estimated
134
from the observed concentration distribution through the estuary following an idealized conceptual model 6 ACS Paragon Plus Environment
Page 6 of 27
Page 7 of 27
Environmental Science & Technology
135
of non-conservative behavior (Figure S1).33 The model assumes that the Hg concentration in the river is
136
greater than that in the coastal ocean (Csea). Accordingly, a linear decrease in concentration between the
137
riverine and marine end-members is interpreted to demonstrate conservative mixing within the estuary.
138
Deviations in the distribution above the conservative mixing line are indicative of inputs within the
139
estuary (e.g. in situ production, benthic mobilization, sediment resuspension). Likewise, deviations below
140
the mixing line suggest removal of Hg (e.g. scavenging). Subtraction of Friv from Fsea yields Festuary
141
(Equation 3), which describes either net addition (positive flux) or removal (negative flux) within the
142
estuary.
143
Friv = C0 × Driv
Equation 1
144
Fsea = C* × Driv
Equation 2
145
Festuary = Fsea – Friv
Equation 3
146
Our field observations of ΣMeHg and ΣHgT rarely followed the idealized patterns of the conceptual
147
model, often with both addition and removal occurring in different reaches of the estuaries. Therefore,
148
three different effective river concentrations were estimated by plotting Hg concentration versus salinity
149
and extrapolating upstream from the high salinity end-member to the y-intercept (S = 0) to estimate
150
effective riverine concentrations identified as either C*max, C*mid, or C*min (Figure 2). The value for C*max
151
was estimated from a line drawn from the marine end-member concentration through the highest observed
152
concentration to the y-intercept. Estimates of C*mid and C*min were derived with the line passing through
153
either the majority of the data points (C*mid) or the sample with the lowest Hg concentration (C*min). In
154
cases of quasi-conservative behavior, C*mid was comparable to the observed C0 in the zero-salinity
155
sample. Both Friv and Fsea were calculated for each of the two sampling periods with the corresponding C0
156
and C*mid, respectively. Annual Friv and Fsea in each estuarine system were estimated as the product of
157
either the mean C0 or C*mid between the two sampling events and the annual river discharge in 2012.
158 159
RESULTS AND DISCUSSION
7 ACS Paragon Plus Environment
Environmental Science & Technology
160
Discharge. River discharge to the nGOM is influenced by natural variations in precipitation and
161
evaporation as well as flow control from dams. Springtime river discharge was significantly (15–90%)
162
greater than during the fall sampling (Wilcoxon Signed Rank Test, p = 0.02) with the exception of the
163
Suwannee River, because fall sampling occurred soon after Hurricane Isaac had passed through the area
164
(Supplemental Table 1). Compared to the long-term average river discharge from USGS records, 2012
165
was a year of relatively low discharge from nGOM rivers.
166 167
Riverine Concentration (C0) and Flux from Rivers to Estuaries. Riverine end-member concentrations
168
(C0) of ΣMeHg varied 20-fold among rivers and were greatest in the Escambia and Pascagoula Rivers
169
(Table 1). Measured concentrations of ΣMeHg were within the range reported in other North American
170
streams.12,34,35,36,37 In four nGOM rivers, C0 was greater during the fall sampling event than in the spring
171
(Suwannee, Mobile, Mississippi and Lower Atchafalaya), while three other systems (Apalachicola,
172
Sabine and Brazos) showed no large difference between sampling periods. The other three rivers
173
(Escambia, Pearl and Trinity) were either sampled only once or the ΣMeHg concentration was greater in
174
the spring (Pascagoula).
175
While concentrations of ΣMeHg differed between sampling campaigns, estimated river fluxes (Friv)
176
were remarkably consistent between spring and fall, in most cases (Table 2). The greatest temporal
177
difference of Friv was estimated for the Brazos River, which had a 10-fold higher flux in spring as a result
178
of greater spring discharge. However, ΣMeHg flux from the Brazos, and the two other Texas rivers
179
(Sabine and Trinity), was insignificant relative to the other rivers. The greatest Friv (~0.1 mol d−1) was
180
delivered from the Mississippi River because of its high flow. The flux of ΣMeHg from the Mobile River
181
(0.012 mol d−1) was about an order of magnitude less than the Mississippi and about 10-fold greater than
182
the nine other rivers. As with the river-specific Friv values, the cumulative flux of ΣMeHg to nGOM
183
estuaries from all rivers was similar between sampling events with input of 0.17 mol d−1 during spring
184
and 0.16 mol d−1 during fall (Table 2). In sum, the flux of river-sourced ΣMeHg to the estuaries was
185
estimated to be about 142 mol y−1, with 59% from the Mississippi River. The Mobile and Pascagoula 8 ACS Paragon Plus Environment
Page 8 of 27
Page 9 of 27
Environmental Science & Technology
186
Rivers were also substantial sources to their respective estuaries. The other rivers each discharged less
187
than 5 mol y−1.
188
Measured ΣHgT C0 were less variable between sampling campaigns than those of ΣMeHg (Table 3)
189
and were also within the range observed in other North American streams.12,34,35,36,37 Concentrations in the
190
Suwannee, Apalachicola, and Brazos Rivers were higher in the fall than during the spring and lower in
191
the Pascagoula and Mobile Rivers. The two highest C0 were in the Escambia (32.8 ± 0.9 pM) and Pearl
192
(30.8 ± 0.04 pM) Rivers, which were not sampled in the spring. The largest temporal difference was
193
observed in the Suwannee River, where the mean fall C0 (22.1 ± 1.7 pM) was approximately fivefold
194
greater than the single measurement taken in the spring (4.8 pM). Rainfall and watershed flushing
195
associated with Hurricane Isaac may have caused the concentration to be much higher in the fall. Mercury
196
concentrations in Hurricane Isaac rainfall were 36 pM at Chassahowitzka (28.7486° N, 82.555° W),
197
located south of the Suwannee watershed, and 25 pM at the Suwannee’s headwaters at Okefenokee
198
(30.7404° N, 82.1283° W; NADP, 2014). Alternatively, increased precipitation related to the hurricane
199
(12.4 cm of rain at Chassahowitzka) may have resulted in increased Hg mobilization from the Suwanee
200
watershed.38,39 The Friv values were greatest in the Mississippi and Lower Atchafalaya Rivers due to their
201
high flow compared to the other rivers (Table 4; Table S1). Differences in ΣHgT flux between sampling
202
periods from the Mississippi and Lower Atchafalaya were driven mostly by variability in freshwater
203
discharge, because their respective spring and fall C0 were similar. The annual Friv for the Mississippi was
204
3100 mol y-1 of ΣHgT, which was about half of the total estimated 6300 mol y−1 cumulative flux from the
205
rivers studied.
206 207
Effective Riverine Concentrations (C*) and Hg Fluxes to the Coastal nGOM. Mercury input and
208
removal processes within estuaries cause non-conservative behavior and complicate estimates of flux to
209
the coastal ocean. We observed non-conservative behavior of ΣMeHg and ΣHgT in most estuaries
210
(Figures S2 and S3), which resulted in large differences among the C*min, C*mid and C*max estimates
211
within some estuaries. Differences among C* estimates were about two to fivefold for ΣMeHg with 9 ACS Paragon Plus Environment
Environmental Science & Technology
212
comparatively less variability among ΣHgT C* estimates (Tables 1 and 3). However, the estuaries were
213
often characterized by complex non-conservative behavior with both input and removal processes
214
indicated in different reaches along the salinity gradient. Therefore, the Hg flux from each estuary to the
215
coastal ocean (Fsea) was estimated as the product of Driv and C*mid (Tables 2 and 4).
216
Despite differences in ΣMeHg C*mid, spring and fall Fsea showed differences on the order of only a
217
few mmol d−1 for most of the estuaries (Table 2). The Mississippi River was the most notable exception
218
exporting nearly 0.4 mol d−1 in the spring and about 0.1 mol d−1 in the fall. This fourfold difference was a
219
function of substantially greater discharge in the spring rather than a difference in C*mid (0.3 pM in spring
220
and 0.2 pM in fall). On an annual basis, the estuaries were a source of 116 mol ΣMeHg y−1 to the Gulf,
221
with the Mississippi accounting for 76% of the flux. The Lower Atchafalaya (12 mol y−1) delivered 10%
222
of the total riverine input, twice that of the Pascagoula at 5.6 mol y−1, the third largest source.
223
Most of the Fsea values for ΣHgT were less than 0.5 mol d−1 (Table 4). The Mississippi (12.1 and 5.4
224
mol d−1 in spring and fall, respectively) and Lower Atchafalaya (1.9 and 1.2 mol d−1) were exceptions.
225
The larger flux from the Mississippi during the spring was due to higher discharge. The sum of ΣHgT flux
226
was estimated to be about 5200 mol y−1, with 70% of the total from the Mississippi and 14% from the
227
Lower Atchafalaya River.
228 229
Estuarine Input and Removal. Estuaries are often assumed to be net sinks for ΣHgT because of internal
230
scavenging on particles while their role in the flux of ΣMeHg to the coastal ocean is more
231
ambiguous.5,12,13,40,41,42 Differences in ΣMeHg fluxes within the estuaries (Festuary) were relatively small
232
and often Friv and Fsea were nearly equivalent resulting in Festuary that were near zero. However, springtime
233
Festuary values in the Mississippi and Lower Atchafalaya Rivers were sufficiently large to suggest estuarine
234
inputs of ΣMeHg (Table 2). Mercury methylation within sediments has been observed in other GOM
235
estuaries with differing morphologies, including the mangroves of Florida Bay42 and the Shark Bay
236
estuary.43 By comparison, the Mississippi River is characterized by a swift current and deep channel. The
237
Lower Atchafalaya, created through diversion of 30% of the Mississippi River’s flow to join the Red 10 ACS Paragon Plus Environment
Page 10 of 27
Page 11 of 27
Environmental Science & Technology
238
River, passes through the largest contiguous swamp in the U.S. before becoming an estuary.44 Except for
239
the Mississippi and Lower Atchafalaya, the studied estuaries were net sinks of ΣMeHg. Overall, nGOM
240
estuaries removed 26 mol of ΣMeHg annually, or about 20% of the riverine input.
241
Values of Festuary for ΣHgT were more variable between sampling periods than for ΣMeHg, and in the
242
Pascagoula and Lower Atchafalaya River estuaries the result was a transition between net input and
243
removal (Table 4). Despite a net input of 500 mol y−1 in the Mississippi River estuary, the nGOM
244
estuaries collectively removed over 1000 mol y−1, which is about 20% of the total annual riverine flux.
245
The most significant sinks for ΣHgT were the Lower Atchafalaya and Mobile River estuaries. Both of
246
these systems are relatively wide estuaries with slow current, allowing for significant sedimentation and
247
Hg scavenging. However, these observations should be interpreted with caution because each system was
248
sampled only once during each season. Higher frequency sampling is needed to better describe the impact
249
of estuarine processes.
250 251
Significance of Fluvial Sources. Mercury in the GOM is derived from three primary sources: the North
252
Atlantic Ocean, atmospheric deposition, and rivers. Atlantic Ocean water arriving through the Yucatan
253
Channel via the Loop Current is the largest estimated source delivering an estimated 78–88% of HgT
254
loadings assuming a ΣHgT concentration of 2.2 pM;16 however, recent measurements suggest that ΣHgT
255
concentrations in North Atlantic surface waters are