Page 1 of 32
1 2 3
Environmental Science & Technology
Dynamic coupling of iron, manganese and phosphorus behavior in water and sediment of shallow ice-covered eutrophic lakes (submitted ES&T as a research article)
4 5
Andrew W. Schroth*1,2, Courtney D. Giles2, Peter D.F. Isles2,3, Yaoyang Xu2, Zachary Perzan4, Gregory K. Druschel5
6 7
1 Department of Geology, University of Vermont, Delehanty Hall, 180 Colchester Ave. Burlington, Vermont 05405
8 9
2. Vermont EPSCoR, University of Vermont, Cook Physical Science Building, 82 University Place, Burlington, Vermont 05405
10 11
3 Rubenstein School of Environment and Natural Resources, University of Vermont, Aiken Center, 81 Carrigan Drive, Burlington, Vermont 05405
12 13
4 Department of Geology, Middlebury College, McCardell Bicentennial Hall, 276 Bicentennial Way , Middlebury College, Middlebury, VT 05753
14 15
5. Department of Earth Sciences, Indiana University Purdue University, Indianapolis, 723 W. Michigan Street, SL118, Indianapolis, Indiana 46202
16
* Corresponding author: Phone: (603) 252 6551 email:
[email protected] Fax: (802) 656 3131
17
18
19
20
21
22
23
24
25
1 ACS Paragon Plus Environment
Environmental Science & Technology
26
Abstract: Decreasing duration and occurrence of northern hemisphere ice cover due to recent
27
climate warming is well-documented; however, biogeochemical dynamics underneath the ice are
28
poorly-understood. We couple time-series analyses of water column and sediment water
29
interface (SWI) geochemistry with hydrodynamic data to develop a holistic model of iron (Fe),
30
manganese (Mn), and phosphorus (P) behavior underneath the ice of a shallow eutrophic
31
freshwater bay. During periods of persistent subfreezing temperatures, a highly reactive pool of
32
dissolved and colloidal Fe, Mn and P develops over time in surface sediments and bottom waters
33
due to reductive dissolution of Fe/Mn(oxy)hydroxides below the SWI. Redox dynamics are
34
driven by benthic O2 consumption, limited air-water exchange of oxygen due to ice cover, and
35
minimal circulation. During thaw events, the concentration, distribution and size partitioning of
36
all species changes, with the highest concentrations of P and ‘truly dissolved’ Fe near the water
37
column surface, and a relatively well-mixed ‘truly dissolved’ Mn and ‘colloidal’ Fe profile due
38
to the influx of geochemically distinct river water and increased circulation. The partitioning
39
and flux of trace metals and phosphorus beneath the ice is dynamic, and heavily influenced by
40
climate-dependent physical processes that vary in both time and space.
41
Introduction: One of the most clear and consistent harbingers of climate change in the
42
Anthropocene has been a gradual reduction in the frequency and duration of lake wintertime ice
43
cover in the northern hemisphere 1-3. In the great lakes, increases in water temperature above
44
those predicted by observed changes in air temperature have confirmed this process and its
45
physical impact4. Using a space-for-time substitution approach that compares otherwise similar
46
Canadian (consistent winter freeze) and Danish (lacking ice cover) lake systems, Jackson et al.
47
2007 suggested that the duration of ice cover has profound impacts on nutrient loading, food web
48
structure, and water column anoxia5. Sediment water interface (SWI) redox chemistry, especially 2 ACS Paragon Plus Environment
Page 2 of 32
Page 3 of 32
Environmental Science & Technology
49
in shallow systems, is linked to overlying water column anoxia, through competitive processes of
50
oxygenation (via O2 entrainment and photosynthesis) v. reductive processes (including
51
O2/alternate electron acceptor respiration)6. Ice cover adds a physical barrier to the rate of O2
52
entrainment, both due to a change in the diffusivity of O2 through ice as compared to water, and
53
due to effectively eliminating wave action as a driver of enhanced O2 entrainment rates 7, 8, both
54
of which should have a profound impact on the position and structure of the redoxcline and
55
related biogeochemical cycling. However, the role of ice cover in lake ecosystem productivity or
56
biogeochemical dynamics remains poorly understood, particularly in shallow systems where
57
impacts on SWI geochemistry are likely most pronounced9. Given that the duration and extent
58
of regional ice cover is predicted to decrease over time as climate continues to warm, it is critical
59
to understand biogeochemical dynamics underneath the ice, and how under ice processes supply
60
or sequester nutrients and pollutants in lake ecosystems.
61
In shallow lakes or bays, an important driver of P dynamics is thought to be the behavior of
62
secondary amorphous Fe and Mn (oxy)hydroxides that are concentrated as nanoparticles at the
63
SWI6, 10, 11. These nanoparticle phases are the product of dissolved Fe2+ and Mn2+ oxidation; Fe2+
64
oxidation by O2 at circumneutral pH is very fast whereas Mn2+ oxidation is slower and involves
65
important Mn3+ intermediates12. The high surface area and positive surface charge of these Fe
66
and Mn (oxy)hydroxide phases (pHzpc ~9) allow them to be highly effective scavengers of
67
orthophosphate and other oxyanions under oxidizing conditions at circumneutral pH typical of
68
many lake systems13. However, these phases are highly susceptible to reductive dissolution under
69
hypoxic and anoxic conditions 14, 15, which are largely driven by biotic processes such as
70
heterotrophic respiration by iron or manganese reducing bacteria, which couple organic molecule
71
oxidation with the reduction of Fe and Mn (oxy)hydroxides16 upon consumption of oxygen and 3 ACS Paragon Plus Environment
Environmental Science & Technology
72
nitrate17. As such, when strong redox gradients are established at the SWI, reduction of
73
nanoparticulate Fe and Mn (oxy)hydroxides bearing significant sorbed P species can result a
74
significant flux of this macronutrient from the sediment into the water column 6, 18. In the
75
summertime, when water column temperatures are sufficient to support plankton blooms,
76
reductive dissolution of these phases is thought to initiate harmful algal bloom development and
77
propagation in many shallow P-limited systems19 20. While this process has been documented in
78
a few shallow systems during periods of high productivity, it is unclear to what extent it also
79
occurs under ice cover, when benthic microbial communities remain active, but oxygen diffusion
80
and hydrodynamic mixing may be highly limited. Thus, the continued occurrence of oxygen-
81
demanding processes, coupled with relatively stable hydrodynamic conditions, would be
82
expected to maintain, and possibly enhance, anoxia at the SWI. Indeed, Gammons et al. 2014
83
observed the build-up of anoxia and a related suite of reduced dissolved species (Mn+2, Fe+2,
84
NH+4, S-2) in under ice bottom waters of a Montana reservoir21. However, no studies have
85
examined the size partitioning or speciation of soluble metal phases in the under ice water
86
column. Speciation often controls the role of metals, particularly Fe, in various freshwater
87
biogeochemical cycles15. The size of metal phases (and thus their surface area) is controlled by
88
external factors influencing particle coarsening rates, such as temperature, salinity, and specific
89
ligands , as well as the timing between formation and any subsequent reductive dissolution22.
90
Furthermore, the response of the under ice biogeochemical system to intermittent external
91
disturbances such as winter rain or thaw periods typical of the northern hemisphere winter is
92
completely unknown, although such events are forecast to increase in frequency and severity in
93
the northern hemisphere due to climate change23, 24. Conceptually, it is intuitive that such events
94
should influence under ice biogeochemistry by impacting redox status and related processes at
4 ACS Paragon Plus Environment
Page 4 of 32
Page 5 of 32
Environmental Science & Technology
95
the SWI, under ice hydrodynamics in the lake, and the input of metal and P loads with
96
potentially unique geochemical composition associated with biogeochemical processes internal
97
to the watershed rather than the lake. As such, comprehensive studies based on empirical data
98
are needed to understand the interaction between the under-ice lake water column, sediment
99
water interface (SWI) and the lake watershed during the northern winter. Such studies can
100
provide a basis for understanding and predicting how overall lake nutrient-metal systems may
101
evolve as ice cover extent and duration decrease, as well as the occurrence, nature and severity of
102
extreme weather events change.
103
Here we took advantage of the relatively severe winter of 2013-14 in northern Vermont to study
104
under ice biogeochemistry in Missisquoi Bay, a shallow eutrophic arm of Lake Champlain. We
105
couple under ice sediment, water column metals and nutrient time-series geochemical data with
106
hydrodynamic data from a continuously-deployed Acoustic Doppler Current Profiler (ADCP)
107
array to develop a comprehensive understanding of the behavior of metals and phosphorous in
108
the under ice water column and SWI across the variable hydrodynamic conditions typical of the
109
northern winter climate. For the first time, the biogeochemical and hydrodynamic behavior of an
110
under-ice water column is linked to the evolution of sediment nutrient and metal pools, and the
111
response of both to climate disturbance.
112
Site Description: The study was conducted in the Missisquoi Bay of Lake Champlain, which
113
spans the border between Vermont, USA and Quebec, Canada (Figure 1A). Missisquoi Bay is
114
eutrophic and has experienced regular blooms of toxic cyanobacteria over the past two decades
115
due to persistent primarily nonpoint source-derived P and N loading from its agriculturalized
116
watershed25, 26 . Missisquoi Bay is uniformly shallow (max depth 5m, mean depth 2.8m) and
117
largely isolated from the main body of Lake Champlain. As such, Missisquoi Bay is similar in 5 ACS Paragon Plus Environment
Environmental Science & Technology
118
configuration to many eutrophic systems where cyanobacteria can access sediment nutrient
119
reserves (shallow, with a high agriculturalized watershed to lake ratio)27, but quite different from
120
most of the rest of the relatively deep and oligo/mesotrophic Lake Champlain28. During a typical
121
winter, the Bay freezes completely during the month of December and thaws sometime during
122
the month of April, although significant interannual variability of the duration of ice cover has
123
been observed, and the ice-over date and decadal frequency of ice-over of Lake Champlain has
124
changed significantly over the last 150 years due to climate warming29. During the winter of
125
2013-14, ice thickness approach meter in thickness by the end of the winter (Figure 1B). The
126
main tributary of the bay is the Missisquoi River, which drains a large (~1000 km2) forested
127
(62%) and agricultural (25%) watershed, is heavily impacted by nonpoint source agricultural
128
pollution26, 30. Our biogeochemical monitoring location is located in the southwest corner of the
129
bay (N 44˚ 59.503’, W 73˚ 06.798’), while our hydrodynamic array was spatially distributed across
130
the system at various depths (Figure 1A).
131
Materials and Methods:
132
Hydrodynamics: Water velocity measurements within the Bay were made using six upward-
133
looking 1,200-KHz acoustic Doppler current profilers (Workhorse Montior ADCPs by Teledyne
134
RDI) mounted to concrete platforms on the bottom of the Bay(Figure 1A). The hydrodynamic
135
sensor array was deployed during May of 2013 in a configuration optimized to study baywide
136
circulation, with one unit proximal to our biogeochemical monitoring site (Figure 1A). The
137
ADCPs were configured to measure vertical current profiles from 1.25 m above bottom to ~0.67
138
m below the water surface in 25-cm intervals. The instrument recorded mean velocity profiles
139
every 30 minutes, using an ensemble of 190 pings. Statistical correlation between individual
6 ACS Paragon Plus Environment
Page 6 of 32
Page 7 of 32
Environmental Science & Technology
140
pings provided an indication of the overall reliability of the calculated velocity; weakly
141
correlated measurements were removed from the dataset.
142 143
Water Sampling and Analyses: Biweekly field sampling trips were conducted throughout the
144
winter period while ice was thick enough to support researchers and equipment, which was
145
assessed based on ice fisherman activity and distribution. We used a Garmin GPS unit to locate
146
the same monitoring site upon each visit (Figure 1A). The depth below the ice surface ranged
147
from 3.4m to 3.1m during the winter, and the ice thickness ranged from 0.45m to 0.8m. Upon
148
reaching the site, 6 randomly distributed holes were drilled using a propane-powered ice auger
149
(8” diam.). Two of the holes were used for depth measurements (by weighted measuring tape)
150
followed by the deployment of water column sensors. A YSI EXO2 sonde was manually
151
deployed into each of two holes to collect measurements of T (oC), pH, DO (ppm), conductivity
152
(uS cm-1), and ChlA/phycocyanin (RFU) at predetermined depths(all data shown in Supporting
153
Information-Figure S1) . The sensors on the sonde were calibrated as specified by the
154
manufacturer (YSI Xylem, Yellow Springs, OH, USA). Unfortunately, our sonde was not
155
equipped with an ORP electrode to further quantify and describe redox dynamics near the SWI.
156
Sensor measurements were taken immediately below the ice, then at 0.5m increments, and
157
finally as close to the SWI as possible without disturbing the lake sediment. The position of the
158
sonde and sampling intake tubing for the ‘bottom water’ sample likely varied slightly relative to
159
each sampling event or sonde cast, which is nontrivial since dramatic concentration and redox
160
front gradients have been observed in those environments6, but this was unavoidable. Duplicate
161
sonde casts and water samples did reveal that differences between bottom water samples were
162
generally less than 10%, and, more importantly, those differences do not impact our 7 ACS Paragon Plus Environment
Environmental Science & Technology
163
interpretations. Another unused hole was accessed for the collection of water samples, which
164
were collected using a Masterflex peristaltic pump at the same depths that the sonde
165
measurements were taken. Acid-cleaned (7.5% hydrochloric acid) tubing was used to collect all
166
water samples. The tubing was flushed with at least 10 full volumes of water prior to sample
167
collection at each depth. Samples to be analyzed for SRP were collected in duplicate at each
168
depth in acid-cleaned 1 L polyethylene bottles. All samples were stored in a cooler and brought
169
back to the laboratory for immediate filtration within a clean laminar flow hood (within 4 hours
170
of sample collection). Trace metals were collected separately and filtered following the
171
published protocols and then analyzed by ICP-MS at the Woods Hole Plasma Facility31. The
172
‘truly dissolved’ size fraction is operationally defined by the