Stable isotopes track biogeochemical 
processes under seasonal ice cover in a 
shallow, productive lake
Authors:  Christopher H. Gammons, William Henne, 
Simon R. Poulson, Stephen R. Parker, Tyler B. 
Johnston, John E. Dore, & Eric S. Boyd 
This is a postprint of an article that originally appeared in Biogeochemistry on August 2014. The 
final publication is available at Springer via http://dx.doi.org/10.1007/s10533-014-0005-z.
Gammons, Christopher H., William Henne, Simon R. Poulson, Stephen R. Parker, Tyler B. Johnston, 
John E. Dore, and Eric S. Boyd. "Stable isotopes track biogeochemical processes under seasonal ice 
cover in a shallow, productive lake." Biogeochemistry 120, no. 1-3 (2014): 359-379. http://
dx.doi.org/10.1007/s10533-014-0005-z
Made available through Montana State University’s ScholarWorks 
scholarworks.montana.edu 
Stable isotopes track biogeochemical 
processes under seasonal ice cover in 
a shallow, productive lake
Christopher H. Gammons & William Henne: Department of Geological Engineering, Montana Tech of The University of 
Montana, Butte, MT, USA 
Simon R. Poulson:  Department of Geological Sciences and Engineering, University of Nevada-Reno, Reno, NV, USA 
Stephen R. Parker & Tyler B. Johnston: Department of Chemistry and Geochemistry, Montana Tech of The University of 
Montana, Butte, MT USA
John E. Dore: Department of Land Resources and Environmental Sciences, Montana State University, Bozeman, MT, USA
Eric S. Boyd:   Department of Microbiology, Montana State Univerity, Bozeman, MT, USA
Biogeochemical dynamics under seasonal ice cover were investigated in the shallow (<10 m) water column of 
highly productive Georgetown Lake, western Montana, USA. This high altitude (1,800 m) reservoir is well-mixed in 
summer, but becomes strongly stratified under ice cover (mid-November–mid-May). A rapid drop in dissolved 
oxygen (DO) concentration and rise in dissolved inorganic carbon (DIC) concentration was observed after the 
onset of ice, with a corresponding increase in δ18O-DO and decrease in δ13C-DIC, likely caused by respiration (R) 
of organic carbon. Photosynthesis/respiration ratios (P/R) estimated from simultaneous measurement of DO and 
δ18O-DO were near unity prior to ice formation but then systematically decreased with time and depth in the lake 
under ice cover. P/R in the water column was higher at a shallower monitoring site compared to a deeper site 
near the dam outlet, which may have been important for over-winter survival of salmonids. By March, the bottom 3 
m of the water column at both sites was anoxic, with the bottom 1 m being euxinic. Elevated concentrations of 
dissolved sulfide, ammonium, phosphate, Fe2+, and Mn2+ in deep water suggest coupling of organic carbon deg-
radation with reduction of a number of electron acceptors (e.g., Fe3+,NO3-, SO24-). The concentrations and δ34S 
values of H2S in the deep water and SO2i in the shallow water were similar, indicating near-complete reduction of 
sulfate in the euxinic zone. Late in the winter, an influx of isotopically heavy DIC was noted in the deep water 
coincident with a buildup of dissolved CH4 to concentrations >1 mM. These trends are attributed to acetoclastic 
methanogenesis in the benthic sediments. This pool of dissolved CH4 was likely released from the lake to the 
atmosphere during spring ice-off and lake turnover.
Abstract
Introduction
Seasonal ice cover is a characteristic feature of high-
latitude and/or high-altitude lakes, and it exerts a
strong control on energy and material transport to/
from and within lake ecosystems. Nevertheless, there
is presently a dearth of empirical information on
winter biogeochemical processes occurring beneath
the ice in such systems (Bertilsson et al. 2013). Given
that climate warming is anticipated to continue to
decrease the duration of seasonal lake ice cover
(Weyhenmeyer et al. 2011; Dibike et al. 2012), it is
important that we understand how lake ecosystems
function in both ice-free and ice-covered seasons, and
how the transitions between these states proceed. Of
particular importance to the present study is the
phenomenon of seasonal dissolved oxygen (DO)
depletion in the hypolimnion of shallow, ice-covered
lakes which, in some cases, can lead to winterkill of
fish (Greenbank 1945; Mathias and Barica 1980;
Babin and Prepas 1985). Depletions in DO are mainly
caused by decomposition of plant matter grown in the
preceding ice-free season that settles to the sediment
interface. Aerobic respiration (R) may result in anoxic
conditions in the winter hypolimnion of the lake, with
a concomitant increase in reduced constituents, such
as H2S and ammonia produced through anaerobic
respiratory pathways (Scidmore 1957). When electron
acceptors such as O2, NO

3 ; SO
2
4 are consumed,
fermentation and methanogenesis may ensue, leading
to accumulations of CH4 in the water column
(Bertilsson et al. 2013). Much of this CH4 may be
released to the atmosphere when the ice-covered
season ends and vertical mixing takes place (Mich-
merhuizen et al. 1996). Several authors (e.g., Bastvi-
ken et al. 2003, 2004; Tranvik et al. 2009) have argued
that emissions of CO2 and CH4 from temperate fresh-
water lakes have important consequences to the global
carbon budget.
Whereas conventional water quality sampling is
sufficient to define the existence of vertical chemical
gradients in ice-covered lakes, an approach that com-
bines chemical analysis and stable isotope analysis has
the potential to yield more information on biogeochem-
ical mechanisms. Stable isotopes of dissolved inorganic
carbon (d13C-DIC) and DO (d18O-DO) have been used
to elucidate processes influencing DIC and DO con-
centrations in lakes (e.g., Quay et al. 1986, 1995;
Herczeg 1987; Wachniew and Rozanski 1997; Wang
and Veizer 2000; Luz et al. 2002; Gu et al. 2004; Myrbo
and Shapley 2006; Assayag et al. 2008; Wang et al.
2008; Bass et al. 2010a, b; Karim et al. 2011; Gammons
et al. 2013). These processes, reviewed by Bade et al.
(2004), include advection of DIC from influent surface
water or groundwater, exchange of CO2 with the
atmosphere, precipitation or dissolution of carbonate
minerals, and biological reactions such as photosynthe-
sis (P), aerobic R, sulfate reduction, and methanogen-
esis. For P and R, changes in the concentration and
isotopic compositions of DO and DIC are inversely
related. P produces DO that has the same isotopic
composition as local water (which is typically depleted
in 18O relative to atmospheric O2) and preferentially
consumes 12C-DIC over 13C-DIC (Guy et al. 1993;
Falkowski and Raven 1997). Thus, P causes [DO] and
d13C-DIC to increase and [DIC] and d18O-DO to
decrease. (here and elsewhere, brackets [] are used to
denote concentration). Aerobic R has the opposite
effect, producing biogenic DIC that is isotopically light
(d13C ranging typically from -20 to -30 %; Clark and
Fritz 1997) while preferentially consuming dissolved
16O–16O over 18O–16O (Angert et al. 2001). The net
effect of aerobic R is, therefore, a decrease in [DO] and
d13C-DIC and a concomitant increase in [DIC] and
d18O-DO (Parker et al. 2005, 2010, 2014; Poulson and
Sullivan 2010; Smith et al. 2011). Other biogeochem-
ical reactions occurring in lakes may involve DO only
(e.g., H2S oxidation), DIC only (e.g., sulfate reduction
or methanogenesis), or may affect both DO and DIC
(e.g., aerobic methane oxidation).
Gammons et al. (2013) recently examined sea-
sonal trends in water chemistry, d18O-DO and
d13C-DIC in a deep, ultra-oligotrophic flooded
mining lake in the mountains of western Montana,
USA that experiences 6 months of seasonal ice cover.
Although DO was depleted in the winter months
Electronic supplementary material The online version of 
this article (doi:10.1007/s10533-014-0005-z) contains supple-
mentary material, which is available to authorized users.
in this low-productivity lake, most of the water
column remained oxic, and vertical gradients in
isotopic and chemical parameters were relatively
subdued. We hypothesize that geochemical gradients
and DO depletion will be much more pronounced in
a shallow, highly-productive lake in a similar
geographic setting that experiences a similar cli-
mate. Therefore, in the present paper we apply a
combined chemical/stable isotope approach to inves-
tigate the biogeochemical cycling of DO and DIC
under seasonal ice cover in Georgetown Lake, a
shallow, productive, high-altitude reservoir in wes-
tern Montana. Vertical and temporal changes in
d18O-DO and d13C-DIC are combined with compli-
mentary data on water isotopes and water chemistry
to (1) calculate the relative rates of below-ice P and
R, and (2) determine seasonal concentration changes
in methane and other compounds characteristic of
anoxigenic chemistry (e.g., H2S, NH
þ
4 ). Although
this lake is known to experience DO depletion in
winter (Stafford 2013), the extent of below-ice P has
not previously been explored, nor has the possibility of
CH4 generation. The methods and results of this study
are transferable to other shallow, fresh-water lakes in
cold climates that experience seasonal ice cover.
Site description
Georgetown Lake (Fig. 1), located in western Mon-
tana at an elevation of 1,960 m, was created in 1885 by
damming of a broad, grassy wetland in a high plateau
surrounded by mountains. The lake was originally
used for generation of hydroelectric power for nearby
mining towns, but more recently has been managed as
a recreational fishery as well as a source of irrigation
water. Georgetown Lake contains large numbers of
rainbow trout (Oncorhynchus mykiss), brook trout
(Salvelinus fontinalis), and kokanee salmon (a land-
locked variant of Onchorhynchus nerka), and is one of
the most heavily recreated water bodies for its size in
Montana. The shoreline is currently experiencing
deforestation, in part from rapid housing development,
and in part from invasion of the mountain pine beetle
(Dendroctonus ponderosae) which has decimated pine
Fig. 1 Location map of
Georgetown Lake. Contours
show depth (m) of water at
full pool. Ys mid-
Proterozoic carbonate and
siliclastic rocks of the Belt
Supergroup, Pzs Paleozoic
sediments, undivided
(includes Mississippian
Madison Group limestone)
forests throughout the northern Rocky Mountain
region (Kurz et al. 2008).
At full pool, Georgetown Lake has a surface area of
1,220 ha, a mean depth of 4.9 m, a maximum depth of
10.7 m, and a volume of 6.0 9 107 m3 (data from
Knight 1981). The surrounding watershed (13,500 ha)
is elongate in a north–south direction, with peaks
ranging up to 3,200 m in elevation. The North Fork of
Flint Creek drains the northern highlands, and exhibits
a large range in flow between spring snowmelt and
baseflow conditions. The southern area is drained by
Stuart Mill Spring, a large natural spring with a year-
round flow of approximately 0.5 m3/s that emerges
300 m from the SE corner of the lake. Numerous
smaller springs are evident along the eastern shore,
including Emily’s Spring (Fig. 1). Geologically, the
lake can be divided into a western half underlain by the
‘‘Middle Carbonate’’ metasedimentary unit of the
mid-Proterozoic Belt Supergroup, and an eastern half
underlain by Paleozoic sediments, including thick
limestone units of the Mississippian-aged Madison
Group (Lonn et al. 2003). These two rock sequences
are separated by the Georgetown Thrust, a low-angle,
west-dipping reverse fault. The Madison limestone is
an important karst aquifer in Montana, and is the
probable bedrock source for many of the springs
surrounding the eastern and southern shore of the lake.
Previous sampling of shallow groundwater wells
around the perimeter of Georgetown Lake (Shaw
et al. 2013) show that these waters are well-oxygen-
ated, pH-neutral, Ca–Mg–HCO3 waters with specific
conductivity (SC) between 156 and 553 lS/cm (aver-
age 326 lS/cm).
Georgetown Lake is highly productive and a lush
growth of submerged macrophytes, including white-
stemmed pondweed (Potamogeton praelongus), up to
5 m in height dominate the lake bottom. Many of these
aquatic plants die back to their root stalks during the
course of the winter. A number of previous studies
have investigated nutrient and DO cycling in the lake
(US Environmental Protection Agency, USEPA 1976;
Garrison 1976; Knight 1981; Garrett 1983; Trabert
1993; Stafford 2013). Whereas the lake is well-mixed
in the summer and fall due to strong winds, the water
column becomes stratified in winter, with decreases in
DO concentration occurring that worsen with depth
and with time during the ice-covered season. Ice
typically forms in mid-November and may not leave
the lake until mid-May. In addition, the lake may
accumulate up to a meter of snow on top of ice. Recent
monitoring (Stafford 2013) suggests that Georgetown
Lake is undergoing a decadal trend to more pro-
nounced hypoxia in late winter, although the reasons
for this trend remain obscure.
Methods
Sample collection
Fourteen sampling events were conducted at George-
town Lake between November 2010 and August 2011.
Water samples were collected near the dam (GT-1) at
the deepest spot in the lake (4612.8520 N/11316.7850 W),
and in Badger Bay (GT-2), 360 m northeast of the
shore at a fishing access known as Comer’s Point
(4610.8950 N/11318.8020 W). Each site was chosen
based on historical and concurrent monitoring by other
groups. Sampling was conducted between 11:00 and
14:00, during the warmest time of the day.
A Hydrolab MS5 datasonde or an In Situ Troll 9000
datasonde was used to measure pH, oxidation–reduc-
tion potential (ORP), DO concentration [DO], water
temperature (Tw), and SC. The datasondes were
calibrated to the manufacturer’s instructions on the
morning of each sampling event using appropriate
standards for the pH, ORP, and SC electrodes and
water-saturated air (corrected to ambient barometric
pressure) for the luminescent DO probe. The approx-
imate accuracy of the field readings were ±0.1 pH
units, ±5 mV (ORP), ±0.2 mg/L (DO), ±0.1 C (Tw)
and ±3 lS/cm (SC). For vertical profiling, the data-
sondes were lowered slowly into the water column
with measurements taken at a vertical spacing of
0.3–0.6 m. Measurements at selected depths were
repeated in reverse order when the datasonde was
retrieved. A light detector (LI-192, LI-COR) was used
to measure the photosynthetically active radiation
(PAR) flux (400–700 nm, lmol/m2/s). In April 2011, a
Hydrolab was deployed less than 1 m below the ice for
48 h to test for diel variation in field parameters.
Water samples were retrieved from discrete depths
using a submersible, 12 V pump and thick-walled
plastic hose at a rate of 5–20 L/min. At least 50 L of
water was pumped from each depth before any
samples were taken. Samples requiring filtration were
passed through a 0.2 lm PES syringe filter using a
60 mL syringe. Samples for alkalinity titration were
collected unfiltered and without preservatives in
250 mL high density polyethylene (HDPE) bottles.
Filtered and acidified (1 % HNO3) samples for metals
analysis were collected in 60 mL HDPE bottles.
Filtered and non-acidified samples for analysis of
nutrients and anions were collected in 60 mL HDPE
bottles. Samples for analysis of d18O and dD of water
were filtered and collected with no head space in
10 mL glass vials with caps containing conical inserts.
Samples for d13C analysis of DIC were filtered into
125 mL glass bottles with no head space. DIC was
precipitated as SrCO3 back in the lab by addition of
SrCl2/NH4OH solution, keeping the bottles tightly
capped with no head space (Usdowski et al. 1979).
Samples for d18O analysis of DO were collected by
puncturing the septum of a pre-evacuated, 125-mL
glass serum bottle with a needle through which a
steady stream of sample water was passed (Wassenaar
and Koehler 1999; Smith et al. 2011). The serum
bottles were immersed in a bucket of water while they
were filled, and were pre-loaded with 50 lL of
saturated HgCl2 as a bactericide. All sample bottles
were stored on ice or in a refrigerator prior to analysis.
Three samples for d34S analysis of dissolved sulfide
were collected near the bottom of the water column at
GT-2 in February and March of 2011, and again in
April of 2013, following the procedures of Carmody
et al. (1998). Aqueous sulfide was extracted immedi-
ately after sample recovery as Ag2S by direct addition
of silver nitrate solution to a 4 L Nalgene bottle filled
with lake water. This precipitate was later filtered and
purified with NH4OH to remove traces of AgCl. In
addition, samples of water from Stuart Mill Spring, the
North Fork of Flint Creek, and Georgetown Lake at
GT-2 at a depth of 1 m were collected for d34S and
d18O analysis of dissolved sulfate. Because of the low
sulfate concentrations it was necessary to take 20 L of
each sample and evaporate to 1 L. The evapoconcen-
trated solutions were filtered (0.2 lm) to remove
calcite and the pH was adjusted to\2 with addition of
HCl. The waters were warmed to 60 C and a 39
excess of BaCl2 solution was added to precipitate all
aqueous sulfate as BaSO4 (Carmody et al. 1998).
Certain trends in d13C-DIC with depth (discussed
below) provided indirect evidence for methanogenesis
near the sediment–water interface. To test this
hypothesis, several sets of samples from the bottom
of the water column at GT-2 were collected for CH4
analysis during the winter of 2013. These samples
were collected using pre-evacuated glass bottles, in the
same manner as described above for quantification of
d18O-DO.
Analytical methods
Alkalinity was measured in the laboratory within 24 h
of sample collection by titrating 100.0 mL of raw,
unfiltered sample with a HACH digital titrator, 1.6 N
sulfuric acid cartridges, and bromcresol green-methyl
red indicator dyes. Precision using this method was
estimated at 2 % based on replicate titrations. Filtered,
unpreserved samples were analyzed for total dissolved
ammonia (NHþ4 ? NH3) and soluble reactive phos-
phorus (SRP) within 12 h of collection using a HACH
portable spectrophotometer and HACH methods 8048
(detection limit 0.3 lM) and 8038 (detection limit
1.4 lM), respectively. During the February sampling,
a strong odor of H2S was noted for samples collected
at the bottom of the lake at GT-2. For future samplings,
bottles and reagents were brought into the field for
collection and stabilization of dissolved sulfide using
the methylene blue method (HACH method 8131).
25 mL of freshly pumped water was pipetted into a
polyethylene bottle and the sulfide reagents were
immediately added. The concentration of total
S2- (H2S ? HS
-) was quantified in the laboratory
by colorimetry within 12 h of sample collection.
A subset of samples from GT-1 and GT-2 were
analyzed for major and trace elements by inductively-
coupled plasma atomic emission spectroscopy
(ICP-AES) and for major anions by ion chromatogra-
phy (IC) at the Environmental Biogeochemistry
Laboratory, University of Montana.
Most of the stable isotope analyses were performed
at the University of Nevada-Reno using a Micromass
Isoprime stable isotope ratio mass spectrometer
(IRMS) interfaced to either a Micromass MultiPrep
device or a Eurovector elemental analyzer. Samples
for d18O-H2O, dD-H2O, d
13C-DIC, d18O-DO,
d18O-SO4, and d
34S-H2S/SO4 measurement were
analyzed after Epstein and Mayeda (1953), Morrison
et al. (2001), Harris et al. (1997), Wassenaar and
Koehler (1999), Kornexl et al. (1999), and Giesemann
et al. (1994), respectively. Isotope values are reported
in units of per mil (%) in the usual d-notation versus
VSMOW for oxygen and hydrogen, VPDB for carbon,
and VCDT for sulfur. Replicate analyses indicated a
reproducibility of ±0.1 % for d18O-H2O, ±1 % for
dD-H2O, ±0.05 % for d13C-DIC, ±0.1 % for
d18O-DO, ±0.4 % for d18O-SO4, and ±0.2 % for
d34S-H2S and d
34S-SO4. A single sample of water
from the bottom of the water column at GT-2 collected
in April 2013 was analyzed for d13C-CH4 using a
Picarro cavity ring-down spectrometer at Montana
State University. The d13C analysis was performed in
triplicate, using three different dilution factors, and
returned a precision of ±0.3 %.
Samples for CH4 analysis were collected in 125 mL
pre-evacuated serum bottles as previously described.
A small headspace was created in these bottles due to
exsolution of dissolved gasses. This headspace was
sampled in the laboratory by puncturing the septum
with a gas-tight syringe and withdrawing a gas sample
(usually 20 lL) followed by analysis using a Thermo
Scientific Trace GC Ultra with a TracePLOT
TG-BOND Msieve 5A column connected to a Thermo
Scientific ITQ 900 mass spectrometer using electron
ionization. The GC method used UHP-He carrier gas,
an inlet temperature of 250 C, a split ratio of 60, split
flow of 18, 0.3 mL/min constant column flow, 30 C
isothermal column temperature, outlet temperature of
250 C and an ion source temperature of 250 C.
Calibration was performed using a standard mixture
containing O2, N2, CO2, CO, CH4 and the balance
being He (Scotty #501697). Solution concentrations of
CH4 were determined using the methane solubility
equations of Wiesenburg and Guinasso Jr (1979). The
pre-sampling CH4 concentration was determined by
mass balance calculations based on the sum of the CH4
in the headspace, the moles of CH4 in solution, and the
total sample and headspace volumes, which were
determined gravimetrically.
Geochemical modeling
DIC was speciated using the program Visual
MINTEQ, version 3.0 (a recent adaption of the original
program described by Allison et al. 1991) based on the
measured pH, alkalinity, and Tw of each sample.
Calculated charge imbalances were used as a quality
control check: in most cases the charge imbalance was
less than 5 %. Visual MINTEQ was also used to
calculate CO2 partial pressures and mineral saturation
indices (SIs). Selected results are discussed below.
Results
Water isotopes and lake hydrology
The isotopic composition of water samples collected
from Stuart Mill Spring and the North Fork of Flint
Creek in November 2010 and June 2011 (Fig. 2a) fall
close to the global meteoric water line (MWL),
indicating that minimal evaporation occurred prior to
entering Georgetown Lake (data in Supplementary
Material). In contrast, all of the lake samples collected
between November 2010 and March 2011 were
moderately evaporated, and define a local evaporation
line (LEL) with the following equation:
dD = 4.86 9 d18O - 49.2 (R2 = 0.979). The slope
of this line correlates to an average relative humidity of
roughly 65 %, and is very close to the slope of the LEL
for nearby Butte, Montana (slope = 5.0, Gammons
et al. 2006). The maximum shift in d18O away from the
MWL for the most evaporated samples was approxi-
mately 4 %, consistent with up to 10–15 % loss of
water to evaporation (Gammons et al. 2006). Much of
this evaporation presumably occurred during the pre-
ceding ice-free period in the summer and fall of 2010.
Although the isotopic compositions of water sam-
ples at GT-1 and GT-2 were similar in November of
2010, the two sites had quite different stable isotope
profiles in March 2011 (Fig. 2b). GT-2 showed a slight
increase in d18O and dD with depth in the lake,
whereas GT-1 showed a distinct shift to lighter
isotopic values with depth. The latter trend is tenta-
tively explained by mixing of isotopically light
groundwater entering the lake from submerged springs
along the eastern shore of Georgetown Lake. Previous
work (Shaw et al. 2013) characterized the east shore of
the lake as a groundwater discharge zone based on
spatial trends in dissolved radon gas. In contrast, the
western bays of the lake, including Badger Bay,
showed no evidence of influent groundwater.
Field parameters
In early November 2010, before the onset of ice cover,
the water columns were well-mixed at both GT-1 and
GT-2, and both stations had similar values of temper-
ature, SC, and [DO]. Ice formed in late November, and
by the next sampling date in December both stations
were stratified, with an increase in temperature and SC
and a decrease in [DO] with depth (Figs. 3, 4). These
trends magnified with time until ice-off occurred on
24–26 May 2011, after which point strong winds
induced vertical mixing. Vertical profiles obtained on
5 June and 11 August 2011 showed minimal changes
in SC and [DO] with depth at both sites (Figs. 3, 4).
This is consistent with studies performed in the 1980s
(Knight 1981) and more recently (Stafford 2013)
which show that Georgetown Lake is vertically mixed
throughout the non-frozen months.
The GT-2 site in Badger Bay showed an increase in
SC with depth, especially near the bottom of the lake,
which became more pronounced as the winter
Fig. 2 Water isotope
results. a Isotope cross-plot:
the global meteoric water
line (MWL) is from Craig
(1961), and the local
evaporation line (LEL) is a
regression of data from this
study. b Variations in water-
d18O with depth at GT-1 and
GT-2 in March 2011
Fig. 3 Changes in water temperature (a, b contours in C) and
specific conductance (c, d contours in lS/cm) with depth and
time. a, c GT-2 (Badger Bay), b, d GT-1 (dam). Contour lines
were generated with Golden Software’s Surfer program. Arrows
in centerline of figure denote dates when vertical profiles were
collected at GT-2 (arrow up), GT-1 (arrow down) or both sites
(arrow both directions)
progressed (Figs. 3, 4). As discussed below, most of
this change in SC is believed to have been caused by
diffusion of solutes, especially Ca2? and HCO3 ; from
sediment pore-water. Both GT-1 and GT-2 developed
a deep anoxic layer which increased in thickness as the
winter progressed. This anoxic layer was thicker and
set up earlier in the winter at GT-2 compared to GT-1
(Fig. 4). Nonetheless, a layer of oxic water in the
overlying water column maintained sufficient [DO]
for trout and salmon to overwinter in Badger Bay. In
contrast, [DO] in shallow water at GT-1 dropped to
values \150 lM, and most of the water column had
[DO] \100 lM, which would have caused severe
stress for trout (Raleigh et al. 1984).
Chemistry
The water columns at GT-1 and GT-2 were highly
stratified in March 2011, with a nearly 3-m thick anoxic
layer extending from mid-depth to the sediment–water
interface (Fig. 5). PAR decreased in the top several
meters, but was still detectable (0.1–0.6 lmol/s/m2 in
the bottom meter of the water column (Fig. 5b, g). On
the day of PAR measurement, the lake had approxi-
mately 0.8 m of ice and 0.2 m of compact, windblown
snow cover. An increase in SC towards the bottom of the
lake at GT-2 was highly correlated (R2 [0.98) with a
parallel increase in [HCO3 ] (Fig. 5c). The GT-1
profile shows a more gradual increase in SC and
[HCO3 ] with depth (Fig. 5h). Despite the increase in
[HCO3 ], pH decreased towards the bottom of the lake
at both sites (Fig. 5a, f) due to an even greater increase
in [CO2] (Fig. 5e, j).
Bicarbonate ion is by far the dominant anion in
Georgetown Lake, with concentrations in the
1–2 mM range, compared to \0.05 mM for
SO24 and Cl
-. Cations are dominated by Ca2? and
Mg2?. Whereas [Mg2?] showed no change with depth
in March 2011, [Ca2?] increased sharply near the
sediment–water interface at GT-2 (Fig. 5e), and to a
Fig. 4 Changes in
dissolved oxygen
concentration (contours in
lM) with depth and time.
a GT-2 (Badger Bay),
b GT-1 (dam). Contour lines
were generated with Golden
Software’s Surfer program.
The hatched area indicates
the presence of H2S. Arrows
in centerline of figure denote
dates when vertical profiles
were collected at GT-2
(arrow up), GT-1 (arrow
down) or both sites (arrow
both directions)
lesser extent at GT-1 (Fig. 5j). Amongst the remaining
solutes, K? and dissolved SiO2 increased in concen-
tration towards the bottom of the lake (Fig. 5e, j),
whereas Na? and Mg2? did not. Concentrations of
most trace metals were below the analytical detection
limit of the ICP-AES. Exceptions include Mn, which
was detected in the deep water samples on most
sampling dates (Fig. 5e, j), and Fe, which was detected
in the deep water at GT-1 (Fig. 5j), but not GT-2
(Fig. 5e).
Concentrations of dissolved NO3 were below the
detection limit of the IC (5 lM) for all samples
analyzed in this study. In contrast, [NHþ4 ] increased
towards the bottom of the lake in late winter, to [25
and [150 lM at GT-1 and GT-2, respectively
(Fig. 5d, i). Concentrations of soluble reactive phos-
phate (SRP) were below detection except near the
bottom of the water column where [SRP] exceeded
0.5 lM at both sampling sites (data in Supplementary
Material).
To test the hypothesis that methanogenesis was
occurring near the sediment–water interface, several
samples were taken in 2013 for quantification of
dissolved CH4. A sharp rise in [CH4] was documented
near the bottom of the water column at GT-2 on all dates
sampled (Table 1). Maximum concentrations of CH4
were in the range of 500–1,000 lM in January, February
and March of 2013.
Fig. 5 Example vertical profiles in selected water column
properties at GT-1 (bottom row) and GT-2 (top row) for March
2011. Water temperature and pH (a, f), photosynthetically
active radiation (PAR) and dissolved oxygen (DO; b, g), specific
conductance (SC) and bicarbonate ion (c, h), ammonium (NHþ4 )
and dissolved sulfide (d, i), other major and trace solutes (e, j).
All concentration scales are linear, except for logarithmic
concentration scales for other major and trace solutes (e, j)
Isotopes of DO and DIC
Prior to ice-on, values of d18O-DO were similar at
GT-1 (average 16.0 %) and GT-2 (average 17.8 %),
and showed no clear changes with depth (Table 2;
Fig. 6a, b). Both sites had d18O-DO values significantly
lower than the expected value of 24.2 % for DO in
equilibrium with atmospheric O2 (Benson and Krause
Jr 1980). Once ice cover was established, d18O-DO
increased with both depth and time, and these trends
were magnified at the GT-1 site. No isotopic data are
available for DO in the deeper portions of the water
columns at GT-1 and GT-2, due to insufficient DO
concentrations for IRMS analysis. Overall, d18O-DO
values increased with decreased [DO]. However, the
slope of the relationship was considerably steeper for
the GT-1 data set compared to GT-2 (Fig. 7). Possible
reasons for these differences are discussed below.
Spatial and temporal changes in d13C-DIC under
ice cover (Table 3; Fig. 6c, d) show more complexity
Table 1 Concentrations of dissolved CH4 in samples from GT-2
Dates Depth (m) Temperature (C) [DIC] (mM) d13C-DIC (%) [CH4] (lM)
26 January 2013 5.79 4.69 na na 16
6.40 4.76 na na 595
16 February 2013 4.42 4.39 3.79 na 24
5.64 4.60 3.54 na 44
6.25 4.34 10.8 na 695
29 March 2013 4.27 4.36 4.85 -5.4 25
4.88 4.46 5.21 -6.1 72
5.49 4.69 5.11 -5.2 121
6.10 5.36 17.3 -1.1 1,020
na Not analyzed
Table 2 Concentrations and isotopic compositions of dissolved oxygen (DO)
GT-1 dam site GT-2 Badger Bay
Depth (m) [DO] (lM) [DO] (% sat) d18ODO (%) Depth (m) [DO] (lM) [DO] (% sat) d18ODO (%)
6 November 2010
0.3 302 95.2 16.0 0.3 316 99.1 17.6
3.0 302 95.0 15.7 2.4 316 99.0 18.2
6.1 301 94.6 16.2 4.6 319 99.7 17.7
24 January 2011
1.5 251 70.1 22.6 1.5 334 63.4 20.5
3.0 152 43.4 24.5 2.7 191 54.5 22.4
4.6 117 33.6 26.2 3.7 67 22.0 20.6
6.1 101 29.0 25.6
7.6 77 22.6 26.9
8.8 41 12.0 29.3
16 March 2011
1.2 116 33.4 26.8 0.6 270 74.0 21.4
1.8 89 25.9 29.0 1.2 246 70.0 22.0
3.0 47 13.7 32.6 1.8 193 55.6 22.3
4.3 32 9.5 31.3 2.4 126 36.8 22.9
5.5 38 11.4 33.2
than the changes in d18O-DO outlined above. Prior to
development of ice, both GT-1 and GT-2 had similar
values of d13C-DIC, near -4.3 to -4.5 %. Early in the
winter, there was a clear trend at both sites of a
decrease in d13C-DIC with both depth and time.
However, by early January this trend reversed in the
deeper water at GT-2; at GT-1, a similar reversal was
observed by early March (see shaded regions in
Fig. 6c, d). The deepest water at GT-2 showed a
dramatic increase in d13C-DIC to values greater than
?1 %, whereas the analogous increase was more
subtle at GT-1. The increase in d13C-DIC at depth
coincided with substantial increases in [DIC] (Fig. 8).
Possible sources of this influx of isotopically heavy
DIC are discussed below. Following ice-off in late
May, most of the changes in d13C-DIC with depth
were homogenized due to mixing of the water column.
Isotopes of sulfide and sulfate
Similar to NHþ4 and SRP, concentrations of dissolved
sulfide increased sharply near the base of the water
column at both GT-1 and GT-2 (Fig. 5d, i), presum-
ably from bacterial sulfate reduction (BSR) in the
shallow sediment, possibly augmented by putrefaction
of organic-S in decaying plants or animals (Dunnette
et al. 1985). The concentration of dissolved sulfide at
GT-2 at 6 m depth (30–60 lM) was similar to that of
dissolved sulfate at 1 m depth on the same day
(Table 4). Euxinic bottom water at GT-2 collected on
Fig. 6 Spatial and temporal trends in the isotopic composition
of dissolved oxygen (a, b) and dissolved inorganic carbon (c,
d) at GT-2 (a, c) and GT-1 (b, d). Shaded areas in a and b show
depths where [DO] was absent (based on data in Fig. 4). Shaded
areas in c and d represent the portion of the water column where
d13C-DIC increased with depth. Contour lines were generated
with Golden Software’s Surfer program. Arrows in centerline of
figure denote dates when samples were collected for isotopic
analysis of DO or DIC at both sites
Fig. 7 The isotopic composition of dissolved oxygen
(d18O-DO) versus DO concentration at GT-1 (circles) and
GT-2 (squares). Linear regressions are for all data from each
respective site
three different dates had d34S-H2S in the range of ?4.8
to ?8.5 % (Table 4). This compares to a single value
of ?9.7 % for d34S of aqueous sulfate in the shallow
water at GT-2 (Table 4). Whereas the two main
tributaries to the lake had d18O-SO4 values of ?2.8 %
(Stuart Mill Spring) and ?0.8 % (North Fork of Flint
Creek), sulfate in the shallow water at GT-2 was
depleted in 18O (-6.4 %).
Table 3 Concentrations and isotopic compositions of dissolved inorganic carbon (DIC) in Georgetown Lake and influent waters
GT-1 dam site GT-2 Badger Bay GT-1 dam site GT-2 Badger Bay
Depth
(m)
[DIC]
(mM)
d13CDIC
(%)
Depth
(m)
[DIC]
(mM)
d13CDIC
(%)
Depth
(m)
[DIC]
(mM)
d13CDIC
(%)
Depth
(m)
[DIC]
(mM)
d13CDIC
(%)
6 November 2010 6 November 2010 15 April 2011 15 April 2011
0.3 1.93 -4.6 0.3 1.69 -3.9 1.2 2.93 -5.5 1.2 2.84 -5.5
3.0 1.83 -4.4 2.4 1.75 -3.9 3.0 3.39 -6.2 2.4 2.96 -6.0
6.1 1.85 -4.4 4.6 1.74 -3.9 4.3 3.52 -6.9 3.7 3.00 -5.7
8 December 2010 8 December 2010 6.1 3.71 -6.8 4.9 3.50 -4.4
0.3 2.03 -4.2 0.3 1.95 -4.0 7.9 3.79 -6.8 5.5 4.20 -3.5
4.0 2.10 -4.6 2.7 1.87 -4.1 9.1 4.23 -6.5 6.1 6.54 ?0.8
7.6 2.49 -5.2 4.6 2.19 -5.0 13 May 2011 13 May 2011
24 January 2011 24 January 2011 0.6 1.23 -4.7 1.2 3.02 -5.3
1.5 2.16 -4.7 1.5 1.97 -4.2 1.8 3.17 -5.6 2.4 3.18 -5.5
3.0 2.45 -5.3 2.7 2.16 -5.2 3.0 3.51 -6.7 3.7 3.55 -5.1
4.6 2.63 -5.6 3.7 2.31 -5.1 4.3 3.70 -7.2 4.9 3.82 -3.9
6.1 2.72 -6.1 5.5 2.56 -5.5 5.5 3.80 -7.5 6.1 7.83 ?1.7
7.6 2.80 -6.7 6.7 3.95 -7.5
8.8 2.97 -6.9 7.9 4.20 -7.1
15 February 2011 15 February 2011 9.1 4.99 -6.2
1.2 2.18 -5.5 1.2 naa -4.8 5 June 2011 5 June 2011
2.1 2.32 -6.1 1.8 na -4.8 1.5 2.56 -5.5 1.8 2.85 -5.4
3.0 2.42 -6.1 2.4 na -5.2 4.6 2.52 -5.7 3.7 2.86 -3.8
4.0 2.52 -6.6 3.0 na -6.0 7.6 2.56 -6.0 5.5 2.87 -3.8
4.9 2.57 -6.9 3.7 na -5.5 11 August 2011 11 August 2011
5.8 2.68 -6.8 4.3 na -5.4 0.3 1.85 -4.7 0.3 1.85 -3.8
6.7 2.75 -7.6 4.9 na -5.4 3.0 1.89 -4.4 3.0 1.84 -3.5
7.6 2.84 -7.7 5.5 na -0.6 6.1 1.88 -4.5 5.6 1.79 -3.8
8.5 2.94 -7.9 9.1 1.91 -4.6
16 March 2011 16 March 2011 Miscellaneous surface samples
1.2 2.39 -5.3 0.6 2.57 -4.5 6 November 2010
1.8 2.59 -5.7 1.2 2.34 -4.2 Stuart Mill Spring 2.96 -11.1
3.0 2.79 -6.4 1.8 2.45 -4.9 North Fork of Flint Creek 2.39 -9.1
4.3 3.03 -6.9 2.4 2.60 -5.6 Emily’s Spring 3.14 -10.1
5.5 3.19 -7.1 3.0 2.74 -5.5 2 August 2011
6.7 3.25 -7.3 3.7 2.91 -5.0 Stuart Mill Spring 3.19 -10.9
7.9 3.40 -7.5 4.3 3.00 -4.9 North Fork of Flint Creek 2.32 -10.1
9.1 3.76 -7.2 4.9 3.08 -4.6
5.5 3.41 -4.3
6.1 5.95 -0.1
a Total DIC not available due to pH meter malfunction
Discussion
Controls on lake chemistry
The establishment of vertical gradients in water
chemistry under ice cover could be attributed to
several processes, including: (1) inputs of groundwa-
ter, (2) solute exclusion during ice formation, and (3)
upward diffusion of solutes from the sediment–water
interface.
In addition to Stuart Mill Spring, Georgetown Lake
receives a significant amount of groundwater in the
form of submerged springs, especially along the
eastern shore of the lake. Shaw et al. (2013) showed
that dissolved radon concentrations in Georgetown
Lake under ice cover were elevated along the eastern
shore, but not in the western portion of the lake (e.g.,
near GT-2). This radon was inferred to come from
upwelling groundwater. Based on sampling of domes-
tic water wells, groundwater to the east of the lake is
Ca–Mg–HCO3 type, oxidized (DO present), with an
average SC of 326 lS/cm. Groundwater entering the
lake in winter would therefore tend to sink (higher
density than most of the water column of the lake).
Consistent with this idea, stable isotope profiles from
March 2011 (Fig. 2) show evidence of mixing of deep,
non-evaporated water (i.e., groundwater) with evap-
orated lake water at GT-1, but not at GT-2. Despite
mixing of presumably oxidized groundwater into the
water column along the eastern shore of the lake, the
biological and chemical oxygen demand of the
sediment and overlying stratified water column was
more than sufficient to maintain anoxic conditions for
deep water by the time it reached the dam outlet.
Because the outlet pipes draw water from the bottom
of the lake, water that is discharged has a chemistry
that matches that of deep water at GT-1. This includes
the presence of H2S, which, in late winter, imparts a
strong odor to Flint Creek in the first several hundred
meters below the dam.
Solute exclusion during freezing of water to ice is
another process that could influence the chemistry of
Georgetown Lake. For water bodies with high salinity
(e.g., some Antarctic lakes and marine estuaries) this
process, termed thermohaline convection, can lead to
Fig. 8 The isotopic composition of dissolved inorganic carbon
(d13C-DIC) versus DIC concentration at GT-2 (a) and GT-1 (b).
Arrows show inferred trends for respiration (negative slope) and
methanogenesis (positive slope)
Table 4 Isotopic compositions of dissolved sulfate and sulfide
Dates Locations Depth
(m)
[RS2-]
(lM)
d34S-H2S
(%)
[SO24 ]
(lM)
d34S-SO24
(%)
d18O-SO24
(%)
15 February 2011 GT-2 6.1 47 4.8 na na na
16 March 2011 GT-2 5.9 34 8.5 na na na
24 April 2013 GT-2 6.1 56 6.4 na na na
24 April 2013 GT-2 2.1 na na 50 9.7 -6.4
24 April 2013 Stuart Mill Spring 0 na na 31 13.6 2.8
24 April 2013 North Fork of Flint Creek 0 na na 31 4.7 0.8
na Not analyzed
permanent meromixus (Gibson 1999). In Georgetown
Lake, most ice forms in December and early January.
However, because of the low salinity of the lake, the
increases in SC of the water column during early
winter were relatively minor (Fig. 3). Exclusion of
dissolved gas during freezing may also have tempo-
rarily increased [DO] of the shallow water, although
no values of [DO] that exceeded atmospheric satura-
tion were measured in this study.
Diffusion of solutes from the sediment pore-water
is probably the dominant process leading to the deep
layers of high-TDS, anoxic water at both GT-1 and
GT-2. Like the overlying water column, the major ion
composition of the deep water is dominated by Ca2?
and HCO3 : Inputs of Ca
2? and HCO3 from the
sediment could occur from dissolution of calcite in
response to elevated CO2 partial pressure:
CaCO3ðsÞ þ H2O þ CO2ðaqÞ ¼ Ca2þ þ 2HCO3 :
ð1Þ
Based on thermodynamic modeling with Visual
MINTEQ, all water samples collected during the
period of ice cover had calcite SIs (SI = log ion
activity quotient divided by the equilibrium constant)
that were significantly negative. At GT-1, the SI range
was -0.68 to -0.97 with an average of -0.86
(n = 18). At GT-2, the SI range was -0.19 to -1.18
with an average of -0.62 (n = 13). Thus, any calcite
in sediment at the bottom of the lake should have been
dissolving. Samples collected in November prior to ice
formation had calcite SI values of -0.23 (GT-2) and
-0.03 (GT-2), and were thus closer to equilibrium. In
a previous study, Knight (1981) suggested that
Georgetown Lake is typically supersaturated with
calcite during the warm summer months, and under-
saturated during winter.
In August 2011, a diffusion sampler (i.e., ‘‘peeper’’)
was used to collect samples of sediment-pore water
from a location in Badger Bay close to GT-2, but in
shallower water (data in Shaw et al. 2013). Although
the location and time of year are different, the results
bear on the current study. Pore water from 10 cm
below the bottom of the lake contained [400 mg/L
HCO3 ; [150 mg/L Ca
2?, [10 mg/L each of K?,
Fe2?, NH4–N, and Si,[3 mg/L Mn2?, and[0.5 mg/L
PO4–P. Concentrations of sulfate were below detec-
tion and total dissolved sulfide was roughly 0.1 mg/L.
The most likely source of K?, NH4–N and PO4–P is
decay of organic matter in the sediment (Shaw et al.
2013), and the high concentrations of Ca2? and HCO3
are consistent with dissolution of calcite in an
environment of high pCO2 : Elevated concentrations
of silica could be related to dissolution of amorphous
silica in the form of diatom frustules. The somewhat
elevated concentrations of Fe2? and Mn2? suggest
reductive dissolution of hydrous Fe and Mn oxides.
Finally, the presence of H2S and absence of SO
2
4 is
consistent with BSR.
Upward diffusion of solutes from a more concen-
trated sediment-pore water (as above) can explain
most of the observed changes in chemistry of the deep
lake under ice cover, including increases in the
concentrations of Ca2?, HCO3 ; K
?, NHþ4 ; PO4–P,
dissolved silica, dissolved Fe and Mn, and H2S. The
lower concentration of dissolved Fe at GT-2 (below
detection) compared to GT-1 may be explained by the
higher concentration of H2S at GT-2, leading to
precipitation of Fe-sulfide. Additional sampling of
sediment pore water during the winter months at GT-2
is planned to quantify the diffusive flux of solutes into
the overlying water column and their cumulative
oxygen demand.
DO cycling under ice cover
Because of the extremely slow diffusivity of gases
through ice, the only appreciable sources of DO to the
water column for a shallow lake under ice cover are
aquatic P and influx of oxygenated groundwater or
surface water. However, there are a number of
possible sinks for DO, including the aerobic microbial
oxidation of organic matter (aerobic R), methane
(methanotrophy), ammonium (nitrification), aqueous
sulfide, and reduced metals such as Fe2? or Mn2?.
Because of the widespread distribution of particulate
organic matter, macrophytes, and higher organisms
(insects, fish) in the lake, aerobic R should occur
throughout the entire oxic portion of the water column.
In contrast, CH4, NH
þ
4 ; H2S, Fe
2? and Mn2? are most
likely generated within the sediment where they
subsequently diffuse upwards into the lake and are
oxidized at the oxic/anoxic interface in the water
column. The production of these reduced species is
likely coupled to the oxidation of organic carbon under
anoxic conditions. Because the concentrations of
sulfide, ammonium, and dissolved Fe/Mn are
relatively low in Georgetown Lake (Fig. 3), the main
sink for DO in winter is most likely oxidation of
organic C. Oxidation of methane diffusing upward out
of the anoxic layer could also be an important sink for
downward diffusing DO. Methane can also rise
quickly to the oxic zone beneath the ice via ebullition
of bubbles generated near the sediment–water
interface.
P produces 18O-depleted DO with a similar O-iso-
tope composition as the surrounding water (in this
case, near -16 %). Aerobic R fractionates O-isotopes
so that d18O-DO of residual DO is shifted to heavy
values as DO is consumed (Kroopnick 1975; Guy et al.
1993), as does abiotic oxidation of aqueous sulfide
(Oba and Poulson 2009a), abiotic oxidation of Fe2?
(Oba and Poulson 2009b), consumption of O2 by
methanotrophs (Mandernack et al. 2009), and possibly
the aerobic oxidation of ammonium (suggested by
Mandernack et al. 2009). Assuming that R of organic
C is the main sink for DO under ice cover, the residual
DO pool should become progressively enriched in 18O
as DO concentrations diminish. This trend is clearly
shown at both GT-1 and GT-2, although much more so
at the deeper, GT-1 site. Previous studies have
documented similar increases in d18O-DO with lake
depth (Luz et al. 2002; Bass et al. 2010a; Brown and
Poulson 2010; Gammons et al. 2013).
Previous workers (Quay et al. 1995; Wang and
Veizer 2000) have shown that simultaneous measure-
ment of d18O-DO and [DO] can be used to estimate the
relative rates of oxygenic P and R in lakes. If the
system is at steady state, and there is no net exchange
of gases across the air–water interface, then the P/R
ratio is given by (equations and values of constants
from Wang and Veizer 2000):
P=R ¼ RDOar  Xg
 
= RW  Xg
 
; ð2Þ
Xg ¼ ag Ratmas  ½DO= DOsat½ ð Þ
RDO
 
= 1  ½DO=ð
DOsat½ Þ; ð3Þ
where ar is the O-isotopic fractionation from R and has
typical values of 0.977–0.98 (e.g. Kroopnick 1975;
Kiddon et al. 1993), although a larger range of values
has been measured (e.g. Lane and Dole 1956; Brandes
and Devol 1997), ag is the ratio of
18O–16O to 16O–16O
gas transfer velocities (0.9972 at 20 C), as is the ratio
of 18O–16O to 16O–16O solubility in water (1.0007),
RDO is the
18O/16O ratio of DO (measured), Ratm is the
18O/16O ratio of O2 in air, RW is the
18O/16O ratio of
water, [DO] is the measured DO concentration, and
[DO]sat is the computed DO concentration at atmo-
spheric saturation at the temperature and pressure of
interest. In the present study, ice creates a physical
barrier to gas exchange with the atmosphere. Although
the observed gradients in [DO] with time show that the
lake was not at steady state with respect to DO on the
time scale of weeks–months, changes in [DO] or
d18O-DO were most likely negligible on the scale of
hours–days. A datasonde placed in shallow water
under the ice in April 2011 showed minimal change in
[DO] (\3 lM) over two 24-h periods.
With the above steady-state caveat in mind, the P/R
equations of Quay et al. (1995) and Wang and Veizer
(2000) were applied to the data in this study. The
results (Fig. 9) show that oxygenic P and R were
almost equal (P/R * 1) at both sites in November,
prior to ice formation. Once ice formed, the expected
trends were shown wherein the lake became more
R-dominated with depth on any given sampling date,
and also with time as winter progressed. At the same
depth and sampling dates, P/R ratios were significantly
higher at GT-2 than at GT-1, especially for the March
2011 data set. This result is consistent with other
Fig. 9 Calculated photosynthesis/respiration (P/R) ratios at
GT-2 (a) and GT-1 (b)
observations that suggest a higher P/R ratio under ice
at GT-2, including higher overall DO concentrations at
shallow depth late in the season.
DIC cycling under ice cover
Internal sources of DIC in winter include production
of biogenic CO2 by aerobic or anaerobic R, meth-
anogenesis, and dissolution of carbonate minerals. R
of organic carbon—both aerobic and anaerobic—
adds CO2 to the water column with a d
13C value that
is similar to that of the carbon source, i.e., -20 to
-30 % for C3 plants in temperate climates (Clark
and Fritz 1997). Methanogenesis occurs via several
pathways, two of which are discussed here as each is
well characterized and their effect on [DIC] and
d13C-DIC is well documented (Whiticar et al. 1986;
Whiticar 1999). One pathway involves reduction of
CO2 by H2:
CO2 þ 4H2 ! CH4 þ 2H2O: ð4Þ
Due to kinetic isotopic effects, CH4 formed by this
pathway is strongly depleted in 13C, resulting in an
increase in d13C of the residual DIC pool as [DIC]
decreases. The second pathway, acetoclastic metha-
nogenesis, produces a 1:1 mixture of isotopically-
heavy DIC (as HCO3 ) and isotopically-light CH4:
CH3COO
 þ H2O ! CH4 þ HCO3 : ð5Þ
Thus, although both reactions promote an increase in
d13C-DIC, acetate fermentation increases [DIC]
whereas CO2 reduction decreases [DIC]. The fact that
the increase in d13C-DIC of deep water in Georgetown
Lake was accompanied by a sharp increase in [DIC]
(Fig. 8) is evidence that reaction (5) was the dominant
pathway for methanogenesis. Previous workers (Whi-
ticar and Faber 1986; Whiticar et al. 1986) have shown
that reaction (5) usually dominates over CO2 reduction
in freshwater sediments. A sample from the bottom of
the water column collected at GT-2 in April 2013
returned a d13C-CH4 value of -0.7 ± 0.3 %. This
value is within the range of d13C reported for methane
formed from acetate fermentation in fresh water
environments (Whiticar et al. 1986).
A typical value for the C-isotopic fractionation of
CO2 formed by reaction (5) in freshwater sediments is
?20 % (Stiller and Magaritz 1974; Gu et al. 2004). If
we adopt this value and assume an average d13C of
-25 % for CO2 derived from aerobic/anaerobic R
(including CO2 produced from BSR), then simple
mass balance dictates that the rate of production of
DIC by acetoclastic methanogenesis [reaction (5)]
must have exceeded the rate of production of DIC by R
for d13C-DIC to increase. In addition, it is possible that
other reactions, such as dissolution of carbonate
minerals, could modify the DIC mass balance. As
mentioned previously, SIs for calcite were negative at
all depths throughout the winter months. Carbonate
mineral dissolution is supported by the increase in
[Ca2?] towards the bottom of the water column at
GT-2 (Fig. 5e). Although CaCO3 has been identified
in shallow sediment from Georgetown Lake in the
form of gastropod shells, no d13C-CaCO3 data are
available at the time of this writing.
An interesting question that is presently unan-
swered deals with the fate of the CH4 produced near
the sediment–water interface in Georgetown Lake.
Because of strong stratification under ice cover, much
of the CH4 remains dissolved in the lower portion of
the water column until ice-off, at which time strong
winds induce vertical mixing and promote evasion of
CH4 to the atmosphere. This scenario has been
documented from other high-latitude lakes, and may
be important in the global CH4 budget (Michmerhu-
izen et al. 1996; Huttunen et al. 2003). It is also
possible that some CH4 may rise to the base of the ice
as bubbles of CO2–N2–CH4 gas mixtures (i.e., ebul-
lition) which then may be frozen into the ice (Anthony
et al. 2010; Boereboom et al. 2012). Oxidation and
assimilation of CH4 by methanotrophs is another
potentially important sink (Bastviken et al. 2003;
Lehmann et al. 2004; Utsumi et al. 1998).
Sulfur cycling under ice cover
The S-isotopic composition of sulfate in the shallow
water column at GT-2 (9.7 %) is similar to the average
S-isotopic composition of sulfate in the major tributaries
to the lake (9.2 %; Table 4). Given the existence of H2S
in the deep water column, the d34S of sulfate might be
expected to increase owing to BSR, which produces
isotopically light H2S. S-isotope fractionation accom-
panying BSR using natural organic matter as substrate
varies from 30 to 40 %, as long as concentrations of
sulfate are non-limiting (Canfield 2001). However, at
low sulfate concentrations (\1 mM), S-isotope frac-
tionation during BSR approaches zero (Harrison and
Thode 1958; Canfield and Raiswell 1999). Given the
very low concentrations of sulfate in Georgetown Lake
(\0.1 mM) coupled with the small observed S-isotope
separation between H2S and SO
2
4 ; it can be assumed
that BSR in the lake is strongly sulfate-limited. If all
sulfate was reduced to sulfide, then d34S of sulfide
would be the same as d34S of sulfate in the overlying
water column. At GT-2, the average of three
d34S-sulfide values (6.7 %) was slightly lower than
the d34S of sulfate in the upper water column (9.7 %;
Table 4). This suggests that the majority of the sulfate
was reduced to sulfide.
In contrast to d34S, sulfate in the shallow water at
GT-2 has a much lower value of d18O (-6.4 %)
compared to sulfate in the major tributaries (2.8,
0.8 %; Table 4). This can be explained by internal
cycling of S within the lake. In winter, much of the
sulfate in the lake is reduced to H2S by BSR. After
ice-off and spring overturn, much of this H2S is
probably re-oxidized to sulfate. This reaction typi-
cally involves formation of SO3 as an intermediate
species. Although the rate of O-isotope exchange
between SO24 and water is negligible at low temper-
ature, a number of studies have shown that exchange
of O-isotopes is rapid between SO3 and water (Lloyd
1968; Seal II 2003). Thus, SO24 formed by oxidation
of H2S may derive a considerable fraction of its O
atoms from water in the lake, which—in the case of
Georgetown Lake—is isotopically light (average
d18O-H2O at GT-2 = -14.7 %).
Conceptual site models
Using the cumulative data presented in this paper, we
have developed conceptual models to explain vertical
changes in water chemistry at both sampling locations
(Fig. 10). Based on the concentrations and isotopic
compositions of DO, oxygenic P under snow and ice
cover was more important at the shallower GT-2 site
compared to GT-1. One possible explanation for this
difference is that the supply of nutrients to the photic
zone at GT-1 is limited by deep currents which
intercept solutes diffusing upward from the sediment,
discharging them at the dam. In any event, [DO]
remained sufficiently high to allow overwinter sur-
vival of salmonids at GT-2, although the fish were
most likely confined to the top meter of the water
column by winter’s end. A second important
difference between the two sampling sites was the
presence of a purple-pigmented layer of putative
S-oxidizing bacteria about 1–2 m above the bottom of
the lake at GT-2, but not at GT-1. If the purple bacteria
are anoxygenic H2S-oxidizing phototrophs, the
absence a purple layer at GT-1 could be related to
the lower H2S concentrations at this site, or to the fact
that the euxinic zone at GT-1 is about 3 m deeper than
at GT-2 and therefore may lie below the photic zone.
Some other differences between GT-1 and GT-2
can be linked to hydrology. As mentioned earlier, the
outlet pipes at Georgetown Lake dam are located near
the bottom of the water column, and consequently they
discharge higher-salinity bottom water. This, com-
bined with a steeper topographic gradient in the
narrow outlet bay of the lake, would create a slow
current of deep, dense water heading towards the
outlet pipes (Fig. 10b). Solutes diffusing upwards
from sediment at GT-1 would have become entrained
into this current and thence advected away, rather than
accumulating in the deep water column, as at GT-2.
This may help to explain why increases in solute
concentrations with depth as well as vertical changes
in d13C-DIC were less dramatic at GT-1 compared to
GT-2. As well, water isotope profiles (Fig. 2) suggest
mixing of lake water and upwelling groundwater at
GT-1, but not at GT-2.
Conclusions
We have combined chemistry and stable isotopes to
better understand biogeochemical processes taking
place beneath ice cover in a shallow, highly productive
lake over the course of a winter season. In the period of
time between ice formation in November and ice-off
in late May, [DO] decreased while [DIC] increased,
consistent with rates of aerobic R exceeding rates of
oxygenic P. Nonetheless, based on an interpretation of
seasonal and vertical trends in [DO] and d18O-DO,
significant below-ice oxygenic P occurred, with higher
P rates at a shallower monitoring site (GT-2) as
compared to a deeper channel near the dam outlet
(GT-1). At GT-2, oxygenic P allowed the top meter of
the water column to remain sufficiently oxic to allow
over-winter survival of fish. Water near the bottom of
the lake at both sampling sites was anoxic, sulfidic,
highly enriched in DIC and ammonium, and slightly
enriched in dissolved Fe, Mn, SRP, and silica. These
trends, which were more pronounced at GT-2 as
compared to GT-1, are explained by upward diffusion
from sediment-pore water where organic matter was
being decomposed.
Stable isotopes of DIC were particularly helpful in
this study to track biogeochemical processes under ice
cover in Georgetown Lake. Values of d13C-DIC
initially decreased in response to net production of
isotopically light CO2 by R, but then showed a sharp
increase in the deep water towards the end of the ice-
covered season. This increase in d13C-DIC, accompa-
nied by a sharp increase in [DIC] and [CH4], is
attributed to acetoclastic methanogenesis near the
sediment–water interface, a process that produces
13C-enriched DIC. Methanogenesis was likely favored
by a lack of competition for organic substrates from
sulfate reducing bacteria, whose activity was con-
strained by very low sulfate concentrations (Lovley
and Klug 1983). Although some CH4 may have been
consumed within the water column by methanotrophs,
an unquantified mass of methane likely escaped to the
atmosphere during spring turnover, which occurred
within 48 h of ice-off.
Seasonal temperature averages and extremes in
western Montana have been increasing two–three
times faster than the global average (Pederson et al.
2010); hence, decreases in the duration of seasonal ice
cover may be particularly rapid among the lakes of this
region. In Georgetown Lake, under-ice P was signif-
icant at shallow depth and possibly critical to over-
Fig. 10 Conceptual models
summarizing hydrological,
geochemical, and biological
processes impacting water
quality at GT-2 (top) and
GT-1 (bottom). A lush
growth of pondweed at
GT-2 in early winter (top
left) dies back to roots in
winter, creating a large mass
of biodegradable organic
carbon at the bottom of the
lake (top right)
winter survival of fish. Nonetheless, Georgetown Lake
was R-dominated in winter, becoming methanogene-
sis-dominated near the sediment–water interface.
Stratification due to ice cover allowed for the devel-
opment of steep vertical gradients in the concentra-
tions of O2, CO2, CH4, NH4
?, NHþ4 ; SO
2
4 ; H2S, Fe
2?,
and Mn2? at the sub-meter scale, creating a complex
hierarchy of potential ecological niches for heterotro-
phic and photosynthetic (both oxygenic and anoxy-
genic) microbes. Ice cover-induced stratification
clearly structures the biogeochemical environment;
however, the consequences of a longer ice-free period
for fisheries depend critically on the P–R balance,
which may be sensitive to other factors such as
temperature and organic carbon loading. Future stud-
ies will attempt to link the observed chemical and
isotopic gradients at Georgetown Lake to spatial and
temporal changes in carbon pools and specific micro-
bial populations and rate processes.
Acknowledgments This research was partly funded by
collaborative Grants 0739054 and 0738912, with follow-up
support from NSF-EPSCoR and the Montana Institute on
Ecosystems. We thank Craig Stafford (University of Montana)
for sharing his ideas and data on Georgetown Lake. This paper
was substantially improved by the comments of two anonymous
reviewers and the Associate Editor, R. Kelman Wieder.
References
Allison JD, Brown DS, Novo-Gradac KJ (1991) MINTEQA2/
PRODEFA2, a geochemical assessment model for envi-
ronmental systems. US Environmental Protection Agency,
EPA/600/3-91/021
Angert A, Luz B, Yakir D (2001) Fractionation of oxygen iso-
topes by respiration and diffusion in soils and its implica-
tions for the isotopic composition of atmospheric O2. Glob
Biogeochem Cycles 15:871–880
Anthony KM, Vas D, Brosius L, Chapin F III, Zimov SA,
Zhuang Q (2010) Estimating methane emissions from
northern lakes using ice-bubble surveys. Limnol Oceanogr
Methods 8:592–609
Assayag N, Je´ze´quel D, Ader M, Viollier E, Michard G, Pre´vot
F, Agrinier P (2008) Hydrological budget, carbon sources
and biogeochemical processes in Lac Pavin (France):
constraints from d18O of water and d13C of dissolved
inorganic carbon. Appl Geochem 23:2800–2816
Babin J, Prepas EE (1985) Modelling winter oxygen depletion
rates in ice-covered temperate zone lakes in Canada. Can J
Fish Aquat Sci 42:239–249
Bade DL, Carpenter SR, Cole JJ, Hanson PC, Hesslein RH
(2004) Controls of d13C-DIC in lakes: geochemistry, lake
metabolism, and morphometry. Limnol Oceanogr 49:
1160–1172
Bass AM, Waldron S, Preston T, Adams CE, Drummond J
(2010a) Temporal and spatial heterogeneity in lacustrine
d13CDIC and d
18ODO signatures in a large mid-latitude
temperate lake. J Limnol 69:341–349
Bass AM, Waldron S, Preston T, Adams CE (2010b) Net pelagic
heterotrophy in mesotrophic and oligotrophic basins of a
large, temperate lake. Hydrobiologia 652:363–375
Bastviken D, Ejlertsson J, Sundh I, Tranvik L (2003) Methane as
a source of carbon and energy for a lake pelagic food web.
Ecology 84:969–981
Bastviken DJ, Cole J, Pace M, Tranvik L (2004) Methane
emissions from lakes: dependence of lake characteristics,
two regional assessments, and a global estimate. Glob
Biogeochem Cycles 18:GB4009
Benson BB, Krause D Jr (1980) The concentration and isotopic
fractionation of gases dissolved in freshwater in equilib-
rium with the atmosphere: 1, oxygen. Limnol Oceanogr
25:662–671
Bertilsson S, Burgin A, Carey CC, Fey SB, Grossart H-P,
Grubisic LM, Jones ID, Kirillin G, Lennon JT, Shade A,
Smyth RL (2013) The under-ice microbiome of seasonally
frozen lakes. Limnol Oceanogr 58:1998–2012
Boereboom T, Depoorter M, Coppens S, Tison JL (2012) Gas
properties of winter lake ice in Northern Sweden: impli-
cation for carbon gas release. Biogeosciences 9:827–838
Brandes JA, Devol AH (1997) Isotopic fractionation of oxygen
and nitrogen in coastal marine sediments. Geochim Cos-
mochim Acta 61:1793–1801
Brown JM, Poulson SR (2010) Biogeochemical controls on
seasonal variations of the stable isotopes of dissolved
oxygen and dissolved inorganic carbon in Castle Lake, CA.
In: AGU, fall meeting 2010, abstract #H54B-02
Canfield DE (2001) Isotope fractionation by natural populations
of sulfate-reducing bacteria. Geochim Cosmochim Acta
65:1117–1124
Canfield DE, Raiswell R (1999) The evolution of the sulfur
cycle. Am J Sci 299:697–723
Carmody RW, Plummer LN, Busenberg E, Coplen TB (1998)
Methods for collection of dissolved sulfate and sulfide and
analysis of their sulfur isotopic composition. US Geologi-
cal Survey, Open File Report 997-234
Clark J, Fritz P (1997) Environmental isotopes in hydrogeology.
Lewis Publishers, Boca-Raton
Craig H (1961) Isotopic variations in meteoric waters. Science
133:1702–1703
Dibike Y, Prowse T, Bonsal B, de Rham L, Saloranta T (2012)
Simulation of North American lake-ice cover characteris-
tics under contemporary and future climate conditions. Int
J Climatol 32:695–709
Dunnette DA, Chynoweth DP, Khalil HM (1985) The source of
hydrogen sulfide in anoxic sediment. Water Res
19:875–884
Epstein S, Mayeda T (1953) Variation of O-18 content of waters
from natural sources. Geochim Cosmochim Acta 4:
213–224
Falkowski PG, Raven JA (1997) Aquatic photosynthesis.
Blackwell Sciences, Oxford
Gammons CH, Poulson SR, Pellicori DA, Roesler A, Reed PJ,
Petrescu EM (2006) The hydrogen and oxygen isotopic
composition of precipitation, evaporated mine water, and
river water in Montana, USA. J Hydrol 328:319–330
Gammons CH, Pape BL, Parker SR, Poulson SR, Blank CE
(2013) Geochemistry, water balance, and stable isotopes of
a ‘‘clean’’ pit lake at an abandoned tungsten mine, Mon-
tana, USA. Appl Geochem 36:57–69
Garrett PA (1983) Relationships between benthic communities,
land use, chemical dynamics, and trophic state in George-
town Lake. PhD Dissertation, Montana State University
Garrison RJ (1976) Sediment chemistry of Georgetown Lake,
Montana. MS Thesis, Montana State University
Gibson JAE (1999) The meromictic lakes and stratified marine
basins of the Vestfold Hills, East Antarctica. Antarct Sci
11:175–192
Giesemann A, Jager HJ, Norman AL, Krouse HP, Brand WA
(1994) On-line sulfur-isotope determination using an ele-
mental analyzer coupled to a mass spectrometer. Anal
Chem 66:2816–2819
Greenbank J (1945) Limnological conditions in ice-covered
lakes, especially as related to winterkill of fish. Ecol Mo-
nogr 15:343–392
Gu B, Schelske CL, Hodell DA (2004) Extreme 13C enrichments
in a shallow hypereutrophic lake: implications for carbon
cycling. Limnol Oceanogr 49:1152–1159
Guy RD, Fogel ML, Berry JA (1993) Photosynthetic fraction-
ation of the stable isotopes of oxygen and carbon. Plant
Physiol 101:37–47
Harris D, Porter LK, Paul EA (1997) Continuous flow isotope
ratio mass spectrometry of carbon dioxide trapped as
strontium carbonate. Commun Soil Sci Plant Anal
28:747–757
Harrison AG, Thode HG (1958) Mechanisms of the bacterial
reduction of sulfate from isotope fractionation studies.
Trans Faraday Soc 53:84–92
Herczeg AL (1987) A stable carbon isotope study of dissolved
inorganic carbon cycling in a softwater lake. Biogeo-
chemistry 4:231–263
Huttunen JT, Alm J, Lukanen A, Juutinen S, Larmola T, Ham-
mar T, Silvola J, Martikainen PJ (2003) Fluxes of methane,
carbon dioxide and nitrous oxide in boreal lakes and
potential anthropogenic effects on the aquatic greenhouse
gas emissions. Chemosphere 52:609–621
Karim A, Dubois K, Veizer J (2011) Carbon and oxygen
dynamics in the Laurentian Great Lakes: implications for
the CO2 flux from terrestrial aquatic systems to the atmo-
sphere. Chem Geol 281:133–141
Kiddon J, Bender ML, Orchardo J, Caron DA, Goldman JC,
Dennett M (1993) Isotope fractionation of oxygen by
respiring marine organisms. Glob Biogeochem Cycles
7:679–694
Knight JC (1981) An investigation of the general limnology of
Georgetown Lake, Montana. PhD Dissertation, Montana
State University
Kornexl BE, Gehre M, Ho¨ffling R, Werner RA (1999) On-line
d18O measurement of organic and inorganic substances.
Rapid Commun Mass Spectrom 13:1685–1693
Kroopnick PM (1975) Respiration, photosynthesis and oxygen
isotope fractionation in oceanic surface water. Limnol
Oceanogr 20:988–992
Kurz WA, Dymond CC, Stinson G, Rampley GJ, Neilson ET,
Carroll AL, Ebata T, Safranyik L (2008) Mountain pine
beetle and forest carbon feedback to climate change. Nat-
ure 452:987–990
Lane GA, Dole M (1956) Fractionation of oxygen isotopes
during respiration. Science 123:574–576
Lehmann MF, Bernasconi SM, McKenzie JA, Barbieri A, Si-
mona M, Veronesi M (2004) Seasonal variation of the d13C
and d15N of particulate and dissolved carbon and nitrogen
in Lake Lugano: constraints on biogeochemical cycling in
a eutrophic lake. Limnol Oceanogr 49:415–429
Lloyd RM (1968) Oxygen isotope behavior in the sulfate–water
system. J Geophys Res 73:6099–6110
Lonn JD, McDonald C, Lewis RS, Kalakay TJ, O’Neill JM,
Berg RB, Hargrave P (2003) Preliminary geologic map of
the Philipsburg 300 9 600 quadrangle, western Montana.
Montana Bureau of Mines and Geology, Open File 483
Lovley DR, Klug MJ (1983) Sulfate reducers can outcompete
methanogens at freshwater sulfate concentrations. Appl
Environ Microbiol 45:187–192
Luz B, Barkan E, Sagi Y, Yacobi YZ (2002) Evaluation of
community respiratory mechanisms with oxygen isotopes:
a case study in Lake Kinneret. Limnol Oceanogr 47:33–42
Mandernack KW, Mills CT, Johnson CA, Rahn T, Kinney C
(2009) The d15N and d18O values of N2O produced during
the co-oxidation of ammonia by methanotrophic bacteria.
Chem Geol 267:96–107
Mathias JA, Barica J (1980) Factors controlling oxygen deple-
tion in ice-covered lakes. Can J Fish Aquat Sci 37:185–194
Michmerhuizen CM, Striegl RG, McDonald ME (1996)
Potential methane emission from north-temperate lakes
following ice melt. Limnol Oceanogr 41:985–991
Morrison J, Brockwell T, Merren T, Fourel F, Phillips AM
(2001) On-line high precision stable hydrogen isotopic
analyses on nanoliter water samples. Anal Chem 73:
3570–3575
Myrbo A, Shapley MD (2006) Seasonal water-column dynamics
of dissolved inorganic carbon stable isotope compositions
(d13CDIC) in small hardwater lakes in Minnesota and
Montana. Geochim Cosmochim Acta 70:2699–2714
Oba Y, Poulson SR (2009a) Oxygen isotope fractionation of
dissolved oxygen during abiological reduction by aqueous
sulfide. Chem Geol 268:226–232
Oba Y, Poulson SR (2009b) Oxygen isotope fractionation of
dissolved oxygen during reduction by ferrous iron. Geo-
chim Cosmochim Acta 73:13–24
Parker SR, Poulson SR, Gammons CH, DeGrandpre M (2005)
Biogeochemical controls on diel cycles in the stable iso-
topic composition of dissolved O2 and DIC in the Big Hole
River, Montana, USA. Environ Sci Technol 39:7134–7140
Parker SR, Gammons CH, Poulson SR, DeGrandpre MD, We-
yer CL, Smith MG, Babcock JN, Oba Y (2010) Diel
behavior of stable isotopes of dissolved oxygen and dis-
solved inorganic carbon in rivers over a range of trophic
conditions, and in a mesocosm experiment. Chem Geol
269:22–32
Parker SR, Darvis MN, Poulson SR, Gammons CH, Stanford JA
(2014) Dissolved oxygen and dissolved inorganic carbon
stable isotope composition and concentrations fluxes
across several shallow floodplain aquifers and in a diffu-
sion experiment. Biogeochemistry 117:539–552
Pederson GT, Graumlich LJ, Fagre DB, Kipfer T, Muhlfeld CC
(2010) A century of climate and ecosystem change in
western Montana: what do temperature trends portend?
Clim Change 98:133–154
Poulson SR, Sullivan AB (2010) Assessment of diel chemical
and isotopic techniques to investigate biogeochemical
cycles in the upper Klamath River, Oregon, USA. Chem
Geol 269:3–11
Quay PD, Emerson SR, Quay BM, Devol AH (1986) The carbon
cycle for Lake Washington—a stable isotope study. Lim-
nol Oceanogr 31:596–611
Quay PD, Wilbur DO, Richey JE, Devol AH (1995) The
18O:16O of dissolved oxygen in rivers and lakes of the
Amazon Basin: determining the ratio of respiration to
photosynthesis rates in freshwaters. Limnol Oceanogr
40:718–729
Raleigh RF, Hickman T, Soloman RC, Nelson PC (1984)
Habitat suitability information: rainbow trout (Oncorhyn-
chus mykiss). US Fish Wildlife Service: FWS/OBS-82/
10.60
Scidmore WJ (1957) An investigation of carbon dioxide,
ammonia, and hydrogen sulfide as factors contributing to
fish kills in ice-covered lakes. Progress Fish Cult
19:124–127
Seal RR II (2003) Stable-isotope geochemistry of mine waters
and related solids. In: JL Jambor, DW Blowes, AIM
Ritchie, eds. Environmental Aspects of Mine Wastes.
Miner Soc Can Short Course Ser 31:303–334
Shaw GE, White ES, Gammons CH (2013) Characterizing
groundwater–lake interaction and its impact on lake water
quality. J Hydrol 492:69–78
Smith MG, Parker SR, Gammons CH, Poulson SR, Hauer FR
(2011) Tracing dissolved O2 and dissolved inorganic car-
bon stable isotope dynamics in the Nyack aquifer: Middle
Fork Flathead River, Montana, USA. Geochim Cosmo-
chim Acta 75:5971–5986
Stafford CP (2013) Long-term trends in the water quality of
Georgetown Lake, Montana. Prepared for Montana
Department of Justice and Montana Department of Envi-
ronmental Quality
Stiller M, Magaritz M (1974) Carbon-13 enriched carbonate in
interstitial waters of Lake Kinneret sediments. Limnol
Oceanogr 19:849–853
Trabert MJ (1993) The depletion of oxygen in Georgetown
Lake, Montana, during the winter months. MS Thesis,
Montana Tech
Tranvik LJ et al. (2009) Lakes and reservoirs as regulators of
carbon cycling and climate. Limnol Oceanogr 54:
2298–2314
US Environmental Protection Agency (1976) Preliminary report
on Georgetown Lake, national eutrophication survey.
CERL, Corvallis
Usdowski E, Hoefs J, Menschel G (1979) Relationship between
13C and 18O fractionation and changes in major element
composition in a recent calcite-depositing spring—a model
of chemical variations with inorganic CaCO3 precipitation.
Earth Planet Sci Lett 42:267–276
Utsumi M, Nojiri Y, Nakamura T, Nozawa T, Otsuki A, Seki H
(1998) Oxidation of dissolved methane in a eutrophic,
shallow lake: Lake Kasumigaura, Japan. Limnol Oceanogr
43:471–480
Wachniew P, Rozanski K (1997) Carbon budget of a mid-lati-
tude, groundwater-controlled lake: isotopic evidence for
the importance of dissolved inorganic carbon recycling.
Geochim Cosmochim Acta 61:2453–2465
Wang X, Veizer J (2000) Respiration–photosynthesis balance of
terrestrial aquatic ecosystems, Ottawa area, Canada. Geo-
chim Cosmochim Acta 64:3775–3786
Wang X, Depew D, Schiff S, Smith REH (2008) Photosynthesis,
respiration, and stable isotopes of oxygen in a large oli-
gotrophic lake (Lake Erie, USA-Canada). Can J Fish Aquat
Sci 65:2320–2331
Wassenaar LI, Koehler G (1999) An on-line technique for the
determination of the delta O-18 and delta O-17 of gaseous
and dissolved oxygen. Anal Chem 71:4965–4968
Weyhenmeyer GA, Livingstone DM, Meili M, Jensen O, Ben-
son B, Magnuson JJ (2011) Large geographical differences
in the sensitivity of ice-covered lakes and rivers in the
Northern Hemisphere to temperature changes. Glob
Change Biol 17:268–275
Whiticar MJ (1999) Carbon and hydrogen isotope systematics of
bacterial formation and oxidation of methane. Chem Geol
161:291–314
Whiticar MJ, Faber E (1986) Methane oxidation in sediment and
water column environments—isotope evidence. Org Geo-
chem 10:759–768
Whiticar MJ, Faber E, Schoell M (1986) Biogenic methane
formation in marine and freshwater environments: CO2
reduction vs. acetate fermentation—isotope evidence.
Geochim Cosmochim Acta 50:693–709
Wiesenburg DA, Guinasso NL Jr (1979) Equilibrium solubili-
ties of methane, carbon monoxide and hydrogen in water
and sea water. J Chem Eng Data 24:356–360