Bio-optical characteristics of the snow, ice, and water column of a perennially ice-covered lake in the High Arctic

Abstract

Lake A is a meromictic, perennially ice-covered lake located at the northern limit of North America (latitude 83°N, Ellesmere Island, Canada). In early June 1999, only 0.45% of incident photosynthetically available radiation (PAR) was transmitted through its 2-m ice and 0.5-m snow cover. Removal of snow from 12 m2 increased PAR under the ice by a factor of 13 and biologically effective ultraviolet radiation (UVR) by a factor of 16 (from 0.4% to 6.3% of incident). The diffuse attenuation coefficient (Kd) for UVR was substantially lower in the ice than in the underlying freshwater (e.g., 50% lower at 320 nm), indicating the exclusion of chromophoric dissolved organic matter (CDOM) during freeze-up or the subsequent degradation of CDOM retained in the ice. Peak phytoplankton concentrations occurred immediately under the ice, and a broad maximum of photosynthetic sulfur bacteria and associated sulfur particles was observed over the depth interval 20–45 m at <0.005% of incident PAR. Climate-induced changes in the overlying snow and ice have the potential to cause major habitat disruption (UV exposure, PAR, temperature, mixing regime) for these stratified, extreme-shade communities.

Full text

Bio-optical characteristics of the snow, ice, and
water column of a perennially ice-covered lake in
the High Arctic
Claude Belzile, Warwick F. Vincent, John A.E. Gibson, and Patrick Van Hove
Abstract: Lake A is a meromictic, perennially ice-covered lake located at the northern limit of North America (latitude
83°N, Ellesmere Island, Canada). In early June 1999, only 0.45% of incident photosynthetically available radiation
(PAR) was transmitted through its 2-m ice and 0.5-m snow cover. Removal of snow from 12 m
2
increased PAR under
the ice by a factor of 13 and biologically effective ultraviolet radiation (UVR) by a factor of 16 (from 0.4% to 6.3% of
incident). The diffuse attenuation coefficient (K
d
) for UVR was substantially lower in the ice than in the underlying
freshwater (e.g., 50% lower at 320 nm), indicating the exclusion of chromophoric dissolved organic matter (CDOM)
during freeze-up or the subsequent degradation of CDOM retained in the ice. Peak phytoplankton concentrations
occurred immediately under the ice, and a broad maximum of photosynthetic sulfur bacteria and associated sulfur
particles was observed over the depth interval 20–45 m at <0.005% of incident PAR. Climate-induced changes in the
overlying snow and ice have the potential to cause major habitat disruption (UV exposure, PAR, temperature, mixing
regime) for these stratified, extreme-shade communities.
Résumé : Le lac A est un lac méromictique couvert de glace en permanence et situé à la limite boréale du continent
nord-américain (latitude 83°N, île d’Ellesmere, Canada). Au début de juin 1999, seulement 0,45 % de la radiation
incidente disponible pour la photosynthèse (PAR) pénétrait la couche de2mdeglacerecouverte de 0,5 m de neige.
Le retrait de la neige sur une surface de 12 m
2
a augmenté la PAR d’un facteur de 13 et la radiation ultraviolette
(UVR) à effets biologiques, d’un facteur de 16 (de 0,4%à6,3%delaradiation incidente). Le coefficient
d’atténuation (K
d
) de l’UVR était considérablement moins élevé dans la glace que dans l’eau douce sous-jacente (e.g.
50 % plus faible à 320 nm), ce qui indique une exclusion de la matière organique dissoute colorée (CDOM) lors du
gel ou la dégradation subséquente de la CDOM retenue dans la glace. Les concentrations maximales de phytoplancton
se trouvaient juste sous la glace et la densité maximale des bactéries sulfureuses photosynthétiques et des particules de
soufre associées s’étalait dans l’intervalle des profondeurs de 20–45 m, à moins de 0,005 % de la PAR incidente. Les
changements dans la couverture de neige et de glace générés par le climat peuvent potentiellement causer des modifi-
cations majeures de l’habitat (exposition à l’UV, PAR, température, régime de brassage) dans ces communautés strati
-
fiées acclimatées à de faibles intensités lumineuses.
[Traduit par la Rédaction] Belzile et al. 2418
Introduction
Perennially ice-covered lakes are an important limnologi
-
cal feature of Antarctica and have proved to be useful model
systems for exploring general concepts in geophysics, geo
-
chemistry, and microbial ecology. In recognition of this
value, a series of such lakes in the McMurdo Dry Valleys
region is now the site of a long-term ecological research pro
-
gram (LTER; Priscu 1998). It is less widely known, how
-
ever, that lakes covered by multiyear ice are also found in
the north polar region. Several lakes on northern Ellesmere
Island are thought to be perennially ice covered (Ludlam
1996) and other lakes in the High Arctic retain their ice
cover throughout summer in some years (e.g., Doran et al.
1996). Our aim in the present study was to evaluate the ice
cover and water-column properties of one such north polar
lake, Lake A in the Canadian High Arctic, with emphasis on
its bio-optical properties relative to analogous lakes in the
south polar region.
Ice and snow cover high latitude lakes for more than
8 months of the year and have a broad range of effects on
the physical and biological properties of these ecosystems.
The presence of ice isolates the water from wind-induced
mixing, and thick snow and ice may also inhibit convective
processes. This reduced mixing in turn affects the vertical
distribution of gases, nutrients, and planktonic organisms
(Wharton et al. 1993; Priscu et al. 1999a). The effects of ice
and snow cover on radiative transfer also regulate biological
productivity by controlling photosynthetically available radi
-
ation (PAR), exposure to ultraviolet radiation (UVR), and
water temperature. Despite this pervasive influence on lim
-
nological processes in northern lakes, little attention has
been given to the bio-optics of freshwater ice and snow.
The High Arctic is currently subject to major changes in
Can. J. Fish. Aquat. Sci. 58: 2405–2418 (2001) © 2001 NRC Canada
2405
DOI: 10.1139/cjfas-58-12-2405
Received May 8, 2001. Accepted October 24, 2001.
Published on the NRC Research Press Web site at
http://cjfas.nrc.ca on December 14, 2001.
J16340
C. Belzile,
1
W.F. Vincent, J.A.E. Gibson, and P. Van
Hove. Département de biologie and Centre d’études
nordiques, Université Laval, Sainte-Foy, QC G1K 7P4,
Canada.
1
Corresponding author (e-mail: claude.belzile@bio.ulaval.ca).
J:\cjfas\cjfas58\cjfas-12\F01-187.vp
Tuesday, December 11, 2001 2:33:16 PM
Color profile: Disabled
Composite Default screen

climate. The most striking evidence of change has been in
records of the extent and thickness of multiyear sea ice
across the Arctic Ocean. Over the period 1978 to 1998, the
multiyear sea-ice cover diminished in area by 14% in winter
(Johannessen et al. 1999) and by 44% in average thickness
during the past three decades (Rothrock et al. 1999). An ex
-
tensive ice shelf (>10-m-thick sea ice) along the northern
coast of Ellesmere Island retreated by 90% over the course
of the 20th century, with evidence of substantial thinning of
the remaining landfast ice in the 1980s and 1990s (Vincent
et al. 2001). Continental High Arctic meteorological records
also show evidence of change. Temperature data from Arctic
stations over the period 1966–1995 indicate a general warm
-
ing trend, with the greatest effects in the western Arctic (up
to 0.7°C per decade; Weller 1998). There is wide regional
variability in precipitation trends, from significant increases
at certain locations (e.g., Spitsbergen; Hanssenbauer and
Forland 1998) to significant decreases at other sites (e.g.,
Alaska; Curtis et al. 1998). This climatic change is likely to
have an impact on lake ecosystems via shifts in snow and
ice cover (e.g., Doran et al. 1996) and highlights the need
for improved understanding of the properties of these lake
ecosystem components.
The High Arctic is also experiencing changes in its inci-
dent UV radiation flux. During spring, the increases in sur-
face erythemal UVR resulting from severe chemical ozone
losses observed in the Arctic during winter are estimated to
be about 22% relative to the values in the 1970s (Madronich
et al. 1998). This increase is likely to worsen in duration and
severity in the future, in part associated with greenhouse gas
effects on stratospheric cooling (Shindell et al. 1998).
Although many modeling and field studies have addressed
the variations of snow and ice albedo, little is known about
the transmission of UVR through snow and freshwater ice,
and even the PAR attenuation characteristics of these media
are poorly understood. Clean, fresh, cold snow has a high
albedo (>0.9) and is therefore relatively opaque to PAR
(Perovich et al. 1998); however, during the aging of the
snow or when it becomes wet or covered by wind-blown
particles, the PAR albedo can drop substantially (Grenfell
and Maykut 1977; Perovich et al. 1998). The UVR albedo of
snow is lower than that for PAR (Perovich et al. 1998). The
measured diffuse attenuation coefficients for PAR, K
d
(PAR),
of snow are highly variable depending on snow density,
grain size, temperature, and wetness (Thomas 1963; Grenfell
and Maykut 1977). To our knowledge, no K
d
(UVR) values
for snow have been published to date. K
d
(PAR) of thick pe
-
rennial ice has been reported to be as low as 0.2 m
–1
for
Antarctic lakes (Howard-Williams et al. 1998), and transmis
-
sion of incident PAR through snow-free, 1.5- to 2.2-m-thick
clear ice was shown to be as high as 21–35% for High Arc
-
tic lakes (Welch and Kalff 1974; Bolsenga et al. 1996). The
limited available literature (Ellis-Evans et al. 1998; fig. 4 in
Howard-Williams et al. 1998) suggests that UVR transmis
-
sion through clear ice can be of the same order of magni
-
tude, or greater, than PAR transmission. This may result
from the strong within-ice attenuation of the red part of the
PAR spectrum. Kepner et al. (2000), however, found low UV
transmission through the ice cover of Lake Hoare in the
McMurdo Dry Valleys (<0.6% at 320 nm), and this was at
-
tributed to the extremely degraded, highly scattering ice sur
-
face and the high concentrations of sediment on and within
this 4- to 5-m-thick ice. Significant under-ice UVR has been
measured in other Antarctic Dry Valleys lakes despite >3-m-
thick ice (Vincent et al. 1998); however, there have been no
equivalent measurements in Arctic freshwater ecosystems.
In the present study, we addressed the hypothesis that
High Arctic lakes may be sensitive to environmental change
by virtue of their overlying snow and ice. We undertook
measurements of UVR and PAR transmission through the
snow and ice cover and of optical properties of the underlying
water column of perennially ice-covered Lake A on northern
Ellesmere Island. We selected Lake A because it lies at the
northern limit of Arctic lakes where future climate changes
and ozone depletion are likely to be pronounced.
Materials and methods
Study site
Lake A is located on northern Ellesmere Island (83°00
′
N,
75°27
′
W; Fig. 1) in a polar desert region with mean annual precipi
-
tation of 157 mm and mean annual temperature of –17.9°C (Ta
-
ble 1). This precipitation is much greater than that in the McMurdo
LTER (
≤
50 mm) at a comparable latitude in Antarctica (78°S), but
the mean annual temperature is similar to that in the McMurdo
lakes region (–17°C to –19°C; Doran et al. 1994). Table 1 summa-
rizes the mean values for climate variables for the period 1951–
1999 at Alert, the nearest available meteorological station, 180 km
from Lake A (no local data are available as yet from the lake ba-
sin). Lake A was first sampled in 1969 by Geoffrey Hattersley-
Smith’s group (Hattersley-Smith et al. 1970) and was the object of
work in 1982 (Jeffries et al. 1984) and 1993 (Ludlam 1996). The
lake depth is >115 m (deepest measured point) and has an area of
4.9 km
2
and a glacier-free drainage basin with an area of 36 km
2
.
A short outlet stream flows from the northernmost point of the lake
to an epishelf lake dammed by the Ward Hunt Ice Shelf. The lake
is 3.3 m above sea level and was cut off from the sea following iso-
static uplift approximately 3000 years ago, thereby producing the
current meromictic conditions (Hattersley-Smith et al. 1970;
Jeffries et al. 1984). The conductivity of the monimolimnion is ap
-
proximately the same as seawater and its ionic composition is that
of modern ocean water although modified by biological activity; it
is depleted in sulfate and enriched in dissolved inorganic carbon by
microbial activity (Jeffries et al. 1984). The monimolimnion and
mixolimnion are separated by a chemocline that extends from 7 to
25 m. A thermal maximum of 8–9°C occurs at ~15 m and the wa
-
ters are anoxic below 14 m (Ludlam 1996). Lake A is covered by
2-m-thick ice, with a candled ice surface resulting from the sum
-
mertime partial melting (Hattersley-Smith et al. 1970). A narrow
moat opens around the ice pan during summer, at least in some
years. On August 5, 2001, the moat was 1–10 m wide, except on
the eastern side of the lake at the site of the main inflowing stream
where the moat was about 100 m wide (C. Belzile and P. Van
Hove, personal observation). RADARSAT data for one of the
warmest summers on record (1998; Vincent et al. 2001) show the
persistence of Lake A ice in late August, whereas the nearby Taco
-
nite Inlet lakes (40 km southwest of Lake A) lost their ice cover
(Fig. 2). However, the moat on the northern shore of Lake A was
much wider (30–120 m; Fig. 2) than observed in 2001. Lake A was
covered by ice during late summer, or candled ice was present un
-
der the snow in early summer (indicating multiyear ice cover), on
each of 11 years of observations during the period 1950–2001.
Sampling
Lake A was sampled during the period of 3–8 June 1999. This
time of the year is characterized by continuous daylight and the solar
© 2001 NRC Canada
2406 Can. J. Fish. Aquat. Sci. Vol. 58, 2001
J:\cjfas\cjfas58\cjfas-12\F01-187.vp
Tuesday, December 11, 2001 2:33:17 PM
Color profile: Disabled
Composite Default screen

© 2001 NRC Canada
Belzile et al. 2407
elevation varies from 13 to 33° over the 24-h cycle. Air temperatures
at Alert varied from –0.5°C to –10.5°C during the 6 days. Water sam
-
ples were obtained from 25 selected depths over the deepest part of
the lake using a Kemmerer sampling bottle (length = 31 cm) and
were stored in dark Nalgene bottles until filtration 1–3 h later at our
base camp. Profiles of temperature, specific conductivity, and dis
-
solved oxygen were made using a Hydrolab Surveyor 3 (Hydrolab
Corp., Austin, Tex.). Data were recorded at discrete depths to allow
stabilization of the instrument; measurements were made every metre
from 2 to 13 m and every 50 cm down to 47.5 m (the full length of
the profiler cable). Specific conductivity readings were converted to
salinity using the algorithm of Fofonoff and Millard (1983). This al
-
gorithm is the equation of state for seawater and is applicable over the
salinity range of 2–40. Terrestrial input, differential diffusion, and bi
-
ological processes will alter the ionic composition of the water, but
with the exception of the surface waters, this variation should be mi
-
nor considering that the water in the lake is largely marine derived
(Jeffries et al. 1984). The thickness of the ice was measured in five
holes distributed over the central region of the lake (spaced by at
most 500 m). Snow depth was measured at the site of transmittance
measurements (n = 15 over a 12-m
2
area) and every 1 m along a 20-
m transect nearby this site. All optical measurements (described be
-
low) were made within5hoflocal noon (solar elevation between 25
and 33°) during clear sky conditions or high light clouds.
Albedo measurements
The snow and cleared-ice albedo (
α
(
λ
)), i.e., the ratios of
upwelling irradiance (E
u
) to downwelling irradiance (E
d
), were
measured using a PUV510 radiometer (Biospherical Instruments,
Inc., San Diego, Calif.) at the same site as the transmittance mea
-
surements (see below). The PUV510 measured cosine-corrected
UVR at 305, 320, 340, and 380 nm (full bandwidth at half maxi
-
mum of 8–10 nm) and PAR (400–700 nm). The radiometer was
mounted on a 3-m-long pole, which allowed the measurement of
E
d
and E
u
at a height of 1 m over the snow or ice surface.
Irradiance transmittance through the ice and snow cover
The transmittance of the snow and ice (T(
λ
)) was calculated as
the ratio of downwelling irradiance of wavelength
λ
at the lower
surface of the ice (E
d
(z
ice
,
λ
)) to the incident irradiance (E
d
(0
+
,
λ
)).
Transmittance depends on the specular and volume components of
the albedo and on the attenuation of irradiance by snow and ice ac
-
cording to Beer’s Law. For these transmittance measurements, the
PUV510 was installed in an unshaded location on the lake ice, and
an underwater radiometer (PUV500, Biospherical Instruments, Inc.)
was positioned to measure simultaneously the cosine-corrected,
under-ice irradiance. The PUV500 recorded irradiance at the same
wavelengths as the PUV510 and also recorded depth and tempera
-
ture. Underwater irradiance measurements were corrected for dark
current by subtracting the value obtained at in situ temperatures
within the water column after fitting the radiometer with a light-
tight Neoprene cap. The PUV500 radiometer was positioned under
the ice through a 65-cm-diameter hole using an articulated arm
designed in our laboratory. The aluminum arm allowed the posi-
tioning of the radiometer flush with the lower ice surface at a dis-
tance of 1 m from the hole. During the measurements, the hole was
tightly covered with an opaque plastic board, 1 m
2
in area, to pre-
vent direct solar radiation from entering the ice sheet through the
hole. Great care was taken to preserve the integrity of the snow
cover over the area where the measurements were made in order to
obtain transmittance through the natural ice and snow cover. The
snow cover was then removed from an area of 12 m
2
, and the mea-
surements were repeated under that area to estimate transmittance
of only the ice.
Snow and ice attenuation coefficients
The attenuation of irradiance in ice calculated from radiative
transfer models is often characterized by the asymptotic extinction
coefficient K
asym
(
λ
) (e.g., Grenfell 1991). In an optically thick me
-
dium, the asymptotic state, where the shape of the light field is axi
-
ally symmetric and does not change, should be approached far
from the upper boundary. K
asym
(
λ
) is difficult to measure in the
field and small-scale variability in the physical, biological, and
thus optical properties of the ice makes it difficult for the light dis
-
tribution to remain in the asymptotic state (see Perovich et al.
1998). One simpler way of characterizing irradiance attenuation in
ice from field measurements is the “effective” extinction coeffi
-
cient (Grenfell 1991). This effective downwelling diffuse attenua
-
tion coefficient (K
d
(
λ
), m
–1
) for ice can be calculated as
(1) K
d
(
λ
)
ice
= –ln(E
d
(
λ
, z
ice
)/E
d
(
λ
,0
–
))/z
ice
where E
d
(
λ
, z
ice
) is the under-ice downwelling irradiance measured
in the absence of snow and E
d
(
λ
,0
–
) is the downwelling irradiance
just below the surface given by
(2) E
d
(
λ
,0
–
)=E
d
(
λ
,0
+
)(1–
α
(
λ
)
ice
)
where
α
(
λ
)
ice
is the albedo of the ice surface (Grenfell 1991).
Knowing K
d
(
λ
)
ice
, snow K
d
(
λ
) can similarly be calculated as
(3) K
d
(
λ
)
snow
= –ln(E
d
(
λ
, z
snow
)/E
d
(
λ
,0
–
))/z
snow
where E
d
(
λ
, z
snow
) is the spectral downwelling irradiance at the
snow–ice interface given by
Fig. 1. Map showing the location of Lake A, northern Ellesmere
Island, Canada.
J:\cjfas\cjfas58\cjfas-12\F01-187.vp
Tuesday, December 11, 2001 2:33:24 PM
Color profile: Disabled
Composite Default screen

(4) E
d
(
λ
, z
snow
)=E
d
(
λ
, z
ice
)/(exp(–K
d
(
λ
)
ice
z
ice
)
and E
d
(
λ
,0
–
) is equal to E
d
(
λ
,0
+
) multiplied by 1 –
α
(
λ
)
snow
.
Water-column irradiance profiles
UVR and PAR under-ice profiles were measured over the deep
-
est part of the lake, with natural snow-cover conditions, using the
same PUV500 radiometer as used for transmittance measurements.
The PUV500 gave ~15 data points per metre during profiling. Un
-
derwater irradiance measurements were corrected for dark current
by subtracting the minimum asymptotic value reached at depth. The
PUV500 radiometer was lowered through the hole used for T(
λ
)
measurements, which was tightly covered witha1m
2
opaque
board to prevent direct solar radiation from entering the water col
-
umn through the hole. K
d
(
λ
) for the surface freshwater (2–5 m, the
0-m depth being the piezometric water level in the hole) was deter
-
© 2001 NRC Canada
2408 Can. J. Fish. Aquat. Sci. Vol. 58, 2001
Mean Minimum Maximum
Precipitation (mm) 157 93 291
Temperature (°C) –17.9 –19.5 –15.8
Onset of melt (date) 17 June 4 June 1 July
Onset of freeze (date) 21 August 7 August 5 September
Length of melt season (days) 65 43 85
Freezing degree-days 6736 5988 7247
Thawing degree-days 200 104 336
Note: Following Rigor et al. (2000), the length of the melt season was computed from the daily average
temperature by using a 2-week running mean filter and defining the onset of melt and freeze as the day that
the filtered data rose or dropped below the melting temperature of snow and freshwater ice (0°C).
Table 1. Mean, minimum, and maximum for meteorological data from Alert over the period 1951–1999.
Fig. 2. RADARSAT-1 images of northern Ellesmere Island acquired on 30 August 1998. On the right, ice-covered Lake A and Lake B
are shown. For comparison, on the left, ice-free lakes of the Taconite Inlet region, 40 km to the southwest of Lake A, are shown.
J:\cjfas\cjfas58\cjfas-12\F01-187.vp
Tuesday, December 11, 2001 2:33:30 PM
Color profile: Disabled
Composite Default screen

mined by linear regression of the natural logarithm of E
d
(
λ
, z) ver
-
sus depth. A vertical profile of the scalar PAR attenuation
coefficient (K
0
(PAR), m
–1
) was calculated from E
0
(PAR) measured
using a PNF300 radiometer (Biospherical Instruments, Inc.).
K
0
(PAR) was calculated for every 0.5-m interval from the surface
to 23 m (the detection limit of the instrument). The beam attenua
-
tion coefficient at 660 nm, beamc(660) (m
–1
), was measured using
a Sea Tech transmissometer (Wet Labs, Inc., Philomath, Oreg.)
measuring transmission of a collimated light beam over a 10-cm
path length.
Chlorophyll a
Water samples collected at every metre in the oxic zone (2–12 m)
were vacuum-filtered through a 25-mm GF/F-equivalent filter (MFS
GF75; Advantec MFS, Inc., Dublin, Calif.). Additional samples
were prefiltered through 2-
µm
Nuclepore filters (Nucleopore Corp.,
Pleasanton, Calif.) and then onto MFS filters to obtain the
picoplanktonic fraction. Filters were stored at –20°C until measure
-
ment of chlorophyll a (Chl a). Chl a was extracted with boiling
95% ethanol (Nusch 1980), and the fluorescence of the extract was
measured with a model 450 Sequoia–Turner fluorometer (Sequoia–
Turner Corp, Mountain View, Calif.). Phaeopigments were cor
-
rected for by acidification. For samples collected in the anoxic
zone, the absorbance of ethanol extracts of pigments collected on
GF/F-equivalent filters was measured using a Hewlett-Packard
8452A diode array spectrophotometer (Hewlett-Packard Corp.,
Palo Alto, Calif.).
Particulate absorption coefficient
Water samples from selected depths (2, 6.5, 12, 22, and 29 m)
were vacuum-filtered in duplicate through 25-mm GF/F filters and
stored at –20°C until measurement (10 days after collection) of
spectral absorption by particles (a
p
). The absorbance of particles
concentrated onto the filters was measured every 2 nm over the
spectral range 390–800 nm according to Roesler (1998) using a
Hewlett-Packard 8452A diode array spectrophotometer equipped
with an integrating sphere (Labsphere RSA-HP-84; Hewlett-
Packard Corp.). Absorbance values were then converted to a
p
using
the algorithm of Roesler (1998).
Chromophoric dissolved organic matter (CDOM)
absorption coefficient
Measurements of the CDOM absorption coefficient, a
CDOM
(m
–1
),
were made on each water sample and also for a melted-ice sample
taken at ~0.7 m below ice surface and slowly thawed in the dark at
4°C in a plastic Ziploc® bag (S.C. Johnson & Son, Inc., Racine,
Wis.). Initial meltwater from this ice sample was discarded to re
-
move any contaminant that may have been transferred during ice
drilling and handling. Samples were filtered through 0.22-
µm
Sar
-
torius cellulose acetate filters (Sartorius, Goettingen, Germany)
and stored at 4°C in acid-cleaned, amber glass bottles until analy
-
sis (within 4 months). a
CDOM
was measured every 2 nm over the
wavelength range 250–820 nm using a 1-cm, acid-cleaned, quartz
cuvette in a Hewlett-Packard 8452A spectrophotometer.
Dissolved organic carbon analysis and CDOM
characterization
Dissolved organic carbon (DOC) concentration was determined
using the UV Digestion and Infra-red Detection method (Environ
-
ment Canada, National Laboratory for Environmental Testing).
CDOM characterization was made using synchronous fluorescence
spectroscopy (Senesi et al. (1991) as modified by J.A.E. Gibson et
al., unpublished data). Synchronous fluorescence spectra were re
-
corded with a Shimadzu FR5000 spectrofluorometer (Shimadzu,
Kyoto, Japan) used in the synchronous mode with a slit width of
5 nm on both sides and a wavelength difference between the exci
-
tation and emission beams of 14 nm, which we have found to be
optimal for resolving differences in CDOM between sources. This
setting also minimizes the overlap between CDOM peaks and the
Raman water peak. Spectra were recorded over the excitation
wavelength range 200–600 nm with a corresponding emission
range of 214–614 nm. Previous analysis of water samples or iso
-
lated fulvic acids from >50 aquatic environments showed that
small, autochthonous-like molecules of CDOM (simple aromatic
rings without further conjugation) usually show a strong fluores
-
cence peak at 293–308 nm and often show secondary peaks around
360 nm, whereas higher molecular weight molecules with more
complex fluorophores (allochthonous-like humic and fulvic materi
-
als) fluoresce at longer wavelengths (J.A.E. Gibson et al., unpub
-
lished data).
Results
Ice- and snow-cover properties
At the time of sampling, Lake A was covered by 1.97 m
(standard deviation (SD) = 0.01 m) of ice. The top 0.12 m
was candled ice with long crystals several centimetres in di
-
ameter, and the rest was completely fused and clear. No
white ice or troughs or ridges at the ice surface were ob
-
served. The maximum height of relief on the smooth ice sur
-
face was of the order of 10 cm. The top metre of clear ice
contained large flattened air bubbles (up to 6 cm long, a
few centimetres wide, but only a few millimetres thick).
Bubbles were rare and smaller deeper in the ice sheet. Sedi-
ment was distributed throughout the top metre of the ice,
forming aggregates up to 2 cm in diameter. The piezometric
water level was ~13 cm below the ice surface. The snow
depths at 1-m intervals along a 20-m transect varied from 39
to 58 cm (average 52 cm, coefficient of variation (CV) =
9%), and the snow surface was flat. At the site of transmit-
tance measurements, the ice was overlaid by 41 cm (CV =
8%, n = 15) of snow. The top ~30 cm was composed of
fresh snowflakes and small-grained snow. The deeper layers
were composed of loose large crystals typical of deep hoar
and indicative of temperature-gradient metamorphism. The
average snow density over the full snow profile was
0.21 g·cm
–3
(as measured gravimetrically using a 6.4-cm-
diameter cylinder). The weight of that snow would depress
the ice surface by 10 cm. Given the piezometric water level
of 13 cm, an average ice density of 0.88 g·cm
–3
can be de
-
duced.
Irradiance transmittance through the ice and snow cover
The transmittance of PAR and discrete UV wavelengths
through the intact ice and snow cover and %T through the
ice for a 12-m
2
area cleared of snow are shown in Fig. 3a.In
presence of the snow cover, 0.45% of incident PAR was
transmitted. In the UVR waveband, %T decreased slightly
with decreasing wavelength, from 0.46% at 380 nm to
0.31% at 320 nm. Under-ice E
d
(305) was below the PUV500
detection limit, preventing %T(305) determination. Re
-
moving the snow greatly increased the transmittance; 5.7–
6.6% of incident irradiance was transmitted through the ice.
The increase was greater at shorter UV wavelengths; this re
-
moval of the snow cover resulted in an 18-fold increase in
UV at 320 nm and 13-fold in PAR transmittance. Under the
conditions of almost eliminated snow cover, the transmit
-
tance of 340 nm and 380 nm UV-A was higher than that for
© 2001 NRC Canada
Belzile et al. 2409
J:\cjfas\cjfas58\cjfas-12\F01-187.vp
Tuesday, December 11, 2001 2:33:30 PM
Color profile: Disabled
Composite Default screen

PAR. To calculate the integrated photoinhibiting irradiance
prevailing just under the ice, we extended the UVR trans
-
mission from the three wavelengths measured by the
PUV500 over the full UV spectrum and applied this mod
-
eled transmission to the spectral incident irradiance prevail
-
ing at noon. This under-ice irradiance was weighted with the
biological weighting function for inhibition of photosynthe
-
sis from Cullen et al. (1992). Under natural conditions, the
weighted under-ice irradiance was only 0.4% of incident.
The almost complete removal of snow cover increased
photoinhibiting irradiance to 6.3% of incident, a 16-fold in
-
crease.
An important part of the difference in %T between natural
snow conditions and cleared ice was caused by changes in
the albedo of the surface. The snow albedo was very high
and showed little spectral dependence, varying from 0.94 at
320 nm to 0.97 for PAR (Fig. 3b). Once the snow had been
removed, the albedo of the candled ice varied from 0.73 at
320 nm to 0.84 for PAR (Fig. 3b). Removing the snow cover
thus decreased the albedo by 20–22% in the UVR and 13%
in the PAR range.
Water-column properties
The vertical profiles of temperature, salinity, and dissolved
oxygen showed that the water column was highly stratified
(Fig. 4). The top 10-m section of the profile was freshwater
with moderately low conductivity (minimum of 0.27 mS·cm
–1
).
A chemocline with increasing salinity extended from 10 m to
~28 m; below this depth, the salinity reached ~90% of seawa-
ter. The oxic zone extended to 13 m; the percent of oxygen sat-
uration was 120–133% in the top 9 m (probably influenced by
oxygen exclusion during ice formation, Wharton et al. 1993)
and dropped to 9% at 13 m. Although water temperature was
near freezing just under the ice, a thermal maximum of 8.75°C
was located at 17.8 m. In the monimolimnion, water tempera-
ture gradually decreased with depth to a measured minimum of
4.0°C. The profiles showed no evidence of a mixed isothermal
or isohaline layer at any depth.
Spectral irradiance attenuation in the water column
Despite low absolute values, complex patterns of variation
of beamc(660) were observed down through the water col
-
umn (Fig. 5). An upper maximum was observed just under
the ice, and a second peak began at 29 m and extended to
45 m. A series of irregular, localized peaks were also mea
-
sured between 12 and 20 m; these peaks are likely associ
-
ated with Schlieren effects caused by changes in refractive
index resulting from the mixing of waters of different salin
-
ity during the lowering of the profiler. Below 60 m,
beamc(660) showed little variability and was slightly higher
than the value encountered at 15–20 m. The insert in Fig. 5
compares our beamc(660) profile to the one of Ludlam
(1996). The large peak measured between 10 and 20 m in
late May 1993 was absent in 1999. Also, in 1993, the near-
surface maximum was observed very close to the lower ice
surface and was only ~1 m thick.
The water column values of K
d
(
λ
) were in the low to
moderate range for lake waters. The K
d
values for UVR in
the surface5mofthewater column were 1.65 (0.28), 1.33
(0.06), and 0.80 (0.10) m
–1
at 320, 340, and 380 nm, respec
-
tively (means of five profiles, SD in parentheses). The
equivalent value for PAR was 0.352 (0.089) m
–1
. In the ab
-
sence of ice, the 1% depth of 320 nm and 380 nm UVR
would be 2.8 m and 5.7 m, respectively, whereas the 1%
depth of PAR would be 13.1 m. The large vertical variabil
-
ity in K
0
(PAR) matched the beamc(660) profile (which is
independent of the natural irradiance field), with maximum
attenuation just under the ice (K
0
(PAR) = 0.347 m
–1
over the
2- to 5-m depth interval) and lower attenuation between 7
and 14 m (Fig. 6). Only 0.03% of incident E
0
(PAR) reached
the top of the photosynthetic sulfur bacteria distribution at
16 m (see below) and less than 0.003% at 29 m, the depth of
the beamc(660) maximum (Fig. 6).
Vertical distribution of optically active particulate and
dissolved organic matter
Variations in a
p
(Fig. 7) were consistent with the vertical
pattern in beamc(660). At 22 m, the a
p
spectra showed a
large absorption peak at 715 nm, typical of bacterio
-
© 2001 NRC Canada
2410 Can. J. Fish. Aquat. Sci. Vol. 58, 2001
Fig. 3. (a) Transmittance through the snow and ice cover (solid
bars) of Lake A and through the ice cleared of snow over a 12-
m
2
area (open bars). Error bars denote standard deviation (SD)
on means from 3 and 4 measurements, respectively. (b) Albedo
of the snow (solid bars) and candled ice (open bars) surfaces of
Lake A. Error bars are SD for 2 and 3 replicated measurements,
respectively.
J:\cjfas\cjfas58\cjfas-12\F01-187.vp
Tuesday, December 11, 2001 2:33:31 PM
Color profile: Disabled
Composite Default screen

chlorophyll e (BChl e). The BChl e identification was con
-
firmed using methanol extracts and showed the red
absorption peak to be centered at 659 nm (Stal et al. 1984).
The absorbance of ethanol extracts indicated the absence of
BChl e above 16 m and maximum concentrations between
20 m and 30 m. The source of this BChl e is likely to be
green photosynthetic sulfur bacteria of the genus Chloro
-
bium (Holt et al. 1993). The absorption maximum found at
29 m (more than two orders of magnitude higher than the a
p
measured at other depths; Fig. 7) is probably due to particles
of elemental sulfur that accumulate outside these bacterial
cells (Holt et al. 1993) superimposed on the BChl e absorp
-
tion peak and possibly also the formation of iron oxides dur
-
ing filtration. The highest a
p
in the upper freshwater section
of the water column was found just below the ice at a depth
of 2 m. The spectral shape of a
p
in this oxic zone was typi
-
cal of phytoplankton, with the red absorption peak at or
close to 672 nm, indicative of Chl a. The primary Chl a ab
-
sorption maximum was in the blue region of the spectrum,
as expected, with an additional broad peak of absorption in
the range 530–580 nm, indicative of carotenoids and possi
-
bly phycoerythrin (epifluorescence microscopy showed the
presence of picocyanobacteria; P. Van Hove et al., unpub
-
lished data). The detailed Chl a profile in the oxic waters
confirmed that Chl a concentration was maximal just under
the ice (0.43
µg
·L
–1
) and declined steadily with depth to
reach 0.07
µg
·L
–1
at 12 m. On average, 62% of the total
Chl a was contributed by the <2
µm
fraction (range 39–
97%).
DOC concentrations were low in the upper freshwaters,
varying from 0.6 to 1.3 mg·L
–1
with maximum concentra
-
tions just under the ice (Fig. 8). In the monimolimnion,
DOC concentrations increased rapidly to reach 3–4 mg·L
–1
;
these DOC concentrations are high considering the extreme
high latitude position of the lake (cf. Vincent et al. 1998).
The vertical distribution of a
CDOM
(320) followed that of DOC
(Fig. 8). However, the DOC-specific a
CDOM
(320) was low in
the freshwater layer (2–3 m
–1
(mg·L
–1
)
–1
) and increased to 4–
5m
–1
(mg·L
–1
)
–1
in the monimolimnion. The much more col
-
ored DOC in the monimolimnion suggests a different com
-
position of the DOC pool given that high DOC-specific
absorption is usually associated with allochthonous DOC
(Morris et al. 1995).
The synchronous fluorescence scans provided further in
-
sights into the chemical composition of the DOC and CDOM
(Fig. 9). A fluorescence peak occurred at 293–308 nm, indi
-
© 2001 NRC Canada
Belzile et al. 2411
Fig. 4. Temperature (thick line), salinity (solid circles), and dissolved oxygen (open circles) profiles in the water column of Lake A.
Salinity and dissolved oxygen were measured using an Hydrolab profiler and temperature was measured using a PUV500.
J:\cjfas\cjfas58\cjfas-12\F01-187.vp
Tuesday, December 11, 2001 2:33:32 PM
Color profile: Disabled
Composite Default screen

cating that the CDOM in the mixolimnion (Fig. 9a) was es
-
sentially from microbial origin, in agreement with the low
DOC-specific a
CDOM
. The CDOM in the ice (0.7 m below
ice surface) showed a large peak centered at 293 nm that
was not found in the scans of samples from the underlying
water column. a
CDOM
(320) in the ice was 1.75 m
–1
; this
value is about half that measured just under the ice and
slightly lower than the minimum value measured in the wa
-
ter column (1.85 m
–1
at 10 m, Fig. 8). From 12 to 20 m,
where the brackish waters were suboxic or anoxic, the fluo
-
rescence spectra showed a large peak with a maximum at
362 nm and high fluorescence per unit DOC (Fig. 9b). Sam
-
ples from 25 m to the bottom had two additional peaks cen
-
tered at 392 nm and 491 nm, indicative of large, complex
molecules (Fig. 9c).
Comparison of snow, ice, and water K
d
Snow K
d
(PAR) was 2.1 m
–1
(Fig. 10a), almost an order of
magnitude lower than values previously reported (Thomas
1963; Grenfell and Maykut 1977). This striking difference
probably results from the low attenuation in the deep hoar
layer of the Lake A snow. Snow K
d
(UVR) increased with
decreasing wavelength to reach 3.3 m
–1
at 320 nm. Attenua
-
tion in the ice was very low compared with that in the snow.
Ice K
d
(PAR) was 0.32 m
–1
and K
d
increased slightly with
decreasing wavelength in the UVR (Fig. 10a). The K
d
(UVR)
for the ice was strikingly lower (up to 53%) than that of the
water column, whereas the K
d
(PAR) in the ice was 45%
higher than in the water column (Fig. 10b). Our method of
calculation resulted in low effective K
d
values for snow and
ice. To calculate K
d
(
λ
)
ice
, some authors (e.g., Fritsen et al.
1992), rather than using eq. 2, assumed that E
d
(0
–
) is deter
-
mined by the specular reflection of E
d
(0
+
), i.e., E
d
(0
–
)isap
-
proximated to be equal to the incident irradiance reduced by
only 2–5%. K
d
(
λ
)
ice
calculated with this E
d
(0
–
) gives values
about two times higher than those calculated using our equa
-
tions, although the resulting ice K
d
(320) is still lower than
that of the water column. In the water column, CDOM was
responsible for most of the UV and blue light absorption; for
example, at 440 nm, CDOM absorption was 0.45 m
–1
just under
the ice, whereas absorption by particles was only 0.04 m
–1
.
Discussion
Life at the northern limit of lakes
In a previous limnological study of Lake A, Ludlam (1996)
suggested that a supersaturated oxygen maximum found just
under the ice was related to phytoplankton photosynthesis
© 2001 NRC Canada
2412 Can. J. Fish. Aquat. Sci. Vol. 58, 2001
Fig. 5. Beam c(660) profile measured on June 7th, 1999. The insert shows the comparison with Ludlam’s (1996) profile measured on
May 26, 1993.
J:\cjfas\cjfas58\cjfas-12\F01-187.vp
Tuesday, December 11, 2001 2:33:36 PM
Color profile: Disabled
Composite Default screen

© 2001 NRC Canada
Belzile et al. 2413
and the deep turbidity maximum was likely due to
anoxygenic phototrophic bacteria. The a
p
spectra presented
here confirm the presence of these hypothesized microbial
assemblages and indicate that the deep maximum is domi
-
nated by Chlorobium. Despite its extreme northern location
and thick perennial ice cover, Lake A is inhabited by a com
-
plex community of prokaryotic and eukaryotic autotrophs.
These autotrophs support an assemblage of small-sized
copepods with a mean density of ~400 individuals·m
–3
in the
top 23 m (Van Hove et al. 2001). The presence of macro
-
zooplankton in Lake A represents one important difference
relative to the McMurdo LTER perennially ice-covered
lakes (Priscu et al. 1999a). The low Chl a biomass that we
measured may be partly due to grazing by copepods, al
-
though it may also be related to our sampling early in the
growing season. The migration of zooplankton through the
water column and their production of fecal pellets is a trans
-
port mechanism for organic material in this lake that is ab
-
sent from the McMurdo lake ecosystems. The bio-optical
properties of Lake A’s snow and ice cover exert a major in
-
fluence on all of these planktonic communities by control
-
ling the biological UV exposure and the availability of solar
energy for photosynthesis. The large bio-optical and other
limnological variations down the water column of Lake A
parallel the conditions found in the McMurdo lakes. In these
Antarctic systems, as in Lake A, there are major changes in
temperature, salinity, and oxygen as a function of depth,
with pronounced maxima in bio-optical variables (e.g.,
Lizotte and Priscu 1992a; Vincent et al. 1998). Deep max
-
ima of photosynthetic sulfur bacteria are also found in some
Antarctic lakes, for example lakes Fryxell and Hoare
(Lizotte and Priscu 1992a). In Lake A, peak concentrations
of sulfur bacteria pigments were found between 20 m and
30 m, although anoxic conditions start at 14 m. The absence
of measurable sulfide in waters above 32 m (J.A.E. Gibson
et al., unpublished data) may explain the location of sulfur
bacteria well below the oxycline.
Low irradiance attenuation in Lake A ice
The K
d
(PAR) for lake ice measured in this study (0.32 m
–1
)is
within the range of ice values for Lake Vanda (0.19–0.67 m
–1
;
Howard-Williams et al. 1998), the clearest of the McMurdo Dry
Valleys perennially ice-covered lakes. The dry climate of the Dry
Valleys induces the formation of a highly scattering white ice
layer at the ice surface as austral summer progresses, decreasing
irradiance transmission through the ice (Howard-Williams et al.
Fig. 6. Attenuation coefficient for scalar photosynthetically available radiation (K
0
(PAR); open circles) and percent of incident scalar
PAR down the water column (thick line). K
0
was determined for 0.5-m depth intervals.
J:\cjfas\cjfas58\cjfas-12\F01-187.vp
Tuesday, December 11, 2001 2:33:37 PM
Color profile: Disabled
Composite Default screen

© 2001 NRC Canada
2414 Can. J. Fish. Aquat. Sci. Vol. 58, 2001
1998; Fritsen and Priscu 1999). Reduction of the transmittance
through Antarctic lake ice also results from layers of wind-
blown sediment present on top and inside the ice as well as nu-
merous gas bubbles trapped in the ice (Wharton et al. 1993; Ad-
ams et al. 1998; Kepner et al. 2000). Sediment was present in
the top metre of Lake A ice but as widely spaced aggregates that
were unlikely to have a strong influence on the bulk radiative
properties of the ice. However, these sediment aggregates are
likely to cause melt lenses during summer and to modify bubble
structure during freeze-down, thereby indirectly affecting local
ice properties (Adams et al. 1998). Gas bubbles were observed
although they appeared to be smaller and less numerous than re
-
ported for Antarctic lake ice (cf. Adams et al. 1998), presumably
because of the lower O
2
and N
2
supersaturation in Lake A. As
-
suming that the sediment load had a negligible effect on ice
buoyancy and given a density ratio of pure ice to water of 0.917
(Adams et al. 1998), the ice-cover density of 0.88 indicates an
average air content of the ice cover of only 4%. The 12-cm-
thick candled ice of Lake A therefore had a disproportionate im
-
pact on irradiance transmission because the remaining ice was so
transparent.
The remarkably low K
d
(UVR) in the ice relative to the
water column of Lake A is likely to be the result of the di
-
lute CDOM content of the ice. We have also measured re
-
duced a
CDOM
in the 2.3-m-thick ice of Char Lake
(Cornwallis Island, 74°42
′
N); a
CDOM
(320) in Char Lake ice
on June 11, 1999, was 39 and 48% lower at 1 m and 1.5 m,
respectively, than in the underlying water column (C. Belzile
et al., unpublished data). Priscu et al. (1999b) reported the
concentration of dissolved organic carbon in bottom accre
-
tion ice of Lake Bonney to be 58% of that of the water col
-
umn, consistent with reduced CDOM in lake ice relative to
water. Because of its high absorption in the UVR, CDOM is
typically the main optical component controlling UVR atten-
uation in northern freshwaters (Laurion et al. 1997). The low
a
CDOM
in the ice thus allows Lake A ice to be relatively
transparent to UVR despite elevated scattering. PAR attenu-
ation is less affected by CDOM absorption owing to the ex-
ponential decrease of a
CDOM
with increasing wavelength.
Accordingly, K
d
(PAR) in the ice was higher than in the wa-
ter column, reflecting the higher scattering in the ice and the
much lesser importance for PAR of the large CDOM differ
-
ences between the ice and water.
Several mechanisms may explain the low a
CDOM
in the
ice. Firstly, freeze-outs of major ions, nutrients, and gases
are known to occur during lake-ice formation and are more
effective during periods of slow ice growth rate (Wharton et
al. 1993; Adams et al. 1998). Ice accretion in Lake A is
likely to be slow owing to the insulation provided by the
thick ice and the relatively deep snow cover. Photochemical
processes may contribute to a decrease of a
CDOM
in the ice
and may contribute to the fluorescence peaks indicative of
small, simple molecules. Despite the short Arctic summer,
the perennial nature of Lake A ice would allow relatively
long exposure to UVR and the possibility of gradual photo
-
degradation of DOC within the ice sheet. Finally, the sedi
-
ment aggregates found in the top metre of the ice may be the
site of microbial activity when liquid water is present, equiv
-
alent to the microbial consortium recently discovered in Ant
-
arctic lake ice (Priscu et al. 1998). Microbial degradation
processes could accelerate the breakdown of CDOM. This
biological activity may also result in autochthonous DOC
production (Priscu et al. 1998), which could further explain
spectral fluorescence peaks of small, simple molecules. In
lakes of the Queen Maud Land (Antarctica), Kaup (1988)
found situations where ice K
d
(PAR) did not exceed that of
Fig. 7. Spectral absorption coefficients of particles at selected depths in the water column of Lake A. Note the different scale for the
29 m spectra (right axis).
J:\cjfas\cjfas58\cjfas-12\F01-187.vp
Tuesday, December 11, 2001 2:33:39 PM
Color profile: Disabled
Composite Default screen

the underlying water column. He noted that the ice crystals
of the perennially ice-covered Lake Untersee can have diam
-
eters of 20 cm and more and can extend through the 3-m-
thick ice. He postulated that these vertically oriented crystals
act as light conductors, resulting in the very low K
d
(PAR)
for this type of ice. This situation may also apply to Lake A
ice and the combination of low scattering and low absorp
-
tion inside the ice crystals would allow relatively high UVR
transmission through the ice.
Although the ice itself is relatively transparent to solar
irradiance, the high albedo and K
d
of the snow considerably
reduced the irradiance reaching the water column. In early
June, although solar irradiance was near its maximum, phyto
-
plankton cells directly under the ice received at best 5
µmol
photons·m
–2
·s
–1
of PAR. This is likely to cause severe light
limitation. For example, even the highly shade-adapted phyto
-
plankton in the McMurdo ice-covered lakes have light satura
-
tion values for photosynthesis of 10–30
µmol
photons·m
–2
·s
–1
(Lizotte and Priscu 1992b).
Importance of CDOM in the water column
The attenuation coefficient of scalar PAR measured by
Ludlam (1996) in the surface waters of Lake A was 0.304 m
–1
,
in good agreement with the value of 0.347 m
–1
reported
here. Water column K
d
(PAR) and K
d
(UVR) were low, al
-
though they were higher than those found, using the same
radiometer, in perennially ice-covered lakes Vanda, Bonney,
and Hoare in the McMurdo region (Vincent et al. 1998).
This is consistent with the lower DOC concentrations re
-
corded in these Antarctic lakes. UV attenuation in Lake A
was similar to that measured in Lake Fryxell by Vincent et
al. (1998), although DOC concentration is at least two times
higher in the latter lake (McKnight et al. 1991; Vincent et al.
1998). It is likely that the DOC in these lakes is less colored
than in Lake A, given their extremely barren catchments and
the dominance by autochthonous processes in controlling
their carbon dynamics. Approximately 5–10% of the ground
around Lake A is covered with vegetation, mostly cushion
plants (Saxifraga oppositifolia) but also some herbs and
shrubs (including Salix arctica and Dryas integrifolia) (E.
Lévesque, Université du Québec à Trois-Rivières, Trois-
Rivières, Québec, unpublished data). These plants likely
contribute to the allochthonous DOC input to the lake. As in
Lake A, DOC concentrations increase with depth in most
Antarctic perennially ice-covered lakes (e.g., from 1.3 to
3.9 mg·L
–1
in Lake Hoare and from 3.3 to 30 mg·L
–1
in Lake
Fryxell; McKnight et al. 1991). McKnight et al. (1991)
showed that the fulvic acid (FA) fraction of the DOC in
lakes Hoare and Fryxell had similar elemental compositions,
carbon distributions, and amino acid contents, indicating that
the chemistry of these microbially derived FA is not strongly
influenced by the chemical environment in the water column.
However, the FA fraction represented only 16–20% of total
DOC in these lakes. The very different fluorescence proper-
ties of CDOM observed through the water column of Lake
A suggest that major differences in the composition of the
DOM pool exist as a function of depth. The processes
responsible for the presence in the suboxic and anoxic layers
of CDOM with fluorescence characteristics typical of large
and complex molecules remains to be elucidated. This fluo-
rescent material may be remnant from the original DOC-
containing ocean water that originally filled the basin or,
more likely, may be the result of chemical (e.g., through sul
-
furisation by reaction with sulfide) or biological transforma
-
tion of the simple material originating from the surface
waters. These presumed modifications of the DOC pool are
conspicuous in Lake A because of the highly stratified na
-
ture of the water column, but they are also likely to occur in
other aquatic ecosystems. CDOM can be an important
precursor of carbon substrates for heterotrophic growth and
may help support mixotrophic phytoplankton during the long
period of winter darkness (Priscu et al. 1999a).
In Lake A, UVR attenuation in the water column and ice
would provide sufficient protection against UV damage to
the photosynthetic sulfur bacteria growing below 13 m. The
phytoplankton growing immediately under the ice are ex
-
posed to higher UVR fluxes, up to 6.3% of incident photo
-
inhibiting solar irradiance under snow-free conditions (based
on our snow-clearing experiment). Moreover, the ice surface
artificially cleared of snow probably had a higher albedo be
-
cause of the small residual amounts of snow trapped in the
irregular ice surface and the influence of the snow at the
edge of the cleared area. The air-filled candled ice was also
likely more scattering than under melting conditions when
liquid water ponding on the ice surface would substantially
reduce the albedo. The irradiance transmission in natural
© 2001 NRC Canada
Belzile et al. 2415
Fig. 8. Dissolved organic carbon concentrations (DOC; solid cir
-
cles) and chromophoric dissolved organic matter absorption coef
-
ficients at 320 nm (a
CDOM
(320); open triangles) down the water
column of Lake A.
J:\cjfas\cjfas58\cjfas-12\F01-187.vp
Tuesday, December 11, 2001 2:33:40 PM
Color profile: Disabled
Composite Default screen

snow-free conditions is therefore likely to be higher than
that measured here, unless significant “ice whitening” simi
-
lar to that observed in Antarctic perennially ice-covered lakes
(Fritsen and Priscu 1999) is occurring.
Climate change
Given the relative transparency of the ice cover, the opaque
nature of the snow, and the strong seasonal variations in in
-
cident irradiance (complete darkness from mid-October until
the end of February), most of the annual input of solar en
-
ergy to the lake must be restricted to a brief period of the
year. Simple model calculations based on K
d
and assuming
no change in surface albedo show that a 20% variation in
snow depth would correspond to a similar percent change in
the amount of PAR entering the lake (from %T = 0.50%
with z
snow
= 41 cm to 0.41% and 0.60% with 49 cm and
22 cm of snow, respectively). However, the removal of snow
over a 12-m
2
area increased %T(PAR) 13 times and would
increase photoinhibiting UVR 16 times. Changes in the tim
-
ing of the disappearance of snow cover would have a major
impact on the annual radiation budget of the lake, with impli
-
cations for stratification, mixing, and biological processes.
There is no statistical trend in the temperature or precipi
-
tation at Alert for the period 1951–1999 (C. Belzile et al.,
unpublished data). However, for the last 32 years of this re
-
cord, there has been a significant warming trend in mean
annual temperature (0.04°C per annum) and a highly signifi
-
cant decline in the number of freezing degree-days per an
-
num (loss of 15 freezing degree-days per annum; Vincent et
al. 2001). These most recent trends are likely to be related,
at least in part, to the Arctic Oscillation, although they may
also be amplified by anthropogenic forcing (Rigor et al.
2000).
There is considerable interannual variability in the Alert
climate record. For example, snow depth at Alert at the end
of the cold season ranged from 9 to 79 cm for the period
1949 to 1988 (Curtis et al. 1998), and a difference of
27 days was shown between the earliest (June 4th) and latest
(July 1st) date of the onset of snowmelt (Table 1). Such vari
-
ations in the thickness and duration of snow cover over the
perennial ice of Lake A would translate into major
interannual differences in the annual solar energy input to
the lake and are likely to cause pronounced year-to-year dif
-
ferences in biological productivity. Despite this variability,
© 2001 NRC Canada
2416 Can. J. Fish. Aquat. Sci. Vol. 58, 2001
Fig. 9. Dissolved organic carbon specific synchronous fluorescence
spectroscopy scans of chromophoric dissolved organic matter from
the lake ice and selected depths in the water column.
Fig. 10. (a) Snow (solid bars), ice (open bars), and water-column
(hatched bars) attenuation coefficients for downwelling
irradiance, K
d
(
λ
). Error bars for water-column K
d
(
λ
) are standard
deviations for the means of five different profiles. (b) Percent
change of K
d
(
λ
) in the ice relative to that in the water.
J:\cjfas\cjfas58\cjfas-12\F01-187.vp
Tuesday, December 11, 2001 2:33:42 PM
Color profile: Disabled
Composite Default screen

however, the ice has remained on the lake on all years of di
-
rect or remote observation.
Changes greater than the interannual variability will ac
-
company future warming trends in the region, as predicted
by global models (Weller 1998). Relatively minor increases
in air temperature beyond the 1998 maximum could cause
the complete loss of ice cover of Lake A. Our optical results
reported here show that such a change would completely al
-
ter the limnological characteristics of this High Arctic eco
-
system.
Acknowledgments
This research was funded by the Natural Sciences and En
-
gineering Research Council of Canada, the Fonds pour la
Formation de Chercheurs et l’Aide à la Recherche of Que
-
bec, the Centre d’études nordiques, and Indian and Northern
Affairs Canada. We thank Polar Continental Shelf Project
for logistic support in the Arctic (this is PCSP publication
number 03101) and Robert Gauthier and the staff of the Ca
-
nadian Center for Remote Sensing for their help in the ac
-
quisition of RADARSAT imagery. We also thank Peter T.
Doran and an anonymous reviewer for insightful comments.
References
Adams, E.E., Priscu, J.C., Fritsen, C.H., Smith, S.R., and Brackman,
S.L. 1998. Permanent ice covers of the McMurdo Dry Valleys
lakes, Antarctica: bubble formation and metamorphism. Antarct.
Res. Ser. 72: 281–295.
Bolsenga, S.J., Evans, M., Vanderploeg, H.A., and Norton, D.G.
1996. PAR transmittance through thick, clear freshwater ice.
Hydrobiologia, 330: 227–230.
Cullen, J.J., Neale, P.J., and Lesser, M.P. 1992. Biological weighting
function for the inhibition of phytoplankton photosynthesis by ul-
traviolet radiation. Science (Washington, D.C.), 258: 646–650.
Curtis, J., Wendler, G., Stone, R., and Dutton, E. 1998. Precipitation
decrease in the western Arctic, with special emphasis on Barrow
and Barter Island, Alaska. Int. J. Climatol. 18: 1687–1707.
Doran, P.T., Wharton, R.A., Jr., and Lyons, W.B. 1994. Paleo
-
limnology of the McMurdo Dry Valleys, Antarctica. J.
Paleolimnol. 10: 85–114.
Doran, P.T., McKay, C.P., Adams, W.P., English, M.C., Wharton,
R.A., Jr., and Meyer, M.A. 1996. Climate forcing and thermal
feedback of residual lake-ice covers in the High Arctic. Limnol.
Oceanogr. 41: 839–848.
Ellis-Evans, J.C., Laybourn-Parry, J., Bayliss, P.R., and Perriss,
S.J. 1998. Physical, chemical and microbial community charac
-
teristics of lakes of the Larsemann Hills, Continental Antarctica.
Arch. Hydrobiol. 141: 209–230.
Fofonoff, N.P., and Millard, R.C. 1983. Algorithms for the calcula
-
tions of fundamental properties of seawater. UNESCO Technical
Papers in Marine Science 44. UNESCO, Paris, France.
Fritsen, C.H., and Priscu, J.C. 1999. Seasonal change in the optical
properties of the permanent ice cover on Lake Bonney,
Antarctica: consequences for lake productivity and phytoplankton
dynamics. Limnol. Oceanogr. 44: 447–454.
Fritsen, C.H., Iturriaga, R., and Sullivan, C.W. 1992. Influence of
particulate matter on spectral irradiance fields and energy trans
-
fer in the eastern Arctic Ocean. Ocean Optics XI SPIE, 1750:
527–541.
Grenfell, T.C. 1991. A radiative transfer model for sea ice with ver
-
tical structure variations. J. Geophys. Res. 96: 16 991–17 001.
Grenfell, T.C., and Maykut, G.A. 1977. The optical properties of
ice and snow in the Arctic Basin. J. Glaciol. 18: 445–463.
Hanssenbauer, I., and Forland, E.J. 1998. Long-term trends in pre
-
cipitation and temperature in the Norwegian Arctic: can they be
explained by changes in atmospheric circulation patterns. Clim.
Res. 10: 143–153.
Hattersley-Smith, G., Keys, J.E., Serson, H., and Mielke, J.E. 1970.
Density stratified lakes in Northern Ellesmere Island. Nature
(London), 225: 55–56.
Holt, J.G., Krieg, N.R., Sneath, P.H.A., Staley, J.T., and Williams,
S.T. 1993. Bergey’s manual of determinative bacteriology. 9th
ed. Williams & Wilkins, Baltimore, Md.
Howard-Williams, C., Schwarz, A.-M., Hawes, I., and Priscu, J.C.
1998. Optical properties of McMurdo Dry Valleys lakes,
Antarctica. Antarct. Res. Ser. 72: 189–203.
Jeffries, M.O., Krouse, H.R., Shakur, M.A., and Harris, S.A. 1984.
Isotope geochemistry of stratified Lake “A,” Ellesmere Island,
N.W.T., Canada. Can. J. Earth Sci. 21: 1008–1017.
Johannessen, O.M., Shalina, E.V., and Miles, M.W. 1999. Satellite
evidence for an Arctic sea ice cover in transformation. Science
(Washington, D.C.), 286: 1937–1939.
Kaup, E. 1988. Solar radiation in the water bodies of Queen Maud
Land. In Limnological studies in Queen Maud Land (East
Antarctica). Edited by J. Martin. Academy of Sciences of the
Estonian SSR, Tallinn, Estonia, USSR. pp. 15–27.
Kepner, R.L., Wharton, R.A., Jr., Collier, R.D., Cockell, C.S., and
Jeffrey, W.H. 2000. UV radiation and potential biological effects
beneath the perennial ice cover of an antarctic lake. Hydro-
biologia, 427: 155–165.
Laurion, I., Vincent, W.F., and Lean, D.R.S. 1997. Underwater ul-
traviolet radiation: development of spectral models for northern
high latitude lakes. Photochem. Photobiol. 65: 107–114.
Lizotte, M.P., and Priscu, J.C. 1992a. Spectral irradiance and bio-
optical properties in perennially ice-covered lakes of the Dry Val-
leys (McMurdo Sound, Antarctica). Antarct. Res. Ser. 57: 1–14.
Lizotte, M.P., and Priscu, J.C. 1992b. Photosynthesis–irradiance re-
lationships in phytoplankton from the physically stable water col
-
umn of a perennially ice-covered lake (Lake Bonney, Antarctica).
J. Phycol. 28: 179–185.
Ludlam, S.D. 1996. The comparative limnology of high arctic,
coastal, meromictic lakes. J. Paleolimnol. 16: 111–131.
Madronich, S., McKenzie, R.L., Björn, L.O., and Caldwell, M.M.
1998. Changes in biologically active ultraviolet radiation reaching
the Earth’s surface. J. Photochem. Photobiol. B, Biol. 46: 5–19.
McKnight, D.M., Aiken, G.R., and Smith, R.L. 1991. Aquatic fulvic
acids in microbial based ecosystems: results from two desert
lakes in Antarctica. Limnol. Oceanogr. 36: 998–1006.
Morris, D.P., Zagarese, H., Williamson, C.E., Balseiro, E.G.,
Hargreaves, B.R., Modenutti, B., Moeller, R., and Queimalinos, C.
1995. The attenuation of solar UV radiation in lakes and the role of
dissolved organic carbon. Limnol. Oceanogr. 40: 1381–1391.
Nusch, E.A. 1980. Comparison of different methods for chloro
-
phyll and phaeopigment determination. Arch. Hydrobiol. Beih.
14: 14–36.
Perovich, D.K., Roesler, C.S., and Pegau, W.S. 1998. Variability in
Arctic sea ice optical properties. J. Geophys. Res. 103: 1193–
1208.
Priscu, J.C. 1998. Ecosystem processes in a polar desert: the
McMurdo Dry Valleys, Antarctica. Antarct. Res. Ser. Vol. 72.
American Geophysical Union, Washington, D.C.
Priscu, J.C., Fritsen, C.H., Adams, E.E., Giovannoni, S.J., Paerl,
H.W., McKay, C.P., Doran, P.T., Gordon, D.A., Lanoil, B.D., and
Pinckney, J.L. 1998. Perennial Antarctic lake ice: an oasis for life
in a polar desert. Science (Washington, D.C.), 280: 2095–2098.
© 2001 NRC Canada
Belzile et al. 2417
J:\cjfas\cjfas58\cjfas-12\F01-187.vp
Tuesday, December 11, 2001 2:33:42 PM
Color profile: Disabled
Composite Default screen

© 2001 NRC Canada
2418 Can. J. Fish. Aquat. Sci. Vol. 58, 2001
Priscu, J.C., Wolf, C.F., Takacs, C.D., Fritsen, C.H., Laybourn-
Parry, J., Roberts, E.C., Sattler, B., and Lyons, W.B. 1999a. Car
-
bon transformation in a perennially ice-covered Antarctic lake.
BioScience, 49: 997–1008.
Priscu, J.C., Adams, E.E., Lyons, W.B., Voytek, M.A., Mogk, D.W.,
Brown, R.L, McKay, C.P., Takacs, C.D., Welch, K.A., Wolf, C.F.,
Kirshtein, J.D., and Avci, R. 1999b. Geomicrobiology of
subglacial ice above Lake Vostok, Antarctica. Science (Washing
-
ton, D.C.), 286: 2141–2144.
Rigor, I.G., Colony, R.L., and Martin, S. 2000. Variations in surface
air temperature observations in the Arctic, 1979–97. J. Clim. 13:
896–914.
Roesler, C.S. 1998. Theoretical and experimental approaches to
improve the accuracy of particulate absorption coefficients de
-
rived from the quantitative filter technique. Limnol. Oceanogr.
43: 1649–1660.
Rothrock, D.A., Yu, Y., and Maykut, G.A. 1999. Thinning of the
Arctic sea-ice cover. Geophys. Res. Lett. 26: 3469–3472.
Senesi, N., Miano, T.M., Provenzano, M.R., and Brunetti, G. 1991.
Characterization, differentiation, and classification of humic sub
-
stances by fluorescence spectroscopy. Soil Sci. 152: 259–271.
Shindell, D.T., Rind, D., and Lonergan, P. 1998. Increased polar
stratospheric ozone losses and delayed eventual recovery owing
to increasing greenhouse-gas concentrations. Nature (London),
392: 589–592.
Stal, L.J., van Gemerden, H., and Krumbein, W.E. 1984. The si
-
multaneous assay of chlorophyll and bacteriochlorophyll in nat
-
ural microbial communities. J. Microbiol. Methods, 2: 295–306.
Thomas, C.W. 1963. On the transfer of visible radiation through
sea ice and snow. J. Glaciol. 4: 481–484.
Van Hove, P., Swadling, K.M., Gibson, J.A.E., Vincent, W.F., and
Belzile, C. 2001. Farthest north lake and fjord populations of
calanoid copepods Limnocalanus macrurus and Drepanopus
bungei in the Canadian High Arctic. Polar Biol. 24: 303–307.
Vincent, W.F., Rae, R., Laurion, I., Howard-Williams, C., and
Priscu, J.C. 1998. Transparency of Antarctic ice-covered lakes
to solar UV radiation. Limnol. Oceanogr. 43: 618–624.
Vincent, W.F., Gibson, J.A.E., and Jeffries, M.O. 2001. Ice-shelf
collapse, climate change and habitat loss in the Canadian High
Arctic. Polar Rec. 37: 133–142.
Welch, H.E., and Kalff, J. 1974. Benthic photosynthesis and respi
-
ration in Char Lake. J. Fish. Res. Board Can. 31: 609–620.
Weller, G. 1998. Regional impacts of climate change in the Arctic
and Antarctic. Ann. Glaciol. 27: 543–552.
Wharton, R.A., Jr., McKay, C.P., Clow, G.D., and Andersen, D.T.
1993. Perennial ice covers and their influence on Antarctic lake
ecosystems. Antarct. Res. Ser. 59: 53–70.
J:\cjfas\cjfas58\cjfas-12\F01-187.vp
Tuesday, December 11, 2001 2:33:43 PM
Color profile: Disabled
Composite Default screen