Hydrobiologia 347: 57–68, 1997.
c 1997 Kluwer Academic Publishers. Printed in Belgium.
Detection and characterization of denitrifying bacteria from a permanently
ice-covered Antarctic Lake
B. B. Ward1 & J. C. Priscu2
Institute of Marine Sciences, University of California, Santa Cruz, CA 95064, USA
(fax: 408-459-4882; e-mail: [email protected])
Department of Biology, Montana State University, Bozeman, MT 59717, USA
(fax: 406-994-5863; e-mail: [email protected])
Received 1 August 1996; in revised form 25 March 1997; accepted 3 April 1997
Key words: denitrifying bacteria, Antarctic bacteria, immunofluorescence
Denitrifying bacterial strains were isolated from Lake Bonney, a permanently ice-covered and chemically stratified
lake in the McMurdo dry valley region of Antarctica, using complex media at 4 C. Three strains, identified
as denitrifiers by their ability to produce nitrous oxide using nitrate or nitrite as a respiratory substrate, were
characterized as to their temperature and salinity optima for aerobic growth in batch culture; all three were
psychrophilic and moderately halophilic. Maximum growth rates of near 0.024 h 1 were measured for all three
strains. Growth rates projected to occur at in situ temperature and salinity imply generation times on the order of
100 h. Species specific polyclonal antisera were prepared against two of the strains, ELB 17 (from the east lobe of
the lake at 17 m) and WLB20 (from the west lobe at 20 m). Both strains were subsequently detected and enumerated
in the lake using the antisera. ELB17 was present in both lobes below the chemocline, while WLB20 was present
in the west lobe below the chemocline but only in surface waters of the east lobe. These distributions are related to
the observed chemical distributions which imply the occurrence of denitrification in the west lobe of the lake and
not in the east lobe.
Although the physiological characteristics of nitrifying
and denitrifying bacteria are apparently incompatible,
there are many environments in which coupled nitrification/denitrification is the rule. These environments
are characterized by physicochemical gradients across
which the substrates and products of the coupled reactions diffuse and which allow the incompatible transformations to be separated in time or space. Examples
of such gradient systems, and the typical chemical
distributions which result, are the Black Sea (Codispoti et al., 1991) and aquatic sediment environments
(Reeburgh, 1983). In the water column of lakes and
oceans, the transition from oxygenated water to anoxic
water occurs over meters to hundreds of meters, while
in sediments, the relevant scale is microns to centimeters. In what may be considered a ‘normal’ situation,
the upper layer contains nitrate (and often oxygen);
below the interface, nitrate and oxygen are both totally
depleted and ammonium and sulfide may accumulate
as the result of anaerobic decomposition. Small peaks
in nitrite and nitrous oxide are often present near the
interface. These chemical distributions are interpreted
as the result of stratification in the activities of nitrifying (aerobic), denitrifying (facultatively anaerobic) and
sulfate reducing (usually anaerobic) bacteria. The lake
where this study was performed has two lobes, both
stably stratified and connected by a shallow sill. The
west lobe exhibits the classical chemical distributions
described here, whereas the east lobe exhibits anomalous nitrogen distributions, specifically, high concentrations of oxidized nitrogen in suboxic water (Priscu
et al., 1996).
Permanently ice-covered Lake Bonney, an Antarctic lake in the Taylor Valley adjacent to McMurdo
Article: hydr 3693 GSB: Pips nr 138772 BIO2KAP
hydr3693.tex; 4/07/1997; 17:57; v.7; p.1
Sound, presents a special situation where turbulence
and upper trophic levels are virtually nonexistent. The
lake is chemically stratified; the chemocline divides
the lake into an upper trophogenic zone and a lower zone where photoautotrophic activity is virtually
absent (Lizotte et al., 1996; Priscu, 1995). Thus, physical processes (mainly diffusion) and microbial activity
dominate biogeochemica] reactions and chemical distributions, particularly below the chemocline, which
coincides with the oxycline. The deep water chemistry
of the two lobes of the lake differs dramatically (Spigel
& Priscu, 1996), probably due ultimately to important
differences in the recent (1000–2000 yr) histories of
the two lobes (Lyons et al., in press). The lobes are
separated by a sill depth of 12–13 m and both lobes
are perennially covered by 3–5 m of ice. Phytoplankton primary productivity in both lobes is restricted to
the layer above the chemocline; 20 m in the east lobe
and 15 m in the west lobe (Priscu, 1995; Lizotte et al.,
1996). Productivity maxima are associated with nutrient gradients within this layer. In the west lobe, the
distribution of inorganic nitrogen species below the
chemocline is consistent with a typical stratified system in which the surface layer is nitrogen-depleted, and
the deep anoxic layer has high ammonium concentrations, but nitrate present at very low concentrations,
apparently depleted by denitrification. Denitrification
has been detected in water from the west lobe by the
acetylene block method (Priscu et al., 1996). The east
lobe also has a nitrogen-depleted surface layer and an
anoxic deep layer. This deep layer, in contrast to the
west lobe, has high concentrations of nitrate, nitrite and
ammonium, making it appear that denitrification has
not occurred at rates high enough to deplete the oxidized nitrogen pool. Denitrification was not detectable
in the east lobe by the acetylene block method (Priscu
et al., 1996). At the oxic/anoxic interface of the east
lobe, nitrous oxide occurs at very high levels, but much
lower levels exist in the west lobe (Priscu et al., 1993;
Priscu et al., 1996).
These observations suggest that denitrification is
occurring below the chemocline in the west lobe but not
in the east and that nitrification, rather than denitrification, dominates the chemical distributions in the east
lobe. This difference could be due simply to absence
of denitrifying bacteria in the east lobe, or to inhibition
of denitrification in the east lobe. Even if the chemical
distributions are historical and do not reflect modern
processes (Priscu, 1995), the exact reason for lack of
denitrification in the east lobe remains in question. A
first step in discerning the underlying mechanisms is to
determine the distribution of bacteria responsible for
the processes of nitrification and denitrification, which
are thought partially to control the nutrient distributions. Denitrifying bacteria are addressed specifically
in this paper. Measurements of rate transformations
(Priscu et al., 1996) and nitrifying bacterial distributions (Voytek & Ward, 1995; Voytek, 1996; Voytek,
Ward & Priscu, in preparation) are presented elsewhere.
Isolations and strains
Two approaches were used for isolation of denitrifying strains from lakewater. Strain ELB17 was isolated
from 17 m in the east lobe by adding a complex carbon
source (peptone, 1 g l 1 ) and nitrate (1 mM KNO3 )
to natura] lake water, which was then incubated at in
situ temperature or at 4 C. Strains WLB20 (20 m in
the west lobe) and WLB35 (35 m in the west lobe)
were isolated by inoculating 1 ml of lake water into
25 ml of sterile artificial lake water medium, which had
been enriched with peptone and nitrate as above. The
artificial lake water medium was made from distilled
water plus 1 ml trace metals solution (Biebl & Pfennig,
1978) plus the following salts (g l 1 ) in concentrations
designed to cover the range of salinities found in the
lake (Spigel & Priscu, 1996): MgSO4 .7H2 0O: 46.3 and
65.9; NaCl: 62.5 and 85; CaSO4 .2H2 O: 4.65 and 5.12,
for WLB20 and WLB35, respectively. Isolates were
streaked to purity on agar plates of the same medium under an atmosphere of dinitrogen gas. In order to
encourage the growth of strains adapted to the environment of origin, enrichments were maintained as close
as possible to the original ambient temperature (4 C)
until pure isolates were obtained. The isolates were
then frozen in 15% glycerol at 70 C until required
for experiments. Subsequent growth of the isolates for
production of antigens took place at a higher temperature in order to obtain more rapid growth (12 C).
Ability of isolates to denitrify was ascertained by
sequential growth in the above medium on oxygen,
then nitrate, then nitrite as electron acceptors. Ability to
grow on NO2 (0.01 to 0.25% w/v) and detection of trace
levels of N2 O (by ECD gas chromatography) in the
headspace of sealed tubes in which liquid cultures were
grown were taken as evidence of ability to denitrify
completely to nitrogen gas.
hydr3693.tex; 4/07/1997; 17:57; v.7; p.2
Figure 1. Hydrographic data (November 1994) from central stations in (A) East and (B) West lobes of Lake Bonney. Temperature (+); oxygen
( ); nitrite ( ); ammonium ( ); nitrate ( ); nitrous oxide ( ).
Growth rate experiments
Inocula for the growth rate experiments were grown in
medium of ambient salt concentration at 12 C until
turbidity was detected. One ml of log phase culture
was then inoculated into each of 16 treatment flasks,
each containing 100 ml of artificial lake water medium of identical composition except for salt concentration. The treatments consisted of 4 different NaCl concentrations representing approximately 0.25, 0.5,
1 and 2 relative to ambient (1). (Ambient salt
[NaCl + MgSO4 + CaSO4 ] levels for the three isolates
were 60, 99, and 136 g l 1 for ELB17, WLB20 and
WLB35 respectively.) One flask of each of the different salt concentrations was incubated at one of four
different temperatures: 4 C, 12 C, 15 C, 25 C.
Flasks were shaken continuously at about 125 rpm. At
least once daily, a 1 ml sample was removed asceptically from each flask and its OD measured in a small
volume cell at 450 nm using a Hitachi dual beam spectrophotometer. For each flask, a log transformation of
the OD data was made and a maximum growth rate
was determined from the slope of the regression equation from the linear portion of the log plot during early
Immunofluorescence and bacterial enumeration
For production of antibodies, cells were grown at 12 C
aerobically in 1-L flasks in medium of ambient salt concentration and harvested by centrifugation. Cell pellets were washed twice with phosphate buffered saline
(PBS; per liter of distilled water: 8 g NaCl, 0.2 g KCl,
1.15 g Na2 HPO4 , 0.2 g KH2 PO4 ), resuspended in a
minimum volume of PBS and frozen. The protein concentration of the antigen preparation was adjusted to
10 mg ml 1 (by dilution with PBS) as measured by the
Coomassie blue assay (Bradford, 1976).
Antibodies for ELB17 were produced in rabbits by
the Berkeley Antibody Company (Berkeley, CA) and
for WLB20 by Rockland, Inc. (Boyertown, PA) following the immunization protocol of Ward and Cockcroft
hydr3693.tex; 4/07/1997; 17:57; v.7; p.3
(1993). The working antibody concentration (800-fold
dilution for ELB17, 400-fold dilution for WLB20) was
chosen as the highest dilution which resulted in a + 4
reaction by immunofluorescence (on an operator subjective scale from undetectable ( ) to maximum fluorescence intensity (+ 4)) and which resulted in + 4
staining of every cell in a sample from a homologous
Specificity of the antibodies was investigated by
immunofluorescence staining of a variety of known
denitrifying (15 strains) and non-denitrifying strains
(11 strains), and a group of recently isolated unidentified denitrifiers (9 strains), using the microfuge staining method described previously (Ward & Carlucci,
1985). The strains we tested are those used by Ward
and Cockcroft (1993) to test the specificity of another
denitrifier antiserum, plus two additional environmental isolates. The antisera were found to be very specific
(no cross reactions with heterologous strains). To minimize nonspecific staining in complex samples, both
antisera were absorbed with whole cells of E. colz
before using them to enumerate natural samples (Belly et al., 1973). Absorption of the antiserum did not
reduce the number or intensity of stained cells in the
pure culture of target cells. Field samples were stained
only with absorbed serum.
Indirect immunofluorescence (IIF) staining methods for water samples have been described previously (Ward & Carlucci, 1985). The fluorescent
antibody, fluorescein isothiocyanate-conjugated sheep
anti-rabbit immunoglobulin G, was obtained from
Miles Laboratories, Inc. (Elkhardt, IN). The microfuge
method was used for cross reaction tests and the filter
staining method (using 0.2 m pore size, 25 mm Nuclepore or Poretics filters, pre-stained in Irgalan Black)
was used for enumeration of formalin preserved (see
below) lake samples (Ward & Carlucci, 1985). Stained
samples were stored in the refrigerator and retained
their initial fluorescence almost indefinitely. Samples
were viewed by epifluorescence microscopy at 1000
power using a Zeiss standard microscope with epifluorescence modification (50 W Hg illumination; 450 DF
55 filter for excitation, 505 DF 35 barrier filter and
505DRLEXT02 dichroic filter). Immunofluorescently stained cells were distinguished in natural samples
by a staining reaction of + 4 or + 3 (compared to the
homologous reaction of + 4). ELB17 cells have a characteristic long thin rod shape and WLB20 cells were
quite distinct in shape, being much shorter and thicker. It was possible to enumerate both kinds of cells
on the same filter by staining with a mixture of the
two sera, but it was simpler to enumerate them separately. For enumeration of water samples, 10–20 ml
samples were counted. For both AODC (see below)
and IIF staining, a total of 300 fields or approximately
200 cells were enumerated for each count. This did not
always result in enumeration of a sufficient number of
cells to exceed the limit of detection (in the case of
IIF, it was not always possible to count 200 cells), but
represents a maximum acceptable amount of time and
effort per sample. Acridine orange stained filters were
prepared from 5–10 ml aliquots of the same samples
and enumerated using the same microscope with the
same filter set (AODC, Porter & Fieg, 1980).
Ancillary chemical measurements and sources of
Samples for enumeration of bacteria from planktonic
communities were collected in November and December of 1992, 1993 and 1994 using Niskin sampling
bottles lowered through holes in the 4-m thick layer
of ice covering the surface of the lake. Approximately
200 ml was collected in previously combusted glass
bottles to which 4 ml filtered (0.22 m) buffered (sodium borate) formalin were added. Bottles were stored
in the dark at 4 C until enumeration.
Oxygen concentrations were determined by Winkler titrations (Parsons et al., 1984). Dissolved nutrients (nitrate, nitrite and ammonium) were assayed
by standard methods (Parsons et al., 1984) with
slight modifications to allow for minimal sample sizes.
Conductivity and temperature were measured with a
Martek 7 instrument (Spigel & Priscu, 1996). N2 O
samples were collected directly from a 5-liter Niskin
bottle connected to a 10 ml glass-barrel syringe with
a 2 cm length of rubber tubing. Every precaution was
taken to ensure that the samples did not contact air or
otherwise degas. Water samples (5 ml) were analyzed
following 2 or 3 equilibrations with high-purity He
(McAuliffe, 1971). N2 O was measured with a PerkinElmer Sigma 4 gas chromatograph fitted with a 2 m
stainless steel column of Chromosorb 102 (100–120
mesh) and operated at a column temperature of 56 C,
with a carrier gas (95% Ar: 5% CH4) flow of 24 ml
min 1 . The instrument was standardized as outlined
by Priscu et al. (1996).
During isolation and culturing, nitrite was assayed
by the spectrophotometric method of Parsons et al.
(1984) using a Hitachi double beam spectrophotometer. Nitrous oxide was detected using a Shimadzu Mini2 gas chromatograph equipped with an electron capture
hydr3693.tex; 4/07/1997; 17:57; v.7; p.4
Figure 2. Total bacterial abundance in East ( ) and West ( ) lobes of
Lake Bonney. Means (+) and replicate counts ( , ) from November
Figure 3. Abundance of ELB 17 from IIF assay. Data from November 1994 (counts from three casts for the east lobe samples). East
lobe ( , , ); West lobe ( ).
detector and a 2 m Poropak Q column run at an injection temperature of 300 C and a column temperature
of 45 C. Gas standards were obtained from Scott Specialty Gases (San Bernardino, CA).
region of 13–15 m in the west lobe and at 17–20 m in
the east lobe. Nutrient and hydrographic distributions
have not varied significantly below the chemocline in
at least ten years (Spigel & Priscu, 1996). The water
level of the lakes has risen perceptibly in that time,
but this has resulted from addition of freshwater to the
upper layer and has not significantly affected the distribution of variables discussed here. As previously noted
(Priscu et al., 1993), the two lobes differ strikingly in
the concentrations of inorganic nitrogen species in the
deep waters. Nitrate and nitrite are absent from subhalocline waters in the west lobe, while both are present
at very high levels at similar depths in the east lobe.
Nitrous oxide concentrations also differed between the
two lobes, being present at very low levels in the west
lobe and at concentrations exceeding 40 M N2 O-N
(Priscu et al., 1996) in the interface region of the east
lobe. Maximum temperatures of 5.9 C and 2.9 C
were observed at 18 and 10 m in the east and west
Lake Bonney consists of two elongate lobes, each with
maximum depth of approximately 40 m, separated by a
sill of 13 m depth. All data presented in this paper were
collected from the central stations in each lobe, designated E30 (East lobe) and W20 (West lobe). Although
there is slight variability among years, the nutrient and
chemical data show consistent permanent stratification
(Figure 1). Depths here refer to depth below the hydrostatic water level (i.e. the water level inside the sampling hole). The halocline and oxycline co-occur in the
hydr3693.tex; 4/07/1997; 17:57; v.7; p.5
Distribution of total bacteria and denitrifying strains
Depth profiles of bacterial abundance (AODC) in both
lobes of Lake Bonney in 1994, the year from which
the most complete data set was obtained, are shown
in Figure 2. Data are from a single sample (duplicate
subsamples counted) for each depth from a profile collected in November 1994. In the east lobe, highest cell
numbers (1 106 ml 1 ) were found below the chemocline and in the west lobe, maximum numbers (nearly
2 106 ml 1 ) occurred in the chemocline near 15 m.
In the east lobe, the bacterial abundance maximum was
broad and extended from about 20 to 30 m. In the west
lobe, the upper boundary of the abundance maximum
was sharply defined near the chemocline and numbers
exceeded 1 106 ml 1 between 13 and 20 m.
The distribution of ELB17, the denitrifier originally
isolated from 17 m in the east lobe, is shown in Figure 3. For the immunofluorescence counts, each point
represents a single count (replicate counts on the same
sample were not made due to the large amount of time
required for staining and counting of 350 fields); each
profile represents samples from a different date over the
period of 4 weeks in November and December 1994.
Very few ELB17 cells were detected in the upper layer
but a broad maximum occurred below the chemocline.
This maximum just above the chemocline coincided
roughly with the maximum in total cell numbers but
highest abundances of ELB17 were in the range of
2.5–4 104 cells ml 1 . ELB17 was also detected in
the west lobe with a broad maximum between 13 and
20 m, corresponding to the shallower chemocline in
WLB20 was detected in both lobes but with very
different distributions (Figure 4). In the west lobe,
WLB20 had a maximum of only 1 103 cells ml 1
near 15 m, just below the chemocline, and was present
throughout the deep layer at the level of a few hundred cells per ml. In the east lobe, WLB20 was present
mainly above the chemocline forming a discrete maximum of up to 1 104 cells ml 1 at 17 m, but present
at generally less than 100 cells ml 1 at both shallower
and deeper depths.
Optimal growth conditions for Lake Bonney
The three isolates all exhibited classical growth curves,
with variable lag times and maximal growth rates,
the magnitude of which depended on temperature and
salinity. An example of the range of growth curves
obtained in the 16 salinity/temperature combinations
is shown for strain WL20 in Figure 5. Although the
isolates were never exposed to temperatures greater
than 4 C during the isolation procedure, and the
temperature of the lake environment from which they
were isolated never exceeded 6 C, maximum growth
rates occurred between 12 and 15 C in all three isolates (Table 1). The optimum temperature (the temperature which allows maximum growth rate) appears
to be in this range because lower growth rates were
observed at both higher and lower temperatures, but
the resolution of temperatures tested did not allow the
identification of the single best temperature. Isolates
were never grown in medium containing a salt concentration different from their environment during the
entire isolation procedure. The absolute salt concentration at which maximal growth rates occurred differed
for the three isolates (Table 1), but in each case was
below the salt concentration of the water from which
they were isolated. Contour plots derived from the
observed growth rates for each isolate at each temperature and salinity combination (Figure 6) suggest
that both WLB20 and WLB35 might have optimal salt
concentrations below those tested in the matrix of culture conditions. It was not determined whether the isolates had an absolute salt requirement; the focus of the
experiments was to relate growth of the organisms to
in situ conditions, which included relatively high salt
Maximum growth rates obtained under optimal
conditions were about 0.024 h 1 for all three isolates,
which translates into generation times of around 40 h.
Growth rates estimated to occur at ambient conditions
could be estimated from the contour plots of growth
rate as a function of temperature and salinity (Figure 6).
Rates estimated for in situ temperature and salinity
ranged from 0.011 h 1 for ELB17 to no growth for
WLB35. These values cannot be estimated precisely because growth experiments were not performed at
temperatures below zero. (For the same reason, the
expected growth rates for WLB20 and WLB35 under
ambient conditions cannot be shown on the contour
plots (Figure 6)). The data indicate, however, that,
growth rates at in situ conditions are most likely to be
much slower than those at optimal conditions.
Both denitrifying isolates for which antisera were produced were subsequently detected in the lake, verify-
hydr3693.tex; 4/07/1997; 17:57; v.7; p.6
Figure 4. Abundance of WLB20 from IIF assay. Data from 1994. (A) East lobe; (B) West lobe. Different symbols represent counts from
Table 1. Growth rates measured at optimal conditions and estimated for in situ conditions for denitrifying isolates from Lake Bonney, Antarctica
Optimal growth conditions
Temp ( C) Salinity (g/l) Specific growth Gen time
Source Origin (m) Optimum Optimum
ing that the immunizing strains were true residents of
the lake. Their growth characteristics were also indicative of organisms adapted to low temperatures. In a
separate study, ELB17 was included in an analysis of
diversity of denitrifying strains from various environments by restriction fragment length polymorphism
(Ward, 1995) and found to cluster apart from known
marine and terrestrial strains. All three Lake Bonney
isolates described here appear to be psychrophilic, but
their optimal temperatures exceed those found in their
native environment. Similarly, the optimal salt con-
In situ growth conditions
Temp ( C) Salinity (g/l) In situ growth Gen time
centrations for all three isolates were below those at
the depths from which they were isolated. Salt optima in the range of 3–5% w/v (Table 1) characterizes
these isolates as moderate halophiles. Using specific abundance information based on enumeration by
immunofluorescence (see below), it is possible to interpret the observed distributions of the organisms in situ
in relation to their optimal growth conditions and their
growth response to different temperature and salinity
conditions. IIF assays are usually very specific, and
published estimates of individual strains as percent-
hydr3693.tex; 4/07/1997; 17:57; v.7; p.7
Figure 5. Growth curves for WLB20, isolate from 20 m in West lobe. Symbols represent individual flasks at 4 different temperatures and 4
different salinities. First number in flask identification refers to temperature of incubation. Second number is salinity relative to ambient level.
ages of the total population are usually in the range of
less than 1%, more commonly less than 0.1% (Dahle &
Laake, 1982; Ward & Carlucci, 1985; Ward & Cockcroft, 1993). Each of the two strains enumerated here
by IIF represented from less than 0.01% up to nearly
5% of the total population as enumerated by AODC.
ELB17 constituted 2 to 5% of the AODC total between
17 and 35 m in the east lobe and reached a maximum
of about 2% at 22 m in the west lobe. These maxima
occurred in the depth intervals where denitrification
would be expected to occur, i.e., near the interface
between oxic and anoxic layers in the water column.
WLB20 represented at most 2% of the AODC total,
and this was just above the chemocline in the east lobe.
While WLB20 was present throughout the anoxic layer
of the west lobe, it never contributed more than 0.1%
of the total. These maximum abundances represent a
significant portion of the total cells present, compared
to the abundances of individual strains enumerated by
immunofluorescence in other systems. For example, in
Monterey Bay and the Southern California Bight, the
abundance of Pseudomonas stutzeri was never greater
than 0.08% and 0.2%, respectively (Ward & Cockcroft,
1993; Bard & Ward, 1997). This may mean that total
species diversity in Lake Bonney may be relatively
low, which is consistent with the general trend towards
lower diversity in extreme environments (polar, high
salinity, cold; Atlas, 1984). It is impossible, however,
to assess the total number of strains present, or the total
number of denitrifying strains in the lake. Therefore,
we cannot estimate the extent to which these strains
might represent the behavior of the total denitrifying
assemblage. Nevertheless, their detection as verified by
strain-specific immunofluorescence demonstrates their
presence in the lake and verifies the presence of denitrifiers in both lobes of the lake. Other strains may
dominate under different conditions at various depths
in the water column, where ELB17 and WLB20 are
minor constituents of the population.
The distribution of ELB17 was correlated with total
bacterial abundance (p 0.05, 0.05, 0.01 for the three
sets of ELB17 counts) in the east lobe, and also in the
hydr3693.tex; 4/07/1997; 17:57; v.7; p.8
Figure 6. Variation in growth rate as a function of temperature and salinity for three denitrifying isolates from Lake Bonney. (A) ELB17;
(B) WLB20; (C) WLB35. Contours were generated by SYSTAT using linear interpolation between observations. The salinity scales are different
because the experimental salinities were varied according to the salinity of origin for each isolate.
west lobe (p 0.05) while the abundance of WLB20
was not correlated (p 0.05) with the total abundance
in either lobe. The distributions of the two strains were
not apparently correlated (p 0.05) with each other
within lobes. Two serotypes of nitrifying bacteria were
also enumerated by immunofluorescence in this lake
(unpublished data) and their distributions also did not
correlate with that of the denitrifiers described here.
ELB17 originated in the east lobe at a depth of
17 m, just above the depth where its maximum abundance was subsequently detected by immunofluorescence. Sodium chloride concentration at 17 m was
hydr3693.tex; 4/07/1997; 17:57; v.7; p.9
about 60 g l 1 , nearly twice the salt concentration that
allowed optimal growth. The temperature at this depth
was about 6 C. The generation time of ELB17 estimated for conditions at this depth is about 91 h. Conditions
between 13 and 22 m in the west lobe where ELB17
had a broad abundance maximum, varied from approximately 2.6 C, 90 g l 1 to 1.8 C, 125 g l 1 . The
isolate WLB20 originated within this range, at 20 m,
where its generation time under in situ conditions is
estimated to be about 125 h.
The optimal growth rates were measured, and therefore these in situ rate estimates were deduced, in cultures where carbon substrate was plentiful and oxygen
was freely supplied. While total dissolved organic carbon levels in Lake Bonney range from about 0.2 to
1.7 mM (J. C. Priscu, unpubl. data), concentrations
of labile organic substrates for bacteria, such as glucose, acetate and amino acids, are in the nanomolar to
micromolar range (J. C. Priscu, unpubl. data). Growth
of bacterial cultures under denitrifying conditions is
generally slower than growth on the same medium in
the presence of oxygen (Boogerd et al., 1984). Both
carbon limitation and growth under denitrifying conditions would therefore probably reduce these growth
rate estimates further. Even a 10-fold reduction in
growth rate would still place these generation times
within the range reported for natural bacterial populations growing at similar temperatures in the waters
of McMurdo Sound (Fuhrman & Azam, 1980): minimum generation times of 53 h, and maximum generation times of thousands of hours. It seems likely,
therefore, that both strains are active members of the
microbial community of the lake. Even at relatively
slow growth rates under in situ conditions, their distributions and relative abundances are consistent with
the suggestion that they are actively growing at rates
sufficient to maintain their presence in the lake. In situ
loss rates are expected to be very slow due to the stable vertical stratification (near absence of sinking for
bacterial sized particles), lack of turbulence and slow
advection (Spigel et al., 1991) and a paucity of grazers
(Priscu, pers. observ.).
Their presence in both lobes, although in different distributions, also suggests that absence of bacteria capable of denitrification is not the reason for the
apparent lack of denitrification in the east lobe. ELB17
is a significant portion (up to 5%) of the total bacterial community in the east lobe where denitrification
is absent, and ELB17 is just one of an unknown number of strains capable of denitrification that might be
present. ELB17 is present in both lobes of the lake
in distributions which suggest it is capable of persistence under conditions of the deep water in both lobes.
The distribution of WLB20 implies that this strain persists at low levels throughout the water column in both
lobes, and that it may be seeded into the east lobe by
transport above the connecting sill depth. The fact that
WLB20 occurs at very low levels below the chemocline in the east lobe suggests that the conditions of
the deep water in the east lobe are prohibitive for its
survival. Alternatively, the density stratification of the
east lobe may be so strong as to prevent sinking and
therefore seeding of the deeper layer by cells trapped
in the upper layer.
If denitrifying bacteria are present in both lobes of
the lake, then there must be some in situ chemical condition which inhibits their activity in one lobe versus
the other. The general circulation pattern of the lake
(Spigel et al., 1991) consists of input from the Taylor Glacier at the western end of the west lobe and
general transport of surface water from the west into
the east lobe. The distribution of WLB20 apparently
reflects this circulation pattern. That is, the distribution
of WLB20 is consistent with this strain being abundant
in the west lobe and entering the east lobe via advection in the surface layer from west to east over the sill.
Exchange of water between the lobes below sill depth
is not possible due to the density stratification, but the
chemocline in the west lobe is slightly shallower than
in the east, so that denitrifying bacteria in the chemocline region could be continually transported from the
west to the east. Data from current meter moorings
in Lake Bonney corroborate this flow pattern (Priscu,
Although both temperature and salinity are important determinants of growth rate in all three strains
tested, simple physical parameters such as temperature
and salinity are not likely to be completely inhibitory:
the west lobe is colder than the east, but denitrification appears to proceed there. The east lobe is saltier,
but based on their behavior in culture, the denitrifying
strains investigated here should all be capable of growing actively under conditions pertaining in the region
of the chemocline and even deeper. ELB17 was also
present in the very salty water below the chemocline
of the east lobe, although its viability there cannot be
ascertained with certainty.
Although the lake is permanently stratified and
Winkler oxygen measurements indicate that oxygen
concentrations are low or negligible, recent electrode
measurements of dissolved oxygen (Priscu, in press)
indicate that the deep water of the east lobe contains
hydr3693.tex; 4/07/1997; 17:57; v.7; p.10
between 1 and 2 mg O2 l 1 and the redox is poised
in an oxidized state: Eh below the chemocline always
exceeded 400 mV. Under these redox conditions, both
oxygen and nitrate are thermodynamically favorable
terminal electron acceptors, and denitrification may not
be an advantageous respiratory mode (see also Priscu
et al., 1996). In this interpretation, although denitrifying bacteria are present, they are metabolizing as aerobic heterotrophs, rather than utilizing oxidized forms
of nitrogen for respiration.
Alternatively, the lack of denitrification in the east
lobe may be part of a more general phenomenon
of unusual bacterial activity in this lobe. Repeated
attempts to measure heterotrophic bacterial production using 3 H-thymidine methods (Priscu, 1992) failed
to detect activity in the region of the total bacterial
abundance maximum below the chemocline in the east
lobe. This and the accumulation of products associated with nitrification led to the hypothesis that nitrifying bacteria predominated in this region. Preliminary enumeration of nitrifying bacteria using antisera
developed for marine ammonium-oxidizing bacteria
did indeed detect nitrifiers in this region, but they are
absent or present at very low abundances in the deep
water (Voytek, Priscu & Ward, unpubl. data). It is possible that the bacterial biomass detected in the deep
water of the east lobe is inactive, that the total microbial population is inhibited by the chemistry of the
water, or that the cells present there are in an essentially preserved state, due to the presence of very high
salt concentrations and low temperatures. Nonviable
biomass might also be able to persist suspended in the
water column due to the low numbers of grazers.
The summary of the recent history of Lake Bonney
provided by Lyons et al. (in press) suggests a basis for
the possible general inhibition of microbial life in the
east lobe, or at least for the differences between the
two lobes. The Taylor Valley apparently sustained a
very dry period between 2000 and 2500 yr BP, causing
some of the lakes to dry down and reduce volume
drastically. Lyons et al. (in press) suggest that the east
lobe of Lake Bonney was isolated from input from the
Taylor Glacier and greatly reduced in volume but did
not completely desiccate. The modern relatively fresh
trophogenic surface layer of the east lobe is contiguous
with and similar to that of the west lobe, but the bottom
waters reflect an entirely different history. The refilling
of the east lobe that began 1000–1200 yr BP has simply
redissolved and diluted the salts that were concentrated
in the smaller lake of the dry period. This reconstituted
deep brine may be inhibitory to bacteria seeded into
it from above. A closer investigation of the chemical,
metal and mineral composition of this water might
explain the biogeochemical differences manifested in
the nitrogen and carbon cycles.
M. A. Voytek, C. Cooper, R. Bartlett and K. Wing
assisted with the field work. M. Geissler and
A. R. Cockcroft assisted both with field work and in
the lab. R. Edwards and the LTER team provided the
nutrient data for 1994. This research was supported
by NSF grants (DPP 91-17907 and ODP 92-11773) to
JCP and to BBW.
Atlas, R. M., 1984. Diversity of Microbial Communities. Adv.
microb. Ecol. 7: 1–47.
Bard, D. G. & B. B. Ward, 1997. A species-specific bacterial productivity method using immunomagnetic separation and radiotracer
experiments. J. Microbiol. Met., in press.
Belly, R. T., B. B. Bohlool & T. D. Brock, 1973. The genus Thermoplasma. Ann. N.Y. Acad. Sci. 225: 94–107.
Biebl, H. & N. Pfennig, 1978. Growth yields of green sulfur bacteria
in mixed cultures with sulfur and sulfate reducing bacteria. Arch.
Microbiol. 117: 9–16.
Boogerd, R. C., H. W. van Verseveld, D. Torenvliet, B. Braster
& A. H. Stouthamer, 1984. Reconsideration of the efficiency of
energy transduction in Paracoccus denitrificans during growth
under a variety of culture conditions. Arch. Microbiol. 139: 344–
Bradford, M. M., 1976. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle
of protein-dye binding. Anal. Biochem. 72: 248–254.
Codispoti, L. A., G. E. Friederich, J. W. Murray & C. M. Sakamoto,
1991. Chemical variability in the Black Sea: Implications of continuous vertical profiles that penetrated the oxic/anoxic interface.
Deep-Sea Res. 38: S69–S710.
Dahle, A. B. & M. Laake, 1982. Diversity dynamics of marine
bacteria studied by immunofluorescent staining on membrane
filters. Appl. envir. Microbiol. 43: 169–176.
Fuhrman, J. A. & F. Azam, 1980. Bacterioplankton secondary production estimates for coastal waters of British Columbia, Antarctica, and California. Appl. envir. Microbiol. 39: 1085–1095.
Lizotte, M. P., T. R. Sharp & J. C. Priscu, 1996. Phytoplankton
dynamics in the stratified water column of Lake Bonney, Antarctica. I. Biomass and productivity during the winter-spring transition. Polar Biol. 16: 155–162.
Lyons, W. B., S. W. Tyler, R. A. Wharton, Jr. & D. M. McKnight, in
press. The late Holocene/Paleoclimate history of the McMurdo
Dry Valleys Antarctic as derived from Lacustrine isotope data.
Paleogeogr. Paleoclimatol. Paleoecol.
McAuliffe, C., 1971. GC determination of solutes by multiple phase
equilibration. Chem. Technol. 1: 46–51.
hydr3693.tex; 4/07/1997; 17:57; v.7; p.11
Parsons, T. R., Y. Maita & C. M. Lalli, 1984. A Manual of Chemical
and Biological Methods for Seawater Analysis. Pergamon Press,
Porter, K. G. & Y. S. Feig, 1980. The use of DAPI for identifying and
counting aquatic microflora. Limnol. Oceanogr. 25: 943–948.
Priscu, J. C., 1995. Phytoplankton nutrient deficiency in lakes of the
McMurdo Dry Valleys, Antarctica. Freshwat. Biol. 34: 215–227.
Priscu, J. C., 1992. Particulate organic matter decomposition in the
water column of Lake Bonney, Taylor Valley, Antarctica. Ant. J.
US Rev. 1992: 260–262.
Priscu, J. C., 1997. The biogeochemistry of nitrous oxide in permanently ice-covered lakes of the McMurdo Dry Valleys, Antarctica.
Global Ch. Biol., in press.
Priscu, J. C., B. B. Ward & M. T. Downes, 1993. Water column
transformations of nitrogen in Lake Bonney, a perennially icecovered antarctic lake. Ant. J. US Rev. 1993: 237–239.
Priscu, J. C., M. T. Downes & C. P. McKay, 1996. Extreme supersaturation of nitrous oxide in a poorly-ventilated antarctic lake.
Limnol. Oceanogr., 41: 1544–1551.
Reeburgh, W. S., 1983. Rates of biogeochemical processes in anoxic
sediments. Ann. Rev. Earth Planet. Sci. 11: 269–298.
Spigel, R. H., I. Forne, I. Sheppard & J. C. Priscu, 1991. Differences
in temperature and conductivity between the east and west lobes
of Lake Bonney: evidence for circulation within and between
lobes. Ant. J. US Rev. 26: 221–222.
Spigel, R. H. & J. C. Priscu, 1996. Evolution of temperature and salt
structure of Lake Bonney, a chemically stratified Antarctic lake.
Hydrobiologia, 321: 177–190.
Voytek, M. A. & B. B. Ward, 1995. Detection of ammoniumoxidizing bacteria of the beta-subclass of the class Proteobacteria
in aquatic samples with the PCR. Appl. envir. Microbiol. 61:
Voytek, M. A., 1996. Detection, Abundance and Diversity of Aquatic
Nitrifying Bacteria. Ph.D. Dissertation: 250 pp.
Ward, B. B. & A. F. Carlucci, 1985. Marine ammonia- and
nitrite-oxidizing bacteria: Serological diversity determined by
immunofluorescence in culture and in the environment. Appl.
envir. Microbiol. 50: 194–201.
Ward, B. B. & A. R. Cockcroft, 1993. Immunofluorescence detection of denitrifying bacteria in seawater and intertidal sediment
environments. Microb. Ecol. 25: 233–246.
Ward, B. B., 1995. Diversity in denitrifying bacteria: Limits of
rDNA RFLP analysis and probes for the functional gene, nitrite
reductase. Arch. Microbiol. 163: 167–175.
hydr3693.tex; 4/07/1997; 17:57; v.7; p.12