9. Microbial Carbon Cycling in Pelagic Ecosystems: Microbial Methods for Ecosystem Scientists
141
in humic substances, nonspecific binding can cause
a great deal of difficulty. Under the epifluorescent
microscope, bacteria are recognized as stainpositive particles that resemble bacteria. Clearly,
some experience and training are needed before a
novice can count proficiently. Many researchers
choose one or the other of the stains for different
situations. Although agreement in counts is expected between DAPI and AD, the agreement is not
always perfect. In a formal test, Suzuki et al. (1993)
found that DAPI counts averaged about 70% of AD
counts in the samples they tested. Where possible,
we recommend checking representative samples
with both stains and choosing the most effective.
For a review of the use of fluorochromes for direct
enumeration of planktonic bacteria, refer to Kepner
and Pratt (1994).
While most counting is still done manually, the
use of image-analysis systems to enhance or partially automate the procedure is gaining use but the
equipment is costly (Sieracki et al. 1989; Psenner
1993).
The sources of variance and statistics of manual
counting have been reviewed by Kirchman (1993)
and Kirchman et al. (1982b). From a practical consideration of trying to optimize both effort and accuracy, Kirchman (1993) suggests taking two preserved samples, making one filtered preparation per
sample, and counting about 10 microscopic fields
per sample. The amount of sample (or field size)
should be adjusted to yield about 30 cells per field.
Less than this results in high variance among fields.
With too many cells per field, the counter's eyes
can become fatigued causing underestimation
(Kirchman 1993).
Sample Preservation
Samples can be preserved at two stages. Investigators usually fix the initial, aqueous, samples upon
collection with formalin (0.5 to 2% final concentration), glutaraldehyde (2.5% final concentration),
or either of these fixatives with buffers. While these
fixatives, with the samples stored cold «5°C), are
usually adequate to prevent further microbial
growth, countable cells tend to decrease over time.
Turley and Hughes (1992) observed a 39% decrease in AD counts over a 40-day period. Better
preservation is achieved by making the complete
slide preparation within 12 hours of sample collection and then storing the slide frozen (Turley 1993).
Count by Flow Cytometry
An exciting recent development has been the use
of high-resolution flow cytometry to count bacterial
samples. Cells are again stained with a nuclear fluorochrome but are counted directly in a flow cytometer. The cytometer can resolve event frequency,
forward scatter, and fluorescence intensity of
stained cells. Using these parameters it should be
possible to obtain information on abundance, size
distribution, and possibly activity (Button and Robertson 1993). Some of the initial difficulties with
the cytometric approach were that different groups
of bacteria stained differently with DAPI due to
differences in adenine + thymine/guanine + cytosine (AT/GC) ratios. Further, because of the lack
of suitable standards, it has proven difficult both to
get size information and to know if the smaller cells
were being counted with high efficiency (Button
and Robertson 1993). Some of these problems have
been overcome by using a different nuclear stain
(SYTD 13) (del Giorgio et al. 1996b). In a calibration study using samples from 18 Canadian lakes,
del Giorgio et al. (1996b) obtained excellent agreement between cytometer cell counts and those done
by traditional epifluorescent microscopy. The obvious advantages of the cytometer are speed and
precision. In the same data set, the mean coefficient
of variation for epifluorescent counts was 14.5%,
while for the flow cytometer it was 2.5%. Bacterial
cell size may in theory be obtained from forward
light scatter (Robertson and Button 1989), although
planktonic bacteria are often close to the limits of
resolution of most instruments (Davey and Kell
1996). In addition to total abundance and cell size,
it is also possible to obtain data on the amount of
DNA per cell and other cellular structural and
physiological characteristics (Davey and Kell1996;
Wallner et al. 1997; see also below).
Active and Inactive Cells
Total direct counts of bacteria vary reasonably, predictably among environments, from <5 X 10 6 to
approximately 5 X 10 8 cells per liter. This variance
tends to track trophic richness and is reasonably
well correlated with the biomass of phytoplankton
141
in humic substances, nonspecific binding can cause
a great deal of difficulty. Under the epifluorescent
microscope, bacteria are recognized as stainpositive particles that resemble bacteria. Clearly,
some experience and training are needed before a
novice can count proficiently. Many researchers
choose one or the other of the stains for different
situations. Although agreement in counts is expected between DAPI and AD, the agreement is not
always perfect. In a formal test, Suzuki et al. (1993)
found that DAPI counts averaged about 70% of AD
counts in the samples they tested. Where possible,
we recommend checking representative samples
with both stains and choosing the most effective.
For a review of the use of fluorochromes for direct
enumeration of planktonic bacteria, refer to Kepner
and Pratt (1994).
While most counting is still done manually, the
use of image-analysis systems to enhance or partially automate the procedure is gaining use but the
equipment is costly (Sieracki et al. 1989; Psenner
1993).
The sources of variance and statistics of manual
counting have been reviewed by Kirchman (1993)
and Kirchman et al. (1982b). From a practical consideration of trying to optimize both effort and accuracy, Kirchman (1993) suggests taking two preserved samples, making one filtered preparation per
sample, and counting about 10 microscopic fields
per sample. The amount of sample (or field size)
should be adjusted to yield about 30 cells per field.
Less than this results in high variance among fields.
With too many cells per field, the counter's eyes
can become fatigued causing underestimation
(Kirchman 1993).
Sample Preservation
Samples can be preserved at two stages. Investigators usually fix the initial, aqueous, samples upon
collection with formalin (0.5 to 2% final concentration), glutaraldehyde (2.5% final concentration),
or either of these fixatives with buffers. While these
fixatives, with the samples stored cold «5°C), are
usually adequate to prevent further microbial
growth, countable cells tend to decrease over time.
Turley and Hughes (1992) observed a 39% decrease in AD counts over a 40-day period. Better
preservation is achieved by making the complete
slide preparation within 12 hours of sample collection and then storing the slide frozen (Turley 1993).
Count by Flow Cytometry
An exciting recent development has been the use
of high-resolution flow cytometry to count bacterial
samples. Cells are again stained with a nuclear fluorochrome but are counted directly in a flow cytometer. The cytometer can resolve event frequency,
forward scatter, and fluorescence intensity of
stained cells. Using these parameters it should be
possible to obtain information on abundance, size
distribution, and possibly activity (Button and Robertson 1993). Some of the initial difficulties with
the cytometric approach were that different groups
of bacteria stained differently with DAPI due to
differences in adenine + thymine/guanine + cytosine (AT/GC) ratios. Further, because of the lack
of suitable standards, it has proven difficult both to
get size information and to know if the smaller cells
were being counted with high efficiency (Button
and Robertson 1993). Some of these problems have
been overcome by using a different nuclear stain
(SYTD 13) (del Giorgio et al. 1996b). In a calibration study using samples from 18 Canadian lakes,
del Giorgio et al. (1996b) obtained excellent agreement between cytometer cell counts and those done
by traditional epifluorescent microscopy. The obvious advantages of the cytometer are speed and
precision. In the same data set, the mean coefficient
of variation for epifluorescent counts was 14.5%,
while for the flow cytometer it was 2.5%. Bacterial
cell size may in theory be obtained from forward
light scatter (Robertson and Button 1989), although
planktonic bacteria are often close to the limits of
resolution of most instruments (Davey and Kell
1996). In addition to total abundance and cell size,
it is also possible to obtain data on the amount of
DNA per cell and other cellular structural and
physiological characteristics (Davey and Kell1996;
Wallner et al. 1997; see also below).
Active and Inactive Cells
Total direct counts of bacteria vary reasonably, predictably among environments, from <5 X 10 6 to
approximately 5 X 10 8 cells per liter. This variance
tends to track trophic richness and is reasonably
well correlated with the biomass of phytoplankton