United States
Environmental Protection
Agency
Water Engineering
Research Laboratory
Cincinnati OH 45268
Research and Development
EPA/600/S2-87/083 Jan. 1988
v>EPA Project Summary
Sampling Regimes and
Bacteriological Tests for Coliform
Detection in Groundwater
B. A. Caldwell and R. Y. Morita
Since conventional procedures can
fail to detect conforms in potable water,
the effects of increased sampling fre-
quency and alternative bacteriological
media on coliform detection were evalu-
ated for samples from groundwater-fed
public water supply systems. For 1,560
drinking water samples collected from
10 small water systems in western
Oregon, the presence-absence (P-A) test
detected significantly more coliform-
positive samples than either the con-
ventional membrane filtration (MF) test
using mEndo agar-LES or the 5- and
10-replicate fermentation tube (FT) test.
No difference was found in coliform
detection based on five samples col-
lected on the same day or throughout
the sampling interval at either the same
or different locations. A fivefold increase
in sampling frequency increased the
incidence of coliform detection 2.9- to
5.0-fold.
An additional study of 600 water
samples demonstrated that 5 alternative
coliform tests, including the presence-
absence test and 4 MF media formula-
tions (m-T7, mTECmod, HABmod, and
mLS) detected significantly more coli-
form-positive samples than the con-
ventional MF test under either aerobic
or anaerobic conditions. Coliform
densities from the m-T7 test were 1.4
to 2.1 times greater than for the other
MF tests.
This Project Summary was developed
by EPA's Water Engineering Research
Laboratory, Cincinnati, Ohio, to an-
nounce key findings of the research
project that Is fully documented In a
separate report of the same title (see
Project Report ordering Information at
back).
Introduction
The routine monitoring of coliform
bacteria in water supplies involves two
critical steps. First, samples must be col-
lected in a manner that allows a suf-
ficiently accurate assessment of coliform
occurrence in the distribution system.
Increases in sampling frequency for small
water supplies beyond the current once-
per-month requirement raise the pos-
sibility of designing sampling regimes
based on spatial or temporal patterns or
both. Second, since conventional coliform
detection procedures can fail to detect
coliforms from an often stressful environ-
ment, alternative media and incubation
procedures must be evaluated.
The two objectives of this investigation
were to determine if a fivefold increase in
sampling using three alternative sampling
regimes would significantly increase coli-
form detection, and to compare coliform
detection by the conventional aerobic
membrane filtration test using mEndo
agar-LES (STD-MF) with the presence-
absence test (P-A) and five alternative
MF procedures. The MF alternatives were
an agar medium (mTECmod) using lauryl
sulfate broth (mLS), m-T7 medium, modi-
fications of the mTEC (mTECmod) medium
for detection of E. coli, an experimental
carrageenan-containing medium
(HABmod), and anaerobic incubation of
mEndo agar-LES (AN-MF).
Materials and Methods
Initially, 1,560 water samples were
collected in the Willamette River Valley of
western Oregon from 10 small public
water supply systems that serve less
than 1,000 people. Sampling was based
on 5-week intervals, using five locations
-------
per water system, to evaluate sampling
from different locations at the same time
or from either the same or different loca-
tions throughout the sampling interval.
At the end of this study, an additional
288 samples were collected weekly at 3
locations within 8 water systems to gen-
erate a total of 600 samples for the
alternative media study.
For the sampling frequency study,
100-mL subsamples were tested for the
presence of coliforms using a 10-replicate
fermentation tube (FT10) procedure, a
membrane filtration test using mEndo
agar-LES (STD-MF), and a presence-
absence (P-A) test as recommended by
Standard Methods for the Examination of
Water and Wastewater (16th Ed., Ameri-
can Public Health Association, 1985). For
the alternative media study, additional
100-mL subsamples were filtered and
incubated aerobically on HABmod, rnLS,
mTECmod, and m-T7 media. A second
mEndo agar-LES plate was incubated
under anaerobic conditions (AN-MF).
After 24 and 48 hours at 35°C, all gas-
positive FT tubes and all acid or acid+gas
P-A bottles were inoculated into 2% bril-
liant green bile broth (BGB) to confirm the
presence of coliforms by gas production
with 48 hours. For the MF tests, after 22
hours incubation for the STD-MF,
HABmod, mLS, mTECmod, and m-T7 tests
and 48 hours for the AN-MF test, typical
colonies were counted and verified as
coliforms by gas production, first in lauryl
tryptose broth (LTB) and then in BGB.
Significant differences in coliform
detection by the different sampling re-
gimes and bacteriological tests were
evaluated using the chi-square statistic
(McNemar'stest). Differences in coliform
density estimated by the MF tests were
evaluated using paired t-tests of the
coliform-positive samples. The goodness-
of-fit between the observed coliform
frequencies and either the negative
binomial or Poisson distribution was
evaluated using the chi-square statistic.
showed 27.8% were detected by all three
tests, 14.2% by FT and P-A, 8.5% by STD-
MF and P-A, 4.5% by FT, and STD-MF,
23% by P-A, 13.4% by FT and 8.5% by
STD-MF.
The confirmation or verification effici-
encies were highest for the FT test (77%),
intermediate for total P-A results (62.9%),
and lowest for the STD-MF (47.4%). The
overall efficiency of the P-A test was
strongly influenced by the very low con-
firmation rate (4.2%) of the 142 acid-only
presumptive P-A bottles.
After partitioning the coliform-positive
samples into the original four sampling
regimes, our data (Table 2) showed that a
fivefold increase in sampling frequency
resulted in a 2.9- to 5.0-fold increase in
coliform detection over the once-per-
interval estimate, depending on the
sampling regime and bacteriological test
used. For any one test there was no
significantly superior sampling regime
although there were some significant dif-
ferences between tests within a given
sampling regime.
Alternative Media Study
The incidence of positive samples and
population characteristics of 600 water
samples processed by 7 bacteriological
procedures are listed in Table 3. The
greatest number of coliform-positive
samples was detected using the m-T7
test, although the P-A results were not
significantly lower. The m-T7 results were
also significantly higher than those pro-
duced by the mTECmod, HABmod, or
mLS tests; which in turn were signifi-
cantly higher than the results on mEndo
agar-LES under either aerobic or anaer-
obic incubations.
The statistics describing the coliform
frequency distributions also varied with
the MF test employed. The observed dis-
tributions for all six MF procedures were
significantly different from a Poisson
distribution. All distributions, except for
HABmod, were adequately explained by
a negative binomial function. The pa-
rameters defining the negative binomial
distribution — mean and coefficient of
aggregation — changed substantially with
the different procedures.
Paired comparisons showed the m-T7
to produce coliform density estimates 1.4-
to 2.1-fold greater than the other MF
tests with both the m-T7 and mTECmod
results being significantly higher than
the STD-MF results.
Conclusions
The results of this project indicate that:
the currently accepted replicate fermenta-
tion tube test and membrane filtration
test using mEndo agar-LES are inade-
quate in detecting either the incidence or
densities of coliforms in groundwater;
increased sampling frequency increases
Table 1. Incidence of Coliform Positive Samples and Results of Individual Tests.
FT5 FT10 MF P-A
Positive Samples
Presumptive
Positive Tests
Efficiency*
165° 21 1b
— 795
— 77.0%
174"
1175
47.4%
259°
142 (acid)
270 (acid+gas)
4.2% (acid)
93.3% (acid+gas)
a-b-c - values followed by different letters were significantly different (p<0.05), by McNemar's Test
* - confirmation or verification efficiency
Experimental Results
Sampling Frequency Study
Analysis of 1,560 samples resulted in
352 coliform-positive samples. The in-
cidence of positive samples by bacterio-
logical test and the results of 2,382
individual tests (FT tube, P-A bottle, or
MF colony) are compiled in Table 1. The
P-A test significantly outperformed the
FT10, which in turn outperformed both
the STDMF and FT5 tests. Additional
breakdown of coliform-positive samples
Table 2. Incidence of Coliform Detection by Four Sampling Regimes.
Test
FTW
STD-MF
P-A
Single
Sample
19
20
20
5 Samples
on Same Day
Different Sites
84
58"
99
5 Samples throughout
5-wk Interval
Same Site
79
77
100
Different Sites
78
69
100"
' - column (test) results were significantly different from others
-------
'able 3.
Incidence of Positive Samples and Population Characteristics of 600 Water Samples
Processed by 7 Bacteriological Procedures.
Membrane Filtration Procedures
mEndo agar-LES
Anaerobic Aerobic
Statistics
Positive
Samples
Mean
Variance
K*
P-A
109c-d
na
na
na
(AN-MF)
58"
0.48
14.8
0.0396
ISTD-MF)
64"
0.74
34.0
0.0362
mTECmod
102°
1.00
32.2
0.0638
HABmod
gjb.c
0.87
31.7
0.0549
mLS
79"
0.65
17.0
0.0543
m-T7
121"
1.48
59.2
0.0686
coliform detection, although there is no
apparently superior strategy for timing or
location of sample collection, and selec-
tion of a specific mathematical model
used to describe the distribution of col>-
forms in groundwater is dependent on
the bacteriological test used in the initial
observations.
The full report was submitted in ful-
fillment of Cooperative Agreement No.
CR 811488-01 by Oregon State University
under the sponsorship of the U.S. En-
vironmental Protection Agency.
_ va/ues followed by different letters were significantly different (p«3.05) by McNemar's Test
na - not applicable to P-A test
* - coefficient of aggregation
Richard Y. Morita and Bruce A. Caldwell are with Oregon State University,
Corvallis, OR 97331.
Eugene W. ft ice and Harry D. Nash are the EPA Project Officers fsee below).
The complete report, entitled "Sampling Regimes and Bacteriological Tests for
Coliform Detection in Groundwater," (Order No. PB 88-107 230/AS; Cost:
$11.95, subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officers can be contacted at:
Water Engineering Research Laboratory
U.S. Environmental Protection Agency
Cincinnati, OH 45268
-------
it;
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use S300
EPA/600/S2-87/083
0000329 PS
*GENCY
I il i! II..!!....!..1.1..11 HliMilitliitliililiilMi>liiiil
------- |