Sunteți pe pagina 1din 13

Environmental Monitoring and Assessment (2006) 121: 275–287

DOI: 10.1007/s10661-005-9121-5 
c Springer 2006




Earth Sciences and Resources Institute at the University of South Carolina 402 Byrnes Building,
Columbia, SC, 29208; 2 LT Environmental, 4400 W. 46th Ave, Denver, CO 80212; 3 Department of
Biology, West Virginia University, Morgantown, WV 26506
(∗ author for correspondence, e-mail:

(Received 5 July 2005; accepted 4 November 2005)

Abstract. The United States Environmental Protection Agency (USEPA) recommends the use of
Escherichia coli (E. coli) and enterococci as indicators of enteric pathogens in fresh waters; however,
fecal coliform analyses will remain important by virtue of the large amount of historic data collected
in prior years. In this study, we attempted, in a real-world situation (i.e., a rural inland watershed
in the Piedmont of South Carolina) to compare different bacterial indicators and methods to one
another. We compared fecal coliforms, enumerated by membrane filtration with E. coli, enumerated
by a commercialized enzyme substrate method and observed E. coli/fecal coliform ratios of 1.63
and 1.2 for two separate tests. In the same watershed, we observed an E. coli/fecal coliform ratio
of 0.84 when we used the commercialized enzyme substrate method for both enumerations. Given
these results, users of such data should exercise care when they make comparisons between historic
membrane filtration data and data acquired through the use of the more modern enzymatic methods.
Some sampling and side-by-side testing between methods in a specific watershed may be prudent
before any conversion factors between old and new datasets are applied.

Keywords: Escherichia coli, fecal coliform, bacteria, indicator, methods, rural, watershed

1. Introduction

Ambient water quality regulations make use of bacterial indicators because the den-
sity of an indicator in the water can be quantitatively linked potential human health
risks (Cabelli, 1983). The indicators themselves are not necessarily pathogenic, but
they are commonly associated with pathogens (viral, bacterial and protozoan) that
have made their way into the water column by way of the feces of humans or other
warm-blooded creatures. The United States Public Health Services (USPHS) first
based coliform indicator criteria on studies that the agency conducted in the 1940’s
and early 1950’s. In the 1960’s the USPHS adopted the fecal coliform standard
by using a fixed ratio (18%) of fecal coliform to total coliform bacteria (USEPA,
1996). Subsequent studies in the late 1970 s showed that in freshwater, E. coli and
enterococci were more reliable predictors of gastric and related illnesses than fecal
coliforms. In 1986, the United States Environmental Protection Agency (USEPA)
276 R. W. KLOOT ET AL.

recommended the use of E. coli and enterococci as indicators in fresh waters and
the use of enterococci in marine waters (USEPA, 1986). By 2002, however, only 13
states had adopted these recommendations (USEPA, 2002). As states move toward
implementing the USEPA recommendations, fecal coliform analyses will remain
important by virtue of the large amount of historic fecal coliform data collected in
prior years (Yakub et al., 2002).
Coliform group bacteria encompass some 19 genera (e.g., Klebsiella, Enter-
obacter, and Serratia) and 80 species; they are found in the intestinal tracts of
warm-blooded creatures and in mollusks, insects, plants and other non-fecal (or
environmental) origins such as soils. (Leclerc et al., 2001). The definitions of col-
iform bacteria are therefore based on common biochemical characteristics (Rompré
et al., 2002); the Standard Methods for the Examination of Water and Wastewa-
ter describes coliform bacteria as lactose fermenters, i.e., they produce gas and
acid from lactose. At a temperature of 35 ◦ C on an endo-type medium, they are
detected as a red colony with a metallic sheen (APHA et al., 1998). Thermotol-
erant (or thermotrophic) coliforms have identical properties to coliform bacteria,
except they are able to survive and grow at 44 ± 0.5 ◦ C (Rompré et al., 2002);
thermotolerant coliforms are commonly known as fecal coliforms. In the 1960’s,
this group was favored over the total coliforms as a bacterial indicator because
it is better associated with fecal origins. Nevertheless, this group contains some
species of an environmental (or non-fecal) origin that have a potential for regrowth
in the environment (USEPA, 1996); in this sense, “fecal coliform” is a misnomer
(Leclerc et al., 2001). Escherichia coli, or E. coli bacteria (EC), are a subset of
fecal coliforms (FC); they are thermotolerant, are considered the most common
gram-negative facultative anaerobe in the intestines of warm-blooded creatures,
and are thus principally associated with fecal contamination (Rompré et al., 2002).
Vernberg et al. (1996) conducted a biotyping exercise of fecal coliform bacteria
in two inlets on the coast of South Carolina, one inlet, Murrell’s Inlet, is highly
urbanized and the other, North inlet, is forested and virtually undeveloped. They
found higher E. coli/fecal coliform ratios (75%> EC/FC >90%) for Murrell’s inlet
while at North inlet, this ratio was much lower (50%> EC/FC >75%); the mi-
crobiodiverity of the fecal bacteria in North inlet was higher, where many more
nonfecal bacteria were present. Vernberg et al. (1996) related the higher incidence
of E. coli, the lower microbiodiversity (hence the higher E. coli/total coliform ra-
tio) to the effects of urbanization. Noble et al. (2004) reported median E. coli/fecal
coliform ratios of 0.88, also in a marine environment. In one rural freshwater envi-
ronment, namely the Shoal Creek watershed in Southwestern Missouri, the median
E. coli/fecal coliform ratio was estimated at 0.86 (Schumacher, 2003).
Traditional methods of enumerating bacteria, namely multiple tube fermen-
tation (MTF) and membrane filtration (MF) are now being supplanted by
more reliable, quicker and less expensive enzymatic methods (Rompré et al.,
2002). For instance, coliform bacteria may be detected through their abil-
ity to degrade ortho-nitrophenyl-b-D-galactopyranoside (ONPG), producing the

yellow-colored (chromogenic) product, onitrophenol. E. coli has the ability to

cleave methylumbelliferyl-b-glucuronide (MUG), that results in the formation of
the fluorescent product 4-methylumbelliferone, detectable under ultraviolet light.
These two properties are simultaneously exploited in the commercially available
Colilert test (IDEXX Laboratories, Inc., Westbrook, Maine) of the presence of to-
tal coliform (ONPG-positive) and E. coli (ONPG, MUG-positive) bacteria. The
IDEXX Quanti-TrayTM 2000 (QT) is a prepackaged tray system developed by
IDEXX. Each tray consists of 48 small and 49 large cells and accepts 100 mL
samples mixed with Colilert reagent. After incubation bacteria are enumerated by
presence/absence of a marker (e.g., ONPG, MUG) in each cell – the number of
positive cells (small and large) may be read off on a chart that provides bacterial
density expressed in Most Probable Number per 100 milliliters (MPN/100 mL).
This system is compared to traditional methods in a number of studies in drink-
ing water (e.g., Fricker et al., 1997; Eckner 1998; Niemela et al., 2003), surface
and bathing water (e.g., Budnick et al., 1996; De Roubin et al., 1997; Francy and
Darner, 2000; Yakub et al., 2002) and marine waters (Geissler et al., 2000; Noble
et al., 2003; Noble et al., 2004; Shibata et al., 2004).
In surface waters, concerns with MF enumerations are often associated with
poor recovery of target bacteria. This problem is worse where bacterial counts,
microbiodiversity, and/or turbidities are high, or where chlorination may stress or
injure bacteria (Jacobs et al., 1986; Brenner and Rankin, 1990; McFeters et al.,
1997a,b; Niemela et al., 2003). On the other hand, literature indicates that the
bacterial recoveries with enzymatic methods are higher than for traditional MF
methods (stensvik, 2000; Niemela et al., 2003). Pisciotta et al. (2002) reported
that, when using the Colilert-18 system, they encountered high false positive rates
for E. coli in marine waters, but no false positives in fresh waters. The resultant
differences in analytical endpoints between MF and emerging enzymatic methods
naturally create the potential for differing results among methods (Noble et al.,
In addition to addressing methods, Yakub et al. (2002) and Noble et al. (2004)
also address the issue of the traditional fecal coliform indicator and its relationship
to E. coli. Yakub et al. (2002) suggested a protocol for enumerating thermotolerant
(or fecal) coliforms by incubating IDEXX Quanti-TraysTM at 44.5 ◦ C. They found
that the results from this protocol are a plausible substitute for fecal coliforms that
were enumerated by membrane filtration. Noble et al. (2004) used E. coli as a local
substitute for fecal coliforms and compared them to fecal coliforms enumerated by
MF and MTF in marine waters off the California coast.
This paper responds to the call for more real world, side by side testing of samples
from the environment (Noble et al., 2003). We took surface water samples from 20
sites in a rural watershed in the South Carolina Piedmont (Figure 1), where a number
of the sites are chronically polluted with bacteria. In this study, we compare E. coli
densities enumerated by the Colilert R
QT system [measured in Most Probable
Number per 100 ml (MPN/100 mL)] to fecal coliform densities enumerated by MF
278 R. W. KLOOT ET AL.

Figure 1. Bush River watershed with sample sites. Included are wastewater treatment plants (WWTP)
locations and land use.

[measured in Colony Forming Units per 100 milliliters (CFU/100 mL)]. We also
compare E. coli enumerations with the method Yakub et al. (2002) suggested for
estimating fecal coliform bacteria with the Colilert
QT system.
All work was conducted in the Bush River watershed which is located in the
South Carolina Piedmont and is some 297 km2 (116 sq. miles) in area (Figure 1)
and is on the South Carolina Department of Health and Environmental Control
(SCDHEC) 303(d) (or impaired water body) listing for the fecal coliform indicator
bacteria (SCDHEC 2001). The watershed is rural, and approximately 56% of the
land is wooded (forest or scrub), 37% is agricultural (pasture and row crops), and
7% is urban (Miller, 2001).

2. Materials and Methods

In the study, we made three comparisons (Table I). Series 1 compared in-house E.
coli enumerated by the Colilert
QT system (ECQT) and fecal coliform enumerated
by membrane filtration in a commercial laboratory (FCMF) for paired samples.
Series 2 compared ECQT and FCMF for split samples, while Series 3 compared
ECQT and in-house fecal coliform enumerated by the Colilert R
QT system (FCQT)
for split samples.
We conducted five sample rounds between October 2003 and October 2004
at the sample sites in the Bush River watershed (Figure 1), and restricted each
sampling round to a one-day period. At each sample site, we took dip samples

Description of each comparison as Series 1, 2 and 3

Sample Samples paired

Series Dates number (n) or split Comparison

1 Oct–Dec 2003 56 Paired ECQT and FCMF

2 April 2004 19 Split ECQT and FCMF
3 October 2003 21 Split ECQT and FCQT

with new, pre-sealed, sterile 120 ml polystyrene bottles, facing the mouths of the
bottles upstream. We immediately placed all samples on ice and returned to the
in-house and commercial laboratories for sample preparation within six hours of
the first sampling (APHA et al., 1998). We generated paired samples by taking two
samples at the same time at each site in separate vessels. For split samples, we took
one sample per site; we then diluted the sample in the laboratory and then split the
sample into separate vessels.


For bacterial enumeration, we used one of three methods namely:

(a) In-house E. coli enumerations (ECQT): These were conducted with the IDEXX
/QT system (ECQT) according to the manufacturer’s instructions. To
avoid saturation of the QT assays (i.e., all wells in the QT are positive, giving
a count of >2419.6 CFU/100 mL); we diluted samples with sterile, deionized
water because of our previous experiences with high bacterial counts. At the
end of the incubation period, we counted all ONPG-MUG positive cells and
read E. coli density from the IDEXX most probable number (MPN) chart pro-
vided. The standard Colilert R
incubation temperature is 35 ◦ C for 24 h, which
we used for Series 2 and 3. For Series 1 however, we used incubation temper-
atures of 44.5 ◦ C for 24 h. Given that E. coli are thermotolerant, the elevated
incubation temperature of 44.5 ◦ C should not affect E. coli enumeration. We
compared E. coli counts with the QT system at 35 and 44.5 ◦ C for 32 samples
and found a small (∼1.5%) reduction in E. coli readings at 44.5 ◦ C. When we
conducted a paired t-test of the log-transformed data, we found this difference
to be insignificant ( p = 0.58).
(b) In-house fecal coliform enumerations (FCQT): The enumerations were con-
ducted according to the protocol described by Yakub et al. (2002) for enu-
meration of fecal (or thermotolerant) coliforms with the IDEXX Colilert R
system. We prepared the samples in exactly the same way as we did for E. coli
enumeration (above), incubating the Quanti-TrayTM 2000 in a water bath at
44.5 ± 0.2 ◦ C for 24 h. We then counted the ONPG-positive cells only.
280 R. W. KLOOT ET AL.

(c) Commercial laboratory fecal coliform estimates (FCMF): The commercial lab-
oratory used standard method SM 9222D (APHA et al., 1998). In short, they
filtered the samples, impregnated the filter media with an enriched lactose
medium, then incubated in a water bath at 44.5 ± 0.2 ◦ C for 24 h before enu-
merating fecal coliform colonies. The laboratory based their filtration volumes
on our estimates of fecal coliform densities. Typical sample volumes based on
the laboratory bench notes were 10 and 25 mL for samples where expected
fecal coliform was <600#/100 mL, and 1 mL and 5 mL for samples expected
to be >600#/100 mL.

Appropriate positive and negative controls and random duplicates confirmed the
reliability of the Colilert
/QT system and the sterility of our laboratory conditions.


For each of the series, we regressed ECQT against FCMF and FCQT, respectively
(all value were log-transformed); we used Pearson’s correlation coefficients as
indicators of the strength of the linear relationships and compared the correlation
coefficients results to previously published results (Yakub et al., 2002; Schumacher,
2003; Noble et al., 2004). For the regressions in each series we sequentially checked
for consistency (i.e., independent of bacterial density) and bias between methods
(Smith and Rose, 1995) first testing for a slope of one (i.e., H0 : β1 = 1) and, given
H0 in the first test is not rejected, by testing for an intercept of zero (i.e., H0 : β1 = 0).
Note that if H0 : β 1 = 1 is rejected, the test for bias (H0 : β0 = 0) is of little value
because the relationship between the two methods is dependent on bacterial density
(Smith and Rose, 1995), therefore the test H0 : β0 = 0 was only conducted if the
test of H0 : β1 = 1 was not rejected.
For each series, we calculated the E. coli/fecal coliform ratio and the upper and
lower 95% confidence limits based on the Wilcoxon Signed Rank test (Hollander
and Wolfe, 1999). Based on other studies, we expected E. coli/fecal coliform ratios
of between 0.5 and 0.9 (Vernberg et al., 1996; Noble et al., 2003; Schumacher,
2003) in our results.
We used both SAS (SAS Institute, Inc. 2000), and MINITABTM (MINITABTM
2000) in our statistical analyses. All hypothesis testing was conducted at α = 0.05
unless otherwise specified.

3. Results and Discussion

As an indication of the relatively high bacterial densities in the surface water in this
watershed, E. coli densities in 97% of the samples exceeded 100 MPN/100 mL and
in 37% of the samples, they exceeded 1,000 MPN/100 mL.

Results from linear regressions for Series 1, 2 and 3, and E. coli/fecal coliform ratios by series

Slope (95% Intercept (95%

Series n r (r 2 ) confidence limits) confidence limits) H0: β1 = 1? H0: β0 = 0?

1 56 0.9 (0.81) 0.73 (0.64; 0.83) 0.54 (0.28; 0.81) Rejected N/A
2 19 0.94 (0.88) 0.86 (0.70; 1.03) 0.3 (−0.17; 0.78) Not rejected Not rejected
3 22 0.98 (0.95) 1.01 (0.91; 1.11) 0.04 (−0.28; 0.37) Not rejected Not rejected

Pearson correlation coefficients (Table II) were 0.9, 0.94 and 0.98 for Series 1,
2 and 3 respectively ( p < 0.0001 for all slopes) and indicated a high degree of
linear correlation in each series between methods and indicators. We attributed the
slightly higher correlation coefficients in Series 2 and 3 to the fact that the samples
were split whereas in Series 1 the samples were paired. The correlation coefficients
compared well with those found by Yakub et al. (2002) and Noble et al. (2004)
who found r-values 0.85, and 0.9, respectively when they compared QT and MF
In Series 1, the regression slope (0.73) is significantly less than one (Table II)
suggesting that the relationship between ECQT and FCMF is in fact dependent on
bacterial density and that the increased difference between ECQT and FCMF with
increased bacterial density (Figure 2) appears to corroborate Brenner and Rankin’s
(1990) observations that MF difficulties (and subsequent under-recovery) became
amplified as bacterial densities increased. The standard membrane filtration method

Figure 2. Plot of Escherichia Coli enumerated by the Colilert/Quanti-TrayTM 2000 system (ECQT)
versus fecal coliforms enumerated by membrane filtration (FCMF) using paired samples (Series 1).
For Figures 1, 2 and 3, data are log-transformed. Apart from the observations the fitted curve and
the upper/lower 95% mean confidence limits are shown. The solid diagonal (1:1) lines in the figures
represent lines with slopes = 1 and intercepts = 0.
282 R. W. KLOOT ET AL.

9222D (APHA et al., 1998) however requires the dilution of samples (also docu-
mented in bench notes) to account for increased bacterial densities to correspond to
the most countable plates, therefore, because of sample dilution, increased bacterial
density ought not to pose a problem in MF.
In Series 2, for the identical comparison (Table II), the tests for consistency
did not reject the null hypotheses (H0 : β1 = 1), suggesting that the ECQT and
FCMF relationship is independent of density. The only difference for Series 2 was
that we first diluted and then split the samples before we delivered the samples
to the laboratories. The dilution rates were 20:1 in Series 2; thus, the bacterial
densities in samples that the third party laboratory received were low (min. 6,
max. 283 CFU/100 ml) in comparison to the undiluted samples we provided them
in Series 1 (min. 80, max. 5,900 CFU/100 ml). We suspected that the third party
laboratory encountered difficulties in preparing the higher density Series 1 (undi-
luted) samples but not the Series 2 (diluted) samples. Our follow-up inquiries with
the third party laboratory personnel and laboratory bench notes did not support
our suspicion. Noble et al. (2003) appear to have had similar issues to ours when
they reported that “our results emphasize the need for strict adherence to quality
control guidelines regarding shaking and sample dilution”. Therefore, while our
doubts were unsupported, we remain skeptical that the third party laboratory did
not encounter some quality assurance problems for Series 1.
The median E. coli/fecal coliform ratio for Series 1 was 1.63 with a lower 95%
confidence limit of 1.39 (Table III) and for Series 2, we observed a reduced median
E. coli/fecal coliform ratio of 1.20 with a lower 95% confidence limit of 1.00
(Table III). Given that E. coli is a subset of fecal coliforms (Leclerc et al., 2001), a
ratio of greater than unity is not possible, therefore these unexpectedly high ratios
caused some concern and resulted in further investigation.
There are previous field studies in marine waters that reported E. coli/fecal co-
liform ratios greater than one. Noble et al. (2003) reported median ECQT/FCMF
ratios of between 1.8 and 11.66 in five separate experiments that compared the
ECQT with FCMF. In the same set of experiments, the ratio of fecal coliform (enu-
merated by MTF)/fecal coliform (enumerated by MF) was between 2.27 and 7.68.
They observed that lower fecal coliform counts were attributable to under-recovery
by MF (rather than over-reporting by MTF or QT) based on reports from third
party laboratories in the study that experienced problems (clumping and filtration

Median E. coli/fecal coliform ratios and 95% confidence limits for Series 1, 2 and 3

Series n Median EC/FC ratio Lower 95% limit Upper 95% limit

1 56 1.63 1.39 1.91

2 19 1.20 1.00 1.55
3 22 0.84 0.73 1.02

difficulties) associated with MF methods (Noble et al., 2003). Mallin et al. (2000)
reported that, specifically for oligohaline salinities, E. coli densities were notice-
ably higher than fecal coliform densities. The fecal coliforms in Mallin et al. (2000)
were enumerated by the standard MF method (APHA et al., 1998) and the E.coli
by the mTEC (MF) method (USEPA, 1995). A possible explanation for the Mallin
et al. (2000) finding is that, under oligohaline conditions, the mTEC method may
outperform the standard fecal coliform MF method in its ability to culture cells.
In each of the above cases, the apparently “impossible” E. coli/fecal coliform ra-
tio is attributable to differences in analytical endpoints between methods used to
enumerate bacteria (Noble et al., 2003).
The greater than expected ratios for ECQT/FCMF in Series 1 and 2 may be
due to false positives that would inflate ECQT results. False positive rates for
ECQT were reported in the range of 5–10% (Yakub et al., 2002; Chao et al., 2004)
for freshwater while Pisciotta et al. (2002) found significantly higher false positive
rates in tropical marine waters (only a 17.1% confirmation rate using fecal coliform
culture from ONPG, MUG – positive cells) but not in freshwater samples. On the
other hand, other references do point to the under-recovery of target bacteria under
non-ideal circumstances (Jacobs et al., 1986; Brenner and Rankin, 1990; McFeters
et al., 1997a,b; Niemela et al., 2003). We were therefore unable to conclude from
these data whether the inflated ECQT/FCMF ratio was caused by false positives
that inflated ECQT results or under-recovery of fecal coliforms in the MF method
reducing the FCMF counts. Investigation into the true cause would require further
investigation. Chao et al. (2004) evaluated the accuracy of a Colilert-18 system by
employing the Analytical Profile Index (API) through API 20 E strips (bioMérieux,
Inc., Hazelwood, Mo.) and fatty acid methyl ester analysis. Similar evaluation
methods on the QT analyses this watershed should shed more light on the high
ECQT/FCMF ratios.
The Series 3 tests for consistency and bias (Table II) did not reject the null
hypotheses, suggesting that the FCQT and ECQT did in fact report similar values
across the range; this is also graphically illustrated in Figure 4 and compared to
Figures 2 and 3. This result is not surprising given the Quanti-TrayTM system was
used for both enumerations. The test for bias, however belies the fact that for 15
out of 21 samples, ECQT densities (incubated at 35 ◦ C) were higher than FCQT
densities (incubated at 44.5 ◦ C). In the Series 3 comparison (October 2004), we
estimated fecal coliforms with the QT system and found a median E. coli/fecal
coliform ratio of 0.84 (Table III). In the Shoal Creek, Missouri study, Schumacher
(2003) compared E. coli and fecal coliforms, both enumerated by MTF, presumably
standard methods 9221 F and 9221 E (APHA et al., 1998). The median E. coli/fecal
coliform ratio calculated in the Shoal Creek watershed was 0.86 which compared
well with the estimated Series 3 median (0.84), also consistent with those found by
Noble et al. (2003) and Vernberg et al. (1996). Given the observations that Vernberg
et al. (1996) made with respect to E. coli, fecal coliforms and urbanization, the
median Series 3 E. coli/fecal coliform ratio should not be considered as “the” ratio
284 R. W. KLOOT ET AL.

Figure 3. Plot of Escherichia Coli enumerated by the Colilert/Quanti-TrayTM 2000 system (ECQT)
versus fecal coliforms enumerated by membrane filtration (FCMF) using split samples (Series 2).

Figure 4. Plot of Escherichia Coli enumerated by the Colilert/Quanti-TrayTM 2000 system (ECQT)
at 35 ◦ C versus ‘fecal coliform’ bacteria enumerated by the Yakub et al. (2002) method (FCQT) using
split samples (Series 3).

for the watershed. The Series 3 ratio does, however appear to be typical of those
found in the literature.
The results from this work suggest that when comparing datasets based on the
traditional MF method with the newer QT method, some caution is required before
making a direct comparison, or even in applying a conversion factor. For example,
if one were to apply a conversion factor that assumes an E. coli/fecal coliform ratio
of 0.63 (USEPA, 2002: Noble et al., 2004) to this dataset, the E. coli densities
would be further inflated. In addition, if E. coli/fecal coliform ratios are dependent

on a host of different factors that may include level of urbanization, source type and
distance from source (Vernberg et al., 1996; Kuntz et al., 2004) a single conversion
factor between old and new datasets for all geographic regions and watersheds may
also be misleading. Given the above findings, if one were attempting to compare
a dataset generated by the MF method with a new dataset generated by the QT
method, a prudent course of action would be to conduct at least some sampling in
a given watershed with side-by-side analyses before assuming a conversion factor
between the two datasets.

4. Conclusions

Some of the results of this work were expected; in all series, the comparison of
log-transformed bacterial densities were linear. In Series 3, where we enumerated
E. coli and fecal coliforms with the same system (QT), the E. coli/fecal coliform
ratio of 0.84 compared well to literature values for this ratio. In Series 1, we ob-
served that the difference between ECQT and FCMF enumerations increased with
increasing bacterial density; this was inconsistent with the results from Series 2 and
we attributed the inconsistency to problems encountered with sample preparation
difficulties associated with the MF method.
Other results were unexpected; in Series 1 and Series 2, when we enumerated
E. coli with QT and fecal coliforms with MF methods, the observed median E.
coli/fecal coliform ratios of 1.63 and 1.20 respectively, presumably caused by the
differing endpoints between the standard MF method and the QT method (Noble
et al. 2003). We were not able to conclude whether the high ECQT/FCMF ratios
were attributable to (1) overestimates of bacteria by QT because of false positives,
(2) under-recovery of bacteria by MF or (3) a combination of both. This uncer-
tainty does warrant further investigation by making use of techniques such as the
Analytical Profile Index (API) through API 20 E strips and fatty acid methyl ester
analysis to assess the accuracy of the QT analyses.
Finally, these results suggest that when comparing historical bacterial data us-
ing the traditional MF method with more modern enzymatic methods, one would
need to exercise care when making direct comparisons or when using conversion


We are grateful to Senator Ernest F. Hollings for his leadership and the South
Carolina Congressional Delegation for their interest and support of this research.
Thanks also to Drs. Dwayne Porter, Geoff Scott, John Shafer and Craig Stowe for
their support, input and suggestions.
286 R. W. KLOOT ET AL.


American Public Health Association, American Water Works Association, Water Environment Fed-
eration, Standard Methods for the Examination of Water and Wastewater (1995) 18th ed., Eaton,
A. D., Clesceri, L. S., and Greenberg, A. E. (eds.), Washington, D.C.
Brenner, K. P. and Rankin, C. C.: 1990, ‘New screening test to determine the acceptability of 0.45-μm
membrane filters for analysis of water’, Appl. Environ. Microb. 56, 54–64.
Budnick, G. E., Howard, R. T. and Mayo, D. R.: 1996, ‘Evaluation of enterolert for enumeration of
enterococci in recreational waters’, Appl. Environ. Microb. 62, 3881.
Cabelli, V. J.: 1983, ‘Water-borne Viral Infections’, in: M. Butler, A. R. Medlen and R. Morris (eds),
Viruses and Disinfection of Water and Wastewater, Surrey Press, Guilford, England, pp. 107–
Chao, K., Chao, C. and Chao,W.: 2004, ‘Evaluation of Colilert 18 for detection of coliforms and
Escherichia Coli in Subtropical Freshwater’, Appl. Environ. Microb. 70, 1242–1244.
De Roubin, Osswald, M. L., Bally, D., Cailas, M., and Joret, J. C.: 1997, ‘Comparison of field and
standardized techniques for the enumeration of total coliforms and Escherichia coli in water’,
in: Proceedings of AWWA-Water Quality Technology Conference, November 7–11, 1997, Denver,
Eckner, K. F.: 1998, ‘Comparison of membrane filtration and multiple-tube fermentation by the
Colilert and Enterolert methods for detection of waterborne coliform bacteria, Escherichia coli,
and enterococci used in drinking and bathing water quality monitoring in southern Sweden’, Appl.
Environ. Microb. 64, 3079–3083.
Francy, D. S. and Darner, R. A.: 2000, ‘Comparison of methods for determining Escherichia coli
concentrations in recreational waters’, Water Res. 34, 2770–2778.
Fricker, E. J., Illingworth, K. S. and Fricker, C. R.: 1997, ‘Use of two formulations of Colilert and
Quantitray for assessment of the bacteriological quality of water’, Water Res. 31, 2495–2499.
Geissler, K., Manafi, M., Amoros, I. and Alonso, J. L.: 2000, ‘Quantitative determination of col-
iforms and Escherichia coli in marine waters with chromogenic and fluorogenic media’, J. Appl.
Microbiol. 88, 280–285.
Jacobs, N. J., Ziegler, W. L., Reed, F. C., Stuckel, T. A. and Rice, E. W.: 1986, ‘Comparison of
membrane filter, multiple-fermentation-tube, and presence-absence techniques for detecting total
coliforms in small community water systems’, Appl. Environ. Microb. 51, 1007–1012.
Leclerc, H., Mossel, D. A., Edberg, S. C. and Struijk, C. B.: 2001, ‘Advances in the bacteriology of
the coliform group: their suitability as markers of microbial water safety’, Annu. Rev. Microbiol.
55, 201–234.
Mallin, M., Williams, K. E., Cartier Esham, E. and Lowe, R. P.: 2000, ‘Effect of human development
on bacteriological water quality in coastal watersheds’, Ecol. Appl. 10, 1047–1056.
McFeters, G. A., Pickett, M., Broadaway, S. C. and Pyle, B. H.: 1997a, ‘Impact of chlorine injury
on reaction kinetics of coliforms and E. coli in Colisure and LTB’, Water Sci. Technol. 35, 419–
McFeters, G. A., Broadaway, S. C., Pyle, Pickett, B. H. and Egozy, Y.: 1997b, ‘Comparative perfor-
mance of Colisure’, J. Amer. Wastewater Assn. 89, 112– 120.
Miller A.: 2001, ‘Total maximum daily load development for Bush River: Stations S-046 and S-102 Fe-
cal Coliform Bacteria’, Technical Report, Bureau of Water, South Carolina Department of Health
and Environmental Control, Columbia, SC.
(accessed Sept 25, 2004).
MinitabTM Statistical Software. Release 13 for Windows R
, MinitabTM , State College, PA. 2000.
Neimi, R. M., Heikella, M. P., Lahti, K., Kalso, S. and Niemela, S. I.: 2001, ‘Comparison of methods
for determining the numbers and species distribution of coliform bacteria in well water samples’,
J. Appl. Microbiol. 90, 850–858.

Niemela, S. I., Lee, J. V. and Fricker, C. R.: 2003, ‘A comparison of the International Standards
Organisation reference method for the detection of coliforms and Escherichia coli in water with
a defined substrate procedure’, J. Appl. Microbiol. 95, 1285–1292.
Noble, R. T., Leecaster, M. K., McGee, C. D., Weisberg, S. B. and Ritter, K.: 2004, ‘Comparison
of bacterial indicator analysis methods in stormwater-affected coastal waters’, Water Res. 38,
Noble, R. T., Weisberg, S. B., Leecaster, M. K., McGee, C. D., Dorsey, J. H., Vainik, P. M. and
Orozco-Borbion, V.: 2003, ‘Storm effects on regional beach water quality along the southern
California shoreline’, J. Water Health, 1, 23–31.
Østensvik, Ø.: 2000, ‘Coliform bacteria and Escherichia coli in Norwegian drinking water sources
– Comparison of methods based on the fermentation of lactose and on the activity of specific
enzymes’, in: Proc. AWWA Water Quality Technology Conference, November, 5–8, 2000, Salt
Lake City, Utah.
Pisciotta, J. M., Rath, Stanek, P. A., Flanery, D. M. and Harwood, V. J., 2002, ‘Marine bacteria cause
false-positive results in the colilert-18 rapid identification test for Escherichia coli’, Appl. Environ.
Microb. 68, 539–544.
Rompré, A., Servais, P., Baudart, J., de-Roubin, M. and Laurent, P.: 2002, ‘Detection and enumeration
of coliforms in drinking water: current methods and emerging approaches’, J Microbiol. Meth.
49, 31–54.
SAS Institute Inc.: 2000, SAS OnlineDoc, Version 8, Cary, NC.
Schumacher, J. G.: 2003, ‘Survival, Transport, and Sources of Fecal Bacteria in Streams and Survival
in Land-Applied Poultry Litter in the Upper Shoal Creek Basin, Southwestern Missouri, 2001–
2002’, Technical Report, USGS Water-Resources Investigations Report 03–4243, Rolla MO.
Available at:, October 26, 2004.
Shibata, T. Solo-Gabriele, H. M., Fleming, L. E. and Elmir, S.: 2004, ‘Monitoring marine recreational
water quality using multiple microbial indicators in an urban tropical environment’, Water Res.
38, 3119–3131.
Smith, E. P. and Rose, K. A.: 1995, ‘Model goodness-of fit analysis using regression and related
techniques’, Ecol. Model. 77, 49–64.
South Carolina Department of Health and Environmental Control: 2001, ‘Water Classifications
and Standards (R.61–68)’, Columbia, SC. Available at:
reg.html, October 29, 2004.
United States Environmental Protection Agency (USEPA): 1986, ‘Ambient Water Quality Criteria for
Bacteria–1986’, U.S. Environmental Protection Agency. EPA-440/5-84-002., Washington DC:
Office of Water.
United States Environmental Protection Agency (USEPA): 2002, ‘Draft Implementation Guidance
for Ambient Water Quality Criteria for Bacteria’, U.S. Environmental Protection Agency. EPA-
823-B-02-003, Washington DC: Office of Water.
Vernberg, W. B., Scott, G. I., Strozier, S. H., Bemiss, J., Daugomah, J. W.: 1996, ‘The effects of
urbanization on human and ecosystem health’, in: F. J. Vernberg, W. B. Vernberg, T. Siewicki
(eds.), Sustainable Development in the Southeastern Coastal Zone, University of South Carolina
Press, Columbia, SC, pp. 221–239.
Yakub, G. P., Castric, D. A., Stadterman-Knauer, D. L., Tobin, M. J., Blazina, M., Heineman, T. N., Lee,
G. Y. and Frazier, L.: 2002, ‘Evaluation of Colilert and Enterolert Defined Substrate Methodology
for Wastewater Applications’, Water Environ. Res. 74, 131–135.