EPA/600/R-05/107
September 2005
Temporal and Spatial Variability of Fecal Indicator Bacteria:
Implications for the Application of MST Methodologies to
Differentiate Sources of Fecal Contamination
by
Marirosa Molina
Ecosystems Research Division
National Exposure Research Laboratory
Athens, GA 30605
U.S. Environmental Protection Agency
Office of Research and Development
Washington, DC 20460
-------
Notice
The information in this document has been funded by the United States
Environmental Protection Agency. It has been subject to the Agency's peer and
administrative review, and it has been approved for publication as an EPA
document. Mention of trade names of commercial products does not constitute
endorsement or recommendation for use.
-------
Abstract
Temporal variability in the gastrointestinal flora of animals impacting water
resources with fecal material can be one of the factors producing low source
identification rates when applying microbial source tracking (MST) methods.
Understanding how bacterial species and genotypes vary over time is highly
relevant when the fecal material used to create a source library is collected under
very different seasonal conditions than the environmental sample. Our objective
was to identify and compare the temporal and spatial variability of fecal indicator
bacteria from a specific host in manure and water samples and evaluate the
implications of such variability on microbial source tracking approaches and
applications. We selected Enterococcus as the model fecal indicator, given the
supposedly high specificity of some of the species of this genus to the host
organism. Cattle was chosen as the model host organism because of the
documented high impact that cattle has on impairment of surface waters. The
sites studied were located at a farm where cattle have unrestricted access to the
stream. Enterococci were isolated monthly from water and manure samples
using membrane-Enterococo/s lndoxyl-(3-D-Glucoside agar (mEI) as described
in EPA method 1600. The isolates were identified using a multiplex PCR
procedure that targets the genus and the species-specific gene superoxide
dismutase. Eight species were identified in cattle manure, of which E.
casseliflavus (37%), faecium (22%) and hirae f18%J were the most abundant.
Nine species were identified in stream samples with E. faecalis (43%),
casseliflavus/flavescens (34%), and hirae ("11 %) being the most abundant.
September exhibited the highest species abundance in manure samples while
March had the highest species abundance in stream water samples. E. assini
and E. malodoratus were only detected in manure samples, but were not
detected in water samples. In contrast, E. durans, gallinarum and sulfureous
were only isolated from the stream samples. In general, the enterococci
distribution pattern and species richness found in manure samples did not
correlate with those found in the stream samples at the individual species level.
However, cluster analysis revealed strong seasonal and spatial variability of
groups of enterococci, and indicated that some clusters that seem specific to
manure can be found in the water only during certain seasons. In addition to the
enterococci library development, 16S rDNA host-specific Bacteroides markers
were also applied to the water samples. The results indicate that data obtained
with the Bacteroides markers (BM) generally agreed with the enterococci data
showing higher occurrence of the cattle BM in areas under obvious cattle impact.
However, no seasonality was identified in conjunction with any of the BMs used.
In addition, the cow marker was also detected at an upstream-of-the-farm
location that was not under obvious cattle influence. This study suggests that in
order to increase the validity of MST methods, it is necessary to consider
temporal variability when designing the sampling scheme of the source material
and constructing source libraries, and increase the specificity and field testing of
DMA-based markers.
-------
Table of Contents
TEMPORAL AND GEOGRAPHIC STABILITY OF FECAL INDICATOR
BACTERIA 6
TEMPORAL STABILITY 6
GEOGRAPHIC STABILITY 8
STUDY SITE AND SAMPLE DESCRIPTION 10
METHODOLOGY 13
PROCEDURE FOR VERIFICATION OF ENTEROCOCCI SPECIES 14
MULTIPLEX PCR PROCEDURE 15
DMA EXTRACTION AND AMPLIFICATION WITH BACTEROIDES PRIMERS 17
DNA Extraction from Fecal and Water Samples 17
Amplification using Bacteroides Primers 17
STATISTICAL ANALYSIS 18
RESULTS AND DISCUSSION 19
TOTAL ENTEROCOCCI COUNTS IN STREAM WATER SAMPLES AND COMPARISON OF FLUORESCENT
ASSAY AND MEMBRANE FILTRATION PROCEDURE 19
COMPOSITION AND TEMPORAL VARIABILITY OF ENTEROCOCCUS SPECIES IN MANURE AND WATER 23
SEASONAL AND SPATIAL VARIABILITY OF ENTEROCOCCI COMMUNITIES IN MANURE AND WATER ... 27
COMPARISON OF BACTEROIDES MARKERS AND ENTEROCOCCI CLUSTERS 33
CONCLUSIONS AND FINAL CONSIDERATIONS 36
REFERENCES 39
-------
List of Tables
Table 1: Description of stream water sampling locations at Chandler Farm,
Georgia 13
Table 2: Enterococcus species isolated from cattle manure and stream water at
a farm in Georgia 25
Table 3: Seasonal % composition (mean ± sd) of the most common
Enterococcus species isolated from cattle manure samples collected at a
cattle farm in Georgia 26
Table 4: Seasonal % composition of the most common Enterococcus species
isolated from water samples collected in a stream located at a beef cattle
farm in Georgia 27
Table 5: Composition and abundance (%) of Enterococcus species in clusters
that appeared 3 or more times in water and manure samples collected at a
Georgia cattle farm from September 2003 through January 2005 28
Table 6: Composition of Bacteroides clusters identified in stream water collected
at a cattle farm in Georgia 35
List of Figures
Figure 1: Study site location 11
Figure 2: Sampling locations at Chandler Farm, Georgia 12
Figure 3: Procedure diagram for counting, isolating, verifying and speciating
enterococci in environmental samples 14
Figure 4: Seasonal enterococci counts in water samples collected at a Georgia
cattle farm using the mEI membrane filtration procedure 20
Figure 5: Percent of isolates with a blue halo isolated from mEI that tested
positive for the genus Enterococcus with a multiplex PCR procedure 22
Figure 6: High % occurrence of Enterococcus clusters (EC) in samples
collected from different sources at a cattle farm in Georgia 29
Figure 7: Low % occurrence of Enterococcus clusters in samples collected from
different sources in a cattle farm 30
Figure 8: Low % occurrence of enterococci clusters during different seasons in
samples collected at a cattle farm in Georgia 31
Figure 9: High % occurrence of enterococci clusters during different seasons in
samples collected at a cattle farm in Georgia 32
Figure 10: Cluster distribution and occurrence (%) per Chandler Farm sampling
site and source 33
Figure 11: Relationship of Bacteroides and enterococci clusters in stream water
samples collected at a cattle farm in Georgia 36
-------
Temporal and geographic stability of fecal indicator bacteria
An ideal source tracking organism must be stable in the environment.
Sampling performed overtime should not reveal significant genotypic or
phenotypic variability within host individuals or within host populations. In
addition, variable environmental conditions such as temperature and pH, and
host factors such as quality and type of feed or antibiotic treatments, etc., should
not affect an ideal indicator. All these conditions affect the host organism and, in
turn, the inside environment that the source indicator bacteria inhabit. To date,
very few microbial source tracking (MST) studies have addressed the temporal
stability of fecal indicators, making it difficult to reliably identify sources over time.
In this study, we sampled manure and impacted manure water monthly at a
Georgia farm site over a year to determine: the temporal variability of various
species of Enterococcus] the spatial distribution and stability of enterococci
species in stream water; and which species might be the most relevant and
promising specific indicators of the host organism, i.e., beef cattle.
Temporal stability
When addressing temporal variability, it seems important to establish the
difference between transient and resident populations of source indicators. This
is of particular relevance if, for example, the range of clones estimated in natural
populations of Escherichia coli (100-1000 per host species) (Selander et al.
1987) are found to be comparable for other fecal indicator bacteria. Caugant et
al. (1981) defined a transient population as one observed at only one sampling
time, while a resident population is observed at more than one sampling time. In
order for MST methods to be effective, the source indicator bacteria selected
-------
should be part of the resident population of the source species. In addition, it
should be part of a clonal population that is stable through time as suggested by
Gordon (2001) for E.coli. In a study performed over an 11 -month period on a
single human host, Caugant et al. (1981) found a significant difference between
the resident and the transient populations of E. coli. The resident population
accounted for only 5.6% of the 53 electrophoretic types identified using
multilocus enzyme electrophoresis. Jenkins et al (2003) also found a rather low
percentage of E. coli ribotypes to be part of the resident population in yearling
steers sampled four times over a 129-day period. Specifically, only 8.3% of 240
ribotypes were determined to be resident in the host species. In addition, no
ribotype was found at all four sampling times or in all of the steers sampled from
a total of 20 resident ribotypes. Also using E coli, Ochman et al. (1983)
observed that the resident population from multiple hosts accounted for only 8%
of all the electrophoretic types identified, and only 5 types were found in more
than 7 hosts. These results suggest that there is a high probability that the
majority of ribotypes obtained from a single host species at any given time
belong to transient populations. This observation has major repercussions
relative to the establishment of host origin libraries, that could require continuous
updating in order for a particular MST methodology to be able to track the host
species (Jenkins et al. 2003) over an extended period of time.
It should be noted that although a general lack of temporal stability seems
to be a big limitation in the identification of suitable source indicators, there are
certain source genotypes that have been recovered from environmental samples
after extended periods of time. These periods range from a few weeks to a year
-------
(Faith et al. 1996; Jenkins et al.2003; Wiggins et al. 2003). Restriction
endonuclease digestion profiles (REDP) performed in dairy cows from 70 farms
in Wisconsin revealed that two isolates exhibited the same REDP even though
they were sampled 7 months apart. Results from the same study also indicated
that a herd or animal can contain isolates of E. coli 0157:H7 that have multiple,
but similar profiles; however, most of these profiles were found to change over
time (Faith et al. 1996). Long-term temporal stability has also been observed for
some indicator organisms using phenotypic tests such as antibiotic resistance
patterns (ARA) (Wiggins et al. 2003).
Geographic stability.
Three main assumptions can be made when investigating the
geographical stability of an ideal source indicator. These are that: a) a bacterial
source indicator exhibits "geographical structure", that is, the similarity of the
bacterial indicator in various populations of a given host animal species is directly
proportional to the geographical distance of the members of such population; b) a
bacterial source indicator sampled from one population of a given host animal
species will be similar to a bacterial source indicator sampled from any other
population of the same host animal species, and a predictive relationship can be
established between the two; and c) a bacterial host indicator sampled from
various populations of a given animal host species separated by great
geographic distances exhibits a high similarity index and accurately tracks the
host species.
Studies indicate that the first assumption regarding "geographical
structure" for populations of the same host animal species is hard to verify for
8
-------
human hosts. This is attributed to the mobility of humans among geographic
areas. On the other hand, isolates from non-domesticated animals seem to
exhibit more "geographic structure" due to their restricted movement patterns
(Gordon 2001). Caugant et al. (1984) reported that "geographic structure" was
hard to demonstrate for E. coli in families living within the same city, where only
6% of the diversity was explained by the geographical separation, and 1 % of the
diversity was explained by the distance separating families living in different
cities. Another possible factor affecting "geographic stability or structure" is that
the host animal digestive system can select for particular resident bacterial
strains, generating a very specific gut flora in each host (Souza et al. 2002),
making it difficult to identify genotypes and/or phenotypes over broad geographic
areas.
An important consideration when trying to assess spatial stability is the
analysis methodology used. In a study performed across a broad geographical
area in Florida, researchers used a one-enzyme ribotyping procedure to
determine the accuracy of this MST methodology to identify beef and dairy cattle,
poultry, swine and human host species using E. coli isolates (Scott et al. 2003).
Although the methodology was accurate differentiating human vs. non-human
hosts, it failed to distinguish among the different non-human host species across
the broad geographical region. In contrast, Hartel et al. (2002) were able to
successfully apply a two-enzyme ribotyping methodology to discriminate among
E coli ribotypes isolated from cattle and horses from two locations (Georgia and
Idaho). The results from this study support the first assumption for geographic
stability, but do not support the second and third assumptions. The latter
-------
researchers indicated that up to a distance of 260 km, there is ribotype sharing
among isolates obtained from horses. Cattle exhibited some ribotype sharing up
to a distance of 350 km, but not at a distance of 2900 km (Georgia and Idaho
isolates). However, for swine and poultry, the percent sharing was rather low
and not significantly different from locations closer together (locations within
Georgia) than far apart (Georgia and Idaho). Using a similar ribotyping method,
human vs. non-human hosts were also accurately identified from E. coli isolates
across an extended area in the Apalachicola region of Florida (Parveen et al.
1999).
For library-based methods, the size of the library seems to be a
determinative factor supporting the second and third assumptions presented in
this section. In a study using ARA, results indicated that merging 6 watershed
libraries to encompass a total of 6,500 isolates produced a library large enough
to be representative and capable of being used to accurately identify enterococci
host species across a broad geographic area (Wiggins et al 2003). These latter
study results also suggested that the minimum size of a library should be about
2,300 isolates in order for it to be representative, in this way, it is possible to
create multiwatershed databases representative enough for the reliable
identification of fecal bacterial sources.
Study Site and Sample Description
Water and manure samples were collected at Chandler Farm (CF), a beef
cattle farm located in Madison County, Northeast Georgia (Figure 1).
10
-------
Savannah River Basin
South Fork Broad River Watershed
CRAWFORD
Chandler Farm Sample Sites
« USEPA Facilities
10 Miles -- Seagraves Farm Sample Sites
o Towns
/\/ Principle Area Roads
South Fork Broad River
Streams SFBR Watershed
SFBR Watershed Boundary
Water Bodies
Figure 1: Study site location
Water samples were collected from a first order stream that crosses the farm
from west to east and is a tributary of the South Fork Broad River (Figure 2).
Seven sampling locations were located along the stream within the Chandler
farm site (Table 1) covering a distance of 2.3 km. One liter water samples were
collected at each location once per month from September 2003 through January
2005. During each sampling campaign, five fresh cattle manure samples were
also collected from different individuals after collecting the water samples.
11
-------
CHANDLER FARM SAMPLE SITES
South Fork Broad River Watershed Georgia
Microbial Indicators of Land Use
Site
Name
CFS1
CFS2
CFS3
CFS4
CFS5
CFS6
CFS7
Stream
Name
Un m d
Un m d
Un m d
Un m d
Un m d
Un m d
Un m d
Miles
Trom
Qngen
0.13
0.55
0.58
0.65
0.82
0.90
0.92
Miles
to
He t
Site
042
0.03
0.07
0.17
0.08
0.02
0.00
Mile?
to
Mouth
1.50
1.09
1.05
0.99
0.81
0.74
0.72
0 2 Miles
A Chandler Farm Sites
Madison County Streams
Figure 2: Sampling locations at Chandler Farm, Georgia.
12
-------
Table 1: Description of stream water sampling locations at Chandler Farm, Georgia.
Site ID
Description of Site
Direct Cattle
Impact on Site
Distance from
Origin of Creek
(km)
CFS-1 Creek headwaters, None
upstream from cattle impact
CFS-2 Stream at cattle crossing High
area
CFS-3 Unrestricted access of cattle High
to creek
0.21
0.89
0.93
CFS-4 Intermittent unrestricted
access of cattle to creek
Medium
1.05
CFS-5 Stream by side of pond,
cattle was never observed
in this location
Low
1.32
CFS-6 Stream at outlet of pond,
cattle was never observed
at this location
Low
1.45
CFS-7 Stream outside of property
fence, no direct access by
cattle
Low
1.48
Methodology
After preparing slurries of the manure samples, both the manure and
stream water samples were processed by membrane filtration to obtain the total
number of enterococci in the water and to isolate enterococci species for library
development. The specific procedure is depicted in Figure 3.
13
-------
Sampling
and
Membrane
Filtration
Filter water samples
Incubate mEI plates
for 24 hours
41 °C
polycarbonate filters
If positive in salt
broth, proceed to
speciate
Colonies w/blue halo
= enterococci
Inoculate BHI Slant
with individual colony
incubate for 24
hours S> 35°C
Bacterades primers
Isolate colonies in BHI
plates, incubate for 24
hours @ 37°C
Inoculate BHI broth
incubate for 24 hours
positive result = enterococci
proceed w/ verification step
from BHI broth or slant
10-20%
Verification
Procedure
Inoculate BHI
broth+ 6.5%
salt, incubate for
48 hours @ 35°C
neg result =
discard isolate
ID species using
Multiplex PCR
so ate co onies in BH
plates, incubate for 24
Inoculate BEA
agar, incubate for
48 hours @ 35°C
Freeze 2-3 colonies in BHI
broth + 30% glycerol
Figure 3: Procedure diagram for counting, isolating, verifying and speciating enterococci
in environmental samples.
Procedure for Verification of Enterococci Species
A modification of EPA Method 1600 was used to count, isolate and verify
enterococci from the environmental samples. Briefly, 1, 5, 10, and 50 ml of
stream water and 10 and 25 ml of a 1 x10"6 dilution of manure slurry were filtered
through 0.45um cellulose membranes and incubated on membrane-
Enterococcus Indoxyl -(3-D-Glucoside (mEI) agar plates at 41 ± 0.5°C for 24
hours. After incubation, all colonies having a blue halo were considered to be
presumptive enterococci. Five colonies from each location water sample and ten
colonies from each manure sample were isolated on brain-heart infusion agar
14
-------
(BHIA) slants and in a tube of brain-heart infusion broth (BHIB), using a 1ul loop.
The BHIA and BHIB samples were then incubated at 35 ± 0.5°C for 48 and 24
hrs, respectively. After incubation, a loop-full (1ul) from each BHIB tube
exhibiting growth was transferred to a tube of BHIB plus 6.5% NaCI, and
incubated at 35 ± 0.5°C for 48 hrs. Any isolate not exhibiting growth on BHIA,
BHIB or BHIB + NaCI was considered to be non-enterococci, and was not used
any further in the procedure.
About 20% of the isolates exhibiting growth on the media mentioned
aboved were further verified as Enterococcus using the following procedure: a
1 ul loop of sample was taken from a BHIA slant and transferred to a Bile-
Esculine Agar (BEA) slant, and a tube of BHIB. The BEA slant was incubated at
35 ± 0.5°C for 48 hrs and the BHIB tube was incubated at 45 ± 0.5°C for 48 hrs.
Finally, a Gram stain was performed on the final isolates. Growth in each
medium combined with and identification of the final isolate as Gram positive
cocci verified the isolate as an Enterococcus.
Multiplex PCR procedure
All polymerase chain reactions (PCR) were conducted within currently
established EPA Quality Assurance/ Quality Control guidelines. The workflow
was conducted such that the opportunity for sample contamination was reduced
as much as possible. It was imperative that reagent preparation, sample
preparation, DMA extraction and PCRs followed a one-directional flow in
separate areas with separate pipettes and equipment to prevent cross-
contamination. All reagents were prepared in working volumes in a positive
pressure room on a clean bench after exposing the bleach-disinfected area to UV
15
-------
light for 10 minutes. DMA extractions were conducted in a separate laboratory.
PCRs were run in a third location, physically separated from the reagent and
DMA prep rooms. Pipettes and lab coats were dedicated to the different steps of
the procedure. Environmental samples were processed following standard
microbiological aseptic techniques. Positive controls for the PCR were used in
each run to insure that the PCR was not inhibited by contaminants. Negative
controls (reagent blanks) were used to insure that amplified DMAs were only
coming from the environmental samples, and not introduced to the samples at
the laboratory. PCR optimization for the Bacteroides work was performed at the
beginning of the study.
Speciation of enterococci isolated from manure and stream water samples
was performed as depicted in the lower left side of Figure 3. After verifying the
isolates as Enterococcus, whole cell templates were prepared in molecular grade
sterilized water. These templates were used for up to three weeks. Seven
master mixes were used to identify up to 23 species of Enterococcus using a
multiplex PCR procedure based on the superoxide dismutase gene (Jackson et
al., 2004). The procedure was performed testing the isolates with the master
mixes in the following order: 1, 2, 6, 4, 3, 5, and 7. The majority of the isolates
could be speciated by applying only the first three master mixes in the sequence,
thereby achieving the best use of resources and the most time and cost effective
approach. PCR products were separated and identified using a 2% 1X TAE
agarose gel containing 2 ug/ ml ethidium bromide. Gel analysis was performed
using a EpiChemi Darkroom Biolmaging System (UVP, Inc.) equipped with a
transilluminator, and fitted with Labworks 4.5 software. Band sizes were
16
-------
identified by comparing the sample DMA to the positive controls included with
each run, and by comparing the band size to a 100 bp DMA ladder. Once the
isolates were identified, the templates were plated again on BHIA and 3 to 4
single colonies were inoculated into BHIB containing 30% glycerol. The
inoculated medium was stored at -80°C.
DMA Extraction and Amplification with Bacteroides Primers
DMA Extraction from Fecal and Water Samples. Manure fecal samples
were stored at -20°C immediately upon arrival at the lab until DMA could be
extracted. DMA was extracted with a MoBio UltraClean® fecal DMA mini kit
using 0.25 gram of fecal material according to the manufacturer's instructions.
Water samples (100 ml aliquots) were filtered through 0.4 urn cellulose filters and
DMA was extracted from the membrane filters using a Qiagen DNeasy® tissue
kit, following the Qiagen protocol for DMA extraction with a micro-centrifuge and
an additional wash of Buffer AW2 (included in the kit).
Amplification using Bacteroides Primers. One general, two cow-
specific and two-human specific-Bactero/ctes primers (Bernhard and Field, 2000)
were used to test water samples. After applying PCR optimization procedures,
the following program was used: initial denaturation at 94°C for 2 minutes,
product amplification by 30 cycles of denaturation at 94°C for 1 minute, annealing
at 53/54°C for 1 minute, and elongation at 72°C for 1.5 minute. Amplification was
followed by a final extension at 72°C for 3 minutes. Bacteroides- PCR products
were identified (presumptive positive result) in a 1% agarose gel containing
ethidium bromide by comparing the band intensities under UV light to the
intensities of a commercially available 100 bp DMA mass ladder.
17
-------
Statistical Analysis
A hierarchical cluster analysis was performed (using Minitab v.12
statistical software) on the165 samples taken from cow manure patties,
upstream-of-the-farm stream water, and on-the-farm stream water. The objective
of this analysis was to group together samples that showed similar relative
abundances of the most common species of Enterococcus. Enterococci species
that were found in only a few of the 165 samples were not included in the
analysis. The following five species were seen frequently enough to be included:
E. casseliflavus, E. faecalis, E. faecium, E. flavescence, E. hirae. In addition, we
included one category that was the sum of all unidentified enterococci species.
In the first step of the clustering algorithm, the two samples with the most
similar Enterococcus species relative abundances are grouped together. These
two observations are now designated as a cluster, and this cluster is represented
by a centroid, or mean value, of the two samples that compose it. In step two, all
remaining samples are examined and the next two that have the most similar
relative species abundance are grouped or clustered. In each subsequent step,
the two samples (or possibly clusters) that exhibit the greatest similarity are
grouped together. Hierarchical clustering requires that a subjective stopping-
point be chosen as the algorithm progresses. If this is not done, the algorithm
will eventually form one large group of all observations. We stopped the
procedure after step 121, prior to the formation of two large clusters. At this
point, 15 clusters had been formed with member species (could be the same or
different species) appearing 3 or more times in each one of the 165 samples; 134
of the 165 total samples were found within these 15 clusters. The other 31
18
-------
samples (19% of the total sample pool) were identified as "outliers", meaning that
their enterococci communities did not match well with communities seen in the
other samples.
After stopping the algorithm, we recorded the centroid of each cluster (i.e.,
the mean relative abundances for the five enterococci species used in the
analysis) plus the general enterococci classification. The centroids for the 15
clusters with member species that appear 3 or more times are given in Table 5.
In our final step, we used the cluster designations for each of the samples to
perform ANOVA and MANOVA analyses to test for differences in the clusters
found for manure, upstream-of-the-farm stream water, and on-the-farm stream
water, as well as changes in the seasonal occurrence of the clusters.
Results and Discussion
Total Enterococci Counts in Stream Water Samples and Comparison
of Fluorescent Assay and Membrane Filtration Procedures. As indicated
previously, total enterococci counts were performed using EPA method 1600.
Accordingly, sample volumes of 1, 5, 10, and 50 ml were used to target total
counts into the suggested range of 20 to 66 colonies/100 ml. A defined-substrate
assay method from IDEXX laboratories that applies a methyl-umbelliferyl-(3-
glucuronide (MUG)-based medium (Enterolert®) for detection of enterococci was
also used on the water samples. The objective of the test was to perform a
comparison of the results obtained from both methodologies and to evaluate their
accuracy. Figure 4 shows the total enterococci counts obtained by the
membrane filtration procedure (mEI) for the samples collected at the seven water
19
-------
sampling locations. The results indicate that the highest counts were always
obtained at locations CFS-2 thru -4, locations directly impacted by cattle,
including a cattle crossing point (CFS-2). Locations CFS-5 thru -7 exhibited one-
to two-fold less total counts than the upstream locations (with summer values
being slightly higher than those in other seasons), indicating a decrease (due to
dilution, settling, dye-off, etc.), of the fecal bacteria in the water column.
Although the highest counts were observed mostly from April through September
that tend to be months of low precipitation, monthly variability of the counts didn't
allow us to establish significant differences.
8000
CFS-1 CFS-2 CFS-3 CFS-4 CFS-5 CFS-6 CFS-7
Chandler Farm Stream Location
Figure 4: Seasonal enterococci counts in water samples collected at a Georgia cattle farm
using the mEI membrane filtration procedure.
The accuracy of the mEI procedure was determined by comparing the
number of isolates originally obtained from the mEI plates per site with the
number of isolates identified as Enterococcus with the multiplex PCR. In mEI,
20
-------
colonies of any color that produce a blue halo are presumptive enterococci. Our
results indicate that in water, an average of 99% of all isolates that produced a
blue halo in the mEI, also tested positive in the salt broth (which is one of the
biochemical tests run to verify the presence of enterococci). However, this
agreement was down to 21% for the manure samples. Those isolates that tested
negative in the salt broth were discarded, as a test of them with the multiplex
PCR demonstrated that they were not of the genus Enterococcus. Of the
isolates that tested positive in the salt broth, 99 and 97.5% were identified as
enterococci in the water and manure samples, respectively, indicating that salt
tolerance was indeed a good indicator for the presence of enterococci in this type
of stream water and manure samples. In contrast, the presence of a blue halo in
the mEI was not a good indicator of the presence of enterococci in manure
samples, since only 43.8% of those isolates with a blue halo were identified as
enterococci by the multiplex PCR. For water, however, 97.6% of all those
isolates exhibiting a blue halo were identified as enterococci by the PCR
procedure (Figure 5).
21
-------
140.0
x 120.0
o
2 100.0
3
g 80.0
^o 60.0
o^
u3 40.0
20.0
0.0
n Manure
n Water
Mar- Apr May Jun Jul Aug Sep Oct Nov Dec Jan-
Month
Figure 5: Percent of isolates with a blue halo isolated from mEI that tested positive for the
genus Enterococcus with a multiplex PCR procedure.
The higher than 100% accuracy indicated for some of the samples in Figure 5 is
due to the presence of additional enterococci colonies mixed with isolates
believed to be only one type of colony when they were originally isolated from the
mEI. The mixed colonies were separated into pure cultures and identified as
Enterococcus by the multiplex PCR.
We compared the precision of the Enterolert® method relative to the mEI
for determination of the total number of enterococci in water samples by
calculating the relative percent difference (RPD) of the total counts per 100 ml
obtained with each method. The RPD was calculated using the following
equation:
RPD= (mEI counts-Enterolert counts/(mEI counts + Enterolert counts)/2) * 100
22
-------
The calculated results indicated that there is an average of 32 ±13.7%
underestimation in the total Enterococcus counts obtained with the Enterolert
procedure relative to the mEI method for the type of water samples used in this
study (data not shown). In addition, no correlation was observed between the
two methodologies. Kinzelman et al. (2003) also reported a lack of correlation
between these two methods; however, they found that the Enterolert® procedure
generated false positive results that produced an overestimation of the actual
number of enterococci, contrary to the underestimation found in our study.
Therefore, it is possible that the performance of the Enterolert® procedure is
highly dependent on the physical/chemical conditions of the environment tested,
and probably more studies are necessary to determine its general efficacy in
freshwater systems.
Composition and Temporal Variability of Enterococcus Species in
Manure and Water. A total of 11 species of enterococci were identified in
water and manure samples collected during our study using a multiplex PCR
procedure (Jackson et al., 2004). E. malodoratus and asini were only found in
manure while E. sulfureus, gallinarum and durans were only found in the water
samples (Table 2). Because E. malodoratus, asini and gallinarum were only
found once during the whole sampling period, they were categorized as transient
species in the system (Caugant et al., 1981). E. durans and sulfureus were
found in several occasions during different seasons (data not shown) and were
23
-------
not believed to be transient species, but were not observed frequently enough as
to be considered important members of the enterococci community.
24
-------
Table 2: Enterococcus species isolated from cattle manure and stream water at a farm in
Georgia.
Enterococcus species
Manure Water
E. casseliflavus E. casseliflavus
E. faecalis E. faecalis
E. faecium E. faecium
E. flavescence E. flavescence
E. hirae E. hirae
E. mundii E. mundtii
E. malodoratus E. sulfureous
E. assini E. gallinarum
E. durans
The % abundances of the most common enterococci species found in
manure and water samples are presented in Tables 3 and 4, respectively. We
found greater variability in the seasonal abundances of individual species in
manure than in water. In manure, E. faecalis was the most, while E
casseliflavus and E flavescens were the least abundant abundant during spring
(Table 3). During summer, E casseliflavus , E. faecium and E flavescens were
all in high abundance (Table 3). It is clear that the relative % abundance of the
individual species in manure varies as a function of season; indeed, the results
indicate that E hirae and E faecium were completely absent during summer and
winter, respectively. In contrast, it was not possible to identify any clear
seasonal trend in the % composition of the different species in water due to the
25
-------
high degree of seasonal variability observed. For instance, E. faecalis was
found in high abundance during the fall in both farm and upstream-of the-farm
locations. In contrast, during spring, it was found at relatively high abundance
only in manure. High seasonal variability was noted in the Enterococcus
populations isolated from water at the upstream-of-the-farm location, indicating
that the background Enterococcus populations (in wildlife and possibly poultry
due to the proximity of chicken houses to this site) are as variable as the
Enterococcus populations isolated at the farm sites. In general, these results
suggest that the five most abundant enterococci species are ubiquitous in the
environment, given the fact that they were found in water samples that are not
supposed to be impacted by cattle (CFS-1). In addition, the general use of
individual species to establish seasonal and/ or source trends is likely to be a
difficult task due to the high degree of variability observed.
Table 3: Seasonal % composition (mean ± sd) of the most common Enterococcus species
isolated from cattle manure samples collected at a cattle farm in Georgia.
Season
Spring
Summer
Fall
Winter
E. casseliflavus
4 ±7*
40 ±43
43 ±31
49 ±29
E. faecalis
34 ± 32*
4±14
12 ±22
4±14
E. faecium
6±14
21 ± 32*
14 ±24
0
E. flavescens
3 ±7*
28 ± 39*
12±15
12±29
E. hirae
22 ±32
0
13 ±27
17 ±27
*Significantly different than winter
26
-------
Table 4: Seasonal % composition of the most common Enterococcus species isolated
from water samples collected in a stream located at a beef cattle farm in Georgia.
Season
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Sample
location
CFS-1
CFS-1
CFS-1
CFS-1
CFS-2 thru 7
CFS-2 thru 7
CFS-2 thru 7
CFS-2 thru 7
E. casseliflavus
27 ±12
17±17
0
11 ±19
27 ±27
27 ±21
10 ±20
14±15
E. faecalis
60 ±35
50 ±44
56 ± 38*
44 ±51
26 ±24
26 ±21
53 ± 32*
9 ±20
E. faecium
0
6±10
7±16
0
7±5
1 ±6
5±8
13 ±25
E. flavescens
0
6±10
14 ±27
28 ±25
3±10
29 ±26
14±18
12±18
E. hirae
7±12
0
11 ±20
8±14
21 ±19
1 ±5
4±11
18 ±24
*Significantly different from Winter; CFS-1: upstream from sites impacted by cattle; CFS-2 thru 7:
farm sites potentially impacted by cattle.
Seasonal and Spatial Variability of Enterococci Communities in
Manure and Water. Cluster analysis was performed on the relative %
abundances of the five most common enterococci groups and the general
enterococci category found in the stream water and manure samples. The
purpose of the analysis was to determine if a community approach could produce
useful information related to developing more reliable MST data analysis. The
analysis produced 15 clusters of species that were identified as being present in
the system three or more times through out the whole sampling period (Table 5).
27
-------
Table 5: Composition and abundance (%) of Enterococcus species in clusters that
appeared 3 or more times in water and manure samples collected at a Georgia cattle farm
from September 2003 through January 2005.
% Composition of each Enterococcus species
Cluster
Composition
E. casseliflavus
E. faecalis
E. faecium
E. flavescens
E. hirae
All other
Enterococcus
E. casseliflavus
E. faecalis
E. faecium
E. flavescens
E. hirae
All other
Enterococcus
E. casseliflavus
E. faecalis
E. faecium
E. flavescens
E. hirae
All other
Enterococcus
Cluster 1
14
43
0
0
43
0
Cluster 6
11
3
82
3
0.9
0.9
Cluster 12
0
0
0
100
0
0
Cluster 2
5
0
0
0
95
0
Cluster 7
84
0
4
9
0
3
Cluster 13
0
100
0
0
0
0
Cluster 3
0
32
5
0
37
26
Cluster 8
18
31
0.3
34
4
11
Cluster 16
28
0
7
22
0
43
Cluster 4
39
21
3
0
13
24
Cluster 9
23
61
0
11
1
3
Cluster 17
0
0
0
0
0
100
Cluster 5
67
0
0
0
33
0
Cluster 11
44
0
0
0
17
38
Cluster 18
0
77
23
0
0
0
Six enterococci clusters were found in high relative % occurrence for the three
different sample sources, i.e., upstream-of-the-farm, at-the-farm sites and
manure (Figure 6). Cluster 4 was only found at the upstream and farm locations
where E. casseliflavus was usually more abundant. Clusters 8, 9, and 13 were
not only present in manure, but were also frequently found in the upstream and
farm samples; E. faecalis was overall the most abundant species in these three
clusters. Clusters 6 and 7 were only found at-the-farm sites and manure
28
-------
samples, and had a higher occurrence in the manure samples. E. faecium and
E. casseliflavus were the most abundant species in these two clusters (Figure 6).
30 -
g 25-
1
3 20 -
o
O
i_
-------
10
-------
observed during spring, while for EC-9 is clearly more abundant during fall
(Figure 9).
1 U
14 -
12 -
8 10-
C
-------
30 -
§
1
^ 20 -
8
0
10 -
0 -
1 1
1
I
1
1
r.
ll
1
^^H Spring
I I Summer
1 1 Fall
^m Winter
EC-4 EC-6 EC-7 EC-8 EC-9 EC-13
Cluster*
Figure 9: High % occurrence of enterococci clusters during different seasons in samples
collected at a cattle farm in Georgia.
Figure 10 shows the distribution of clusters and % occurrence of each cluster per
sampling site. The figure shows that only one cluster, although widely observed
in the water samples, could not be found in the manure samples (EC-4).
Likewise, EC-12 and EC-17 were only found in manure but not in the water.
These clusters are composed of E. flavescence and enterococci that could not
be speciated. Although a variety of clusters could be found at any given time at
each water sampling station, the most clusters identified per site was 8, while 14
clusters were identified in the manure samples, which indicates a much larger
diversity of enterococci communities in the cattle Gl system. The upstream site
(CFS-1) with only 4 clusters, had the least diversity found in the system. The
32
-------
clusters present at CSF-1 are part of the background composition of the system
because they are present at all the water sampling sites. In almost all the farm
stream sites, the cluster diversity was higher, probably reflecting the effect that
the manure added to the system.
100%
8
I
o
o
o
(A
O
Chandler Farm Sources
• EC-18
• EC-17
• EC-16
• EC-13
D EC-12
• EC-11
• EC-9
DEC-8
• EC-7
DEC-6
DEC-5
DEC-4
DEC-3
• EC-2
DEC-1
Figure 10: Cluster distribution and occurrence (%) per Chandler Farm sampling site and
source.
Comparison of Bacteroides Markers and Enterococci Clusters. The
Bacteroides markers (BM) were organized in five different clusters (Table 6) and
this information compared to the presence of the enterococci clusters (EC) in the
water samples (Figure 11). The human-BM was found twice concurrent with EC-
8 and once with EC-9. These two clusters had high abundance of E. faecalis
and E. flavescens. The cow-BM was found concurrent with 6 different ECs, but
33
-------
again most frequently with EC-8 and EC-9. In addition, both BMs were found
with EC-1 at least once. No spatial trend for either BM could be established,
which means that the markers were found at various locations in the stream
through out the year. Two possible conclusions can be drawn from these results.
The fact that the human-BM was found in various locations in the farm stream
water suggested that some of the E. faecalis and E. flavescence in the water
may not be coming only from cattle or wildlife, but also from human
contamination. The sources for this contamination could be leaky septic systems
given the rural aspect of the location where the samples were obtained.
Alternatively, these results could suggest that the human-BM was amplifying
Bacteroides DMA from sources other than human. This latter hypothesis also
applies to one of the cow-BMs that was amplified at CFS-1, the site upstream
from obvious areas of cattle contamination. This site could be affected by run-off
coming from various chicken houses located in fairly close proximity to the
stream headwaters (see Figure 2). These chicken houses could also be
responsible for the higher-than-expected enterococci diversity in the stream
water.
34
-------
Table 6: Composition of Bacteroides clusters identified in stream water collected at a
cattle farm in Georgia
Bacteroides Cluster* General Marker
1 Yes
2 Yes
3 Yes
4 Yes
5 No
Presence of marker1
Cow Marker2
No
No
Yes
Yes
No
in cluster
Human Marker2
No
Yes
No
Yes
No
Reference for all Bacteroides markers: Bernhard and Field, 2000.
2 The two human markers and two cow markers were combined to develop clusters 2 and 3,
respectively
35
-------
4 -
Ł
"w
_2
O
w
CD
;o
2
CD
-i-»
o
CD
00
2 -
EC-1
EC-2
EC-4
EC-6
EC-7
EC-8
EC-9
EC-13
234
Frequency of Bacteroides Clusters
Figure 11: Relationship of Bacteroides and enterococci clusters in stream water samples
collected at a cattle farm in Georgia.
Conclusions and Final Considerations
The general conclusions for this study follow:
• From a total of 11 Enterococcus species that were identified at Chandler
farm, 2 were only found in cattle manure, but were not recovered in
stream water. This makes such species unreliable markers of cattle fecal
contamination in surface waters since they do not survive in this
environment.
• The relative abundance of individual Enterococcus species isolated from
cattle manure that were also observed in the stream samples exhibited a
high degree of seasonal variability. This finding suggests that when
tracing back cattle contamination, season should be an important
consideration to include in the criteria to select the species that can be
used as tracer. However, the high degree of seasonal variability in some
of the most common species makes it very difficult to establish significant
differences between seasons and /or the sampled sources.
36
-------
• The 5 most common enterococci species identified were found in the
water samples at-the-farm and upstream-the- farm locations, suggesting
that these species are widely spread in the environment. Wildlife, an
adjacent-to-the-farm chicken house, and a few scattered single-family
houses could be contributing these same species in high numbers,
therefore creating high background concentrations.
• Cluster analysis seems to be a good approach to identify species groups
or enterococci communities that are specific to a location or source, and
suggests that a community fingerprint rather than an individual species
could be an alternative approach to trace back stream fecal contamination
to its source.
• Results with the Bacteroides markers generally agreed with the
enterococci data in that water sampled from stream locations CFS-2 thru 4
was highly impacted by cattle contamination, while locations CFS-5 thru 7
had occasional hits apparently affected by the season of the sampling
event. However, the cow marker was also detected at location CFS-1 that
was not under obvious cattle influence. The human Bacteroides marker
was also detected occasionally throughout all stream locations, except for
CFS-1, indicating either human fecal contamination in parts of the stream
or non-specific amplification of the human- and cow-bacteroides markers
due to other sources, such as poultry manure which is frequently used to
fertilize cattle pasture sites.
• The two methodologies applied in this study differ greatly in terms of cost
effectiveness and turn-around time of results. Building an enterococci
library is a time-consuming, expensive approach that has the potential to
provide a great deal of information when the proper statistical analytical
approach (in this case it was cluster analysis) is used to interpret the
results. Time availability (when are results expected or needed) and
funding support (large quantities of consumable laboratory supplies are
needed) are two important considerations to keep in mind when a library-
dependent method for microbial source tracking is planned. Application of
a library-independent approach, such as the Bacteroides markers allows
for a much faster and possibly less expensive results. However, the need
still exists for highly specific, reliable markers that will allow one to
separate specific sources and not only human vs. non human
contamination. In the case of Bacteroides, there remains a lack of
thorough temporal, spatial and specificity analyses of the few genetic
markers available so far.
37
-------
Acknowledgements
I like to thank Julie Maimes and Jared Fisher for the technical support offered on
the construction, maintenance and speciation of the enterococci library. I like to
acknowledge Dr. Charlene Jackson and Benny Barrett for the multiplex PCR
technology transfer and Paul Smith, Dr. Caragwen Bracken and Lourdes Prieto
for all the field and sample processing support. Thanks are also extended to
Jorge Santodomingo and his laboratory staff for support with the 16S rDNA
analysis. Finally, I like to thank Dr. Mike Cyterski and the UGA Statistical
Consulting Office for providing statistical support for data analysis.
38
-------
References:
Bernhard, A. E., and K. G. Field. 2000. A PCR assay to discriminate human
and ruminant feces on the basis of host differences in Bacteroides-
Prevotella genes encoding 16S rRNA. Appl. Environ. Microbiol. 66:4571-
4574.
Caugant, D. A., B. R. Levin, and R. K. Selander. 1981. Genetic diversity and
temporal variation in the E. coli population of a human host. Genetics
98:467-490.
Caugant, D. A., B. R. Levin, and R. K. Selander. 1984. Distribution of
multilocus genotypes of Escherichia coli within and between host families.
J. Hyg 92: 377-384.
Faith, N.G., J. A. Shere, R. Brosch, K. W. Arnold, S.E. Ansay, M. S. Lee, J. B.
Luchansky, and C.W. Kaspar. 1996. Prevalence and clonal nature of
Escherichia coli 0157:H7 on dairy farms in Wisconsin. Appl. Environ.
Microbiol. 62: 1519-1525.
Gilmore, M.S. 2002. The enterococci: pathogenesis, molecular biology, and
antibiotic resistance. ASM Press, Washington, D.C. 439p.
Gordon, D. M. 2001. Geographical structure and host specificity in bacteria and
the implications for tracing the source of coliform contamination.
Microbiology 147:1079-85.
Gordon, D.M. 1997. The genetic structure of Escherichia coli populations in feral
house mice. Microbiology 143: 2039-2046.
Hartel, P.G., J. D. Summer, J. L. Hill, J. V. Collins, J. A. Entry, and W. I.
Segars. 2002. Geographic variability of Escherichia coli ribotypes from
animals in Idaho and Georgia. J Environ Quality 31:1273-1278.
Jackson, C. R., P. J. Fedorka-Cray, and J. B. Barrett. 2004. Identification of
enterococci using a genus and species specific multiplex PCR. J Clin
Microbiol. 42:3558-3565
Jenkins, M. B., P. G. Hartel, T. J. Olexa, and J. A. Stuedemann. 2003.
Putative temporal variability of Escherichia coli ribotypes from yearling
steers. J. Environ. Quality 32: 305-309.
Kinzelman, J. C. Ng, E. Jackson, S. Gradus, R. Bagley. 2003. Enterococci as
indicators of Lake Michigan recreational water quality: Comparison of two
methodologies and their impacts on public health regulatory events. Appl.
Environ. Microbiol 69:92-96.
Ochman, H., T. S. Whittam, D. A. Caugant, and R. K. Selander. 1983. Enzyme
polymorphism and genetic population structure in Escherichia coli and
Shigella. J Gen Microbiol 129 (Pt 9):2715-26.
Parveen, S., K. M. Portier, K. Robinson, L. Edmiston, and M. L. Tamplin.
1999. Discriminant analysis of ribotype profiles of Escherichia coli for
differentiating human and nonhuman sources of fecal pollution. Appl
Environ Microbiol 65:3142-3147.
Scott, T. M., S. Parveen, K. M. Portier, J. B. Rose, M. L. Tamplin, S. R.
Farrah, A. Koo, and J. Lukasik. 2003. Geographical variation in ribotype
39
-------
profiles of Escherichia coli isolates from humans, swine, poultry, beef, and
dairy cattle in Florida 68:1089-1092.
Selander, R. K., D. A. Caugant, and T. S. Whittam. 1987. Genetic structure
and variation in natural populations of Escherichia coli. P. 1626-1648. In
F.C. Neidhart (ed) Escherichia coli and Salmonella typhimurium cellular
and molecular biology. Am. Soc. Microbiol., Washington, DC.
Souza, V., M. Travisano, P. E. Turner, L. E. Eguiarte. 2002. Does experimental
evolution reflect patterns in natural populations? E. coli strains from long
term studies compared with wild isolates. Antonie Van Leeuwenhoek
81:143-153.
Wiggins, B. A., P. W. Cash, W. S. Creamer, S. E. dart, P. P. Garcia, T. M.
Gerecke, J. Han, B. L. Henry, K. B. Hoover, E. L Johnson, K. C.
Jones, J. C. McCarthy, J. A. McDonough, S. A. Mercer, M. J. Noto, H.
Park, M. S. Philips, S. M. Purner, B. M. Smith, E. N. Stevens, and A. K.
Varner. 2003. use of antibiotic resistance analysis for representativeness
testing of multiwatershed libraries. AEM 69:3399-3405.
40
------- |