Assessmentoffecal pollution sources ina small northern-plainswatershedusingPCRand phylogenetic analysesofBacteroidetes16S rRNAgeneRegina Lamendella1, Jorge W. Santo Domingo2, Daniel B. Oerther1, Jason R. Vogel3 &Donald M. Stoeckel4

1Department of Environmental Engineering, University of Cincinnati, Cincinnati, OH, USA; 2US Environmental Protection Agency, Office of Research

and Development, National Risk Management Research Laboratory, Cincinnati, OH, USA; 3US Geological Survey, Lincoln, NE; and 4US Geological

Survey, Columbus, OH, USA

Correspondence: Jorge W. Santo Domingo,

US Environmental Protection Agency, Office

of Research and Development, National Risk

Management Research Laboratory, Martin

Luther King Drive, Cincinnati OH.

Tel.: 1513 569 7085; fax: 1513 569 7328;

e-mail: santodomingo.jorge@epa.gov

Received 31 March 2006; revised 23 June 2006;

accepted 11 July 2006.

DOI:10.1111/j.1574-6941.2006.00211.x

Editor: Michael Wagner

Keywords

microbial source tracking; Bacteroidetes ;

16S rRNA gene; water quality; fecal pollution.

Abstract

We evaluated the efficacy, sensitivity, host-specificity, and spatial/temporal dynamics

of human- and ruminant-specific 16S rRNA gene Bacteroidetes markers used to

assess the sources of fecal pollution in a fecally impacted watershed. Phylogenetic

analyses of 1271 fecal and environmental 16S rRNA gene clones were also performed

to study the diversity of Bacteroidetes in this watershed. The host-specific assays

indicated that ruminant feces were present in 28–54% of the water samples and in all

sampling seasons, with increasing frequency in downstream sites. The human-

targeted assays indicated that only 3–5% of the water samples were positive for

human fecal signals, although a higher percentage of human-associated signals

(19–24%) were detected in sediment samples. Phylogenetic analysis indicated that

57% of all water clones clustered with yet-to-be-cultured Bacteroidetes species

associated with sequences obtained from ruminant feces, further supporting the

prevalence of ruminant contamination in this watershed. However, since several

clusters contained sequences from multiple sources, future studies need to consider

the potential cosmopolitan nature of these bacterial populations when assessing fecal

pollution sources using Bacteroidetes markers. Moreover, additional data is needed

in order to understand the distribution of Bacteroidetes host-specific markers and

their relationship to water quality regulatory standards.

Introduction

The microbiological impairment of water is traditionally

assessed by monitoring concentrations of fecal indicator

bacteria such as fecal coliforms and enterococci. These

microorganisms are associated with fecal material from

warm-blooded animals and their presence in water indicates

the potential incidence of enteric pathogens that could cause

illness in exposed individuals (Dufour, 1984). Screening for

members of the Bacteroidetes phylum has advantages over

screening for traditional indicators, including fecal coli-

forms and enterococci. For example, Bacteroidetes are more

abundant in the feces of warm-blooded animals than

Escherichia coli, constituting up to one half of fecal bacterial

community (Holdeman et al., 1976; Fiksdal et al., 1985).

Additionally, Bacteroidetes might be useful in predicting

recent fecal contamination, as they are obligate anaerobes

and are assumed to survive for only short periods of time

outside the intestinal tract, while E. coli and enterococci are

facultative anaerobes able to proliferate in soil, sand and

sediments (Duerden, 1980; Allsop & Stickler, 1985; Franks

et al., 1998; Suau et al., 1999; Sghir et al., 2000; Hopkins

et al., 2001; Hold et al., 2002).

Current standard methods for measuring fecal pollution in

water cannot be used to identify sources of pollution, as

coliforms are common to all mammalian hosts (Gordon &

Cowling, 2003). Recently, numerous phenotypic, genotypic

and chemical based microbial source tracking (MST) meth-

ods have been developed with the intent of determining the

source of fecal pollution in various environments (U.S.

Environmental Protection Agency 2000; Scott et al., 2002;

Simpson et al., 2002). In general terms, MST tools could be

classified as library and nonlibrary-based methods, where

library-based methods require the development of fingerprint

databases of cultured bacteria, while library-independent

methods do not rely on the development of large culture

FEMS Microbiol Ecol xx (2006) 1–10 c� 2006 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. No claim to orginal US government works

collections. However, limited research has been conducted in

testing the efficacy of these MST tools on a spatial and

temporal scale (U.S. Environmental Protection Agency 2005).

Several studies have suggested that fecal members of the

phylum Bacteroidetes are ecologically diverse, numerically

important members of the intestinal microbial communities

and exhibit some host specificity (Salyers, 1984; Gherna &

Woese, 1992; Wood et al., 1998; Boone & Castenholz, 2001;

Daly et al., 2001; Leser et al., 2001; Hold et al., 2002). Recently,

Bernhard & Field (2000a, b) developed PCR-based, 16S rRNA

gene assays targeting Bacteroidetes which were capable of

distinguishing between human and ruminant sources (Bern-

hard & Field, 2000a, b; Bernhard et al., 2003; Field et al.,

2003). While these assays are rapid, specific and relatively

sensitive, the spatial and temporal analysis and host-specifi-

city of these PCR-based Bacteroidetes assays have not been

widely tested for their applicability to different large-scale

geographic scenarios. Furthermore, phylogenetic analysis of

the gene sequences recovered from a diverse array of environ-

mental samples is needed to further clarify host distributions

of fecal Bacteroidetes bacteria in environmental systems.

Plum Creek, located in south-central Nebraska, contains

elevated concentrations of fecal coliform bacteria and is one

of the main tributaries contributing to the 303(d) listed,

impaired Platte River (http://www.deq.state.ne.us/Publica.

nsf/Pages/WAT021). Cattle are among the potential primary

sources of fecal pollution; however, this has not been con-

firmed with any of the available source tracking methods.

Other animals (e.g. wildlife, horse and human) may also be

contributing to the pollution in this watershed. Cattle fecal

pollution is of concern, as some of the microorganisms shed

in cattle feces can be pathogenic to humans, such as E. coli

O157:H7, Listeria spp., Salmonella spp. Cryptosporidium

parvum and Giardia spp. (Cotruvo et al., 2004).

The goal of this study was to assess the applicability of

host-specific, Bacteroides-targeted, 16S-rRNA gene PCR

assays in identifying sources of fecal pollution in the Plum

Creek watershed by challenging total, ruminant and human-

targeted 16S rRNA gene Bacteroides markers against fecal,

water and sediment samples collected over a 6-month

period. The specificity of host-specific Bacteroides-targeted

16S rRNA gene assays was determined against target and

nontarget fecal samples, as well as by performing sequencing

analyses of clone libraries from fecal, water and sediment

samples.

Materials and methods

Study site: Plum Creek

Plum Creek drains 224 square miles of rural land and is a

perennial tributary to the middle of the Platte River, which

is located near the town of Smithfield, in south-central

Nebraska. Segment MP2-20000 of the Platte River was listed

on the 2004 Nebraska Section 303(d) list, impaired for

primary contact recreation due to excessive levels of fecal

coliform bacteria (http://www.deq.state.ne.us/Publica.nsf/

Pages/WAT021). Fecal coliform data from 2001 indicated

that Plum Creek was the most contaminated of the peren-

nial tributaries to this segment of the middle of the Platte

River (P. Obrien, USGS, pers. commun.). Grassland was the

most extensive land cover in the Plum Creek watershed,

with irrigated row crops, hay and small grains comprising

secondary land use. Six sampling sites were established along

Plum Creek. Site 1 was located near the headwaters of the

perennially flowing portion of Plum Creek and was used to

record background levels of pollution, due to the previously

observed lower levels of fecal indicator bacteria at this site.

Sites 2 through 6 were located intermittently downstream

on the main stem of Plum Creek, with Site 6 located in close

proximity to the confluence of Plum Creek with the Platte

River. An estimated 1200 cattle were located near the

sampled portion of Plum Creek, while humans (c. 15 septic

systems representing o 100 humans), horses (o 10) and

wildlife (e.g. deer, waterfowl, raccoon and opossum) are

among the other likely fecal sources possibly impacting this

watershed.

Sample collection

All fecal samples were collected aseptically during the last

week of July 2004 between Site 1 and Site 2 and transferred

into cryogenic-safe tubes, transported to the laboratory in

coolers and stored at � 80 1C prior to analyses. A total of

101 individual cattle fecal samples (collected between Site 1

and Site 2), 66 wildlife fecal samples, six horse fecal samples

and 10 septic dips (from 10 different households represent-

ing c. 50 humans) were used to evaluate the specificity of the

assays. Wildlife fecal samples were collected in the riparian

area directly adjacent to Plum Creek and likely comprise

samples from raccoons and opossums. Water samples were

collected in duplicate, biweekly from May 2004 through

November 2004 at each site along the Plum Creek

(n = 26–28 per site) following previously published standard

operating procedures (U.S. Geological Survey 2000). Briefly,

water samples were collected in sterilized Nalgene bottles and

transported to the laboratory in ice coolers. Water samples

(i.e. 30–100 mL) were then filtered on to 0.45mm polycarbo-

nate filters which were stored at � 80 1C until genomic DNA

extractions were performed. Water blanks were collected at

all 14 sampling dates as part of the quality assurance protocol

to determine the potential of cross-contamination due to

sample handling. Sediment bed cores were collected from

three different intervals (0–10 cm, 10–20 cm and 20–30 cm)

at Site 2 during the latter half of the sampling period (i.e.

August – November, 2004) using USGS sampling protocols

FEMS Microbiol Ecol xx (2006) 1–10c� 2006 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. No claim to orginal US government works

2 R. Lamendella et al.

(Edwards & Glysson, 1999). Sediment samples were stored

at � 80 1C until further processing. Enumeration of E. coli

(CFUs 100 mL�1) on mTEC agar was performed for each of

the six water-sampling sites along the Plum Creek to

determine level of fecal pollution (U.S. Environmental

Protection Agency, 2000).

Genomic DNA extraction

Prior to genomic DNA extractions, fecal slurries were made

consisting of 0.10 g of fecal sample and 3.5 mL of magne-

sium phosphate rinse buffer (American Public Health

Association, 1998). Fecal genomic DNA extractions were

performed with the UltraClean Fecal DNA Isolation Kit

according to the manufacturer’s instructions (MO BIO

Laboratories, Inc., Carlsbad, CA) using 250mL of each fecal

slurry. For water samples, DNA was extracted directly from

whole filters using UltraClean Soil DNA Isolation Kit (MO

BIO Laboratories, Inc.). Sediment genomic DNA was also

extracted using the Ultraclean Soil DNA Isolation Kit.

Genomic DNA was eluted in 50 mL of 10 mM Tris and

quantified using a NanoDrops ND-1000 UV spectrophot-

ometer (NanoDrop Technologies, Wilmington, DE). Fecal

genomic DNA was diluted using ultrapure water to achieve a

final concentration of 1.5 ng mL�1. To determine the poten-

tial impact of PCR inhibitors in the environmental DNA

extracts, PCR was performed with 10-fold dilutions (using

ultrapure water) of genomic DNA extracted from the water

and sediment samples. To test for possible cross-contamina-

tion, control blanks for genomic DNA extractions were

performed using ultrapure water samples as templates.

Extracted genomic DNA was stored at � 20 1C until further

processing.

Bacteroides 16S rRNA gene PCR assays

PCR analyses were performed on all fecal, water and

sediment samples using four different primers sets, targeting

bacteria closely related to total Bacteroides and Prevotella

(Bac32; herein referred as total Bacteroides-like) and using

primers specific to ruminant Bacteroides (CF128 and

CF193) or human Bacteroides (HF134 and HF 183). All

reactions were performed using Bac708R as the reverse

primer as previously described (Bernhard & Field, 2000b)

with the following modifications. The final PCR solutions

contained 2.5 mL of Takara Ex Taq 10X buffer (20 mM

Mg21), 2 mL of dNTP mixture (2.5 mM each), 0.4mL of 4%

BSA, 17 mL of UltraPure water, 0.5 mL of primer at

25 pmolmL�1 concentration and 0.625 U of Ex Taq DNA

polymerase (TAKARA Mirus Bio., Madison, WI). Reactions

were conducted on a DNA Engine 2 Tetrad thermalcycler

(Bio-Rad Laboratories, Inc., Hercules, CA). Template con-

centrations for fecal samples were adjusted to 0.25 ng mL�1

and 1.5 ng mL�1 for the total Bacteroides-like and the host-

specific Bacteroides assays, respectively. The PCR conditions

for the total Bacteroides-like PCR assay included a denatur-

ing step at 94 1C for 2 min, followed by 30 cycles (for fecal

samples) or 35 cycles (for sediment and water samples) of

94 1C for 1 min, 53 1C for 1 min and 72 1C for 1 min,

followed by a final extension step at 72 1C for 7 min. The

PCR conditions for CF128 and CF193 ruminant Bacteroides

assays included denaturing at 94 1C for 2 min followed by 30

cycles (fecal samples) or 35 cycles (sediment and water

samples) of 94 1C for 1 min, 62 1C for 1 min and 72 1C for

1.5 min, followed by a final extension step of 72 1C for 7 min.

The PCR conditions for the human assays, HF134 and

HF183, were similar to the aforementioned ruminant assays

with the exception that the annealing step was performed at

63 1C for 1 min and extension at 72 1C for 1.5 min.

All PCR products were visualized using 1% agarose

gels and GelSTAR nucleic acid stain (Cambrex BioScience,

East Rutherford, NJ). Detection limit assays were conducted

for the general Bacteroides-like, ruminant and human

16S rRNA gene targeted primer sets using composite

samples from each potential fecal source type in order to

determine the lowest concentration of detectable fecal

pollution (Table 1). To determine assay sensitivity, known

concentrations of fecal DNA composites were serially di-

luted using ultrapure water. Aliquots (1mL) of the serial

dilutions were then used as template in 35 PCR cycles for

each primer set tested.

PCR purification, cloning and sequencinganalyses

All PCR products were purified using the QIAquick PCR

Purification Kit according to the manufacturer’s instruc-

tions for the microcentrifuge protocol (QIAGEN, Valencia,

CA). Representative PCR products from total Bacteroides-

like and ruminant CF128 Bacteroides assays were cloned into

a pCR4.1 TOPO vector and transformed into Mach-T1

chemically competent E. coli cells as described by the

manufacturer (Invitrogen, Carlsbad, CA). Clone libraries

were developed from PCR products generated with genomic

DNA from septic tanks, cattle, wildlife and horse fecal

samples, water samples from all six Plum Creek sampling

sites and sediment cores from Site 2. Individual E. coli clones

were then subcultured into 300 mL of Luria Broth containing

50 mg mL�1 ampicillin and screened for inserts using M13

PCR.

M13 PCR products were submitted to Children’s Hospital

(Cincinnati, OH) for sequencing. Sequencing data was

generated using Big Dye sequencing chemistry (Applied

Biosystems, Foster City, California), M13 forward and

reverse primers and an Applied Biosystems PRISM 3730XL

DNA Analyzer. Sequences were manually cleaned for base

calls and aligned, using SEQUENCHER 4.5 software. Potential

FEMS Microbiol Ecol xx (2006) 1–10 c� 2006 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. No claim to orginal US government works

3Assessment of fecal pollution sources

chimeric sequences detected using Bellerophon (Hugenholtz

& Huber, 2003) and the Chimera Check (Cole et al., 2003)

were not included in further analyses. Sequences were also

submitted to BLAST homology search algorithms in order to

assess our sequence similarity to entries available in this

public database (Altschul et al., 1990). Phylogenetic analysis

used ARB software and trees were inferred from 527

sequence positions using two tree-building methods: neigh-

bor-joining using a Kimura correction and maximum

parsimony using the Phylip DNAPARS tool (Ludwig et al.,

2004). In order to statistically evaluate the confidence of

branching, bootstrap values were obtained from a consensus

of 100 parsimony trees, as the ARB neighbor-joining boot-

strap tool was not functional at the time of analysis. The 16S

rRNA gene sequence of Porphyromonas gingivalis (accession

# L16492) was considered as the outgroup for all con-

structed trees (Paster et al., 1994).

Diversity of the clone libraries was also investigated by

rarefaction analysis (Hurlbert, 1971; Heck et al., 1975;

Simberloff, 1978). Rarefaction curves were produced

using the Hurlbert analytical approximation algorithm and

95% estimated confidence intervals. Calculations were per-

formed with the freeware program aRarefactWin (Holland,

2003).

Nucleotide sequence accession numbers

Representative sequences generated in this study have been

deposited in the GenBank database under accession num-

bers DQ886077-DQ886236.

Results and discussion

Escherichia coli densities and 16S rRNA gene PCRassays

Microbiological assessment of surface waters from

Plum Creek indicated the geometric mean densities of

E. coli ranged from 55 CFU 100 mL�1 at Site 1, to

534 CFU 100 mL�1 at Site 4 (Table 2).

Using cattle fecal DNA as the target, the detection limits

of the general Bacteroidetes, ruminant CF128 and CF193

assays were found to be 1� 10�14, 1� 10�12 and 1� 10�11 g

fecal genomic DNA, respectively (Table 1). The lower

sensitivity of the ruminant assays is in agreement with the

fact that these assays only target a fraction of the total

Bacteroides-like bacteria (Bernhard & Field, 2000b). The

ruminant CF128 assay also exhibited a higher amplification

percentage for fecal (96%) and water samples (53%) than

the CF193 primer (90% and 28%, respectively), which is

indicative of the higher sensitivity of the CF128 primer and

consistent with previous findings (Bernhard & Field,

2000b). Detection limits for the Bac32, HF134 and HF183

assays performed on septic samples were 1� 10�12, 1� 10�9

and 1� 10�10 g of fecal DNA, respectively. These results

indicate that the HF183 marker is more sensitive than the

HF134 marker, findings that are in agreements with Bern-

hard & Field (2000b). Detection limits with wildlife and

horse fecal DNA extracts using the Bac32 assay were

1� 10�13 and 1� 10�14 g of fecal DNA, respectively.

All water-collection field blanks and fecal extraction

blanks were negative using all PCR assays. None of the no-

template controls (i.e. �10% of the total number of

reactions) produced amplification products, implying the

lack of artifacts introduced by cross-contamination. All

positive controls (�2% of the total number of reactions)

produced the expected amplification products. More than

90% of the cattle fecal samples produced positive signals

with the general Bacteroidetes and the ruminant-specific

PCR assays (Table 2). Interestingly, only one-third of the

wildlife fecal samples amplified using the general Bacteroi-

detes marker, suggesting that Bacteroides-like populations

may not be as prominent in the intestinal microbiota of

certain wildlife or that the tested primers cannot universally

amplify all members of this fecal bacterial group. When

ruminant specific assays were challenged against nontarget

fecal samples, o 3% showed to cross amplify with wildlife

and septic source samples, although it should be noted that

all six horse fecal samples amplified with both ruminant-

specific PCR assays. Cross amplification with the horse

samples suggests that the ruminant assay is not capable of

discriminating between horse and cattle pollution in this

particular watershed. Similar to cattle, horses are large

mammalian herbivores and it is therefore possible for them

to host similar Bacteroidales populations. Both human-

targeted Bacteroides PCR assays produced positive amplifi-

cation signals with 40% of the septic samples. Cross

Table 1. Limit of detection in fecal sample composites using general and host-specific Bacteroidetes primer sets

Bac32 Bac708 CF128 Bac708 CF193 Bac708 HF134 Bac708 HF 183 Bac708

Cattle (10)� 1� 10�14w 1� 10�12 1� 10�11 – –

Septic (10) 1� 10�12 – – 1� 10�9 1� 10�10

Horse (6) 1� 10�14 – – – –

Wildlife (10) 1� 10�13 – – – –

�Number of samples in composite used for detection limit assays.wAll detection limits are in the units of grams of fecal DNA.

FEMS Microbiol Ecol xx (2006) 1–10c� 2006 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. No claim to orginal US government works

4 R. Lamendella et al.

amplification with the human HF134 assay was low when

challenged against cattle and wildlife fecal DNA (i.e. 4% and

6%, respectively), but frequent with horse samples (i.e. four

of six fecal samples). The HF183 assay appeared to be

slightly more specific than the HF134 assay, as of all

nontargeted samples only one cattle sample cross-amplified

with the former assay.

Application of host-specific Bacteroides 16S rRNA gene

markers uncovered interesting spatial and temporal trends

of fecal pollution in the Plum Creek watershed. Nearly all of

the water samples (i.e. 94%) analyzed in this study produced

positive signals using the general Bacteroidetes assay, while

54% and 28% of the water samples were positive for the

ruminant-specific assays CF128 and CF193, respectively

(Table 2). In all sites that CF193 produced positive signals,

the water samples were also positive when using the CF128

assay, further suggesting the presence of ruminant fecal

pollution in these sites. Among the six sites sampled along

the Plum Creek, water samples from Sites 1 and 2 were

associated with the lowest number of positive ruminant

specific signals. The low amount of positive ruminant

signals at Site 2 is in contrast with the fact that cattle were

present nearby this site. However, water samples down-

stream from Site 2 showed an increase in amplification

percentages when using ruminant-specific assays, with sam-

ples from Site 5 achieving 100% amplification using the

CF128 assay. On a spatial scale, these PCR-based assays

indicated ruminant-fecal pollution incidence generally in-

creased along the downstream tributary gradient of Plum

Creek (Fig. 1). This observation may suggest an accumula-

tion of ruminant-associated fecal pollution along the length

of Plum Creek reflected by the increasing amount of

potential cattle source contributions to Plum Creek along

its entirety. Ruminant-associated contamination was ob-

served in all sampling seasons, with summer water samples

showing the highest number of positive ruminant signals

after short rainfall events that occurred during July 2004

(Fig. 1).

PCR amplification using the HF134 primer (i.e. human-

specific assay) was only positive for a limited number of

water samples from Sites 4, 5 and 6 (i.e. 5% of all water

samples). The HF183 assays supported these results as a

limited number of water samples from Site 4 and 6 amplified

as well. This low percentage of human-specific amplification

may be indicative of the relatively low ratio of humans (i.e.

o 500) to cattle (i.e. 2000–5000) existing in Plum Creek

watershed. However, it should be noted that in this study the

human assays amplified only 40% of the septic samples,

suggesting that these assays might be underestimating the

overall contribution of human sources in this watershed.

The low efficiency of the human-targeted Bacteroides-based

assays has previously been reported, suggesting their some-

what limited value in MST studies (Carson et al., 2005;

Fogarty & Voytek, 2005). In contrast, PCR results for

sediment showed that nearly 24% and 19% of the sediment

samples were positive for the HF134 and HF183 PCR assays,

respectively, while only 14% of the sediment samples were

positive with the ruminant-specific assays, even though

nearly 81% of the sediment samples produced a PCR

product with the total Bacteroides-like specific assay. Sedi-

ment-derived PCR results may be biased, as sediment cores

were only collected from one of the six sampling sites along

the Plum Creek. No amplification trends were evident

among samples from different sediment depths, although

all depths amplified at least once with all assays. Higher

levels of human-associated fecal pollution in sediment

Table 2. Escherichia coli densities and PCR results using total Bacteroides-like and host-specific markers

Samples (n) Bac32 CF128 CF193 HF134 HF183 E. colic

Cattle fecal samples (101) 97� 96 90 4 1 –

Horse fecal samples (6) 100 100 100 67 0 –

Wildlife fecal samples (66) 32 3 2 6 0 –

Septic samples (10) 90 0 0 40 40 –

Sediment samples (21)w 81 14 14 24 19 –

Water samples Site 1 (26) 96 12 8 0 0 55

Water samples Site 2 (26) 96 19 15 0 0 270

Water samples Site 3 (26) 92 38 19 0 0 381

Water samples Site 4 (28) 89 68 21 14 11 534

Water samples Site 5 (26) 96 100 54 4 0 466

Water samples Site 6 (26) 93 85 50 8 8 302

Total water samples (158) 94 54 28 5 3 –

�Percent positive total Bacteroides-like, ruminant and human-targeted PCR amplification for 101 cattle fecal samples, 66 wildlife fecal samples, six

horse fecal samples, 10 septic tank samples representing human fecal samples, 158 total water samples (n = 26–28 per sampling site along Plum Creek).

Water samples were collected in duplicate, thus PCR results for each replicate were averaged.wSediment samples were collected from Site 2.cGeometric mean densities of E. coli for Sites 1 through 6 measured in CFU 100 mL�1.

FEMS Microbiol Ecol xx (2006) 1–10 c� 2006 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. No claim to orginal US government works

5Assessment of fecal pollution sources

samples may also reflect differences in survival rates of

Bacteroides species from different hosts. Survivability of

Bacteroides species in water is impacted by dissolved oxygen

concentration and temperature (Gerba & McLeod, 1976;

Howell et al., 1996; Jones et al., 1980; LaLiberte & Grimes,

1982; Gordon & Cowling, 2003); however, the survivability

of the different Bacteroides species in sediment is poorly

understood. Thus, survivability and accurate identification

of fecal bacteria from sediment environments may be

imperative in understanding sediment’s contribution to

presumed fecal loads in the water column (Hood & Ness,

1982). If indeed populations from different sources have

different survival rates in sediments, these populations

might not be the best targets to track recent fecal contam-

ination events.

Sequence and phylogenetic analysis

To further understand and assess environmental host-dis-

tribution of Bacteroides-like populations in the Plum Creek

watershed, clone libraries were constructed using DNA from

cattle, horse, wildlife, human (septic samples) hosts, water

samples from each site and sediment samples from Site 2. A

total of 142 sequences were identified as potential chimeric

sequences and were removed from the phylogenetic analysis.

The final phylogenetic analysis contained 1271 sequences

from fecal samples (405 cattle, 174 wildlife, 167 septic tanks,

87 horse), water (319) and sediment (119) samples. Phylo-

genetic relationships were inferred from partial 16S rRNA

gene sequences (527 positions) (E. coli bases 151 to 678) of

clones recovered from Plum Creek watershed environmental

samples, cultured members of the Bacteroides and Prevotella

genera and previously identified uncultivated fecal bacteria

of the order Bacteroidales bacteria from different hosts

(Paster et al., 1994; Simpson et al., 2004; Dick et al., 2005).

Among all the unique clone sequences from the different

library types analyzed in this study, 19–64% exhibited high

sequence similarity (Z97%) to Bacteroidales-like16S rRNA

gene sequences in the publicly available databases, while

0–28% showed relatively low sequence similarity (o 93%)

(Fig. 2).

The phylogenetic analysis of the clones indicated some

interesting relationships among host fecal bacterial distribu-

tions in fecal, water and sediment matrices (Fig. 3). Most

notably, almost half (46%) of all the sequences in this study

(581 sequences) exclusively derived from cattle feces and

water samples comprised a cluster of their own, forming a

remote sister group to cultured Bacteroides genera. This

latter group also clustered with previously identified cattle

fecal members (i.e. CF123 and CF151 clusters) (Bernhard &

Field, 2000b; Dick et al., 2005). This observation of endemic

host-distribution of ruminant Bacteroides is supported by

Dick et al. (2005) in which ruminant Bacteroidales sequences

shared the same cluster and were not present in clusters of

other host species. The unique anatomy and physiology of

the ruminant digestive system may have provided a different

evolutionary pathway for Bacteroides-like organisms to live

than those inhabiting nonruminant hosts (Dick et al., 2005).

This large ruminant-associated cluster also provides further

evidence of cattle fecal pollution present in the Plum Creek,

as several cattle fecal-derived sequences shared 100% se-

quence identity to sequences acquired from water sample

clone libraries. Additionally, 57% of all water-derived

sequences clustered with cattle-derived sequences (Fig. 3).

Sequences obtained from water samples from all six

sites were represented in this large supercluster, while

some sequences from exclusively water clone libraries

0

10

20

30

40

50

60

70

Cattle Horse Septic Wildlife Water SedimentSample Type

Per

cen

t o

f B

acte

roid

ales

-lik

e S

equ

ence

s 97-100%96-93%< 93%

Fig. 2. Percent sequence similarity of Plum Creek watershed sequences

to Bacteroidales-like bacteria. Sequences from cattle, human (septic),

sediment and water-derived libraries were divided among three se-

quence similarity categories: 97–100%, 96–93% and o 93%.

0

10

20

30

40

50

60

70

80

90

100R

um

inan

t-sp

ecif

ic A

mp

lific

atio

n%

(C

F12

8 m

arke

r)

Site 1 Site 2 Site 3 Site 4 Site 5 Site 6Plum Creek Tributary Gradient

AUTUMN

SUMMER

SPRING

Fig. 1. Spatial and temporal distribution of ruminant host-specific

positive PCR amplification using the CF128 marker, along the down-

stream Plum Creek tributary gradient for spring, summer and autumn

2004 (n = 8–10 per season, n = 26–28 per sampling site along the Plum

Creek.

FEMS Microbiol Ecol xx (2006) 1–10c� 2006 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. No claim to orginal US government works

6 R. Lamendella et al.

formed clusters outside of this supercluster, indicating

even this extensive fecal clone library construction did

not represent all fecal Bacteroidales present in the water

environment. Additionally, because fecal samples used

in this study represent only a portion of the hosts contribut-

ing to fecal loadings in Plum Creek, fecal samples may

not have adequately represented the overall molecular

diversity associated with this watershed. Representative

sampling of both host fecal material and environmental

samples is critical to ensure fecal bacterial diversity is well-

characterized.

To a lesser extent, phylogenetic analysis also uncovered

relationships among wildlife and human (i.e. septic sam-

ples) derived sequences in both water and sediment samples

from Plum Creek. Interestingly, 54% of the wildlife se-

quences affiliated with cultured members of the Bacteroides

spp. group. Most notably, 60 wildlife clones clustered with

Bacteroides ovatus, while small exclusive wildlife-derived

clades formed with Bacteroides acidofaciens and Bacteroides

eggerthii. Human and avian clones have been previously

observed to be affiliated to these cultured Bacteroides; (Dick

et al., 2005) however, this is the first study in which

nonwaterfowl wildlife derived sequences are shown to be

associated with these Bacteroides clades (Fig. 3). Nearly 60%

of all septic clones clustered with cultured Bacteroides spp.,

with 89 clones composing an exclusive cluster with Bacter-

oides thetaiotaomicron, a bacteria species primarily asso-

ciated with human fecal contamination (Holdeman et al.,

Bacteroides spp.

Prevotella spp.

Unidentified Env.,Fecal, and HUMAN

Unidentified Env., CattleFecal, and CATTLE

Cattle Fecaland ELK

Horse Fecal, UnidentifiedEnv., and HORSE

B. merdae, Fecal, andUnidentified Env.

0.10

Fig. 3. Schematic representation of the host distribution of Bacteroides 16S rRNA gene sequences in Plum Creek based on a neighbor-joining

algorithm. Information in parenthesis refers to the number of sequences per library type (i.e. fecal, water and sediment) as compared to cultured

Prevotella and Bacteroides species, as well as previously identified nonculturable CATTLE, ELK, HUMAN, CAT, DOG and HORSE Bacteroidales bacterial

sequences (Paster et al., 1994; Simpson et al., 2004; Dick et al., 2005). Unidentified Env. refers to bacterial sequences obtained from Plum Creek water

and sediment samples.

FEMS Microbiol Ecol xx (2006) 1–10 c� 2006 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. No claim to orginal US government works

7Assessment of fecal pollution sources

1976). Many of the remaining wildlife and septic clones

clustered in unidentified environmental clades, distinct from

the Bacteroides and Prevetolla spp. clades, clustering with 86%

of all sediment-derived sequences and nearly one-third of

water-derived clones. For example, in an unidentified envir-

onmental and fecal clade, 30 wildlife and 16 septic-derived

clones affiliated with 32 and 23 sediment and water clones,

respectively. Compared to the cattle-like sequences, clones

generated in this study from septic and wildlife were found in

numerous clades, suggesting a relatively cosmopolitan life-

style of many of these bacterial populations. In addition,

novel clades outside the Bacteroides and Prevotella-related

groups containing sequences from septic, wildlife, sediment

and water samples were identified, further suggesting that the

Bacteroidales is a large and complex bacterial group.

Overall, fewer clone sequences clustered with Prevotella-

related species. One water sample clone from Site 6 was

100% similar to Prevotella ruminicola, a group commonly

found in the rumen and hindgut of mammalians. Although

13 of the 87 horse-derived bacterial sequences were members

of a cosmopolitan clade related to the Bacteroides group, the

vast majority of the horse clones clustered with Prevotella

spp. and Prevotella-like groups. A horse-specific cluster

containing 58 sequences was previously documented by Dick

et al. (2005). Additionally, three sediment sequences pre-

viously described as specific to cat, dog and human formed a

small clade within the Prevotella group. Very few (two

clones) cattle fecal-derived sequences were closely related to

cultured Prevotella species. These trends were noted by Dick

et al. (2005), as sequences from nonruminant feces fell within

the genera Bacteroides and Prevotella, while most sequences

from ruminant hosts did not cluster with any known species.

Another notable feature of the phylogenetic analysis was

the apparent clustering of sediment-derived sequences to

culturable Bacteroides genera and septic sequences as seen in

the Bacteroides uniformis and Bacteroides ovatus-related

groups. These clusters provide some evidence of human-

fecal contamination in the sediment samples perhaps origi-

nating from surrounding septic systems used in this wa-

tershed. This hypothesis however becomes confounded, as

the B. ovatus cluster also contains several wildlife bacterial

clones. These observations may be explained by the cosmo-

politan lifestyle of cultured fecal Bacteroidales with respect

to their host, as explained by the horizontal transmission of

fecal bacteria among hosts that may have more interaction

with each other, resulting in higher rates of horizontal

transfer of fecal bacteria (Dick et al., 2005).

The overall topology of the neighbor-joining phyloge-

netic analysis of Bacteroides-like 16S rRNA gene sequences

was supported by parsimony trees with 100 resamplings

(data not shown). The branch of the tree containing a

majority of the cattle fecal sequences and water-derived

sequences was supported in 85% of the parsimony boot-

strap-resamplings, while two branches of the tree containing

cultured Bacteroides species (B. acidofaciens and Bacteroides

fragilis) and fecal sequences were not supported by high

bootstrap values. The low bootstrap values may be a result of

the use of partial 16S rRNA gene sequences limiting phylo-

genetic resolution in a comparative analysis (Hopkins et al.,

2001). All other superclusters in the parsimony analysis were

supported by bootstrap values of 50% and higher.

Rarefaction analysis was conducted to evaluate whether

screening 1271 clones was sufficient to estimate diversity of

Bacteroides-like sequences derived from fecal and environ-

mental clone libraries from Plum Creek (Fig. 4). The

calculated rarefaction curves did not reach saturation, in-

dicating that analysis of an increasing number of clones

would have revealed further diversity. Rarefaction analysis

did indicate that the richness of species appeared to be higher

in sequences derived from water clone libraries (120 opera-

tion taxonomic units, herafter OTUs) as compared to cattle

fecal sequences (85 OTUs) and other fecal sequences (102

OTUs). This observation may indicate that our fecal source

sampling was not extensive enough to capture the overall

Bacteroides-like diversity in water samples. These results

further revealed that the molecular diversity of Bacteroides-

like bacteria in the environment is high. Attempts to quantify

specific groups using techniques like qPCR will benefit from

a better understanding of the molecular diversity of fecal

bacteria. Quantification of fecal sources is an important step

towards determining fecal loads and for the development of

accurate microbial risk assessment models (Simpson et al.,

2002; U.S. Environmental Protection Agency 2005).

The results from the Bacteroidetes host-specific assays and

the phylogenetic analyses were able to confirm the preva-

lence of ruminant contamination in Plum Creek. The results

also suggest that other sources (e.g. human, horse and

wildlife) are potentially contributing to the fecal pollution

of this watershed. However, it should be noted that these

host-specific PCR assays do have some limitations, as the

assays did show some cross-amplification with a small

percentage of wildlife fecal samples and most horse fecal

samples. Despite these inherent limitations, assessment of

fecal pollution in Plum Creek watershed highlighted the

applicability of 16S rRNA gene PCR assays and phylogenetic

analysis as tools to differentiate among ruminant and non-

ruminant fecal source pollution in fecal, water and sediment

matrices. Furthermore, notable spatial and temporal varia-

tions of fecal pollution could be indicative of ‘hot spots’

associated to particular sources of fecal pollution. This is true

only if some of the specific clusters are endemic. For example,

in this study ruminant-specific Bacteroides sequences illu-

strated endemism, while nonruminant sequences were less

host-specific. While culture-independent PCR-based assays

presents the opportunity of characterizing microbial ecosys-

tems independent of microbial cultivation, understanding

FEMS Microbiol Ecol xx (2006) 1–10c� 2006 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. No claim to orginal US government works

8 R. Lamendella et al.

the host distribution of host-specific markers and its relation-

ship to water quality regulatory standards are necessary to

effectively apply these assays under regulatory scenarios.

Acknowledgements

This research was supported in part by an Augmentation

Award to J.S.D. from the National Center for Computational

Toxicology of the U.S. EPA, Office of Research and Devel-

opment, by a grant from the Nebraska Department of

Environmental Quality and by the USGS Cooperative Water

Program. We are grateful for the cooperation of the land-

owners in the Plum Creek watershed during this project and

for the technical assistance from Cathy Kelty, Jinrang Lu and

Christopher Luedeker. Any opinions expressed in this paper

are those of the author(s) and do not, necessarily, reflect the

official positions and policies of the U.S. EPA. Any mention

of products or trade names does not constitute recommen-

dation for use by the U.S. EPA or the USGS.

References

Allsop K & Stickler DJ (1985) An assessment of Bacteroides fragilisgroup organisms as indicators of human fecal pollution. J ApplBacteriol 58: 95–99.

Altschul SF, Gish W, Miller W, Myers EW & Lipman DJ (1990)Basic local alignment search tool. J Mol Biol 215: 403–410.

American Public Health Association (1998) Standard MethodsFor the Examination of Water and Wastewater. APHA/AWWA/WEF, Washington, DC.

Bernhard AE & Field KG (2000a) Identification of nonpointsources of fecal pollution in coastal waters by using host-specific 16S ribosomal DNA genetic markers from fecalanaerobes. Appl Environ Microbiol 66: 1587–1594.

Bernhard AE & Field KG (2000b) A PCR assay to discriminatehuman and ruminant feces on the basis of host-differences inBacteroides-Prevotella genes encoding 16S rRNA. Appl EnvironMicrobiol 66: 4571–4574.

Bernhard AE, Goyard T, Simonich MT & Field KG (2003)Application of a rapid method for identifying fecal pollutionsources in a multi-use estuary. Water Res 37: 909–913.

Boone DR & Castenholz RW (2001) Bergey’s Manual ofSystematic Bacteriology, Vol. 1. Springer-Verlag, New York, NY.

Carson CA, Christiansen JM, Yampara-Iquise H, Benson VW,Baffaut C, Davis JV, Broz RR, Kurtz WB, Rogers WM & FalesWH (2005) Specificity of a Bacteroides thetaiotaomicronmarker for human feces. Appl Environ Microbiol 71:4945–4949.

Cole JR, Chai B, Marsh T, Farris R, Wang Q, Kulam S, Chandra S,McGarrell D, Schmidt T, Garrity G & Tiedje J (2003) TheRibosomal Database Project (RDP-II): previewing a newautoaligner that allows regular updates and the newprokaryotic taxonomy. Nucleic Acids Res 31: 442–443.

Cotruvo JA, Dufour A, Rees G, Bartram J, Carr R, Cliver DO,Craun GF, Fayer R & Gannon VPJ (2004) WaterborneZoonoses: Identification, Causes and Control. IWA Publishing,London, UK.

Daly K, Stewart CS, Flint HJ & Shirazi-Beechey SP (2001)Bacterial diversity within the equine large intestine as revealedby molecular analysis of cloned 16S rRNA genes. FEMSMicrobiol Ecol 38: 141–151.

Dick LK, Bernhard AE, Brodeur TJ, Santo Domingo JW, SimpsonJM, Walters SP & Field KG. (2005) Host distributions ofuncultivated fecal Bacteroidales bacteria reveal genetic markersfor fecal source identification. Appl Environ Microbiol 71:3184–3191.

Duerden BI (1980) The isolation and identification of Bacteroidesspp. From normal human fecal flora. J Med Microbiol 13:69–78.

Dufour AP (1984) Bacterial indicators of recreational waterquality. Can J Public Health 75: 49–56.

Edwards TK & Glysson GD (1999) Field methods formeasurement of fluvial sediment. U.S. Geological SurveyTechniques of Water-Resources Investigations, pp. 89.

Field KG, Bernhard AE & Brodeur TJ (2003) Molecularapproaches to microbiological monitoring: fecal sourcedetection. Environ Moni Assess 81: 313–326.

Fiksdal L, Maki JS, LaCroix SJ & Staley JT (1985) Survival anddetection of Bacteroides spp. Prospective indicator bacteria.Appl Environ Microbiol 49: 148–150.

Fogarty LR & Voytek MA (2005) Comparison of Bacteroides-Prevotella 16S rRNA genetic markers for fecal samples fromdifferent animal species. Appl Environ Microbiol 71:5999–6007.

Franks AH, Harmsen HJM, Raangs GC, Jansen GJ, Schut F &Welling GW (1998) Variations of bacterial populations inhuman feces measured by fluorescent in situ hybridizationwith group-specific 16S rRNA-targeted oligonucleotideprobes. Appl Environ Microbiol 64: 3336–3345.

0

20

40

60

80

100

120

140

0 50 100 150 200 250 300 350 400 450

Number of Unique 16S rDNA Sequences

Exp

ecte

d N

um

ber

of

OT

U's

WaterCattle FecalSedimentOther Fecal

Fig. 4. Rarefaction curves for unique 16S rRNA gene library clones.

Rarefaction curves were calculated using the analytical approximation

algorithm described by Hurlbert (1971) and 95% confidence intervals

estimated as described by Heck et al. (1975). The expected number of

operational taxonomic units is plotted vs. the number of unique 16S

rRNA gene clones. Rarefaction curves were calculated for cattle fecal,

other fecal (wildlife, horse and human), water and sediment clone

libraries. The error bars indicate the 95% confidence intervals.

FEMS Microbiol Ecol xx (2006) 1–10 c� 2006 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. No claim to orginal US government works

9Assessment of fecal pollution sources

Gerba CP & McLeod JS (1976) Effects of sediments on the

survival of Escherichia coli in marine waters. Appl Environ

Microbiol 32: 114–120.Gherna R & Woese CR (1992) A partial phylogenetic analysis of

the ‘‘Flavobacter-Bacteroides’’ phylum: basis for taxonomic

restructuring. Syst Appl Microbiol 15: 513–521.Gordon DM & Cowling A (2003) The distribution and genetic

structure of Escherichia coli in Australian vertebrates: host and

geographic effects. Microbiology 149: 3575–3586.Heck KL Jr, Van Belle G & Simberloff D (1975) Explicit

calculation of the rarefaction diversity measurement and the

determination of sufficient sample size. Ecology 56: 1459–1461.Hold GL, Pryde SE, Russell VJ, Furrie E & Flint HJ (2002)

Assessment of microbial diversity in human colonic samples

by 16S rDNA sequence analysis. FEMS Microbiol Ecol 39:

33–39.Holdeman LV, Good IJ & Moore WEC (1976) Human fecal flora:

variation in bacterial composition within individuals and a

possible effect of emotional stress. Appl Environ Microbiol 31:

359–375.Holland S (2003) aRarefactWin 1.3 program. University of

Georgia, Athens. www.uga.edu/strata/software/Software.html.Hood MA & Ness GE (1982) Survival of Vibrio cholerae and

Escherichia coli in estuarine waters and sediments. Appl

Environ Microbiol 43: 578–584.Hopkins MJ, Sharp R & Macfarlane GT (2001) Age and disease

related changes in intestinal bacterial populations assessed by

cell culture, 16S rRNA abundance and community cellular

fatty acid profiles. Gut 48: 198–205.Howell JM, Coyne MS & Cornelius PL (1996) Effect of sediment

particle size and temperature on fecal bacteria mortality rates

and fecal coliform/fecal streptococci ratio. J Environ Qual 25:

1216–1220.Hugenholtz P & Huber T (2003) Chimeric 16S rDNA sequences

of diverse origin are accumulating in public databases. Int J

Syst Evol Microbiol 53: 289–293.Hurlbert SH (1971) The nonconcept of species diversity: a

critique and alternative parameters. Ecology 52: 577–586.Jones DT, Robb FT & Woods DR (1980) Effect of oxygen on

Bacteroides fragilis survival after far-ultraviolet irradiation. J

Bacteriol 144: 1179–1181.LaLiberte P & Grimes DJ (1982) Survival of Escherichia coli in

lake bottom sediment. Appl Environ Microbiol 43: 623–628.Leser TD, Amenuvor JZ, Jensen TK, Lindecrona RH, Boye M &

Moller K (2001) Culture-independent analysis of gut bacteria:

the pig gastrointestinal tract microbiota revisited. Appl Environ

Microbiol 68: 673–690.

Ludwig W, Strunk O, Westram R et al. (2004) ARB: a softwareenvironment for sequence data. Nucleic Acids Res 32:1363–1371.

Paster BJ, Dewhirst FE, Olsen I & Fraser GJ (1994) Phylogeny ofBacteroides, Prevotella and Porphyromonas spp. and relatedbacteria. J Bacteriol 176: 725–732.

Salyers AA (1984) Bacteroides of the human lower intestinal tract.Annu Rev Microbiol 38: 293–313.

Scott TM, Rose JB, Jenkins TM, Farrah SR & Lukasik J (2002)Microbial source tracking: current methodology and futuredirections. Appl Environ Microbiol 58: 5796–5803.

Sghir A, Gramet G, Suau A, Rochet V, Pochart P & Dore J (2000)Quantification of bacterial groups within human fecal flora byoligonucleotide probe hybridization. Appl Environ Microbiol66: 2263–2266.

Simberloff D (1978) Use of rarefaction and related methods inecology. Biological Data in Water Pollution Assessment:Quantitative and Statistical Analyses (Dickson KL, Cairns JJ &Livingston RJ, eds), pp. 150–165. American Society for Testingand Materials, Philadelphia, PA.

Simpson JM, Santo Domingo JW & Reasoner DJ (2002)Microbial source tracking: state of science. Environ Sci Tech 36:5279–5288.

Simpson JM, Santo Domingo JW & Reasoner DJ (2004)Assessment of equine fecal contamination: the search foralternative bacterial source-tracking targets. FEMS MicrobiolEcol 47: 65–75.

Suau ARB, Sutren M, Godon JJ, Gibson GR, Collins MD & Dore J(1999) Direct analysis of genes encoding 16S rRNA fromcomplex communities reveals many novel molecularspecies within the human gut. Appl Environ Microbiol 65:4799–4807.

U.S. Environmental Protection Agency (2000) ImprovedEnumeration Methods for the Recreational Water QualityIndicators-Enterococci and Escherichia coli. Office of Scienceand Technology, Washington, DC. pp 27.

U.S. Environmental Protection Agency (2005) Microbial SourceTracking Guide Document. Office of Research andDevelopment, United States Environmental ProtectionAgency, 143pp.

U.S. Geological Survey (2000) National field manual for thecollection of water-quality data, U.S. Geological SurveyTechniques of Water-Resources Investigations, pp. A1-A9.

Wood J, Scott KP, Newbold CJ & Flint HJ (1998) Estimation ofthe relative abundance of different Bacteroides and Prevotellaribotypes in gut samples by restriction enzyme profiling ofPCR-amplified 16S rRNA gene sequences. Appl EnvironMicrobiol 64: 3683–3689.

FEMS Microbiol Ecol xx (2006) 1–10c� 2006 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. No claim to orginal US government works

10 R. Lamendella et al.