Enterococcus species distribution among human and animal hosts using multiplex PCR


  • Present addressL. H. Lam, Department of Microbiology & Immunology, Stanford University, Stanford, CA 94305, USA.

Alexandria Boehm, Environmental & Water Studies, Stanford University, The Jerry Yang and Akiko Yamazaki Environment and Energy Building, Mail code 4020, Stanford, CA 94035, USA. E-mail: aboehm@stanford.edu


Aims:  This study evaluated the use of Enterococcus species differentiation as a tool for microbial source tracking (MST) in recreational waters.

Methods and Results:  Avian, mammalian and human faecal samples were screened for the occurrence of Enterococcus avium, Enterococcus casseliflavus, Enterococcus durans, Enterococcus gallinarum, Enterococcus faecium, Enterococcus faecalis, Enterococcus hirae and Enterococcus saccharolyticus using multiplex PCR. Host-specific patterns of Enterococcus species presence were observed only when data for multiple Enterococcus species were considered in aggregate.

Conclusions:  The results suggest that no single Enterococcus species is a reliable indicator of the host faecal source. However, Enterococcus species composite ‘fingerprints’ may offer auxiliary evidence for bacterial source identification.

Significance and Impact of Study:  This study presents novel information on the enterococci species assemblages present in avian and mammalian hosts proximate to the nearshore ocean. These data will aid the development of appropriate MST strategies, and the approach used in this study could potentially assist in the identification of faecal pollution sources.


Bacterial pollution at recreational beaches is an increasing global problem (Bartram and Rees 2000) that threatens human health as well as the tourism and recreation industries (Freeman 1995). Microbial water quality is measured and regulated based on faecal indicator bacteria (FIB), including total coliforms, Escherichia coli and enterococci. The FIB standards for United States (US) recreational waters were promulgated based on a handful of epidemiological studies conducted at sewage-impacted beaches (Boehm et al. 2009). High FIB levels result in tens of thousands of beach closures and advisory days each year (Dorfman and Stoner 2007); however, it is well understood that most nonhuman animal faeces also contain FIB (Ashbolt et al. 2001; Aarestrup et al. 2002). A recent epidemiological study conducted at a beach where poor microbial water quality was attributed to nonhuman sources indicated no correlative relationship between FIB and human health risks (Colford et al. 2007). Thus, there is growing concern that beach closures and advisories can be overprotective when FIB is from nonhuman sources.

To determine the origin of FIB in environmental waters, numerous microbial source tracking (MST) methods have been developed in recent years, each with its own set of strengths and drawbacks (Simpson et al. 2002; Field and Samadpour 2007; Boehm et al. 2009). According to an international panel of water quality experts recently convened by the US Environmental Protection Agency, more research is needed to evaluate the efficacy of existing MST methods and their applicability to different geographical regions (USEPA 2007). The present study addresses this need by investigating the usefulness of identifying Enterococcus to the species level for MST, a method proposed by several authors (Wheeler et al. 2002; Harwood et al. 2004; Kuntz et al. 2004). Enterococcus faecalis and Enterococcus faecium are the predominant enterococci in human faeces and sewage (Murray 1990; Ruoff et al. 1990; Manero et al. 2002). However, these species are also present in the faeces of nonhuman animals (Devriese et al. 1987; Aarestrup et al. 2002; Kühn et al. 2003), including wildlife (Mundt 1963). Efforts to validate the use of Enterococcus speciation for MST applications have included the Atlantic, Pacific and Gulf coasts of the US, particularly Florida (Harwood et al. 2004; Bonilla et al. 2006), Southern California (Ferguson et al. 2005), and Georgia (Wheeler et al. 2002). Most Enterococcus speciation studies rely on time-intensive strain isolation and biochemical tests. A relatively new technique using enrichment cultures to screen for an Enterococcus DNA target has been gaining more widespread use (Scott et al. 2005; Betancourt and Fujioka 2009; Layton et al. 2009). The present study builds on previous work by screening enrichment cultures for Enterococcus species-specific DNA.

Multiplex PCR is a relatively new addition to the MST toolbox. Recent studies have developed multiplex PCR assays to detect mitochondrial (Caldwell et al. 2007) and viral (Wolf et al. 2008) targets of faecal origin. The present study sought to combine existing MST techniques with an established multiplex PCR protocol to rapidly and simultaneously identify multiple Enterococcus species from faecal samples. The study had two objectives: (i) to compare Enterococcus species assemblages present in human and nonhuman animal faeces using multiplex PCR and (ii) to investigate the usefulness of differentiating Enterococcus species as a source tracking method in Northern California. This work contributes to a basic understanding of enterococcal diversity in animal faeces. Furthermore, this is the first study to use exclusively molecular methods to identify Enterococcus species distribution among hosts.

Materials and methods

Sample collection

Eighty-four faecal samples were collected from six host species in the San Francisco Bay Area during June–August 2007 (nonhuman animals) and November 2007 (humans). The faecal donors included 12 healthy human volunteers, 16 dogs at a local dog park, 16 horses from a local farm, 22 seagulls at an urban beach, and 4 seals and 14 sea lions at The Marine Mammal Center in Sausalito, CA. These hosts were chosen because they are commonly present at or near Northern California beaches and could potentially contribute to bacterial pollution in the nearshore ocean. In addition, two grab samples of primary-treated sewage effluent were collected in January and July 2007 from the Palo Alto Water Quality Control Plant in Palo Alto, CA.

Sample processing

Sewage samples were membrane-filtered in triplicate onto mEnterococcus [Difco/BD, Sparks, MD, USA; Standard Method 9230C (APHA 1992)] or mEI agar (EPA Method 1600). Approximately 0·5 g (wet weight) of each animal faecal sample was combined with Dulbecco’s PBS (GIBCO/Invitrogen, Grand Island, NY, USA) to form a slurry, and then 10-fold dilutions of each faecal slurry were membrane-filtered onto mEnterococcus agar (Difco/BD). The membrane filtration apparatuses were ultraviolet (UV)-disinfected for 5 min in a UV cross-linker (UVP, model CL1000, delivers c. 3000 μW cm−2) between samples, and filtration blanks were run. The effectiveness of this disinfection method has been demonstrated previously (Layton et al. 2009). Agar plates were incubated according to standard methods (EPA Method 1600 and Standard Method 9230C). Membrane filters showing growth of presumptive Enterococcus colony-forming units (CFU) (typically 2–3 filters per faecal sample) were placed in 20 ml tryptic soy broth (TSB) and enriched at 41°C for 4–6 h as described previously (Scott et al. 2005). This step generated 2–3 unique enrichments from each host animal. DNA was extracted from one millilitre of each enrichment culture. Extractions were performed with a QIAamp Min-Elute DNA Spin Kit (Qiagen Inc., Valencia, CA, USA) using the protocol for Gram-positive bacteria.

Control strains

Cultures of eight Enterococcus species were obtained from the American Type Culture Collection (ATCC): Enterococcus avium (ATCC# 14025), Enterococcus casseliflavus (ATCC# 25788), Enterococcus durans (ATCC# 19432), Enterococcus faecalis (ATCC# 19433) Enterococcus faecium (ATCC# 19434), Enterococcus gallinarum (ATCC# 49573), Enterococcus hirae (ATCC# 8043) and Enterococcus saccharolyticus (ATCC# 43076). In addition, the following species were used as negative controls: Streptococcus bovis (ATCC# 33317), Staphylococcus aureus (ATCC# 25923), Salmonella typhimurium LT2 and Enterobacter aerogenes (ATCC# 13048). Purified DNA from monocultures of these strains was used as control templates in PCR.

Enterococci speciation multiplex PCR

Multiplex PCR assays confirming the presence of the genus Enterococcus and identifying eight distinct Enterococcus species were performed as described previously (Jackson et al. 2004), with minor modifications. Several of the species-specific primers from Jackson et al. (2004) were shifted up- or down-stream a few bases to improve specificity, which was assessed using Amplify 3× software (ver. 3.1.4; http://engels.genetics.wisc.edu/amplify) (Table 1). Candidate primer pairs were tested using the control strains. Primers that were modified from Jackson et al. are specified with a ‘B’ at the end of the primer name. The genus primers E1 and E2 (forward and reverse, respectively (Deasy et al. 2000)) were included in the multiplex groups, as well as primers for the following common Enterococcus species: Ent. avium (AV1 and AV2B), Ent. casseliflavus (CA1B and CA2), Ent. durans (DU1B and DU2B), Ent. faecalis (FL1 and FL2), Ent. faecium (FM1B and FM2B), Ent. gallinarum (GA1 and GA2), Ent. hirae (HI1 and HI2B) and Ent. saccharolyticus (SA1 and SA2). These eight species were chosen because of their previously documented presence in animal faeces (Devriese et al. 1987; Fogarty et al. 2003) and the nearshore environment (Ferguson et al. 2005; Bonilla et al. 2006). Species and genus primer pairs were multiplexed as follows: Group 1: E1, E2, FL1, FL2, FM1B, FM2B; Group 2: E1, E2, CA1B, CA2, GA1, GA2; Group 3: E1, E2, SA1, SA2; Group 4: DU1B, DU2B; Group 5: E1, E2, AV1, AV2B, HI1, HI2B. All reactions were performed using Invitrogen Platinum Taq polymerase, an annealing temperature of 55°C, and 1 μl of purified, mixed-template DNA (200–300 ng μl−1); all other reaction conditions were as previously described (Jackson et al. 2004). An Applied BioSystems GeneAmp 9700 thermocycler (Foster City, CA, USA) was used for all reactions. A no-template control was included in each PCR run, and duplicate reactions were performed with each DNA sample. PCR products were analysed on a 1·5% agarose gel containing ethidium bromide and visualized and photographed using a GelDoc imaging system (Bio-Rad Laboratories, Hercules, CA, USA) and QuantityOne software (ver. 4.6.3, Bio-Rad Laboratories). Each host individual was assigned a positive result for a given Enterococcus species if at least one of the unique enrichments from that sample produced an amplicon.

Table 1.   PCR primers and multiplex groups used in this study
Primer name*Sequence (5′-3′)Multiplex groupTarget taxonTarget genePurposeAmplicon length (bp)Position†
  1. *Primers E1 and E2 are from Deasy et al. 2000; all others are from, or based on, Jackson et al. (2004). The primers with names ending in ‘B’ were modified for this study.

  2. †The position of the primer on the sodA gene relative to the leading strand in the 5′-3′ direction. Positions in parentheses are the locations of the original primers published in Jackson et al. (2004). Priming positions were determined using the following reference sequences in GenBank (all are 438 bp): AJ387906-7; AJ387911-3; AJ387915-6; and AJ387920.

E1TCA ACC GGG GAG GGT1, 2, 3, 5Enterococcus16S rRNAGenus733
AV1GCT GCG ATT GAA AAA TAT CCG5Enterococcus aviumsodASpeciation36170–90
CA1BACA GTT GAA GAA TTA TTA GCA GAC TTT T2Enterococcus casseliflavussodASpeciation269106–133 (87–107)
DU1BACT GAT ATT AAG ACA GCG GTA CAA4Enterococcus duranssodASpeciation286145–168 (142–162)
DU2BAGA TAG GTG TTT GGC CTG TCA T409–430 (418–438)
FL1ACT TAT GTG ACT AAC TTA ACC1Enterococcus faecalissodASpeciation36049–69
FM1BACA ATA GAA GAA TTA TTA TCT G1Enterococcus faeciumSodASpeciation214106–127 (100–121)
GA1TTA CTT GCT GAT TTT GAT TCG2Enterococcus gallinarumsodASpeciation190118–138
HI1CTT TCT GAT ATG GAT GCT GTC5Enterococcus  hiraesodASpeciation186121–141
HI2BAAA TTC TTC CTT AAA TGT TGC286–306 (287–307)
SA1AAA CAC CAT AAC ACT TAT GTG3Enterococcus saccharolyticussodASpeciation35037–57

Multiplex mixed-template control

The ability of the PCR multiplexes to detect their targets in a mixed culture was assessed. Ten microlitres each of stationary-phase cultures of Ent. avium, Ent. casseliflavus, Ent. durans, Ent. faecalis, Ent. faecium, Ent. gallinarum, Ent. hirae and Ent. saccharolyticus were combined and diluted 10-fold in PBS. Three appropriate dilutions (producing 101–103 colonies) were membrane-filtered in duplicate and grown on mEnterococcus agar (Difco). Each of the six filters was then enriched in TSB (Difco) as described above. DNA was extracted from the enrichment cultures and screened with multiplex PCR for all eight Enterococcus species.

Isolates from filters

To determine whether an individual Enterococcus species initially present on the membrane filter could still be detected in the resulting enrichment culture, random isolated colonies of presumptive enterococci were picked from six sewage filters prior to enrichment, being careful to leave biomass from each colony on the filter. The six filters chosen ranged in density from 10 to 99 CFU. The selected colonies were streaked on mEnterococcus agar and then subcultured into TSB. DNA was extracted from the isolates, using the same protocol as for the enrichment cultures, and screened with multiplex PCR.

Statistical analyses

Fisher’s exact tests were used to examine the observed distributions of Enterococcus species presence between sample types. The results were deemed statistically significant if < 0·05. Similarity scores between Enterococcus species ‘fingerprints’ present in host faeces were calculated using Dice’s coefficients and assembled into a dendrogram using the UPGMA algorithm. The Fisher’s exact tests were performed using R (ver. 2.10.1; http://www.R-project.org); all other statistical analyses were conducted with Matlab (ver. 7.3.0 R2006b; The Mathworks, Inc., Natick, MA, USA).


Preliminary work in silico revealed several potential cross-reactions between previously published multiplex primer sets (Jackson et al. 2004) and nontarget sequences. For example, the FM and DU primer pairs both had potential to bind to the Ent. casseliflavus sodA sequence, with the potential DU1/DU2 amplicon just 11 bp larger than the intended Ent. durans target amplicon. Similarly, CA1 and GA2 (which are in the same multiplex group) showed potential for binding to the Ent. faecium sequence, possibly producing amplicons of intermediate size between the intended Ent. casseliflavus and Ent. gallinarum amplicons (31 bp larger than the latter). While these primers’ affinity for nontarget sequences was certainly weaker than for the intended priming sites, and therefore not guaranteed to produce nonspecific amplicons in vitro, these potential reactions were of concern because of the highly heterogeneous nature of the DNA templates used in the present study. Accordingly, the positions of several primers were adjusted slightly to eliminate potential nonspecific amplification (Table 1). When the new primers were tested in multiplex reactions against the eight Enterococcus target species as well as various nontarget species, their specificity was satisfactory (Fig. 1).

Figure 1.

 Electrophoresis gels of Enterococcus species multiplex PCR products demonstrating the performance of the modified primers. (a, b): Representative mixed-template control culture and American Type Culture Collection controls for each multiplex group. (c, d): Sewage samples showing amplification with the Enterococcus saccharolyticus and Enterococcus durans primers. Duplicate reactions were run in adjacent lanes. Multiplex groups are indicated at the bottom of each gel. The topmost band at ∼733 bp in (a–c) is the product of the Enterococcus genus primers (Table 1). L, 100 bp ladder; NTC, no template control; M, mixed-template control; S, sewage; Efm, Enterococcus faecium; Efl, Enterococcus faecalis; Eca, Enterococcus casseliflavus; Ega, Enterococcus gallinarum; Esa, Ent. saccharolyticus; Edu, Ent. durans; Eav, Enterococcus avium; Ehi, Enterococcus hirae; *nonspecific amplification; †primer dimers. The black/white contrast of the gel images was digitally reversed for clarity in reproduction.

When the ability of the species-specific PCR multiplexes to detect their targets in a mixed-template control was assessed, all targets except Ent. durans and Ent. saccharolyticus amplified (Fig. 1a,b). Nevertheless, the Ent. durans and Ent. saccharolyticus targets were detected in enrichment cultures from faeces (Ent. durans) and sewage (Ent. durans and Ent. saccharolyticus) (Fig. 1c,d). Fewer Enterococcus species were detected in mixed-template control enrichments that originated from membrane filters with <30 CFU. Further post hoc analysis of all samples revealed a logarithmic relationship between the number of CFU enriched and the number of species detected (Fig. 2). Based on this model, enriching at least 30 CFU captured a reasonable estimate (∼70%) of the potential Enterococcus species detectable in the sample. Therefore, a minimum threshold of 30 CFU was required for a sample to be included in the analysis. Multiple unique enrichment cultures from each host individual produced consistent results in each species assay c. 84% of the time. All filtration blanks, DNA extraction blanks and PCR no-template control reactions were negative.

Figure 2.

 Number of colony-forming units enriched vs number of species detected. The relationship of enterococci CFU enriched vs the number species detected in each sample was modeled using a natural logarithm. The data were binned by half-logs of the number of CFU enriched per individual. Within these bins, means were taken in both x (no. CFU) and y (no. species detected). The error bars on each point show the 95% confidence intervals around the means. The model (= 1·2886 * ln(x) + 2·4998) was then fitted to these points, and the R2 value was 0·928. The dotted and dashed lines indicate the weighted 95% confidence intervals of the model and the prediction curves, respectively.

Enterococcus enrichment cultures from two grab samples of primary-treated sewage were screened with the species-specific multiplex PCR assays. Six Enterococcus species (all but Ent. avium and Ent. saccharolyticus) were detected in the January sample, and all eight species were detected in the July sample. A total of 80 Enterococcus isolates were obtained from six sewage membrane filters prior to enrichment. The number of isolates collected from each filter ranged from 15 to 100% of the total CFU present on the filter. In each case, 100% of the Enterococcus species detected among the isolates (three to four species per filter) were also detected in the resulting enrichment culture. In addition, when the number of isolates selected from the filter represented <∼55% of the total CFU present, 1–2 additional species that were not captured among the isolates were detected in the enrichment.

Of the 84 host faecal samples collected, 62 produced at least 30 presumptive Enterococcus colonies on selective media: 13 (81%) dogs, 12 (55%) gulls, 16 (100%) horses, 8 (57%) sea lions, 4 (100%) seals and 9 (75%) humans. Only results from these 62 samples were included in the analysis and results. The overall faecal enterococci density in each host group was reported previously (Layton et al. 2009).

Seven of the eight Enterococcus species targets were detected in faecal enrichment cultures (Table 2). When all faecal samples (human and nonhuman) were considered in aggregate, the most frequently detected Enterococcus species were Ent. faecium and Ent. faecalis; Ent. hirae was also present in 44% of samples. All samples were positive with the genus-level Enterococcus primers.

Table 2.   Presence of enterococci species in enrichment cultures from sewage, human faeces and nonhuman animal faeces as determined by multiplex PCR. The n for each faecal source refers to the number of samples with ≥30 CFU of presumptive enterococci
Enterococcus speciesFaecal Source (n)Fisher’s exact P value
Dog (13)Gull (12)Horse (16)Sea Lion (8)Seal (4)Human (9)Sewage (2)
Enterococcus avium6 (46%)3 (25%)1 (6%)2 (25%)3 (75%)1 (11%)1 (50%)0·036
Enterococcus casseliflavus2 (15%)3 (25%)14 (88%)0 (0%)2 (50%)0 (0%)2 (100%)0·000
Enterococcus durans4 (31%)1 (8%)1 (6%)5 (63%)2 (50%)3 (33%)2 (100%)0·005
Enterococcus faecalis9 (69%)11 (92%)0 (0%)6 (75%)4 (100%)7 (78%)2 (100%)0·000
Enterococcus faecium9 (69%)4 (33%)11 (69%)2 (25%)1 (25%)9 (100%)2 (100%)0·003
Enterococcus gallinarum5 (38%)1 (8%)3 (19%)2 (25%)1 (25%)3 (33%)2 (100%)0·183
Enterococcus hirae7 (54%)1 (8%)16 (100%)2 (25%)0 (0%)1 (11%)2 (100%)0·000
Enterococcus saccharolyticus0 (0%)0 (0%)0 (0%)0 (0%)0 (0%)0 (0%)1 (50%)0·031

Enterococcus species presence varied between sources. For example, Ent. hirae was detected in all horses; Ent. faecalis was found in all seal faeces; Ent. casseliflavus was not detected in sea lions; and Ent. saccharolyticus was detected only in sewage. The distributions of Ent. avium, Ent. casseliflavus, Ent. durans, Ent. faecalis, Ent. faecium, Ent. hirae and Ent. saccharolyticus were significantly different across hosts (< 0·05, Fisher’s exact test; Table 2). Ent. gallinarum was detected at consistent frequencies in each host; hence, its distribution was not significantly different (= 0·183, Fisher’s exact test).

Each sample was assigned a binary ‘fingerprint’ based on the presence or absence of each Enterococcus species. Given the results of the Fisher’s exact tests, several fingerprint schemes were constructed, excluding species with the least significantly different host distributions. To compare these different schemes, mean within-host distance scores (Dice’s coefficients) were calculated (Table 3). The smallest distance (maximum similarity) scores for all sources were obtained when both Ent. gallinarum and Ent. avium were excluded from the fingerprints; these species had the largest Fisher’s exact P-values of all eight species (0·183 and 0·036, respectively; Table 2).

Table 3.   Mean within-host distance scores using three different fingerprinting schemes. The scores are the means of the Dice’s coefficients for each possible pair of individuals in each host group; a distance score of zero would mean that all fingerprints in that host group were identical. For each scheme, the average distance score across all hosts was also calculated
Fingerprint schemeDogGullHorseSea lionSealHumanSewageMean
All eight Enterococcus species0·5340·4290·2300·5960·4570·3410·1430·390
No Enterococcus gallinarum0·5110·4200·1920·5990·4200·2860·1670·371
No Enterococcus avium or Ent. gallinarum0·4950·3790·1740·5630·4110·2610·0910·339

Accordingly, fingerprints based on the remaining six species were compared across all samples and assembled into a dendrogram (Fig. 3). Three groups and two singletons that are <50% similar to each other emerged. Two sea lions did not cluster into any of the groups (singletons); these samples had amplification with few or none of the species-specific markers. Group 1 is composed of 8/12 gulls and 7/12 marine mammals, as well as 4/13 dogs. Group 2 contains all 16 horses and no other samples. Group 3 consists of 9/9 human, 2/2 sewage and 9/13 dog fingerprints, as well as 3/12 marine mammal and 4/12 gull fingerprints. While Group 2 was the only exclusively host-specific group, Group 1 contained fingerprints from primarily marine hosts (gulls, sea lions and seals), and Group 3 contained predominantly human, sewage and dog fingerprints.

Figure 3.

 Dendrogram showing similarity between enterococci species presence profiles in faeces and sewage. Distance scores were calculated using Dice’s coefficient, and the dendrogram was assembled using the UPGMA linkage algorithm.


To date, the Enterococcus species diversity in avian and mammalian hosts adjacent to coastal areas has not been fully characterized. This study provides a snapshot of the predominant Enterococcus species populations in gulls, dogs, horses, seals, sea lions and humans proximate to the Northern California coast. These data underscore the ubiquitous nature of enterococci in animal faeces and sewage, while revealing host-specific variations in the composition of species assemblages. Nearly every species of Enterococcus that was tested for was found in each host examined, which suggests that no single species of Enterococcus is a reliable indicator of human faeces.

Significantly different patterns of Enterococcus species presence between hosts were identified, although when individual ‘fingerprints’ were assembled into a dendrogram, horses were the only host group that formed an exclusive cluster. The observed groupings may be driven by similarities in the hosts’ diet and/or habitat. For example, high similarities were seen among the marine hosts as well as between humans and dogs. The observed variation between the sewage samples may be because of temporal separation: fewer species were detected in the winter than in summer. Repeated faecal sampling across seasonal timescales would be informative, as the diet and/or habitat (and therefore the intestinal microbiota) of the hosts may change with time. The groupings could also be verified by screening a broad range of host species in various locations.

The robustness of the fingerprints could be further optimized by evaluating the host distribution of other Enterococcus species beyond the eight considered here, as it was evident that additional Enterococcus species were likely present. For example, one avian (seagull) faecal sample did not produce any PCR products when screened with the species-specific primers, although it did generate amplicons with the Enterococcus genus primers. This suggests that this individual carried enterococci other than the species targeted in this study; possibilities include Enterococcus columbae or Enterococcus cecorum, which have been detected previously in avian hosts (Aarestrup et al. 2002). Inclusion of these or other Enterococcus species could potentially improve the observed fingerprint clustering among the gulls (Fig. 3).

The results from the mixed-template control experiment highlight a drawback of using enrichment cultures followed by PCR for Enterococcus speciation. Different species may have different growth or recovery rates on selective media and TSB (Lleo et al. 2001), leading to some species out-competing others in the enrichment. In addition, the various PCR assays may have different sensitivities in complex mixtures of target and nontarget DNA. However, the enrichment technique is consistent with the method being used for the Ent. faecium esp gene human faecal marker assay (Scott et al. 2005; Yamahara et al. 2007; Layton et al. 2009) and provides a straightforward way to rapidly and simultaneously detect multiple species in a sample. While the PCR multiplexes used in this study provide a reasonable estimate of the species present in a particular sample, the mixed-template control experiment indicated that false negative results for Ent. durans and Ent. saccharolyticus are possible.

Despite the drawbacks of the method, multiple unique enrichment cultures from each host individual produced consistent results in each species assay a great majority (84%) of the time. Given the expected heterogeneity among organisms initially present in replicates of membrane filtration, this level of agreement is quite satisfactory. The validity of the method was further confirmed by subculturing isolates from sewage membrane filters prior to enrichment. Each species that was isolated was also detected in the corresponding enrichment culture, and for more densely populated filters, additional species were detected in the enrichment culture that were not among the randomly selected isolates. These results suggest that despite the potential for enrichment bias, the method still provides a reasonable snapshot of the enterococci species present in a given sample.

The distribution of Enterococcus species presence among hosts observed in the present study differs from previous works. For example, Wheeler et al. (2002) reported very low frequencies of Ent. gallinarum in humans and dogs, whereas the PCR multiplexes used here detected Ent. gallinarum in 38% of dogs and 33% of humans (Table 2). Similarly, Wheeler et al. (2002) did not detect Ent. avium or Ent. durans in dogs or humans, while both of these Enterococcus species were observed in dogs and humans in this study. However, other studies of Enterococcus species differentiation (Wheeler et al. 2002; Harwood et al. 2004; Ferguson et al. 2005) utilized biochemical assays; thus, it may not be possible to directly compare those results to the data presented here. To date, this is the first study to use exclusively molecular methods to determine Enterococcus species distribution in human and nonhuman animal faeces.

It is notable that Ent. casseliflavus was detected in 21 (34%) of 62 host individuals, as well as both sewage samples, as this species is considered epiphytic (Mundt and Graham 1968; Müller et al. 2001). It is conceivable that this organism may be incorporated into the microbiota of the digestive tract after consumption. The present study supports the idea that the presence of Ent. casseliflavus in faeces could be attributable to host diet: Ent. casseliflavus was detected most frequently in horses, the only exclusive herbivore that was sampled. Nevertheless, these data illustrate that detection of Ent. casseliflavus in the environment cannot be attributed exclusively to nonfaecal sources.

The analysis shown here suggests that Enterococcus species assemblages may be conserved within host species, which supports previous findings (Kühn et al. 2003). Additional insight into host-specific Enterococcus assemblages – including relative abundances – could be obtained by extracting DNA directly from water or faeces, with no enrichment step. The method could then potentially be adapted for quantitative PCR or high-throughput sequencing. The host specificity of Enterococcus assemblages suggests possible MST applications for Enterococcus species fingerprints; however, this prospect is confounded by several factors. Both differential survival of Enterococcus species in the environment and variable recovery from the VBNC state have been observed (Lleo et al. 2001, 2005); thus, the ability to detect each species could vary depending on environmental conditions and the duration of environmental exposure. Additional work is needed to determine whether host-specific fingerprints could be discerned among various ratios of mixed faecal inputs and ‘background’ enterococci in the environment. The ability to parse mixed fingerprints in the environment could be further complicated by the presence of all eight species in sewage (Table 2); detection of all eight species in the environment could indicate either sewage or a mixture of animal inputs. This problem could potentially be rectified by utilizing a different combination of Enterococcus species in the fingerprints. Applying species fingerprinting as an MST method would likely require the creation of a fingerprint library for different geographical locations and thus may be prohibitively expensive. Lastly, species fingerprints could only offer information on the faecal source and not the risk to health.

The work presented here offers valuable information on faecal Enterococcus assemblages that is pertinent to recreational water quality monitoring. It is evident from this and other studies that no single species of Enterococcus is an accurate indicator of human faecal contamination and that detection of a given species can neither confirm nor rule out a specific source. Finally, these data illustrate the cosmopolitan nature of the genus in hosts proximate to the nearshore environment.


This project was funded through the NOAA Oceans & Human Health initiative (Grant NA04OAR4600195) and the National Science Foundation (Grant OCE-0742048); BAL is also supported by a National Science Foundation Graduate Fellowship. The authors are grateful to Tracey Goldstein and Phil Bobel for their assistance in obtaining faecal and sewage samples.