Environ. Sci. Technol. 2007, 41, 6090-6095
Incidence of the Enterococcal Surface Protein (esp) Gene in Human and Animal Fecal Sources RICHARD L. WHITMAN,† KATARZYNA PRZYBYLA-KELLY, DAWN A. SHIVELY, AND M U R U L E E D H A R A N . B Y A P P A N A H A L L I * ,† U.S. Geological Survey, Great Lakes Science Center, Lake Michigan Ecological Research Station, 1100 Mineral Springs Road, Porter, Indiana 46304
The occurrence of the enterococcal surface protein (esp) gene in the opportunistic pathogens Enterococcus faecalis and E. faecium is well-documented in clinical research. Recently, the esp gene has been proposed as a marker of human pollution in environmental waters; however, information on its relative incidence in various human and animal fecal sources is limited. We have determined the occurrence of the esp gene in enterococci from human (n ) 64) and animal (n ) 233) fecal samples by polymerase chain reaction using two primer sets: one presumably specific for E. faecium (espfm) and the other for both E. faecalis and E. faecium (espfs/fm). We believe that this research is the first to explore the use of espfs/fm for the detection of human waste in natural environmental settings. The incidence in human sources was 93.1% espfm and 100% espfs/fm in raw sewage influent; 30% for both espfm and espfs/fm in septic waste; and 0% espfm and 80% espfs/fm in active pit toilets. The overall occurrence of the gene in animal feces was 7.7% (espfs/fm) and 4.7% (espfm); animal types with positive results included dogs (9/43, all espfm), gulls (10/34, espfs/fm; 2/34, espfm), mice (3/22, all espfs/fm), and songbirds (5/55, all espfs/fm). The esp gene was not detected in cat (0/34), deer (0/4), goose (0/18), or raccoon (0/23) feces. The inconsistent occurrence, especially in septic and pit toilet sewage, suggests a low statistical power of discrimination between animal and human sources, which means a large number of replicates should be collected. Both espfm and espfs/fm were common in raw sewage, but neither one efficiently differentiated between animal and other human sources.
Introduction Escherichia coli and enterococci have been recognized as fecal indicators of recreational water quality worldwide; however, identifying their sources (i.e., human versus nonhuman) has been a challenging task. The traditional culturing methods used in monitoring programs are not designed to differentiate contamination sources. As a result, numerous alternate methodsspopularly referenced as microbial source tracking (MST)shave been proposed recently for identifying the sources of contaminants in recreational † R. L. Whitman and M. N. Byappanahalli contributed equally to this work. * Corresponding author. Phone: (219) 926-8336; Fax: (219) 9295792.
6090 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 17, 2007
water. Their applications in routine beach monitoring programs have been hampered, because, typically, they are expensive, require well-equipped labs and trained personnel to run the assays, need reference libraries, are limited regionally, and have had variable success (1, 2). Consequently, it would be advantageous to have simple and inexpensive tools for differentiating human and animal sources (3). Alternate methods that circumvent reference libraries have been tried with varying degrees of success; specific examples include virulent markers, coliphage analysis, and humanspecific bacterial groups (e.g., Bacteroides) (4-9). A recent addition to this list is the enterococcal surface protein (esp) gene (6, 9, 10). Enterococci are gram-positive bacteria found in the gastrointestinal tract of humans and warm-blooded animals. In human feces, Enterococcus faecalis and E. faecium comprise almost 90% of Enterococcus spp. present. They are opportunistic pathogens, often associated with nosocomial infections. The esp gene was originally discovered in E. faecalis (11) and subsequently in E. faecium (12, 13). This gene has been linked with pathogenesis, possibly through tissue colonization and biofilm formation, increased virulence, and immune evasion (11-16). Recently, it has been posited by Scott et al. (9) that the presence of the E. faecium esp gene (espfm) is a reliable indication of the presence of sewage because it is rarely found in animal feces. While the early investigations are encouraging, studies that confirm and extend this hypothesis to a wide range of animals and human fecal sources are lacking. The present study estimates the relative incidence of the esp gene in enterococci derived from humans, domesticated animals, and wildlife of the southern Lake Michigan area. We used two sets of primers: one specific for targeting the esp gene derived from E. faecium (espfm) (9), and one generic for targeting the esp gene derived from both E. faecalis and E. faecium (espfs/fm) (11). To our knowledge, the use of the latter has been limited to clinical settings and has not been explored for potential application in water quality. Overall, a total of 594 assays on 233 animal and 64 human fecal samples were performed to detect espfm and espfs/fm.
Materials and Methods Sample Collection. Samples from human and animal fecal wastes were collected between May 2006 and March 2007. All samples were transported to the laboratory on ice and were analyzed within 4-6 h of collection. Human Waste. A total of 29 sewage influent samples (∼250 mL each) were collected from sewage treatment plants located in Indiana (Chesterton, South Haven, Indiana Dunes State Park, Valparaiso, Gary, Michigan City, LaPorte, Westville, and Kingsford Heights). In addition, individual final effluent samples were collected from the Chesterton, Valparaiso, Gary, and Indiana Dunes State Park treatment plants (postultraviolet treatment for the first three). Fifteen samples from pit toilets were collected from several camping and recreational areas at the Indiana Dunes State Park in Chesterton. Samples were collected by submerging a cotton swab into the pit waste, swirling several times to achieve a homogeneous sample, and transferring the swab directly into a sterile 15-mL centrifuge tube that contained 6 mL of phosphate-buffered water (PBW). In addition, 20 250-mL samples were collected from four different septic trucks (3-8 replicates from each) at a nearby treatment plant. Animal Waste. All animal fecal samples were collected between June 2006 and August 2006. Thirty-four cat and 43 dog fecal samples were collected from animal boarding 10.1021/es070817t Not subject to U.S. copyright. Publ. 2007 Am. Chem. Soc. Published on Web 08/08/2007
facilities located in Michigan City, IN and Valparaiso, IN. Fecal samples from wildlifesspecifically, deer (n ) 4), mice (n ) 22), and raccoons (n ) 23)swere collected from forested areas in and around the Indiana Dunes National Lakeshore in Porter, IN. In addition, fecal samples were collected from shoreline birds (seagulls (n ) 34) and geese (n ) 18)) and songbirds (n ) 55; 19 species from order Passeriformes and 3 species from order Piciformes). Songbirds were initially trapped in mist nets and transferred to holding bags, and fecal droppings that were collected from the bags were then transferred to 15-mL centrifuge tubes that contained 6 mL of PBW. Samples from shoreline birds were collected by gently rolling a cotton swab over fresh fecal deposits several times; the swab was then transferred to a 15-mL centrifuge tube that contained 6 mL of PBW. Sample Processing and Microbiological Analysis. Sewage and septic waste samples were diluted as necessary to achieve enterococci counts generally in the range of 150-500 colonyforming units, cfu/filter for later enrichment (see below) in azide dextrose broth (ADB) (Difco Laboratories, Sparks, MD). Animal feces from dogs, cats, raccoons, geese, gulls, and deer were processed as follows. Three swabs of the fecal material (∼0.3-0.5 g) were aseptically transferred to 15-mL centrifuge tubes that contained 6 mL of PBW, vortexed vigorously, and then diluted as necessary to achieve desirable counts (see above) of enterococci. For mice samples, several fecal pellets were transferred to 15-mL centrifuge tubes that contained 6 mL of PBW; the contents were vigorously mixed and then diluted as necessary to achieve desirable counts of enterococci. Sewage and fecal samples were analyzed for enterococci by two different methods: (1) membrane filtration (MF) followed by enrichment in ADB (n ) 297); and (2) direct enrichment in ADB, without initial MF (see below; n ) 224). It should be emphasized that ADB is one of the commonly used media for isolating/enriching enterococci from environmental samples (17). MF Method. Aliquots (0.5-5.0 mL) of wastewater and diluted fecal samples were first filtered through 0.45-µm membrane filters. The filters were then placed onto m Enterococcus Agar (Difco Laboratories) supplemented with indoxyl (0.75 g/L) and incubated at 41.5 °C for 24 h. Filters with typical enterococci coloniessdistinguishable by blue halos around the coloniesswere then transferred into 15mL tubes that contained ADB and were further enriched for 4 h at 35 °C. From these tubes, 1-mL aliquots were used as a template for polymerase chain reaction (PCR) (see below). Direct Enrichment Method. Aliquots of raw, undiluted, well-homogenized fecal material (0.1-1.0 mL) were added to 6 mL of single- or double-strength ADB and incubated at 35 °C for 24 h. An aliquot (1 mL) of the enriched broth was later used as a template for PCR. Template for PCR Reactions. Aliquots (1 mL) of enriched enterococci cultures from ADB broth were transferred to 1.5mL centrifuge tubes and centrifuged at 12 000 rpm for 5 min. After discarding the supernatant, resulting pellet (in each tube) was washed (two times) with PCR-grade water and later suspended in 300 µL of AE elutriation buffer (Qiagen Gmbh, Hilden, Germany). The cells were then thoroughly homogenized by vigorous mixing. A quantity of 5 µL of this homogenized suspension was used as the template for PCR. PCR Primers and Reaction Conditions. In this study, two sets of primers were used for PCR amplification of the esp gene. The primer design and PCR protocols have been explained in detail elsewhere (9, 11). The primer sequences were esp11 5′-TTG CTA ATG CTA GTC CAC GAC C-3′ (forward) and esp12 5′-GCG TCA ACA CTT GCA TTG CCG AA-3′ (reverse) for espfs/fm (11), and 5′-TAT GAA AGC AAC AGC ACA AGT T-3′ (forward) and 5′-ACG TCG AAA GTT CGA TTT CC-3′ (reverse) for espfm (9).
PCR reactions were performed in a 50-µL reaction mixture. For espfs/fm: 1× PCR buffer, 2.5 mM MgCl2, 200 µM each of the four dNTPs, 0.2 µM of each forward and reverse primers, 2.5 U of AmpliTaq DNA polymerase (Promega Corporation, Madison, WI), and 5 µL of the template. For espfm: 1× PCR buffer, 1.5 mM MgCl2, 200 µM each of the four dNTPs, 0.3 µM of each primer, 2.5 U of AmpliTaq DNA polymerase, and 5 µL of the template. The PCR amplification conditions were as follows: initial denaturation for 2 min at 95 °C, followed by 30 cycles of denaturation, 94 °C for 45 s, annealing, 63 °C for 45 s, and extension, 72 °C for 4 min (espfs/fm); and initial denaturation at 95 °C for 10 min, followed by 35 cycles of denaturation, 94 °C for 1 min, annealing, 58 °C for 1 min, and extension, 72 °C for 1 min (espfm). Positive and negative control strains were included with each set of amplifications. For espfs/fm, the positive and negative strains were 594 and OG1RF (kindly provided by Nathan Shankar, University of Oklahoma, Oklahoma City, OK). For espfm, the positive control was either sewage, where this gene has consistently been found (9), or E. faecium C68 (kindly provided by Joan Rose, Michigan State University, East Lansing, MI); the no-template condition (5 µL of PCRgrade water) served as the negative control. PCR products were separated on a 1.5% agarose gel (prepared in 0.5% Tris-Borate-EDTA, TBE buffer) stained with ethidium bromide and viewed under UV light. The PCR products for espfs/fm and espfm were 680 and 950 bp, respectively. Gel images were captured using the Kodak Gel Logic 100 Imaging System. Band positions were determined with respect to the standard, 1-kb DNA ladder. Statistical Analyses. The esp gene incidence rates were superior (for both espfs/fm and espfm) if the samples were first selectively cultured by the MF method, followed by enrichment in ADB, rather than culturing samples directly in ADB without the initial MF step (refer to microbiological analysis previously described). Note that all statistical analyses and illustrations (the table and graphs in this manuscript) have been performed on data obtained for samples that were processed via the MF method, followed by enrichment in ADB. All statistical analyses were performed using SPSS v.12 and SYSTAT 11 statistical software. Fisher’s exact test was performed to compare the esp frequency between sample sources; data were weighted prior to analysis. Individual p-values (P < 0.05) were adjusted according to sequential Bonferroni (18). For fecal wastes derived from human sources, we chose to set the expected probability of its occurrence at the level of at least 70%. Consequently, in fecal wastes from nonhuman sources (e.g., animals and wildlife), the expected probability of its occurrence would be up to 30%. In this study, the 70% (+)/30% (-) cutoff is a compromise between discrimination power and practice. Higher cutoff (for instance, a requirement that 90% were positive to confidently classify that a sample had a human source component) would have been impractical, requiring far too many samples to discriminate sources (see confidence intervals in Figure 2, presented later in this work). A one-tailed binomial distribution test was used to calculate the probability of esp incidence of 0.7 (H0: p > 0.7, where p is the probability of esp occurrence in human samples) and 0.3 (H0: p < 0.3, where p is the probability of esp occurrence in animal samples). Exact binomial 95% confidence intervals were calculated according to Clopper and Pearson (19) using the formula
(θ; P[Bin(n;θ) e X] g R/2) ∩ (θ; P[Bin(n;θ) g X] g R/2) where X is the number of successes observed in the sample and Bin(n;θ) is a binomial random variable with n trials and VOL. 41, NO. 17, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6091
TABLE 1. Results of Fisher’s Exact Test on Frequency of the espfm and espfs/fm Gene between Tested Fecal Sourcesa source
number of samples
number of positivesb
29 15 20 34 4 43 18 34 22 23 55
27 0 6 0 0 9 0 2 0 0 0
29 15 20 34 4 43 18 34 22 23 55
29 12 6 0 0 0 0 10 3 0 5
sewage influent
pit toilets
septic trucks
cats
deer
S
dogs
geese
gulls
mice
raccoons
songbirds
NS
S
S
S
S
S
NS
NS
S
S
S
S
S S
S NS
S S
S S
S
S S
S NS
NS
NS
S
S
espfm Gene sewage influent pit toilets septic trucks cats deer dogs geese gulls mice raccoons songbirds
S
S
S S S S S S S S S
S S
S S S S S S S S
espfs/fm Gene sewage influent pit toilets septic trucks cats deer dogs geese seagulls mice raccoons songbirds
S S S S S S S S
S
S
S S S S S S S S S
S S S S NS S S S
S
S represents a significant P value adjusted by sequential Bonferroni: P < 0.05 for the espfm gene and P < 0.046 for the espfs/fm gene, whereas NS represents a nonsignificant P. Empty cells represent inappropriate or redundant comparisons. b For the espfm gene, the number of positives is represented by espfm+; for the espfs/fm gene, the number of positives is represented by espfs/fm+. a
probability of success θ. To determine the statistically reliable sample size, power analysis was performed using a onesided single proportion model with R ) 0.05, β ) 0.8.
Results The esp Gene in Human Fecal Sources. The overall incidence in human fecal samples was 73.4% (47/64) for espfs/fm and 51.6% (33/64) for espfm. The relative frequency of espfs/fm and espfm among the three fecal sources (i.e., sewage influent, pit toilets, and septic trucks) ranged from 30% to 100% and 0% to 93.1%, respectively. The frequencies of both espfs/fm and espfm in all human-derived samples were significantly different from one another (Table 1). Interestingly, there was no detection of the espfm signal from pit toilet samples. Also, both espfs/fm and espfm occurred in the same samples collected from four septic trucks, and their occurrence was incidental among the replicates: 3/3, 1/6, 1/8, 1/3. Based on the binomial distribution, we calculated the probability of esp incidence making an assumption of esp occurrence in human sources at 0.7. The espfs/fm positive samples from sewage influent and pit toilets and for espfm samples from influent met this criterion (P < 0.05). The esp Gene in Animal Fecal Sources. Among 233 animal fecal samples, the overall incidence was 7.7% (espfs/fm) and 4.7% (espfm). The incidence of esp among animals varied: for dogs, 9/43 (all espfm); for gulls, 10/34 (espfs/fm), 2/34 (espfm); for mice, 3/22 (all espfs/fm); and for songbirds, 5/55 (all espfm/fs) (see Table 1). The esp gene was not detected in cat (0/34), deer (0/4), goose (0/18), or raccoon (0/23) feces. Among the discrepancies of espfs/fm incidence in animal fecal sources, a significant difference was observed in 14/18 (78%) comparisons; in espfm, discrepancies were observed in 10/13 comparisons (77%) (P < 0.05) (see Table 1). Based on the assumption that the expected proportion of esp occurrence in animal feces would be 0.05). 6092
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 17, 2007
FIGURE 1. Comparison of the observed incidence of the espfm and espfs/fm genes in different fecal samples. The diagonal line is for visual purposes only and has no statistical purpose. The esp Gene in Human versus Animal Sources. The relative probabilities of the espfs/fm and espfm occurrence in human and animal fecal sources are depicted in Figure 1. The animals in the group, including cats, deer, geese, and raccoons, were all negative for both espfs/fm and espfm. Two human fecal sources (i.e., sewage effluent and septic truck) had the same incidence rates for espfs/fm and espfm. The incidence rates of both espfs/fm and espfm suggest that animal fecal sources and sewage influent were different from one another. However, overlapping (i.e., at 95% confidence intervals) occurred among all the samples to a different extent,
FIGURE 2. Relative incidence and binomial 95% confidence intervals of (A) the espfm gene and (B) the espfs/fm gene in different fecal sources. creating more uncertainty about correctly classifying the origin of the sample, but the most evident and important overlaps would be the espfm incidence in septic truck samples versus dog feces (Figure 2a) or the espfs/fm incidence in septic truck samples versus gull feces (Figure 2b). Considering the occurrence of espfs/fm exclusively, statistical analysis showed that there was a higher frequency of this gene in human fecal sources, such as sewage influent and pit toilets, relative to all animal fecal sources (P < 0.05) (see Table 1). Furthermore, the espfs/fm prevalence in septic truck samples was significantly different from all animal fecal sources, except gulls (P > 0.05). The espfm frequency was significantly higher in sewage influent and septic truck samples than in the animal fecal sources (P < 0.05). For all comparisons of espfm incidence, human-derived samples were significantly different from animal fecal sources (P < 0.05). Enterococci Culturing Protocols on esp Detection. Two different culturing methods were used for enterococci analysis: the traditional MF technique (mEI), followed by enrichment in ADB; and the direct inoculation of ADB. The method of MF followed by ADB enrichment resulted in a higher esp detection rate for both espfs/fm and espfm, relative to the direct ADB enrichment method: 15.1% vs 8.3% (espfs/fm) and 6.5% vs 0% (espfm) for the two methods, respectively.
Discussion The potential role of the esp gene in opportunistic pathogens (E. faecalis and E. faecium) is well-documented in clinical research. The esp gene seems to be associated with other virulence markers on a pathogenicity island in both species (20, 21), but E. faecalis strains harbor significantly higher virulence determinants than E. faecium (11, 22). Furthermore, E. faecalis is responsible for 80%-90% of all enterococci infections, and E. faecium is responsible for most of the remaining 10%-20%. Recently, there is growing interest in developing PCRbased assays for detection of esp in environmental waters, mainly espfm, for potential use as a marker of sewage contamination (6, 9). Heuer et al. (23) stated that distinguishing human from animal enterococci may be difficult, as revealed in previous findings of Poeta et al. (24), Harada et al. (25) and Hammerum and Jensen (26), who found E. faecium and E. faecalis esp genes among both animal and
human isolates. Our findings confirm the uncertainties associated with the presumptive exclusiveness of the esp gene to human fecal sources. The esp Gene in Human Fecal Sources. The espfm Gene. In the present study, the overall frequency of the espfm gene in human fecal samples was somewhat low (51.6%), perhaps because of the high variation between sample types. Although Scott et al. (9) reported 100% detection rate of espfm in sewage and 80% in septic tanks samples, numerous clinical studies have revealed great variation in espfm detection rates performed on clinical isolates, ranging from 0% to 78% (11, 22, 26-30), which suggests great inconsistency in gene detection. Unfortunately, studies that examine the relative distribution of the espfm gene in nonclinical samples are limited. Our study revealed a 93.1% incidence rate of espfm in sewage influent, but there was none in samples from active pit toilets, where detection would be expected. Perhaps the espfm gene is more commonly detected in sewage, which is a mixture of human waste from a large population. In contrast, wastes from septic trucks and pit toilets represent a small population (i.e., constricted area) and, therefore, have a much lower probability of the esp gene occurrence. Compared to sewage influent samples, the effluent samples were mostly negative for culturable enterococci; one of the five samples (i.e., 20%) was positive for both espfm and espfs/fm. Inconsistency of the espfm or espfs/fm detection in samples obtained from the same septic truck (i.e., one positive per eight replicates) implies that low gene detection may result if an insufficient number of samples are collected. In a related unpublished study, we did not find espfm upstream or downstream of sewage treatment plant after a severe storm even though enterococci counts were reaching 140000 cfu/100 mL downstream. Thus, the reliability of espfm as an indicator of human feces remains unresolved. The espfs/fm Gene. In contrast to espfm, our detection rates for espfs/fm in human samples were higher (i.e., 73.4%), perhaps because of the amplification of the esp gene in both E. faecalis and E. faecium. Although we did not speciate our enterococci, the higher detection rates of espfs/fm in samples from pit toilets (80%) may suggest that concentrations of E. faecalis were higher in this source. Wheeler et al. (31) stated that E. faecalis occurrence was limited to humans and a few other animals, and its ratio might vary substantially between individuals (e.g., the percentage of isolates from human VOL. 41, NO. 17, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6093
samples ranged from 0% to 95% for E. faecalis and 0% to 63% for E. faecium). High variation in E. faecalis esp in clinical studies has also been demonstrated, with detection ranging from 29% to 71% (11, 22, 26, 27, 32). The esp Gene in Animal Fecal Sources. The Espfm Gene. Previous studies on the esp gene occurrence in animals have focused mostly on pets such as cats and dogs. Espfm has previously been detected in 5% of isolates obtained from 39 dogs (25), whereas our findings showed almost 21%. Its incidence rate in animals is difficult to ascertain because of differences in assay procedures. In clinical studies, esp incidence has been determined on pure Enterococcus isolates, in contrast to studying the incidence in a subset of the population (9). Our methods differed in that we analyzed enriched entire fecal subsamples using primers developed by Scott et al. (9), whereas Harada et al. (25) assayed speciated isolates using primers developed by Shankar et al. (11). Approximately 89% of the dogs that tested positive were not strays, which suggests a potential association between owner and pet. Oancea et al. (33) found that esp was easily transferable between enterococci strains. Even dogs from the same household differed in the occurrence of the espfm gene, which suggests that its presence may be incidental and not directly related to host habits. Detection of the espfm gene in the feces of 2/34 seagulls perhaps is related to their (i.e., birds) close association with anthropogenic waste. To date, fecal matter from poultry, swine, dairy cattle, beef cattle, Canada geese, seagulls, pelicans, wild birds (species not cited) (9), cats, and dogs (25) has been examined; only dogs tested positive. Our study contributed deer, mice, and raccoons and detected an espfm signal in 20.9% of dogs and 5.9% of seagull feces. The occurrence of this gene in nonhumans may compromise its host specificity and suggests transfer from one host to another. The Espfs/fm Gene. The majority of work reported in the literature on E. faecalis esp has been conducted on speciated isolates. The esp gene related to E. faecalis was detected in 8% of pigs (26), 4.1% of wild birds (1 bird of prey and 2 owls) (24), 22% of dogs, and 20% of cats (25). In our study, the esp gene was detected in 29% of seagulls, 14% of mice, and 9% of the 22 songbird species. The espfs/fm positive birds were the American robin, the red-headed woodpecker, the redwinged blackbird, the eastern wood-pewee, and the American goldfinch that inhabited a black oak savanna woodland complex. Positive espfs/fm signals in seagull feces could be linked to their contact with human waste; we can only speculate on the gene’s origin in songbird and mouse feces. These animal feces were collected within National Park Service boundaries with no known exposure to human waste. It can be inferred that the esp gene in E. faecalis is rather common among animals, although E. faecalis itself is considered to be a limited host range bacterium, occurring mostly in humans and a few animals (31, 34). The Esp Gene in Human Sources versus Animal Sources. The percentage of waste samples from human sources in which the espfm gene was found far exceeds that of animal feces (51.6% vs 4.7%); however, the question remains: is the gene a reliable marker of human waste? This answer may depend more on the particular sources being compared; sewage represents community waste, which includes human and pet waste and sometimes urban runoff, whereas pit toilets generally have only human waste. Septic trucks are integrated family, perhaps pet, and household wastes but nonetheless should include qualified human markers. For instance, no espfm was found in active pit toilets, and there was low incidence in septic waste, which implies that entire enriched samples do not universally contain this esp gene. A better understanding of the distribution of this gene in humans would be possible with increased sample size and replication. For instance, statistical power analysis demonstrates that 6094
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 17, 2007
we would need to take only 15 samples from sewage influent to confirm their human origin (>0.7 probability of espfm occurrence). Alternately, 141 dog fecal samples would be needed to confirm nonhuman origin of the sample (
