ABSTRACT
In 2012, the U.S. EPA suggested that coastal and Great Lakes states adopt enterococci as an alternative indicator for the monitoring of recreational water quality. Limited information, however, is available about the presence and persistence of enterococci in Lake Superior. In this study, the density, species composition, and persistence of enterococci in sand, sediment, water, and soil samples were examined at two sites in a Lake Superior watershed from May to September over a 2-year period. The genetic diversity of Enterococcus faecalis isolates collected from environmental samples was also studied by using the horizontal, fluorophore-enhanced repetitive PCR DNA fingerprinting technique. Results obtained by most-probable-number analyses indicated that enterococci were present in 149 (94%) of 159 samples and their densities were generally higher in the summer than in the other months examined. The Enterococcus species composition displayed spatial and temporal changes, with the dominant species being E. hirae, E. faecalis, E. faecium, E. mundtii, and E. casseliflavus. DNA fingerprint analyses indicated that the E. faecalis population in the watershed was genetically diverse and changed spatially and temporally. Moreover, some DNA fingerprints reoccurred over multiple sampling events. Taken together, these results suggest that some enterococci are able to persist and grow in the Lake Superior watershed, especially in soil, for a prolonged time after being introduced.
INTRODUCTION
Fecal contamination of recreational waters is a widespread problem across the Great Lakes region of the United States. Because of difficulties and high costs associated with detection and quantitation of fecal pathogens (1), fecal indicator bacteria (FIB) were chosen to assess the potential presence of pathogens. Traditionally, Escherichia coli and fecal coliforms have been used as FIB in freshwater systems, while enterococci were initially used as indicators in marine waters. Previous studies have shown that E. coli can become naturalized to the microbial community in tropical, subtropical, and temperate soil and sand (2–4). This likely limits the use of this bacterium as an indicator of water quality. Moreover, these culture-based methods cannot differentiate among sources of fecal bacteria (5).
The U.S. Environmental Protection Agency (EPA) has suggested that coastal and Great Lakes states adopt enterococci as an alternative indicator of fecal contamination (6). Epidemiological studies have shown that the enterococcal concentration has a strong positive correlation with the risk of gastroenteritis associated with swimming in contaminated freshwater (7). However, the potential advantages of enterococci over E. coli as the indicator of choice to detect fecal contamination of waterways and the environment warrant further investigation (8), especially in the cold climate associated with the Lake Superior watershed.
It has been suggested that some Enterococcus species are primarily of environmental origin (5). Some species, including Enterococcus faecalis, E. faecium, E. casseliflavus, E. hirae, E. mundtii, E. gallinarum, E. durans, E. avium, and E. sulfureus, have been repeatedly isolated from sand, sediment, water, or plants (9–12). In addition, evidence of the persistence of enterococci in municipal oxidation ponds and in microcosms simulating environmental conditions has been previously documented (13, 14). The high level of fecal bacteria in sand, sediment, soil, and submerged vegetation has raised public health concerns (2, 15, 16). Since matrices harboring enterococci may protect these bacteria from inactivation by sunlight and from protozoan predation or offer a range of surfaces for attachment and nutrient acquisition (17–20), it has been suggested that they may serve as reservoirs for FIB (21). However, the majority of studies on the ecology of enterococci have been done in tropical and subtropical environments or in warmer Great Lakes states.
Enterococci are nearly ubiquitous and have been isolated from a variety of substrates (4, 10, 15). Species recovered from marine environments included mainly E. faecalis, E. faecium, E. casseliflavus, E. hirae, and E. mundtii (10, 15, 22). Of these species, E. casseliflavus has been shown to replicate and persist in vegetation-containing microcosms (23). Since the ecology of enterococci in water, sand, and soil may be affected by abiotic and biotic factors (21), it is likely that the concentration and species composition of enterococci in temperate freshwater environments differ from those found in other environments. In the few studies carried out in temperate freshwater environments, enterococci were recovered from backshore beach sand in Lake Michigan from early fall to early summer of the next year, and the dominant species (92%) recovered was E. faecium (9). Another study carried out at Lake Huron found that the density of enterococci in wet sand was higher than that in water (16). But none of the previous studies examined the species composition over time and in different substrates at the study sites.
The bacterium E. faecalis has received much attention because of its ability to survive and grow under a variety of harsh conditions (24). This bacterium is the primary Enterococcus species inhabiting the intestinal tracts of humans and some animals (25) and is the predominant species causing hospital-acquired infections (26). The majority of studies on the survival of E. faecalis have been done in medical fields (27–29), and many environmental studies have examined the survival and genetic diversity of a limited number of Enterococcus species (15, 30). Therefore, there is very limited information on the survival of E. faecalis in the environment. One of the few studies done under lab conditions found that E. faecalis survived longer than E. coli did in sand microcosms (31). Moreover, there is currently little information on the genetic relatedness among E. faecalis strains isolated from different habitats.
The objective of the studies reported here was to examine the occurrence of enterococci in sand, sediment, water, and soil in a Lake Superior watershed. Two different sites were included: the new Duluth Boat Club (DBC) beach site and the Kingsbury Creek bank site. Moreover, the studies also examined changes in Enterococcus species composition over considerable spatial and temporal ranges. The E. faecalis isolates identified were genotyped to examine their genetic diversity, and since naturalized E. coli strains have been isolated from both sites (2, 32), we further determined if enterococci could also persist in these two habitats.
MATERIALS AND METHODS
Sampling site description.Two sampling sites in a Lake Superior watershed were used in these studies: the new DBC beach in the Duluth-Superior Harbor and the Kingsbury Creek bank where it intersects Stark Road (KS) in Proctor, MN (Fig. 1A). The two sites were previously described (2, 32). At the DBC site, samples were collected from submerged sediment located 5 m from the waterline (S5), wet sand at the shoreline (SL), wet sand located 1 m upshore from the SL (NS), and dry sand located 8 m upshore from the waterline (US) (Fig. 1B). The Minnesota Lake Superior Beach Monitoring Program reported a high number of beach advisory days based on E. coli concentrations at the DBC in the summer of 2011 (33). At the KS site, samples were collected 5 m (KS5) and 14 m (KS14) from the creek (Fig. 1C). On 23 May 2011, three exclosure boxes (referred to as KS14I), made from 32-gallon trash cans (50-cm diameter) as described previously (34), were buried at the KS14 location in order to exclude external enterococcal sources: runoff and feces deposition from animals. Four mesh-covered windows were placed into the exclosure boxes to facilitate air exchange. The exclosure boxes were buried in the soil at a depth of 10 cm, and the original trash can lids covered the tops of the boxes.
Study sites at Lake Superior. (A) Sampling sites in the Lake Superior watershed. (B) Sampling areas at the DBC. (C) Sampling areas at the KS site. Legend: DBC = Duluth Boat Club beach; KS = Kingsbury Creek bank. Panels A and C are modified from reference 2.
Sample collection.Triplicate samples, located 1 m apart, were collected from the top 10-cm layer of sand, submerged sediment, and soil. Samples were taken by a shovel or with core tubes disinfected with 70% ethanol. Samples were stored in Whirl-Pak bags at 4°C until they were processed in the lab. In 2011, water (W) and dry sand (US) samples were also collected from the DBC. All samples were processed by the day after sampling. Sample temperature was measured in 2011 by inserting a thermometer to a depth of 10 cm in the sampling area. Sampling stopped at the KS14 site in August 2011 because of road construction that resulted in the destruction of the sampling areas. While two of the exclosure boxes were damaged in August, one was not damaged severely and samples were collected on 24 August 2011. The soil in the very top layer in the exclosure boxes became dry over time, so it was removed when sampling. In addition, runoff flowed through a recently installed drainage culvert into our sampling area (KS5) in September 2011.
Enumeration and isolation of cultivable enterococci.Samples were thoroughly mixed before processing. A 10-g aliquot of each sample and 100 ml of extraction buffer (35) amended with 0.01% Tween 20 were added to a 160-ml wide-mouth milk dilution bottle (Corning, Tewksbury, MA) containing 10 g of 3-mm glass beads (Fisher Scientific, Pittsburgh, PA). The mixture was shaken at 280 oscillations/min for 40 min on a horizontal, reciprocating shaker. The bottle contents were allowed to settle for 30 min. The supernatants from the May 2010 samples were filtered through 0.45-μm membranes that were placed onto the surface of m-EI Agar by the U.S. EPA method 1600 membrane filtration (MF) technique (36). However, because of the low enterococcal densities in samples and the high turbidity of the extractant, no enterococci were recovered and observed on m-EI Agar in May 2010 (detection limit, 1 CFU/2 g of original sample). Thereafter, standard method 9230 B, a multiple-tube most-probable-number (MPN) method, was used (37). Aliquots consisting of 10 ml, 1 ml, and 0.1 ml of the supernatants of May, June, and September samples (detection limit, 1.8 MPN/10 g of original sample) or 1 ml, 0.1 ml, and 0.01 ml of the supernatants of July and August samples (detection limit, 1.8 MPN/1 g of original sample) were inoculated into azide dextrose broth. Other steps were followed according to method 9230 B. To isolate enterococci, 20- to 25-g samples were treated as described above. The appropriate amount of extractant was filtered or directly spread onto m-EI Agar plates. The MF method was used to enumerate enterococci in water samples. Colonies on m-EI Agar with a blue halo were picked and cultivated in 150 μl Enterococcosel broth (Difco-BBL, Franklin Lakes, NJ) in 96-well microtiter plates. After growth, 40 μl of 80% glycerol was added to the wells exhibiting a brownish black color and plates were stored at −80°C until used.
Enterococcus isolate identification to the species level.Enterococcus isolates were stamped from glycerol stocks onto LB agar or Trypticase soy agar plates with a 48-pin replicator (Boekel Scientific, Feasterville, PA). Overnight colonies were suspended in 100 μl of double-distilled H2O (ddH2O). Multiplex PCR was used to identify eight species of enterococci: E. faecalis (FL), E. faecium (FM), E. casseliflavus (CA), E. hirae (HI), E. mundtii (MU), E. gallinarum (GA), E. durans (DU), and E. avium (AV) (38). American Type Culture Collection strains E. faecalis ATCC 19433, E. faecium ATCC 19434, E. casseliflavus ATCC 25788, E. hirae ATCC 8043, E. gallinarum ATCC 49573, E. durans ATCC 19432, and E. avium ATCC 14025 and an E. mundtii strain previously isolated from the environment by the lab were used as positive controls. A tube not containing any template DNA served as the negative control. The primers and PCR parameters used were described previously (38).
HFERP DNA fingerprinting.Isolates identified as E. faecalis by multiplex PCR were subjected to biochemical tests for verification. The biochemical tests included arginine hydrolysis and arabinose, raffinose, and sorbitol fermentation with the basal medium as previously described (25). Isolates that exhibited arginine hydrolysis and were sorbitol fermentation positive and arabinose and raffinose fermentation negative were verified as E. faecalis. The E. faecalis isolates were transferred into new 96-well microplates for horizontal, fluorophore-enhanced repetitive PCR (HFERP) DNA fingerprinting. This technique is similar to repetitive-sequence PCR-based DNA fingerprinting but uses fluorescently labeled primers and size markers to more adequately examine genetic population structures among bacteria (39).
DNA, extracted using a GenElute bacterial genomics DNA kit (Sigma-Aldrich), was used as the template for HFERP DNA fingerprinting of the 2010 E. faecalis isolates. DNA was extracted with the GenElute Bacterial Genomics DNA kit (Sigma-Aldrich). For isolates obtained in 2011, however, a new rapid colony method was developed. Approximately 0.5 μl of each colony was suspended in 100 μl of ddH2O and frozen at −20°C overnight, and 2 μl of the thawed suspension was used directly as the PCR template. Both methods were tested on the same strains and found to yield identical results. The BOXA2R primer (5′-ACGTGGTTTGAAGAGATTTTCG-3′) was used for DNA fingerprinting (40). The PCR protocol was modified on the basis of previous reports (23, 41). The master mixture for 96 reaction mixtures was prepared first. Each reaction mixture contained 5 μl of 5× Gitschier buffer (42), 0.25 μl of 100% dimethyl sulfoxide, 1 μl of 50 μM BOXA2R primers (50% unlabeled primers and 50% 6-carboxyfluorescein-labeled primers), 0.625 μl of 25 mM deoxynucleoside triphosphate, 0.2 μl of 20 mg/ml bovine serum albumin, 0.4 μl of 5-U/μl Taq (Denville Choice Taq), and 15.525 μl of nuclease-free H2O. Two microliters of template DNA (100 ng of DNA for 2010 isolates) was added to each reaction mixture in a final volume of 25 μl. PCR was carried out in a PTC 100 or PTC 200 thermal cycler (Bio-Rad MJ Research, Hercules, CA) by the protocol described previously (41), except that 30 cycles was used instead of 35 cycles. E. faecalis strain OG1RF and ddH2O served as positive and negative controls, respectively. DNA fingerprint data were analyzed as previously described (39).
Statistical analysis.MPN values were determined as described previously (37). When all tubes were negative (the MPN index was <1.8 MPN/100 ml), 0.1 MPN/100 ml was used for log transformation and statistical analysis. When all tubes were positive (the MPN index was >1,600 MPN/100 ml), a value of 1,600 MPN/100 ml was used. Enterococcal density was expressed as MPN/100 g of oven-dried sample (sand, submerged sediment, and soil samples) or CFU/100 ml (water samples). Sample moisture was expressed as the ratio of the mass of water to the mass of the original sample.
To satisfy the assumption of a normal distribution, the density data were log transformed for all statistical analyses. Multiple comparisons of densities were carried out by using unprotected Fisher's least significant difference at α = 0.05 (R software). The precipitation data for the KS site were obtained from the Lake Superior-Duluth streams website (www.lakesuperiorstreams.org/). The data recorded at the Duluth Lift Bridge were used for the DBC site because of its close proximity (a straight-line distance of about 1,100 m). The daily mean temperatures at the KS and DBC sites in 2010 were obtained at the Thompson Hill I-35 mile post 248 (a straight-line distance of about 3,600 m) and Duluth Sky Harbor (a straight-line distance of about 5,500 m) weather stations, respectively. The correlation between the sample temperature and the enterococcal density in 2011 was studied by using the Pearson product-moment correlation coefficient. HFERP banding patterns were analyzed by using Bionumerics software (version 3.0; Applied Math, Inc.) as described previously (43).
RESULTS
Enterococcal density.Enterococci were not detected in May 2010 at the DBC and KS sites by the MF technique (detection limit, 1 CFU/2 g of original sample). However, enterococci were ubiquitous in samples analyzed by the MPN technique. This is probably due to the different detection limits of these two methods. By the MPN technique, enterococci were detected in 94% of the samples, with concentrations ranging from 3 × 105 to 5.6 × 105 MPN/100 g of sample (dry weight of sand, sediment, and soil samples and ml of water samples) (Fig. 2). Even in upshore sand samples at the DBC, where the moisture ranged from 0.3 to 4.5%, 14 of 15 samples (93%) contained enterococci, with concentrations ranging from 2 × 101 to 1.6 × 104 MPN/100 g of sample. The enterococci were not found in some of the May, June, and September samples (detection limit, 1.8 MPN/10 g of original sample). Generally, the densities were greater when the temperatures were higher. This is supported by the positive correlations between enterococcal densities and sample temperatures in 2011 (DBC, r = 0.57, P = 9.4 × 10−8; KS, r = 0.51, P = 0.01).
Density of enterococci in the Lake Superior watershed. The densities and temperatures are shown as bar and scatter-line plots, respectively. Error bars represent standard errors. The same letter in more than one bar indicates that there is no significant difference (P > 0.05). BDL1 indicates that enterococcal densities were below the detection limit (1 CFU/2 g of original sample) of the MF technique. The BDL2 indicates that the enterococcal densities were below the detection limit (1.8 MPN/10 g of original sample) of the MPN technique. The numbers (5, 6, 7, 8, and 9) on the x axis represent sampling months (May, June, July, August, and September, respectively). Samples: S5, submerged sediment located 5 m from the waterline; SL, wet sand located at the waterline; NS, wet sand located 1 m upshore from the waterline; US, dry sand located 8 m upshore from the waterline; W, water; KS5, soil located 5 m from the creek water; KS14, soil located 14 m from the creek water; KS14I, soil in exclosure boxes.
At the DBC, the comparisons of monthly enterococcal densities across different matrices showed that the nearshore sand samples harbored 7 to 87 times more enterococci than did water throughout the sampling period (all P values, <0.05). Moreover, the enterococcal density in water was positively correlated with those in SL sand samples (r = 0.62, P = 0.01) and submerged sediment samples (r = 0.64, P = 0.01), suggesting that the bacteria might be transported among these three matrices.
The density of enterococci varied from 7.5 × 101 to 5.6 × 105 MPN/100 g of oven-dried sample at the KS site over the study period (Fig. 2, KS5 and KS14). The data obtained in 2010 showed no significant difference in the overall enterococcal concentration between the KS5 and KS14 samples at α = 0.05 (t test, P = 0.30). In 2011, however, the overall enterococcal density at the KS5 site was significantly greater than that at the KS14 site (t test, P = 4.0 × 10−3). This may be due in part to the fewer samples taken at the KS14 site. When monthly comparisons of the KS and DBC sites over the 2-year period were carried out, the overall mean enterococcal densities in soil at the KS site (combining KS5 and KS14 samples) were greater than those at the DBC site (combining sand and sediment samples) in July and August (both P values are <0.05).
Diversity of Enterococcus species composition.The species status of 2,441 enterococcal isolates was determined by multiplex PCR and biochemical analyses. The majority of the isolates (97.8%) could be assigned to one of eight species: E. faecalis, E. faecium, E. casseliflavus, E. hirae, E. mundtii, E. gallinarum, E. durans, or E. avium. The most abundant species at the DBC were E. hirae (36.4%), E. faecium (27.6%), E. faecalis (14.5%), and E. mundtii (12.0%). In contrast, the dominant species at the KS site were E. faecalis (48.8%), E. mundtii (20.0%), E. casseliflavus (14.2%), and E. faecium (10.8%). Further analyses showed that the Enterococcus species composition varied both spatially and temporally (Fig. 3).
Diversity of Enterococcus species composition in the Lake Superior watershed. *, the number of isolates analyzed was less than 24; ND, no data available as densities were below the detection limit; NA, data not accessible; S5, submerged sediment located 5 m from the waterline; SL, wet sand located at the waterline; NS, wet sand located 1 m upshore from the waterline; US, dry sand located 8 m upshore from the waterline; W, water; KS5, soil located 5 m from the creek water; KS14, soil located 14 m from the creek water; KS14I, soil in exclosure boxes.
Genetic diversity of E. faecalis in the Lake Superior watershed.Over the 2-year study period, 309 and 227 E. faecalis isolates from the DBC and KS sites, respectively, were subjected to HFERP DNA fingerprinting. Since the fingerprints of positive-control strain E. faecalis OG1RF had a minimum similarity of 95% over repeated analyses (including multiple PCR analyses using DNA and colonies as templates), fingerprints with ≥95% similarity were regarded as the same genotype (data not shown).
The E. faecalis population was diverse in the Lake Superior watershed over the 2-year period and was composed of unique isolates, groups of a few isolates, and large groups of clonal isolates (Fig. 4). Further examination of the dendrogram showed that large groups usually contained a large proportion of isolates collected in July and August. Among the 536 isolates examined, 148 genotypes were identified. Their similarity values ranged from 9.8 to 94.9%, with the majority of isolates being ≥60% similar to each other. The Shannon diversity index value was 4.08, suggesting a high level of diversity within the total population.
Partial dendrogram generated from DNA fingerprints of E. faecalis strains isolated from the Lake Superior watershed. A cutoff value of 85% was selected in order to display the dendrogram. The number next to a cluster is the number of isolates in that cluster.
The genetic diversity of E. faecalis isolates at the DBC site was greater than that at the KS site. At the DBC site, 108 genotypes were identified among the 309 isolates examined. Seven of the genotypes contained ≥10 isolates. However, the majority of DNA fingerprint patterns were unique. In contrast, 46 genotypes were detected among the 227 isolates obtained from the KS site. The Shannon diversity indices of E. faecalis at the KS and DBC sites were 2.87 and 3.84, respectively. Further examination of the dendrogram generated with isolates from the KS site revealed that two large groups, accounting for 38% of the isolates, contained 52 and 34 isolates, respectively. The total population diversity was also examined by discriminant analyses. As shown in Fig. 5, multivariate analysis of variance (MANOVA) indicated that the E. faecalis population exhibited spatial and temporal variability, although there was some overlap because of the relative relatedness of different groups. This finding was further supported by Jackknife analysis ( Table 1). Jaccard similarity values ranging from 1.8 to 6.8% suggested that very few genotypes were shared by the different groups.
MANOVA of all HFERP DNA fingerprints generated from environmental E. faecalis isolates grouped by year and site. The discriminants are shown by the distance along the x and y axes. The percentage of variation each discriminant accounts for is shown, and the number before the site name indicates the year the isolates were collected.
Jackknife analyses of DNA fingerprints from E. faecalis strains grouped by site and year
Recurrence of some E. faecalis fingerprints.Further examination of the dendrogram showed that some E. faecalis DNA fingerprints occurred over multiple (≥2) sampling events at each sampling site. For example, 21 of 25 KS5 isolates in August 2010, 21 of 25 KS14 isolates in August 2010, and 8 of 10 KS14 isolates in September 2010 clustered together with similarity values of ≥97% (Fig. 6). These isolates were considered to be of the same genotype. In total, 12 and 8 genotypes obtained from the DBC and KS sites, respectively, recurred over multiple sampling times, accounting for 25 and 52% of the isolates collected at the two sites, respectively. Moreover, MANOVA (Fig. 7A) showed that the majority of these isolates, especially those isolated from KS soil, clustered and were separate from the others, suggesting that some of these isolates likely persist in these environments.
Partial dendrogram generated from HFERP DNA fingerprints of some E. faecalis isolates collected at the KS site. The terms on the right of the dendrogram indicate the number of strains clustered (for example, 1/25 is 1 out of 25 strains), the sampling site, and the sampling time.
MANOVA of HFERP DNA fingerprints generated from environmental E. faecalis isolates grouped by site and their frequency. Graph A contains all of the isolates from the DBC and KS sites; graph B contains only the isolates from the KS site. “Recurrent” indicates that the genotype appeared over multiple sampling events, “unique” indicates that the genotype only appeared once. KS14I indicates the isolates collected from the exclosure boxes at the KS14 site.
Enterococci in exclosure boxes.The enterococcal density in the exclosure boxes at the KS14 site initially decreased below the detection limit (1.8 MPN/10 g of original sample) in June 2011 but later increased to as much as 3.7 × 103 MPN/100 g of oven-dried sample as the temperature increased in August 2011 (Fig. 2, KS14I-2011). As expected, the moisture of the samples in the exclosures was slightly lower than that of soil exposed to the environment (June, KS14, 11%, KS14I, 10%; July, KS14, 17%, KS14I, 15%). Isolates of E. faecalis , E. casseliflavus, E. mundtii, and E. hirae were consistently isolated from exclosure boxes from July to September, after the sampling areas were protected (Fig. 3, KS14I-2011).
Since the exclosure boxes limited direct enterococcal input from external sources, it was of interest to examine the genetic diversity of the E. faecalis isolates in the exclosure boxes at the KS14 site. The population of E. faecalis in the exclosure boxes was relatively diverse, and 17 genotypes, with similarity values ranging from 49.0 to 92.5%, were identified among 47 isolates. One large group contained 18 isolates that were collected in August 2011. No fingerprints appeared over multiple sampling events. Moreover, MANOVA showed that these isolates tended to be separate from other isolates from the KS site (Fig. 7B). Because of disruption of the sampling area by road construction, the isolates collected in September 2011 were excluded from the MANOVA.
DISCUSSION
The goals of this study were to examine the population structure of enterococci in a Lake Superior watershed and to determine if these bacteria can persist in the extraintestinal environment, as was reported for E. coli at the same sites (2, 32). The main finding of this study was that enterococci can persist in this Great Lakes environment for a prolonged time after being introduced and likely grow when environmental conditions (moisture and temperature) become favorable.
The densities of enterococci were positively correlated with sample temperatures (DBC, r = 0.57, P = 9.4 × 10−8; KS, r = 0.51, P = 0.01). Since our samples were taken from the top layer of the sampling areas around noon, the sample temperature was quite close to the air temperature, which could reach around 33°C in July. This is quite close to the optimum growth temperature for enterococci (35°C). Moreover, nutrients, including sea grass debris carried onto the beach sand by wave action and natural grass at the KS site, might also favor the persistence and growth of these bacteria in the summer months.
To test if enterococci could persist for a prolonged time or become “naturalized” (2) to the environment examined, we covered the sampling area at the KS14 site to avoid direct contamination from external sources. We consistently isolated enterococci within exclosure boxes. Since the sampling areas were covered, the isolates were likely independent of recent contamination events and represented persistent enterococci in the environment. The disappearance and reappearance of enterococci and the inconsistent Enterococcus species composition in our study might be due to the heterogeneous character of the microbial community in the soil environment and the limited number of culturable enterococci studied.
The presence of persistent enterococci at the study site was also supported by the presence of recurrent E. faecalis DNA fingerprints over multiple sampling events, especially those strains isolated at the KS site. There is limited human activity at the KS site, and there was not consistent precipitation before sampling (www.lakesuperiorstreams.org/). This indicated that at least some of these isolates were not related to direct input from human or animal feces and recent runoff from rain events. Moreover, some of these recurrent isolates were very abundant in samples at the KS site. For example, 21 (84%) of 25 E. faecalis isolates collected at the KS14 site in August 2010 had the same fingerprint. In addition, isolates collected in July and August usually clustered as large groups. Since we did not observe any animals or feces in the sampling area at the KS site over the sampling period, our data suggest that these E. faecalis isolates were persistent after being introduced into the extraintestinal environment. Moreover, these isolates likely grew in the summer months, especially in July and August, when environmental conditions, including temperature, moisture, and nutrients, become favorable and support microbial growth (8). This likely partially explained the elevated enterococcal concentration and high percentage of E. faecalis bacteria in the summer of 2010 at the KS site. We also found that only a very limited number of strains could be repeatedly isolated over the 2 years. Taken together, these results suggested that some enterococci are able to persist in the Lake Superior environment, especially in soil, but because of extreme cold temperatures and nutrient depravation, they might not become naturalized to the environments examined.
We also found that enterococcal densities in sand, sediment, and soil samples were high. On the basis of a mass unit calculation, the enterococcal concentration in a majority of the samples exceeded the standard of 35 CFU/100 g of sample (6). Exceedances are expected if we express the concentration as MPN/100 ml of interstitial water, since the moisture of a majority of the samples was below 50%. If the high density of enterococci at the sites we examined was due mainly to their persistence and growth, it may lead to unnecessary beach closures. Therefore, assessment of the public health risks of illnesses associated with exposure to the matrices examined at the DBC site is suggested.
Monthly enterococcal densities in nearshore sand samples at the DBC site were 7 to 87 times greater than those found in water throughout the sampling period on the basis of a mass unit. A similar phenomenon was also observed at some Lake Huron beaches and marine beaches (16, 44). Compared with water, sand and soil provide relatively favorable environments for the bacteria to survive (18, 21). Considering that our sampling time was generally near noon, when water was exposed to strong sunlight, a lower enterococcal concentration in water samples than in sand may be responsible for some of the noted disparity in values.
Consistent with previous research (10, 15, 45), the abundant species identified in sand, sediment, and soil samples of the freshwater environments examined included E. hirae, E. faecalis , E. faecium, E. mundtii, and E. casseliflavus. However, unexpectedly, there was a very low percentage of E. faecalis and a high percentage of E. faecium in the water column at the DBC site. Previous research reported relatively high percentages of both E. faecalis and E. faecium in marine water in the United States (10, 22) and in lake water in Russia (45), and some studies reported that the most abundant species in sewage was E. faecium (46, 47). Further studies using microbial source tracking techniques need to be carried out in order to better understand the source of enterococci in water.
The spatial and temporal dynamics of Enterococcus species composition reflected the effects of environmental factors, such as runoff from rain events. The percentages of E. faecalis bacteria in submerged sediment, SL, and nearshore sand samples at the DBC were related to antecedent precipitation 24 h prior to sampling (submerged sediment, r = 0.67, P = 0.05; SL sand, r = 0.55, P = 0.12; nearshore sand, r = −0.44, P = 0.24). This suggested that runoff from rain events might contain E. faecalis and transport this bacterium from nearshore sand into water or into matrices having contact with water. The Enterococcus species composition shifted dramatically at the KS site in 2011, perhaps because of disruption from road construction. The species composition at this site is also likely influenced by vegetation. For example, some yellow-pigmented enterococci, including E. casseliflavus and E. mundtii, are considered to associate mainly with plants (48–50), and the KS site contained extensive vegetation. A previous study also suggested that rainfall and gravity could transport the bacteria from plant leaves to soil (48). Since this study aimed to uncover the species composition of these bacteria at the sites, instead of tracking their source, further studies on the possible sources of enterococci at these sites are needed.
At the DBC site, the positive correlation of enterococcal densities in water, submerged sediment, and SL sand samples; the diverse Enterococcus species composition in water and sand samples; and the co-occurrence of some E. faecalis fingerprints in different matrices support the hypothesis that external forces are involved in the transport of these bacteria. Previous research reported seiche tides in the Duluth-Superior Harbor occur every several hours, with amplitudes ranging from 3 to 25 cm (51, 52). It is possible that seiche mixed the enterococci among SL sand, submerged sediments, and water. Considering that nearshore sand samples were wet throughout all of the sampling events, seiche might also transport enterococci from nearshore sand into water. Wave action and runoff from rain events were suggested to transport bacteria from sand or soil to adjacent water (21), elevating their populations in water and confounding their use as fecal indicators.
The E. faecalis strains isolated in the Lake Superior watershed were genetically diverse, with a Shannon diversity index of 3.84 at the DBC and 2.87 at the KS site. Brownell et al. (30) reported that the Shannon diversity indexes of enterococci ranged from 1.88 to 2.69 in raw sewage, pristine river water, storm water-impacted sediments, and water on the basis of repetitive-sequence PCR-based DNA fingerprinting with BOXA2R primers. The discriminant analyses done here showed that the genetic diversity of all of our E. faecalis isolates varied spatially and temporally and was likely influenced by the diverse sources of these bacteria, similar to what was seen in a previous study of E. coli (53).
Understanding of the occurrence and persistence of enterococci in freshwater environments is important before all Great Lakes and coastal states decide to use enterococci as the fecal indicator bacteria for the assessment of recreational water quality. Moreover, since Lake Superior watersheds are different in temperature and beach composition from those of the other Great Lakes, it is also important to examine the presence and species distribution of enterococci near Lake Superior. In contrast to some previous studies, our results provide in-depth information on the distribution, genetic diversity, and persistence of enterococci in freshwater environments. Our results also showed the seasonal change in the enterococcal concentration in the watershed, which was partially due to the persistence and growth of enterococci in the environment after their introduction. The diversity of Enterococcus species composition and the genetic diversity of E. faecalis isolates likely reflect diverse input sources and multiple environmental factors influencing the growth and distribution of enterococci.
ACKNOWLEDGMENTS
This work was funded in part by the Minnesota Sea Grant College Program, supported by the NOAA Office of Sea Grant, United States Department of Commerce, under grant no. R/CC-02-10. This paper is journal reprint no. JR 604 of the Minnesota Sea Grant College Program.
We thank Matthew Hamilton and John Ferguson for help with HFERP DNA fingerprint analyses and Jessica Eichmiller for her assistance with sampling. We also thank Alexandria Boehm and Dawn Manias for providing strains and Charlene Jackson for multiplex PCR support.
FOOTNOTES
- Received 17 December 2012.
- Accepted 22 February 2013.
- Accepted manuscript posted online 1 March 2013.
- Copyright © 2013, American Society for Microbiology. All Rights Reserved.
