Occurrence and Persistence of Bacterial Pathogens and Indicator Organisms in Beach Sand along the California Coast

California coast sand survey sites and sample collection.Sand was collected at 53 California beaches between the Mexico and Oregon borders (Fig. 1; see also Table S1 in the supplemental material) on four separate outings between 16 and 29 October 2009. The climate in California is Mediterranean, with distinct dry and wet seasons, and sampling was conducted prior to the onset of the rainy season. In the 3 days prior to sampling, coastal counties reported precipitation of less than 2.5 cm (data not shown) (http://cdec.water.ca.gov). Beaches represented a wide range of natural and anthropogenic conditions, including sand grain size, sand organic carbon content, presence of a putative pollution point source (river, creek, or storm drain), surrounding land use, and degree of shelter from waves; many sites from Yamahara et al. (77) were included. At each beach, two samples of dry, exposed sand were collected from (i) within 1 m above the high tide line, here termed “dry samples,” and (ii) from a location likely to be polluted (e.g., near a flock of birds, storm drain, river, sea wall, or a beach path), referred to here as “targeted samples.” The samples were out of the tidal range during collection, but these sites presumably could be inundated during spring tides or large-wave events. Each sand sample was collected by compositing 10 subsamples (25 ml) to obtain a total volume of 250 ml. Samples were stored on ice and processed within 24 h of collection.

• Open in new tab
• Download powerpoint

Fig 1

Occurrence of Campylobacter (A), the human-specific fecal marker in Bacteriodales (HF marker) (B), S. aureus, including MRSA (C), and Salmonella (D) in beach sand along the California coast. Inset pie charts indicate the percentages of beaches positive for each organism. n = 53 for all assays (except n = 37 for S. aureus).

Microcosms to study pathogen and fecal indicator persistence and mobilization in sand.Microcosm experiments were conducted to document microbe survival profiles and to test the hypothesis that the persistence of microbes in dry beach sand increases when the sands are subjected to periodic wetting (e.g., spring tides). Acrylic column microcosms were constructed to mimic a vertical section of sand at the beach (78). Columns consisted of a single acrylic tube 22.9 cm in length with an inner diameter of 2.3 cm. Stainless steel mesh (mesh pore size = 0.0625 mm) was used at the top and bottom of the columns to retain sand but allowed water to flow through and help maintain aerobic conditions. Sand collected from Lovers Point, CA (36°37′29″N, 121°54′59″W), was used to pack the columns. Beach sand was collected in sterile 1-liter Whirl-Pak (Nasco, Fort Atkinson, WI) bags within 1 m above the high tide line; the sand was collected from 10 different locations along the shoreline of the beach and homogenized (78). The sand was not washed or sterilized. The sand was then seeded with primary treated sewage (filtered through a 250-μm-pore-size sieve) from the Palo Alto Regional Water Quality Control Plant. A total of 4.5 liters of sewage was added to approximately 6,650 g (∼4.75 liters) of sand. The sewage-sand mixture was mechanically homogenized using a sterilized stainless steel mixing paddle for 5 min, allowed to rest for 30 min, and transferred to a sterile draining container, which allowed excess sewage to drain. Seeded sand was rehomogenized as previously described (78) and packed into 44 columns using a tap and fill procedure. The total mass of sand in each column was ∼133 g (dry weight). To determine the initial concentration of microorganisms, a 100-g sample of the composite sand was removed after packing every seventh column, and the average was used to estimate the initial concentration (n = 6).

Columns were kept in the dark at 22°C for up to 32 days. Equal numbers of columns were designated treatments and controls. Each treatment column was subjected to wetting with 50 ml of seawater (collected at Lovers Point on the day of use and filtered using a 0.22-μm-pore-size filter) and subsequent draining to simulate the natural wetting and draining process that occurs at the upper reaches of the beach during the highest spring tides (78). Here this process is referred to as “watering.” Prior to application of the filtered seawater to the columns, it was tested for all microorganisms assayed during this experiment (see below). Watering was performed in June 2010 on each day that the tide at Lovers Point exceeded 1.85 m (relative to mean sea level) (http://tbone.biol.sc.edu/tide). This datum was chosen as it was at the upper reaches of the tidal excursion at Lovers Point. The tides exceeded the datum on days 10 and 11 of the microcosm experiment. During the watering events, 50 ml of water was applied to the treatment columns and allowed to drain to a catch basin under the influence of gravity. The leachates from all treatment columns were collected and combined for analysis of microorganisms. The concentration of microorganisms in the leachate was multiplied by the collected leachate volume and divided by the dry weight of sand in the column to normalize the number of organisms mobilized to the mass of the sand.

One control and one treatment column were sacrificed (i.e., the entire sand volume of each column was emptied into its own sterile 1-liter beaker) daily for the first 7 days of the experiment. Thereafter, control and treatment columns were sacrificed concurrently approximately every 2 to 3 days. On the 2 days when the columns were watered, treatment columns were sacrificed before watering. Sacrificed sand was homogenized by hand mixing with a sterile spatula for 10 min. Homogenized sand was divided into 2 subsamples: (i) approximately 90 g for elution and membrane filtration and (ii) 20 g for moisture content and organic carbon analyses. All elutions and filtrations were performed immediately after a column was sacrificed.

Methods to analyze microorganisms.Sand collected during the beach survey was analyzed for levels of E. coli, enterococci, somatic coliphages, a Bacteroidales human-specific fecal marker (HF marker), Salmonella spp., Campylobacter spp., S. aureus, and methicillin-resistant S. aureus. Sand microcosms were analyzed for E. coli, enterococci, F⁺ and somatic coliphages, the HF marker, Salmonella spp., and Campylobacter spp.

Some targets were measured by culture-dependent methods only (E. coli, somatic coliphage, and F⁺ coliphage) and some by culture-independent methods only (HF marker). Other targets were measured using culture methods supplemented by PCR verification (S. aureus and MRSA), and some targets were measured by both culture-based and QPCR methods (enterococci, Salmonella spp., and Campylobacter spp.). E. coli and enterococci measured by culture-based methods were designated cEC and cENT, respectively. Enterococci, Salmonella spp., and Campylobacter spp. measured by QPCR methods were labeled tENT, total Salmonella, and total Campylobacter, respectively. Note that the analysis for S. aureus was carried out in only 37 beach samples due to analytical constraints (Fig. 1).

For all assays, microorganisms from sand aliquots were eluted using a ratio of eluant volume to sand mass of 10:1 (6). During the beach sand survey and the microcosm experiments, the masses of sand eluted were 60 g and 90 g, respectively. Sand was eluted by a previously described method of shaking by hand using nanopure water (6) or 1× phosphate-buffered saline (PBS) for S. aureus prior to membrane filtration (28). Eluant was membrane filtered through 0.45-μm-pore-size HAWG filters (Millipore, Billerica, MA) for culture methods and through a 0.22-μm-pore-size polycarbonate filter (GE Osmonics, Minnetonka, MN) for QPCR.

Microbial analyses by culture-dependent methods.EPA methods 1604 (67) and 1600 (66) were used to enumerate cEC and cENT, respectively. Survey samples used 50 ml of sand eluant, and microcosm samples used 0.1 to 50 ml, adjusted to ensure a countable number of colonies. All concentrations are reported as CFU per gram of dry weight of sand.

Coliphages were enumerated using a modified version of a previously published protocol for water samples (62). Briefly, MgCl₂ (0.05 M final concentration) was added to 100 ml of sand eluants and mixed for 5 min. Sand eluants amended with MgCl₂ were membrane filtered through 0.22-μm-pore-size GSWP filters (Millipore, Billerica, MA). Filters were placed face down in 300 μl of 50% glycerol–phosphate-buffered saline and stored at −80°C until analysis. Coliphages were eluted from filters using 2 ml of 3% beef extract (MP Biomedical, Solon, OH) (pH 9.0). A double-agar-layer method using E. coli F-amp (ATCC 700891) for F⁺ coliphages and E. coli CN13 (ATCC 700609) for somatic coliphages was used to enumerate coliphages by plating 1-ml volumes of beef extract eluants with the appropriate coliphage host. Plaques were counted after 24 h of incubation at 37°C. Only somatic coliphages were enumerated during the coastal sand survey. In the column persistence experiments, both somatic and F⁺ coliphages were enumerated by aseptically cutting a single membrane filter in half. Coliphage concentrations are reported as PFU per gram of dry weight of sand.

S. aureus enumeration and PCR verification for sand samples were performed according to previously published protocols (28). Briefly, sand eluants (28) were filtered and incubated on CHROMagar Staph aureus (SCA) or CHROMagar MRSA (C-MRSA) (BD Diagnostic, Franklin Lake, NJ) to enumerate S. aureus and MRSA, respectively. SCA plates were incubated at 37°C for 24 h, and C-MRSA plates were incubated at 37°C for 48 h. Filters were refrigerated overnight for color development before counting the number of mauve colonies on the filters was performed. A representative number of mauve colonies (typically 5 to 10 colonies) were picked from each filter and streaked for isolation to verify the presence of S. aureus by appearance (mauve with matte halo). For SCA, this combination of visual identification of colonies on the filter with visual identification of isolated colonies provided a positive agreement (sensitivity) value of 84%, negative agreement (specificity) value of 95%, and positive predictive accuracy value of 99% (28). Further verification of isolate identification was provided by PCR. The source of DNA was crude lysate obtained by heating (95°C, 10 min) an isolated colony in molecular-analysis-grade water with lysostaphin (Sigma catalog no. L7386) (final concentration, 50 μg/ml). Identification of S. aureus was verified by PCR confirmation of the presence of the clfA gene, and MRSA identification was verified by PCR confirmation of the presence of both the clfA and mecA genes (48). Amplification reactions were carried out according to Goodwin and Pobuda (28).

The presence of culturable Salmonella and Campylobacter in each survey sample was assessed by enriching (i) membrane-filtered microorganisms eluted from sand (100-ml filtration [60 g of sand and 600 ml of eluant]) or (ii) an equivalent mass of sand (6 g). Enumeration in microcosm samples was achieved by the most-probable-number (MPN) method (see below) by membrane filtration of triplicate sand eluant volumes (100, 10, and 1 ml) and placing aseptically rolled filters into 25 ml of enrichment broth.

Salmonella detection and enumeration used a modified version of EPA method 1682 for enumeration in biosolids by the MPN method (3, 60, 68). Membrane filters or sands were enriched in 25 ml of tryptic soy broth (TSB) at 37°C for 24 h. TSB enrichments (30 μl) were dropped onto modified semisolid Rappaport-(MSRV) medium (BD Diagnostic) and incubated at 42°C for 24 h. Motile organisms were streaked and incubated (37°C, 24 h) on xylose lysine deoxycholate agar (XLDA) (BD Diagnostic). Colonies displaying typical Salmonella morphology (red-pink or red with black colonies) underwent biochemical confirmation on lysine iron agar (LIA) (BD Diagnostic) and triple-sugar iron agar (TSIA) (BD Diagnostic). Putative positive isolates were archived in 50 μl of 1× Tris-EDTA (TE) buffer and stored at −80°C for confirmation by PCR. DNA was obtained by lysing archived isolates at 85°C for 10 min followed by centrifugation and collection of the supernatant. Lysate supernatants were amplified for the Salmonella genus-specific invA gene by the use of Qiagen Hotstar Plus Master Mix (Qiagen Inc., Valencia, CA) following the PCR conditions in Malorny et al. (46). PCR positives were matched to the appropriate TSB tube, and MPNs were determined using a 3-tube MPN table (3).

Campylobacter detection and enumeration used a modified version of the method of Khan and Edge (38) in which membrane filters or sands were enriched in 25 ml of Bolton broth (catalog no. CM0983; Remel, Lenexa, KS) supplemented with Campylobacter Selective Supplement (catalog no. SR0183; Remel, Lenexa, KS) under microaerophilic conditions at 42°C for 48 h using GasPak 100 systems with EZ Campy Container System sachets (BD Diagnostic). Bolton broth enrichments were streaked onto modified Karmali agar (MKA) (catalog no. CM0935; Remel, Lenexa, KS) supplemented with Campylobacter Selective Supplement Karmali (catalog no. SR0167; Remel, Lenexa, KS) and incubated under microaerophilic conditions at 42°C for 48 h. Colonies displaying typical Campylobacter morphology (white to gray colonies) were picked as presumptive positives. DNA from presumptive positives was obtained as described above for Salmonella, with PCR confirmation targeting Campylobacter 16S rRNA (43). Corresponding positives were matched to the appropriate TSB tube, and MPNs were determined using a 3-tube MPN table (3).

Microbial analyses by culture-independent methods.For all QPCR assays, the membrane filter from a 100-ml volume of sand eluant was folded into a 2.0-ml FastDNA Lysing Matrix E microcentrifuge tube (MP Biomedical, Solon, OH) and stored at −80°C until DNA extraction. Bacterial DNA was extracted using the manufacturer's instructions for the FastDNA Spin kit for soils (MP Biomedical, Solon, OH) (78) as modified by first pulverizing the frozen filters by bead milling (Mini-Beadbeater-1; Biospec Products, Bartlesville, OK) (15 s at 4,800 oscillation/min) and next centrifuging the lysing tubes for 60 s at 24,000 × g. DNA extraction proceeded using a FastPrep instrument (MP Biomedical, Solon, OH) with the addition of a 10-min settling step using binding matrix solution and an additional wash step performed with FastDNA Spin kit SEWS-M solution. DNA was eluted in two 50-μl fractions, for a total of 100 μl. DNA was stored at −80°C until used for downstream PCRs.

QPCR assays (Table 1) used previously published protocols (32, 35, 40, 45, 47). Reaction mixtures consisted of 1× TaqMan Universal Mastermix (Applied Biosystems, Foster City, CA), forward and reverse primers and probe (Table 1), 0.2% bovine serum albumin fraction V (Gibco, Carlsbad, CA), and 2 μl of template DNA (20-μl final volume). Triplicate reactions were run on a StepOnePlus real-time PCR system (Applied Biosystems, Foster City, CA) under the following thermal cycling conditions: 2 min at 50°C and 10 min at 95°C, followed by cycles of 15 s at 95°C and 1 min at the corresponding annealing temperatures and numbers of cycles shown in Table 1. Fluorescence and quantification cycles for each assay were manually adjusted to compare standard curves across runs. Slopes, y intercepts, and reaction efficiencies are presented in Table 1.

Genomic DNA (gDNA) standards were generated using Enterococcus faecium (ATCC 19434), Salmonella enterica subsp. enterica serovar Typhimurium LT2 (ATCC 19585), and Campylobacter jejuni (ATCC 700819). Pure cultures of ATCC stocks were grown to the log phase, and DNA was extracted using a QIAmp DNA Mini kit (Qiagen, Valencia, CA). Plasmid DNA standards were created for the HF marker by cloning PCR amplicons into a pCR2.1-TOPO vector, transformation into competent E. coli following manufacturer's protocols (Invitrogen, Carlsbad, CA), and purification using a Qiagen Plasmid Maxi kit (Qiagen, Valencia, CA). Genomic and plasmid DNA was quantified using a NanoDrop ND-1000 spectrophotometer (Nanorop Technologies, Wilmington, DE) and serially diluted to create standards that ranged from 10⁰ to 10⁶ copies (Table 1). Copy numbers and cell equivalents (CE) were calculated utilizing the total genome size and copy numbers of the amplified gene in Enterococcus spp. (6 copies of 23S rRNA [51]), Salmonella spp. (1 copy of the ttrRSBCA locus [49]), and Campylobacter spp. (3 copies of 16S rRNA [55]). For plasmid-generated standards, the total plasmid size was calculated and cell equivalents (CE) were calculated using 6 rRNA operons/genome for the HF marker (71).

QPCR inhibition tests were performed on sand survey samples by comparing bacterial concentrations in undiluted and 1:10-diluted DNA extracts for each assay. QPCR inhibition tests for microcosm samples with enterococci and Campylobacter assays were performed by spiking sample extract with an aliquot of a standard (10² or 10³ copies). The inhibition factor (IF) was calculated as previously described in Yamahara et al. (78). The microcosm matrix was fixed over the course of the experiment; therefore, tests for inhibition (10% of the samples) were distributed evenly over the duration of the experiment (n = 7; Table S3 in the supplemental material provides details).

Negative controls and blanks.Culture-dependent and -independent filtration blank experiments were performed with every 6 samples (18 total) for each of the membrane filtration methods used during the California beach survey. During the column microcosm study, a single filtration blank experiment was performed each day (23 total) for each of the culture-dependent and -independent membrane filtration assays. Filtration blank experiments for culture-dependent assays were performed by membrane filtration of 50 ml of 1× PBS rinse solution, and filter blanks were processed as previously described. For culture-independent assays, 50 ml of DNase/RNase-free water was membrane filtered as previously described for the DNA-based assays. An extraction blank was run alongside every 12 samples during DNA extractions (4 total). No-template controls were run on each PCR and QPCR plate.

Determination of beach and sand characteristics.The moisture content (θ_m) of each sand sample from the beach survey and microcosm columns was determined by drying preweighed sand at 110°C for 24 h. Moisture content was used to normalize all bacterial densities per gram of dry sand. Microcosm sand porosity (φ) and grain size distribution were calculated and used to estimate sand hydraulic conductivity (K) using the Kozeny-Carman method (23).

The organic carbon content by mass (C_m) of sand from the survey was determined for each sand sample using the loss-on-ignition technique (37). Dry sand was combusted at 550°C for 6 h, cooled in a desiccator, and then weighed. The error associated with triplicate C_m measurements was 0.3%.

The grain size of each sand sample from the beach survey was determined by sieving. Percent mass of sand retained on and passing through no. 35 (0.5 mm), no. 60 (0.25 mm), and no. 120 (0.125 mm) sieves was calculated after shaking for 5 min at 4,000 rpm. Percent fine grains (fines) was determined by adding the mass of sand retained and passing through the no. 120 sieve and dividing by the total mass analyzed.

Land cover within a 10-km-radius circular buffer around each beach was determined using the 2001 National Land Cover Data set (USGS [http://seamless.usgs.gov]) and ARCMAP software (ESRI, Redlands, CA) (see Table S2 in the supplemental material). The various land cover classes can be found at http://www.mrlc.gov/nlcd01_leg.php. Land covers were classified as follows: urban (covers 21 to 24), agricultural (covers 81 and 82), and forested (covers 41, 42, 43, 51, 52, 71, and 72). Wetland land covers, barren rock, and unconsolidated shore were characterized as “other.” The fractions of the land that were urban, agricultural, forested, and other were calculated for each of the sampling sites. Urban and agricultural land covers were added to obtain the total developed land cover value; forested land cover was deemed undeveloped.

Decay rate calculations.For the column microcosm studies, decay rates of microorganisms were calculated using the Chick-Watson model (16, 72) as follows: C(t) = C₀e^k₁(t), where C(t) is the concentration at a point in time (t), C₀ is the initial concentration, and k₁ is the first-order decay constant. Decay rates were calculated using a linear curve fit between the natural log-transformed normalized concentration (ln C/C₀) and time. Only data points within the upper and lower detection limits were included in the calculation of decay rates (see Table S4 in the supplemental material). The following were compared: (i) decay rates of individual microorganisms in treatment versus control microcosms and (ii) decay rates between various organisms as measured in the control microcosms. Comparison of regression slopes (decay rates) was accomplished by regressing the two individual first-order curves combined with a dichotomous interaction term as presented in Neter et al. (50).

We tested whether a biphasic decay model could be used to estimate the concentrations of microorganism during the persistence studies (see the supplemental material for methods and results) and found that the biphasic model generally did not improve the fit of the data compared to the Chick-Watson model. Comparisons between the biphasic and Chick-Watson models were based on adjusted R² and corrected Akaike's Information Criterion (AICc) values (34). Thus, only results from the Chick-Watson model are presented here.

Statistical methods.Statistical analyses were carried out using PASW Statistics, release 18.0.0 (IBM, Chicago, IL). For the beach survey, microbial variables were dichotomous (Salmonella and Campylobacter) or continuous (cENT, tENT, cEC, somatic coliphages, S. aureus, and the HF marker). Values below the lower limit of detection for the continuous variables were assigned a value of zero. Nonparametric statistical methods (Kruskal-Wallis test and Spearman's rank correlation) were used to compare fecal indicators and pathogens measured during the survey. N-way analyses of variance (ANOVA) were used to identify beach and sand characteristics that controlled microbe population variability. Ranked microbial concentrations were used as the dependent variable. The presence of waves, quartile of θ_m, quartile of C_m, quartile of developed land cover, and presence of a putative source were used as independent variables; no interaction terms were included. Terms were eliminated from the models one by one in cases in which they did not represent a significant percentage of the variance of the dependent variable. All results and relationships were deemed significant at P < 0.05. In some cases, results with P < 0.1 are presented. The mean concentrations reported represent the arithmetic averages, and errors reported represent standard deviations about the mean, unless otherwise noted.