Quantitative Microbial Risk Assessment of Pathogenic Vibrios in Marine Recreational Waters of Southern California

Abstract

This study investigated the occurrence of three types of vibrios in Southern California recreational beach waters during the peak marine bathing season in 2007. Over 160 water samples were concentrated and enriched for the detection of vibrios. Four sets of PCR primers, specific for Vibrio cholerae, V. parahaemolyticus, and V. vulnificus species and the V. parahaemolyticus toxin gene, respectively, were used for the amplification of bacterial genomic DNA. Of 66 samples from Doheny State Beach, CA, 40.1% were positive for V. cholerae and 27.3% were positive for V. parahaemolyticus, and 1 sample (1.5%) was positive for the V. parahaemolyticus toxin gene. Of the 96 samples from Avalon Harbor, CA, 18.7% were positive for V. cholerae, 69.8% were positive for V. parahaemolyticus, and 5.2% were positive for the V. parahaemolyticus toxin gene. The detection of the V. cholerae genetic marker was significantly more frequent at Doheny State Beach, while the detection of the V. parahaemolyticus genetic marker was significantly more frequent at Avalon Harbor. A probability-of-illness model for V. parahaemolyticus was applied to the data. The risk for bathers exposed to recreational waters at two beaches was evaluated through Monte Carlo simulation techniques. The results suggest that the microbial risk from vibrios during beach recreation was below the illness benchmark set by the U.S. EPA. However, the risk varied with location and the type of water recreation activities. Surfers and children were exposed to a higher risk of vibrio diseases. Microbial risk assessment can serve as a useful tool for the management of risk related to opportunistic marine pathogens.

INTRODUCTION

Vibrios are Gram-negative, motile bacteria that can cause diseases in humans. They are commonly found in marine coastal ecosystems, where their population changes with seawater temperature, increasing with warmer temperatures and algal blooms and decreasing with cooler temperatures (1, 2). Vibrio vulnificus and Vibrio parahaemolyticus were two of the most common vibrio infections reported in the United States between 1997 and 2006, responsible for the most vibrio-related hospitalizations and deaths (3). V. parahaemolyticus and V. vulnificus are known to cause infection and sepsis if they reach the blood (3). The hemolysin enzyme produced by the tdh (temperature direct hemolysin) toxin gene of V. parahaemolyticus can lead to the destruction of red blood cells, while the lipopolysaccharide toxin of V. vulnificus can produce diarrhea and blistering dermatitis. V. vulnificus infection has a mortality rate of 50%, with the majority of patients dying within the first 48 h of infection (4). Outside the developed world, the most common form of vibrio pathology is the gastrointestinal disease cholera, caused by Vibrio cholerae. These bacteria release the cholera toxin, an enterotoxin that causes an increase in the secretion of sodium in the intestine, leading to diarrhea and dehydration. Cholera epidemics are a serious threat to the developing world, with several hundred thousand cases reported to the World Health Organization annually (5).

Vibrios have been isolated from the marine environments of many geographic regions, such as the African coast, Australia, and the coasts of both North and South America, demonstrating a global distribution (3, 5, 6). A 2004 survey of coastal waters near the Conero River in Italy showed a variety of vibrios, including pathogenic strains (7).

The U.S. Centers for Disease Control and Prevention released a survey of recreation-related waterborne diseases in the United States between 2005 and 2006. This report examined 189 cases of vibrio infection due to recreational water activity, 18 of which resulted in death (9.5% mortality rate) (8). Thus, the development of an understanding of vibrio diseases through marine recreational water exposure is important for human health protection. Furthermore, the fecal indicator bacteria (FIB) used to protect human health during ocean water recreation are not good indicators of vibrios (9, 10), because the occurrence and concentration of indigenous aquatic bacteria like vibrios are governed by the environmental conditions rather than fecal pollution from external sources. There has not been a risk assessment model to estimate the health risk of vibrio diseases associated with marine beach water recreation.

Here we report the detection of V. cholerae, V. parahaemolyticus, V. vulnificus, and the V. parahaemolyticus toxin gene tdh by PCR assays of seawater samples from two popular recreational beaches in Southern California. A quantitative microbial risk assessment (QMRA) model was applied to estimate the risk of vibrio diseases during water recreation.

MATERIALS AND METHODS

Sample collection and vibrio enrichment.

Sixty-six water samples were taken at Doheny State Beach, CA, from the surface of the water column at 5 locations (locations A to E) from 25 May through 4 July 2007 (Fig. 1). Ninety-six beach samples were also taken at 3 locations (locations A to C) at Avalon Harbor, Catalina Island, CA, from 27 July through 3 September 2007 (Fig. 1). The sampling locations on each beach were designed based on their distance from the source of fecal pollution. Locations C and D at Doheny State Beach had historically high counts of FIB because of pollution from San Juan Creek. Locations A and B at Avalon Harbor had relatively higher counts of FIB. At the Avalon Harbor site, approximately half of the samples (n = 45) were collected from the surface of the water column at ankle-deep water, and the other portion (n = 51) was collected from the surface of the water column at chest-deep water. The terms ankle deep and chest deep for measuring distance from the shore are commonly used in water quality assessments by the Orange County Health Department in California.

Fig 1.

Sampling sites used in this study. Sixty-six water samples were taken from Doheny State Beach (DSB), CA, at the surface of the water column at 5 locations (locations A to E) between 25 May and 4 July 2007. Ninety-six samples were taken at 3 locations (locations A to C) at Avalon Harbor (AH), Catalina Island, CA, between 27 July and 3 September 2007.

Bacteria in water samples (50 to 200 ml) were filtered onto a 0.45-μm-pore-size nylon membrane filter (Whatman, Maidstone, Kent, United Kingdom). Filters were placed into 50 ml alkaline peptone water (APW) (10 g/liter of peptone and 10 g/liter of NaCl [pH 9.2]) and incubated at 37°C overnight to encourage vibrio growth. Aliquots of 1-ml enrichment cultures were taken and stored in a −80°C freezer for future analysis.

Genomic DNA extraction.

Two methods of bacterial genomic DNA extraction were evaluated: a cetyltrimethylammonium bromide (CTAB)-chloroform-phenol method and a boiling cell lysis method. The CTAB-chloroform-phenol method was performed according to protocols described in Molecular Cloning: a Laboratory Manual (11). Briefly, cell pellets were resuspended in 1× TE buffer (10 mM Tris Cl, 1 mM EDTA [pH 7.5]) with protein kinase K and an SDS solution and incubated at 37°C for 1 h. Next, CTAB in NaCl (10% CTAB in 0.7 M NaCl) was added, and the mixture was heated to 65°C for 10 min. Genomic DNA was extracted by using isoamyl alcohol, chloroform, and phenol (1:24:25 dilution). The DNA pellet was washed and dried in a Speed Vac concentrator (model SVC100H; Sevant) and resuspended in 1× TE buffer. DNA concentration and purity were analyzed spectrometrically (DU 7400 spectrometer; Beckman Coulter, Fullerton, CA).

For the boiling lysis method, enriched samples were allowed to thaw to room temperature from an −80°C freezer. A subsample of 100 μl was removed and centrifuged at 10,000 rpm for 2 min to pellet the bacteria. The pellet was resuspended in 30 μl 1× TE buffer, and the sample tube was heated to boiling for 10 min and then cooled to room temperature. The lysate was diluted 1:10 with deionized water for PCR assays. A comparison of the CTAB-chloroform-phenol DNA extraction technique and the boiling lysis method showed similar efficiencies for PCR amplification when using positive controls (data not shown). The boiling lysis technique proved faster, produced less waste, and was less expensive. Therefore, the CTAB-chloroform-phenol technique was replaced by the boiling cell lysis method for later part of the sample analysis.

PCR.

Primers used for the detection of Vibrio species and toxins are listed in Table 1. We used primer pairs specific for the thermolabile hemolysin (tlh) gene to detect V. parahaemolyticus, and a primer set that targeted the temperature direct hemolysin (tdh) gene was used to detect toxic V. parahaemolyticus (6). V. cholerae was detected by using primers that target the intergenic spacer (its) between the 16S and 23S ribosomal subunits (12). The primers for V. vulnificus target the cytotoxin-hemolysin gene cth, also called vvh, the V. vulnificus hemolysin gene (6, 13). All samples were tested for V. vulnificus, V. cholerae, and V. parahaemolyticus. Samples that tested positive for V. parahaemolyticus were also analyzed for the tdh toxin gene. The PCR mixture was composed of 21 μl of master mix and 4 μl of extracted DNA or deionized water (negative control). Positive and negative controls were run in conjunction with samples to check for contamination and PCR success and for band size comparisons. The master mix was composed of 11.2 μl deionized water, 2.5 μl 10× buffer (Lucigen Corp., Middleton, WI), 2.5 μl 25 mM MgCl₂ solution (Lucigen Corp., Middleton, WI), 2 μl (25 μM solution) each of forward and reverse primers (Sigma-Aldrich, St. Louis, MO), 0.5 μl deoxynucleoside triphosphates (dNTPs) (Fisher Scientific, Pittsburgh, PA), 0.2 μl Taq polymerase (Lucigen Corp., Middleton, WI), and 0.1 μl 100× purified bovine serum albumin (BSA) (New England BioLabs, Ipswich, MA). PCR was performed by using a GeneAmp 2700 PCR system (Applied Biosystems, Foster City, CA). PCR conditions and cycling temperatures are shown in Table 2. After PCR amplification, samples were run on a 2% agarose gel in 1× TAE buffer (40 mM Tris acetate, 2 mM Na₂EDTA·2H₂O [pH 8.5]) and imaged by using a gel documentation system and Alpha Ease FC software, version 6.0.0 (Alpha InnoTech, San Leandro, CA). Results were scored as positive if the expected amplification sizes were observed.

Table 1.

Primers used for PCR amplification of the genetic markers for V. cholerae, V. parahaemolyticus, the V. parahaemolyticus toxin gene, and V. vulnificus

Target	Primer name	Primer sequence	Tm (°C)a	Amplicon size (bp)	Reference
its, V. cholerae	PVC-F2	TTAAGCSTTTTCRCTGAGAATG	57.0	300	12
	PVCM-R1	AGTCACTTAACCATACAACCCG	61.6
tlh, V. parahaemolyticus	tlhF	AAAGCGGATTATGCAGAAGCACTG	68.9	450	6
	tlhR	GCTACTTTCTAGCATTTTCTCTGC	61.3
tdh, V. parahaemolyticus toxin	tdhF	GTAAAGGTCTCTGACTTTTGGAC	59.7	269	6
	tdhR	TGGAATAGAACCTTCATCTTCACC	64.0
cth, V. vulnificus	VVF	TTCCAACTTCAAACCGAACTATGAC	65.8	205	13
	VVR	ATTCCAGTCGATGCGAATACGTTG	69.5		6

^aT_m, melting temperature.

Table 2.

PCR amplification conditions for V. cholerae, V. parahaemolyticus, the V. parahaemolyticus toxin gene, and V. vulnificus

Primer pair(s)	Conditions
	Denaturing cycle	Repetitive cycle			Elongation cycle
		1	2	3
PVC-F2/PVCM-R1	94°C, 5 min	94°C, 0.5 min	60°C, 1 min	72°C, 0.5 min	72°C, 7 min
tlhF/tlhR and VVF/VVR	94°C, 3 min	94°C, 1 min	65°C, 1 min	72°C, 1 min	72°C, 5 min
tdhF/tdhR	94°C, 3 min	94°C, 1 min	60°C, 1 min	72°C, 1 min	72°C, 5 min

QMRA. (i) QMRA model framework for water recreation-related vibrio illness.

The QMRA steps and model framework are presented in Fig. 2. The individual component within the model framework is described below. The model was constructed and implemented by using MATLAB R2010a (The Mathworks Inc., Natick, MA). QMRA was performed only for V. parahaemolyticus because there was no previous evidence of cholera disease due to recreational exposure to contaminated water in spite of reports of a wide distribution of V. cholerae in coastal oceans. Risk analysis was performed by using PCR data for samples collected at Doheny State Beach and Avalon Harbor, regardless of the sampling dates at each beach.

Fig 2.

Schematics of the QMRA model framework.

(ii) PCR data treatment.

Given that our PCR results had only two outcomes (positive or negative detection), a binomial probability model was used for each recreational beach. Specifically, an m × n (rows × columns) binary matrix containing 1's and 0's was generated by using the randerr function of MATLAB to represent “positive” and “negative” PCR results, respectively. The percentage of 1's in each row was selected randomly from the binomial distribution of the PCR detection results for each beach, where the probability of selecting a certain percentage was highest at the distribution's mode and decreasing toward its lower and upper confidence levels.

Cell identifications (IDs) in the m × n binary matrix were also randomly generated by using the uniformly distributed pseudorandom integer generator randi in MATLAB and were used to represent the location of each cell in the matrix. As both the values in the binary matrix and the cell IDs were generated randomly, they served to simulate the stochastic process occurring in nature.

(iii) Probability distribution function of vibrio concentration.

The probability distribution function (PDF) curve of the V. parahaemolyticus concentration at Doheny State Beach and Avalon Harbor was generated by referring to its concentration distribution in the coastal environment compiled from data reported in the literature (14–18) (see Table S1 in the supplemental material for data). As these data were obtained by using different methods (e.g., PCR or culture methods), they have different lower and upper detection limits. In order to fully utilize the available data, data points below or above the detection limits were defined by using their lower and upper detection limits, respectively. Data were compiled and binned by using a log-scale interval for generating histograms. Random samplings under the distribution curve were conducted by using the emprand function (courtesy of Durga Lal Shrestha) written for MATLAB. The integrity of the function was validated by comparing and cross-checking histograms of the simulated data with the corresponding histogram of the original data. All of the PDF curves in this study were produced through kernel density estimation.

It should be noted that our model referred only to the V. parahaemolyticus concentration from a literature survey when a randomly selected sample from the binary matrix was “1,” denoting “positive” detection (see Fig. 3 for the steps taken to generate various PDF curves). Each of the simulated PDF curves/histograms consists of 15,000 iterations, which were compared to each other for assessing the degrees of variance among them. In our study, three replicates were produced for each simulated PDF curve.

Fig 3.

Algorithm for generating the probability distribution of the exposure dose.

(iv) Probability distribution function of water volume ingested.

The probability distribution of water volume ingested by people during water recreational activities was adapted from data reported in literature. Different distributions of water volume ingested were observed for different water recreational activities. Swimming and surfing were the two main recreational activities at the beach, and therefore, distributions for water volume ingested during swimming and surfing were estimated in this study. The amount of water ingested by people during ocean swimming was estimated based on a survey conducted previously by Schets et al. (19). Three distributions were reported in their study, which applied to men, women, and children (≤15 years old), respectively. The volume of water ingested by men during swimming was described by the gamma distribution as (I_oral) = G {r = 0.45, λ = 60} ml/swim event, whereas (I_oral) = G {r = 0.51, λ = 35} ml/swim event was given for women and (I_oral) = G {r = 0.58, λ = 55} ml/swim event was given for children. Similarly, the amount of water ingested during surfing was estimated based on a survey conducted previously by Stone et al. (20). However, only adults at least 18 years old were included in the study by Stone et al. The volume of water ingested during surfing fits the log-normal (LN) distribution (I_oral) = LN{μ = 3.54, σ = 1.80} ml/surf event. Data points used for subsequent Monte Carlo analyses were sampled from 95% confidence intervals of each swimming distribution and 90% confidence intervals of the surfing distribution according to previous reports.

(v) Exposure assessment.

The exposure of surfers and swimmers to V. parahaemolyticus was estimated by using the joint probability of volume water ingested during water recreation and the V. parahaemolyticus concentration encountered at the time of exposure, which is expressed in equation 1:

$$D_{\text{exp}}=I_{\text{oral}}\times P$$ (1)

where D_exp is the exposure dose, P is the pathogen density distribution, and I_oral is the volume of water ingested.

The procedure for generating the exposure distribution curve was similar to that used to generate the PDF curve of vibrio concentrations. Figure 3 shows the flowchart of the algorithm for generating the probability distribution of the exposure dose.

(vi) Dose-response model.

The probability of illness per surfing or per swimming event was estimated by using a beta-Poisson dose-response model proposed by the U.S. Food and Drug Administration for V. parahaemolyticus, which is presented in equation 2 (21):

$$P_{\text{ill}}=1-{\left(1+D_{\exp }/\beta \right)}^{-\alpha }$$ (2)

where P_ill is the probability of illness from a single vibrio exposure during water recreation and α and β are model parameters for V. parahaemolyticus defined by the U.S. FDA dose-response model. Values for α and β used for this study are 0.60 and 1.31 × 10⁶, respectively, which are the best parameter estimates among the other likely parameter estimates.

RESULTS

Occurrence of vibrios and toxins.

For the 66 water samples collected at Doheny State Beach, the overall frequency for the genetic marker of V. cholerae was 40.1%, the overall frequency for the genetic marker of V. parahaemolyticus was 27.3%, and one sample tested positive for the V. parahaemolyticus toxin gene, with a frequency of 1.5% (Table 3). For the 96 water samples collected at Avalon Harbor, the overall frequency for the genetic marker of V. cholerae was 18.7%. The detection frequencies for the genetic marker of V. parahaemolyticus and the V. parahaemolyticus toxin gene were 69.8% and 5.2%, respectively (Table 3). No samples from either site tested positive for the V. vulnificus genetic marker (Table 3). A comparison of results from Avalon Harbor and Doheny State Beach showed that significantly more samples tested positive for the V. cholerae gene marker at Doheny State Beach (P = 0.033 by one-tailed t test, assuming unequal variance) with a 95% confidence interval, while significantly more samples were positive for the V. parahaemolyticus gene marker at Avalon Harbor (P = 0.004 by one-tailed t test, assuming unequal variance) with a 95% confidence interval (Table 3). No significant difference was detected when samples were analyzed by individual sampling location on each beach or the date of collection during the 4 months of study with sampling every 5 to 7 days (Fig. 4). Further statistical analysis of samples from Avalon Harbor demonstrated no significant difference when samples positive for a vibrio gene marker, either V. parahaemolyticus or V. cholerae, at an ankle-deep location were compared to samples from the chest-deep location (P = 0.53 by two-tailed t test, assuming unequal variance) with a 95% confidence interval (Table 4). Thus, the vibrio data at each beach were combined for QMRA regardless of the sampling dates and the locations on the beach.

Table 3.

Detection of V. cholerae, V. parahaemolyticus, the V. parahaemolyticus toxin gene, and V. vulnificus in water samples from Doheny State Beach and Avalon Harbor, CA

Sampling site (no. of samples)	No. (%) of samples positive for:
	V. cholerae	V. parahaemolyticus	V. parahaemolyticus toxin gene	V. vulnificus
Doheny State Beach
A (15)	3 (20.0)	4 (26.6)	0 (0)	0 (0)
B (14)	5 (35.7)	1 (7.1)	0 (0)	0 (0)
C (7)	6 (85.7)	4 (57.1)	1 (14.3)	0 (0)
D (15)	5 (33.3)	4 (26.7)	0 (0)	0 (0)
E (15)	8 (53.3)	5 (33.3)	0 (0)	0 (0)
Total (66)	27 (40.1)	18 (27.3)	1 (1.5)	0 (0)
Avalon Harbor
A (32)	5 (15.6)	22 (68.8)	2 (6.3)	0 (0)
B (32)	6 (18.8)	22 (68.8)	1 (3.1)	0 (0)
C (32)	5 (15.6)	23 (71.9)	2 (6.3)	0 (0)
Total (96)	16 (16.7)	67 (69.8)	5 (5.2)	0 (0)

Fig 4.

Frequency of detection of Vibrio cholerae, Vibrio parahaemolyticus, and the Vibrio parahaemolyticus toxin gene at both Doheny State Beach and Avalon Harbor, presented by sampling date (day/month) in 2007.

Table 4.

Detection of V. cholerae, V. parahaemolyticus, and the V. parahaemolyticus toxin gene in water samples collected from ankle-deep and chest-deep water at Avalon Harbor, CA

Sample site (no. of samples)	No. of samples positive for:
	V. cholerae	V. parahaemolyticus	V. parahaemolyticus toxin gene
Aankle (17)	5	12	1
Achest (15)	0	10	1
Bankle (17)	4	10	1
Bchest (15)	2	12	0
Cankle (17)	3	15	2
Cchest (15)	2	8	0
Totalankle (51)	12	37	4
Totalchest (45)	4	30	1

QMRA of vibrio exposure during marine water recreation.

Figure 5 shows the PDF of the V. parahaemolyticus concentration at Doheny State Beach and Avalon Harbor. The three replicates of simulated PDF curves all converged after 15,000 iterations and matched the histogram (Fig. 5). A higher probability of a V. parahaemolyticus concentration of 10² to 10³ organisms/100 ml was predicted at Avalon Harbor, while a flat long tail of up to 10⁵ organisms/100 ml was observed by the PDF for Doheny State Beach. A V. parahaemolyticus concentration of 10⁻² organisms/100 ml was most frequently observed at both sites.

Fig 5.

Probability density function and normalized histogram for Vibrio parahaemolyticus concentrations at Doheny State Beach (DSB) and Avalon Harbor (AH). Three replicates were produced for each graph to show the degrees of variance among each curve. Sim, simulated.

Joint probability distributions of water volume ingested and the concentration of vibrios showed that human exposure to V. parahaemolyticus varied with the location (Fig. 6). In general, the exposures to V. parahaemolyticus at Doheny State Beach peaked at between 10⁻³ and 10⁻² organisms/event and decreased sharply before flattening out, resulting in a low flat shoulder toward the exposure to the higher pathogen concentration. At Avalon Harbor, the exposures to V. parahaemolyticus followed a similar trend as that of Doheny State Beach but peaked, decreased slowly, and flattened out with a relatively higher shoulder toward the higher dose of exposure (Fig. 6). The shapes of the exposure dose distributions were not significantly different between the surfers and the swimmers, but the minimum exposure of surfers shifted to the right, showing a 2-order-of-magnitude difference in the minimum exposures between surfers and swimmers. Children also experienced a slightly higher rate of exposure at a low dose during ocean swimming, but the elevation was insignificant at a higher dose (Fig. 6).

Fig 6.

Probability density function for dose of Vibrio parahaemolyticus ingested during surfing or swimming at Doheny State Beach (DSB) and Avalon Harbor (AH).

The risk outcomes due to surfing and swimming at the two Southern California beaches are summarized as a cumulative density function (CDF) of the illness risk in Fig. 7. The U.S. EPA (1986) acceptable ocean recreational illness rate of 19 per 1,000 bathers was used as a point of reference for interpreting the risk outcomes. The CDF curve shifted to the right for the surfers and children swimmers at both locations, suggesting a relatively higher illness risk per ocean recreation event for the two categories. However, the predicted risks at both beaches were below U.S. EPA ocean recreational illness risk benchmarks. Table 5 shows the averages and ranges of vibrio disease risk predicted at each beach for each category of subject. Surfers experienced the highest risk among all categories. The maximum risk for surfers to experience vibrio disease was ca. 13 per 1,000 bathers per surfing event at Avalon Harbor. Children were the next group experiencing a higher risk at Avalon Harbor, with a maximum risk of ca. 6 children per 1,000 bathers per swimming event.

Fig 7.

Cumulative density function of illness risk at Doheny State Beach (DSB) and Avalon Harbor (AH) due to swimming or surfing.

Table 5.

Summary descriptors of illness risk per recreational event

Site and group	Illness risk (per recreational event)
	Range		Avg	95th percentile
	Min	Max
Doheny State Beach
Surfers	1.57 × 10−10	1.07 × 10−2	3.15 × 10−5	9.80 × 10−5
Swimmers
Men	5.83 × 10−13	3.40 × 10−3	1.19 × 10−5	2.73 × 10−5
Women	9.16 × 10−13	3.66 × 10−3	8.32 × 10−6	1.97 × 10−5
Children	3.57 × 10−12	2.99 × 10−3	1.30 × 10−5	4.26 × 10−5
Avalon Harbor
Surfers	1.58 × 10−10	1.28 × 10−2	8.20 × 10−5	4.36 × 10−4
Swimmers
Men	5.98 × 10−13	5.20 × 10−3	3.10 × 10−5	1.61 × 10−4
Women	9.16 × 10−13	3.25 × 10−3	1.98 × 10−5	1.03 × 10−4
Children	3.58 × 10−12	5.59 × 10−3	3.68 × 10−5	2.00 × 10−4

DISCUSSION

Current recreational marine safety guidelines call for the sampling of FIB such as Escherichia coli and enterococci (22) to protect the public from exposure to human pathogens from fecal waste contamination. However, marine-indigenous microorganisms also play a role in recreation water health. As was recently documented by CDC surveillance for recreational-water-related illnesses, vibrio diseases from recreational exposure pose important health risks (8). So far, there has not been a good understanding of the risk associated with exposure to vibrios in marine water or a risk framework to estimate the risk.

Vibrios are important members of the microbial community in coastal waters. Diverse Vibrio spp. were reported in North American coastal waters, including pathogenic species of V. cholerae, V. parahaemolyticus, and V. vulnificus (10, 14, 18, 23–28). However, quantitative studies of different pathogenic groups and associated toxins are rare, which is due largely to the limitations of current detection methods. Direct PCR quantification of vibrios using a concentrated microbial community from seawater was attempted in this study, but the quantitative PCR results were inconsistent. This could be due to (i) the presence of PCR interferences and large amounts of nontarget DNAs in the samples and (ii) the low concentrations of the target organisms. In contrast, the culture enrichment samples were highly reproducible for the V. cholerae and V. parahaemolyticus assays. A second round of PCR using the first PCR products did not yield additional positive results for selected samples. Thus, a positive PCR was considered to be the presence of at least 1 organism per volume of water sample. The detection limit of this assay ranged between 5 and 20 organisms per liter of water.

Samples from Doheny State Beach contained significantly more V. cholerae, and samples from Avalon Harbor contained significantly more V. parahaemolyticus, while neither location tested positive for V. vulnificus. Many factors could be associated with the differences in bacterial types by location, including abiotic factors, such as salinity, temperature, nutrient content and type, and wastewater runoff, or biotic factors, such as predation by protozoans (2). V. cholerae is also known to enter into a viable-but-nonculturable state (29) at cooler water temperatures. Further studies could explore the relationship between species and geographical locations.

The lack of V. vulnificus at both study sites may imply a limitation in enrichment because APW may not favor the selection of V. vulnificus (30) or the low occurrence of V. vulnificus in California coastal waters. V. vulnificus is prevalent in warm ocean waters of the Gulf of Mexico and the Atlantic coast. For example, Panicker et al. previously analyzed oysters from the Gulf of Mexico, and all samples tested positive for vvh, the marker for V. vulnificus (6). However, the rate of detection is low for the West Coast of the United States: only 5.7% of 527 samples were positive for V. vulnificus in California estuarine waters (26). The cases of V. vulnificus infections in California are related to the consumption of shellfish collected in the Gulf of Mexico (31). Similarly, in a 2007 study, Masini found that only 2% of water samples tested positive for V. vulnificus off the Italian coast near the Conero River (7). The occurrence of V. vulnificus seems to be highly dependent on geographical location.

The results of QMRA indicated that the recreation risks of exposure to vibrios were below the acceptable illness benchmark of 19 per 1,000 bathers (22) at two Southern California Beaches. However, it is important to note that both the vibrio concentration and the type of recreational activity are equally significant parameters in predicting the probability of illness. Surfers and children are at higher exposure levels and thus are higher-risk groups for vibrio diseases. The different risk outcomes for two close-by California marine recreational beaches also imply the importance of empirical risk assessment at local beaches, as both pathogen concentrations and recreational activities vary at different locations.

The risk assessment performed in this study considered only a single type of marine bacterium. There are hundreds of thousands of other indigenous marine bacteria in water that are ingested during surfing and swimming. Many of them, including other vibrios, Aeromonas spp., and Pseudomonas spp., etc., are opportunistic human pathogens. The compounding effect of these microbes is likely to weaken the human immune system, which leads to disease that may not be triggered by the dose of a single organism. Human pathogens from fecal pollution, although not assessed in this study, also play a compounding role in recreational disease outcomes. Thus, risk assessment is much limited by the availability of models and systematic understandings of disease outcomes.

Risk prediction, like the true nature of human health risk, is dynamic in nature. Similar to previous reports of health risks using average values (19), the average risk reported in this research should not be taken at face value. Distributions of the illness risk should be considered to reveal the full spectrum of risk. The predicted risk may over- or underestimate the true risk due to multiple factors. For example, vibrio infection through wound infection is well known (32) but was not included due to the lack of an exposure model. The parameters used for the beta-Poisson dose-response model were calculated based on human clinical trial data, which may be inadequate for environmental samples. Moreover, the current model considers the responses of healthy adults only, which underestimates the higher risks presented for sensitive populations, such as the immunocompromised, children, and the elderly.

Although the recreational health risks from vibrio exposure were below recreational safety standards at the two California beaches, the risks could be very different at other locations where these indigenous bacteria are more prevalent and are present in higher concentrations. The QMRA offers a framework for identifying health risk priorities and marine policy gaps and for human health protection.

Conclusions.

V. cholerae and V. parahaemolyticus were detected at two popular marine recreational beaches in Southern California. The prevalences of Vibrio species were different at the two locations. The QMRA results suggest that the illness risks from vibrio exposure during water recreation were below the recreational illness risk benchmark set by the U.S. EPA. However, surfers and children were exposed to relatively higher risks of vibrio diseases. The QMRA offers a tool for identifying marine recreational safety priorities.