APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Dec. 2007, p. 7589–7596
Copyright © 2007, American Society for Microbiology. All Rights Reserved.
Vol. 73, No. 23
Variability of Total and Pathogenic Vibrio parahaemolyticus Densities
in Northern Gulf of Mexico Water and Oysters?
A. M. Zimmerman,1A. DePaola,2J. C. Bowers,3J. A. Krantz,2J. L. Nordstrom,2
C. N. Johnson,1and D. J. Grimes1*
University of Southern Mississippi, Gulf Coast Research Laboratory, Ocean Springs, Mississippi1; Food and Drug Administration,
Gulf Coast Seafood Laboratory, Dauphin Island, Alabama2; and Food and Drug Administration, College Park, Maryland3
Received 24 July 2007/Accepted 26 September 2007
Vibrio parahaemolyticus is indigenous to coastal environments and a frequent cause of seafood-borne gas-
troenteritis in the United States, primarily due to raw-oyster consumption. Previous seasonal-cycle studies of
V. parahaemolyticus have identified water temperature as the strongest environmental predictor. Salinity has
also been identified, although it is evident that its effect on annual variation is not as pronounced. The effects
of other environmental factors, both with respect to the seasonal cycle and intraseasonal variation, are
uncertain. This study investigated intraseasonal variations of densities of total and pathogenic V. parahaemo-
lyticus organisms in oysters and overlying waters during the summer of 2004 at two sites in the northern Gulf
of Mexico. Regression analyses indicated significant associations (P < 0.001) between total V. parahaemolyticus
densities and salinity, as well as turbidity in water and in oysters at the Mississippi site but not at the Alabama
site. Pathogenic V. parahaemolyticus organisms in Mississippi oyster and water samples were detected in 56%
(9 out of 16) and 78% (43 out of 55) of samples, respectively. In contrast, 44% (7 out of 16) of oyster samples
and 30% (14 out of 47) of water samples from Alabama were positive. At both sites, there was greater
sample-to-sample variability in pathogenic V. parahaemolyticus densities than in total V. parahaemolyticus
densities. These data suggest that, although total V. parahaemolyticus densities may be very informative, there
is greater uncertainty when total V. parahaemolyticus densities are used to predict the risk of infection by
pathogenic V. parahaemolyticus than previously recognized.
Vibrio parahaemolyticus is the leading cause of Vibrio-asso-
ciated gastroenteritis in the United States (15, 19, 20, 32) and
has been isolated from oysters, blue crabs, finfish, and plank-
tonic copepods (6, 7, 18, 22). Vibrio infections are most com-
mon in individuals living in states bordering the Gulf of Mexico
(23) and are usually associated with the consumption of raw
shellfish, primarily oysters (9, 20). Recent outbreaks (4, 5, 10,
14) raised the awareness of public health officials concerning
shellfish throughout coastal states and prompted the Interstate
Shellfish Sanitation Conference to develop an Interim Control
Plan for regulating shellfish harvest areas based on V. para-
haemolyticus densities in shellfish (31). The Interim Control
Plan employs a colony lift technique using DNA probes that
target the species-specific thermolabile hemolysin (tlh) gene
and the thermostable direct hemolysin (tdh) gene associated
with pathogenic V. parahaemolyticus strains (32). A similar
DNA probe colony hybridization method has been developed
to target the tdh-related hemolysin gene (trh), which is also
associated with pathogenic strains of V. parahaemolyticus (27).
While these colony lift techniques make it possible to investi-
gate the distribution of pathogenic V. parahaemolyticus strains
directly, they have a relatively high limit of detection (LOD).
Thus, the colony lift method is typically sensitive enough to
quantify the more abundant tlh-positive V. parahaemolyticus
strains from water and oysters but is less effective for the rarer
tdh- or trh-positive strains.
V. parahaemolyticus density in oysters has been shown to be
positively correlated with water temperature, and higher den-
sities of V. parahaemolyticus are normally detected during
warmer months (8, 15). However, V. parahaemolyticus densi-
ties vary considerably even at optimal temperatures, and pos-
sible links between this variability and other environmental
factors remain unclear (29, 35). The U.S. Food and Drug
Administration (FDA) released a V. parahaemolyticus risk
assessment in which densities of total V. parahaemolyticus
organisms in oysters at harvest were predicted based on water
temperature measurements obtained from National Oceano-
graphic and Atmospheric Administration (NOAA) buoys (2).
Water temperature was estimated to be associated with ap-
proximately 50% of the annual variation in total V. parahaemo-
lyticus densities in oysters. In the risk assessment, the effects of
additional environmental factors impacting V. parahaemolyticus
densities were not incorporated due to insufficient data. While
predicted V. parahaemolyticus densities based on water tem-
perature generally agree with average densities determined by
microbiological examination and provide an appropriate basis
for aggregate seasonal and regional predictions, there re-
mains considerable unexplained intraseasonal and intraregional
To address these data gaps, we investigated potential in-
traseasonal associations between selected non-temperature-
based environmental parameters (levels of salinity, chloro-
phyll, and turbidity) and the densities of total and pathogenic
V. parahaemolyticus organisms in oysters and in overlying wa-
* Corresponding author. Mailing address: University of Southern
Mississippi, Gulf Coast Research Laboratory, 703 East Beach Drive,
Ocean Springs, MS 39564. Phone: (228) 872-4210. Fax: (228) 872-4204.
?Published ahead of print on 5 October 2007.
ters. Turbidity and chlorophyll levels have been previously
hypothesized to correlate with V. parahaemolyticus densities
(21, 22, 25, 35). Temperature effects were minimized by con-
ducting the study during those months that provide near-opti-
mal temperatures for V. parahaemolyticus growth (May
through August). If strong and predictive environmental sig-
natures can be identified based on water quality parameters,
then remote sensing of these parameters may prove useful as a
means of effectively monitoring estuarine environments and
limiting human exposure to dangerous V. parahaemolyticus
Since the densities of pathogenic (tdh- or trh-positive) V.
parahaemolyticus organisms in the environment are usually
below the LOD (10 CFU g?1) for the DNA probe colony
hybridization method, previous studies have not adequately
characterized the distribution of pathogenic V. parahaemo-
lyticus densities (15); it appears that a high number of samples
contain nondetectable V. parahaemolyticus. To address this
inaccuracy, a real-time PCR assay was applied for the simul-
taneous detection of tlh, tdh, and trh genes in overnight en-
richments of oysters in alkaline peptone water (APW) (28).
MATERIALS AND METHODS
Sample collection and preparation. Surface water samples (n ? 102) and
oyster samples (n ? 32) were collected at two sites from May through August
2004. One site was a pier at the FDA Gulf Coast Seafood Laboratory (GCSL) in
Dauphin Island, AL (30°15?52?N, 88°06?47?W), and the other was a pier at the
University of Southern Mississippi Gulf Coast Research Laboratory (GCRL) in
Ocean Springs, MS (30°15?53?N, 89°06?48?W). Water samples (four 1-liter sam-
ples from the Alabama site and one 1-liter sample from the Mississippi site) were
collected on average three times per week in sterile 1-liter wide-mouth bottles
(Nalgene, Rochester, NY) as recommended previously (1). Oyster samples
(?1-m depth) were collected once per week using tongs at the Alabama site; at
the Mississippi site, oysters were collected by hand from tethered holding baskets
resting on the bottom. Six (Mississippi) and 12 (Alabama) oysters were analyzed
within 1 hour of collection. Surface and bottom water temperature and salinity
were measured using a YSI model 85 salinometer (Yellow Springs, OH) or by
using a calibrated thermometer and handheld refractometer. A water sample was
also taken for chlorophyll and turbidity analyses using a benchtop fluorometer
(model TD-700; Turner) and turbidimeter (model 2020; LaMotte).
Oysters were cleaned and shucked as recommended previously (1). Oysters
(200 to 250 g) were blended with an equal weight of phosphate-buffered saline
(PBS; 7.65 g NaCl, 0.724 g Na2HPO4[anhydrous], 0.21 g KH2PO4per liter of
distilled water, pH 7.4). A 1:10 dilution was made in PBS by adding 20 g of the
1:1 homogenate to 80 ml of PBS.
Direct plating/colony hybridization. Analysis of the water and oyster samples
for V. parahaemolyticus adhered to the FDA protocol (32). Specifically, 1 ml of
water was spread plated on triplicate T1N3agar plates (1% tryptone, 3% NaCl,
2% agar, pH 7.2). Aliquots of oyster homogenate equal to 0.1 g (0.2 g of the 1:1
homogenate) and 0.01 g (100 ?l of the 1:10 dilution) were spread plated on
triplicate T1N3agar plates.
After overnight incubation at 35°C, bacterial colonies were lifted from the
T1N3agar plates using Whatman no. 541 filters (8.5-cm diameter; Whatman Inc.,
Florham Park, NJ), and filters were analyzed by DNA hybridization using alka-
line phosphatase-labeled gene probes (DNA Technology, Aarhus, Denmark) for
the detection of the tlh, tdh, and trh genes, as previously described (12, 27, 32).
Enrichment and real-time PCR–most probable number (MPN) analysis. Vol-
umes of 1 liter, 100 ml, and 10 ml of Alabama waters were enriched with 110 ml,
11 ml, and 1.1 ml, respectively, of 10? APW (10% peptone, pH 8.5 ? 0.2). These
large sample sizes were used because enrichment followed by real-time PCR
detection improved sensitivity, especially for pathogenic V. parahaemolyticus, which
is usually below the LOD for colony hybridization, in which only 1 ml is exam-
ined. Many of the Mississippi water samples had a salinity of ?10 ppt, and 10?
APW supplemented with 10% NaCl was used in these samples for optimal
growth for V. parahaemolyticus. All water sample enrichments were performed in
Four sterile 100-ml bottles containing 80 ml of APW were inoculated with 20 g
of the 1:1 homogenate described above. Three were used as 10-g oyster enrich-
ment samples, and the fourth bottle was used to inoculate 10 ml of homogenate
into triplicate tubes for the 1-g samples.
After incubation, 100 ?l from each enrichment tube or bottle were placed into
0.5-ml Microfuge tubes, which were heated to 100°C for 10 min in a model 200
Peltier thermal cycler (MJ Research, Inc., Watertown, MA) and immediately
frozen at ?20°C until analysis. Enrichments were analyzed by real-time PCR–
MPN, as described by Nordstrom et al. (28), using a multiplex assay for the
detection of the tlh, tdh, and trh genes of V. parahaemolyticus. This assay includes
an internal amplification control (IAC) for the detection of possible matrix
inhibition. A positive control (tlh?tdh?trh?V. parahaemolyticus strain) and a
negative control (distilled water) were included for each PCR master mix.
Statistical analyses. Generalized linear mixed-model (GLMM) regressions
were used to estimate the distributions and correlations between total (tlh) and
pathogenic (tdh and trh) V. parahaemolyticus densities in oyster and water sam-
ples, as well as their relationship to environmental parameters. The significance
of contemporaneous associations with environmental factors as predictors of V.
parahaemolyticus densities was evaluated using a stepwise backward selection
procedure. In univariate and multivariate analyses, the distribution of V. para-
haemolyticus densities was assumed to be lognormal, with mean log10densities
being either constant or linearly related to environmental parameters. GLMM
regression parameters were estimated by considering the results of colony hy-
bridization or real-time PCR–MPN assays from multiple dilutions obtained from
the same sample as repeated measures, marginally distributed as either Poisson
or binomial outcomes, respectively, based on the volume of sample examined.
The approach of using the GLMM (11, 33) with discrete mixed-type response
variables and latent (underlying) lognormal distribution was considered an ap-
propriate method for combining the plate count and real-time PCR–MPN data
in a manner that appropriately weighted the outcomes of the different methods
in an inverse proportion to their inherent measurement errors. The resulting
estimates of the variation in log10V. parahaemolyticus densities were thereby
corrected for the most prominent factor determining measurement error (i.e.,
assay or dilution volume). For univariate GLMM analyses, a log transformation
of the scale parameter was used to improve numerical stability and asymptotic
properties of statistical estimates and derived confidence intervals. Similarly, for
multivariate GLMM analyses, a spherical parameterization of the variance-
covariance matrix was used (30). Because measurements of pathogenic V. para-
haemolyticus (tdh and trh) densities were frequently below the LOD, the pro-
portion of the variation explained by environmental parameters was evaluated
using Nagelkerke’s pseudo-R2statistic (26) to estimate partial coefficients of
determination, as described by Lipsitz et al. (24). The fit of all GLMMs to the
data were evaluated using the deviance statistic as a goodness-of-fit measure, and
all fits were found to be adequate.
For graphical presentation of the data, a generalized linear model ap-
proach was used to obtain maximum likelihood estimates of the densities of
tlh, tdh, and trh organisms corresponding to each sample based on the com-
bined outcomes of both enumeration methods. For samples enumerated by
real-time PCR alone, this corresponds to the usual estimates for three-tube
MPN format data. For samples enumerated by colony hybridization and
real-time PCR, the resulting estimate is a weighted combination of the
estimates that would be obtained by considering the outcome of each method
alone. This weighting is based on the maximum likelihood principle. When
densities were below the LOD by both methods, half the LOD of the most
sensitive method (generally real-time PCR) was used for graphing purposes.
Improbable MPNs in the real-time PCR data were corrected prior to the
GLMM analysis by excluding the questionable outcome (i.e., based on the
inhibition of an internal amplification control) in lower dilutions. The signif-
icance of site differences in the distribution of salinity, turbidity, and chlo-
rophyll was evaluated by the Mann-Whitney-Wilcoxon test. All statistical
analyses were conducted using the SAS statistical software (SAS Institute,
Cary, NC); significance was defined as a P of ?0.05, and relations with P
values in the range of 0.05 to 0.10 were considered marginally significant.
Water temperatures were similar at the Mississippi and
Alabama sites (ranging from 22.4 to 33.8°C), but there were
statistically significant (P ? 0.001) site differences in the
distributions of salinity, turbidity, and chlorophyll (Fig. 1).
The salinity range was higher at the Alabama site (10 to 28
ppt) than at the Mississippi site (4 to 13 ppt) from May
7590 ZIMMERMAN ET AL.APPL. ENVIRON. MICROBIOL.
through June, but the ranges were similar during July and
August. Turbidity and chlorophyll exhibited day-to-day vari-
ability at both sites, with higher levels being observed at the
Mississippi site; median turbidity and chlorophyll levels
were 10.9 m?1and 13.5 mg/m3at the Mississippi site, com-
pared to 6.1 m?1and 7.9 mg/m3at the Alabama site. At both
sampling sites, V. parahaemolyticus densities were generally
higher (?2 logs) and more consistent in oysters than in
water (Fig. 2).
A summary and inferential statistics of V. parahaemolyticus
measurements in water and oysters are shown in Table 1. With
the exception of one water sample collected at the Alabama
site and examined only by DNA probe colony hybridization
(LOD ? 1 CFU ml?1), all samples contained detectable levels
of total V. parahaemolyticus (tlh) organisms. Because of the
infrequency of tdh?and trh?V. parahaemolyticus strains in
environmental samples, most of the information on total V.
parahaemolyticus (tlh) densities was obtained using colony hy-
bridization, and most of the information on pathogenic V.
parahaemolyticus (tdh and trh) densities was obtained from
The previously described IAC was used to differentiate PCR
inhibition from negative signals (28). Complete inhibition was
observed only with a single MPN tube inoculated with a 10-g
portion of oyster homogenate from Mississippi, and partial
inhibition (threshold cycle, 21 to 30) was generally limited to
oysters collected from Mississippi. No adjustments were made
to data from MPN tubes in which the IAC was partially inhib-
ited because often one or more of the V. parahaemolyticus
targets (10/15 for tlh, 3/15 for tdh, and 1/15 trh) were detected
in these samples.
Pathogenic V. parahaemolyticus (tdh?and/or trh?) organ-
isms were detected in 56% and 78% of Mississippi oyster
and water samples and in 44% and 30% of Alabama oyster
and water samples, respectively. Mean densities of total V.
parahaemolyticus (tlh) were higher in oysters than in water
at both sampling sites (Fig. 3). Mean densities of pathogenic
V. parahaemolyticus (tdh?and trh?) organisms were similar
FIG. 1. Environmental parameters (temperature, salinity, turbidity, and chlorophyll) measured at the Alabama and Mississippi sampling
VOL. 73, 2007DENSITY OF V. PARAHAEMOLYTICUS IN WATER AND OYSTERS7591
in oyster samples from Mississippi and Alabama and lower
in water samples from Alabama. Estimated standard devia-
tions indicate greater variability among tdh?and trh?V.
parahaemolyticus densities than among tlh?V. parahaemo-
lyticus densities in both water and oyster samples at both
sites (Table 1).
Based on multivariate GLMM analysis, total V. parahaemo-
lyticus (tlh) densities in oysters and water were significantly
correlated at the Mississippi site (r ? 0.89; P ? 0.001) but not
at the Alabama site. Similar analyses comparing tdh and trh
organism densities indicated positive correlations for each
sample type and site. The estimated correlations were statisti-
cally significant (P ? 0.001) when tdh and trh organism densi-
ties in water were compared but only marginally significant
(P ? 0.10) when tdh and trh organism densities in oysters were
compared. These positive associations between tdh and trh
organism densities in water and oysters are clearly evident in
The association of water temperature, salinity, turbidity, and
chlorophyll with V. parahaemolyticus densities differed at the
two sampling sites. Water temperature varied from 22.4 to
33.8°C and was not significantly correlated with either total or
pathogenic V. parahaemolyticus densities at either site. No sig-
nificant associations between V. parahaemolyticus and these
environmental parameters were evident at the Alabama site. In
contrast, at the Mississippi site, salinity and turbidity were
positively associated with V. parahaemolyticus densities in wa-
ter and oysters. The results of the GLMM regression analyses
for the Mississippi data are shown in Table 2. As indicated by
estimated regression coefficients, the associations between to-
tal V. parahaemolyticus (tlh) densities and these two parame-
ters were consistent across the sample types (water versus
oyster). The associations were highly significant (P ? 0.001),
except the association between turbidity and V. parahaemo-
lyticus (tlh) densities in oysters, which was only marginally signif-
icant (P ? 0.081). Based on the regression coefficients, point
estimates of the effects that salinity (in parts per thousand)
had on total numbers of V. parahaemolyticus (tlh) organisms
were 0.10 log10ml?1and 0.12 log10g?1for water and oys-
ters, respectively. That is, a 10-ppt change in salinity was
estimated to correspond to a 1.0- and a 1.2-log10change in
total numbers of V. parahaemolyticus (tlh) organisms in wa-
FIG. 2. Vibrio parahaemolyticus (tlh, tdh, and trh) levels in water and oysters at the Alabama and Mississippi sampling sites.
7592 ZIMMERMAN ET AL.APPL. ENVIRON. MICROBIOL.
ter and oysters, respectively. For turbidity, the regression
coefficient estimates differed by smaller amounts: 0.043
log10ml?1and log10g?1m?1for water and oysters, respec-
tively. Consistent with the differences in effect sizes, salinity
corresponded to more of the variation in V. parahaemo-
lyticus (tlh) densities than did turbidity. Partial coefficients of
determination (i.e., pseudo-R2’s) summarizing the amount of
variation explained by each factor are shown in Table 2. Based
on these statistics, salinity was estimated to be associated with
ca. 64% and 76% of the variation of V. parahaemolyticus (tlh)
densities in water and oysters, respectively. Similarly, turbidity
was estimated to be associated with ca. 26% and 18% of the
variation. A graphical summary of the correlations of turbidity
with V. parahaemolyticus (tlh) densities in water is shown in
Fig. 4. Although similar patterns of statistical significance and
magnitude of effect sizes were evident for salinity and turbidity
versus pathogenic V. parahaemolyticus (tdh and trh) densities at
the Mississippi site, the results were inconsistent between
sample types. The proportion of variation in pathogenic V.
parahaemolyticus densities associated with salinity and tur-
bidity was generally much smaller than that estimated for
total V. parahaemolyticus (tlh) densities (Table 2).
The present study examined intraseasonal relationships be-
tween selected environmental parameters (chlorophyll, turbid-
ity, and salinity) and the densities of total and pathogenic V.
parahaemolyticus organisms in the northern Gulf of Mexico
over a sampling period during which the effects of water tem-
perature were minimal. The determination of predictive in-
traseasonal associations may help identify the environmental
conditions under which V. parahaemolyticus is most likely to
persist at high densities. If strong environmental signatures
exist and can be identified to be predictive of high and persis-
tent V. parahaemolyticus abundance, then monitoring of water
quality via remote sensing may prove to be an effective tool for
controlling or mitigating human exposure.
The present study identified marked and persistent differ-
ences between geographic sites and between ratios of total
organisms to pathogenic V. parahaemolyticus organisms.
Pathogenic V. parahaemolyticus densities in both Alabama and
Mississippi oysters and in overlying waters fluctuated more
than did total V. parahaemolyticus densities. Additionally,
higher total V. parahaemolyticus densities were observed in
Alabama samples, but higher pathogenic V. parahaemolyticus
densities were observed in Mississippi samples. Quantitative
data on pathogenic V. parahaemolyticus densities have been
limited in previous studies and lacked the precision needed to
address the issue of spatial variation (e.g., site-to-site differ-
ences). Furthermore, pathogenic V. parahaemolyticus esti-
mates have been based on the ratios of pathogenic to total V.
parahaemolyticus isolates grown on thiosulfate citrate bile su-
crose agar (TCBS) after overnight APW enrichment (2). These
ratios could have been affected by differing growth rates in
APW or differing plating efficiencies on TCBS between patho-
genic and nonpathogenic V. parahaemolyticus strains. The use
of the real-time PCR–MPN format in the present study en-
hances sensitivity by allowing for the inoculation of large sam-
FIG. 3. Maximum likelihood estimates and 95% confidence inter-
vals for mean log10Vibrio parahaemolyticus (tlh, tdh, trh) densities in
Alabama versus in Mississippi oysters and water.
TABLE 1. Observed and estimated distributions of total (tlh) and pathogenic (tdh and trh) V. parahaemolyticus organisms in oysters and
water collected from Mississippi and Alabama sites
unit of measure
No. of positive
% of samples that
MS water tlh/ml
0.65 (0.45, 0.86)
?1.90 (?2.12, ?1.69)
?2.57 (?2.87, ?2.26)
0.68 (0.53, 1.15)
0.71 (0.52, 0.98)
0.79 (0.54, 1.16)
MS oyster tlh/g
1.90 (1.50, 2.29)
?1.34 (?2.03, ?0.65)
?2.87 (?4.76, ?0.97)
0.70 (0.43, 1.10)
1.03 (0.55, 1.91)
2.00 (0.68, 5.86)
1.18 (1.06, 1.29)
?4.13 (?4.68, ?3.59)
?4.52 (?5.26, ?3.79)
0.37 (0.29, 0.47)
0.88 (0.54, 1.44)
1.04 (0.62, 1.76)
2.91 (2.79, 3.03)
?2.12 (?3.66, ?0.58)
?1.74 (?2.57, ?0.92)
0.21 (0.14, 0.32)
1.90 (0.91, 3.98)
0.97 (0.45, 2.11)
aND, not detected.
bEstimated values are followed by parenthetical 95% confidence intervals for log10V. parahaemolyticus (tlh, tdh, trh) densities based on GLMM regression analyses.
VOL. 73, 2007DENSITY OF V. PARAHAEMOLYTICUS IN WATER AND OYSTERS7593
ple portions (?3 liters of water and ?30 g of oysters) and the
examination of many V. parahaemolyticus cells (log104) from
the APW enrichment by real-time PCR without the tedious
and resource-intensive approach of colony isolation, identi-
fication, and characterization on TCBS or other growth me-
Differences in salinity, turbidity, and chlorophyll levels be-
tween the two sampling sites during the study period may
account for some of the discrepancies in V. parahaemolyticus
densities. A significant correlation between salinity and total
(tlh) V. parahaemolyticus densities in water and oysters was
identified at the Mississippi site. The lack of significance for
salinity at the Alabama site is likely a consequence of a higher
and narrower range of salinity levels at that site than at the
Mississippi site. Salinities at the Mississippi site were generally
?10 ppt during the first half of the study, which is well below
the reported optimum salinity of 23 ppt for V. parahaemolyticus
growth (2). When salinity increased at the Mississippi site in
mid-July, densities of total V. parahaemolyticus organisms in-
creased ca. 10-fold to levels typical of the Alabama site. Pre-
vious studies indicated that V. parahaemolyticus densities de-
crease as salinity increases (2, 14, 15). In the present study, the
densities of pathogenic V. parahaemolyticus organisms were
generally positively associated with increasing salinity. The
GLMM regression analyses indicated no significant association
between V. parahaemolyticus densities and water temperature
at either site. This finding is consistent with the relatively
narrow range of temperature variations during the study time
Turbidity was generally higher in Mississippi than in Ala-
bama, and regression analysis indicated a positive association
between turbidity and V. parahaemolyticus densities at the Mis-
sissippi site. Lower levels and less variability in turbidity may
have obscured the effects of turbidity at the Alabama site. The
association between turbidity and V. parahaemolyticus densities
at the Mississippi site was generally consistent and statistically
FIG. 4. Relationship of total numbers of V. parahaemolyticus (tlh)
organisms in water to salinity and turbidity at the Mississippi site.
(A) Correlation plot of log10numbers of tlh organisms/ml versus sa-
linity; (B) partial residual plot of tlh numbers/ml, adjusted for salinity,
TABLE 2. Measures of association between numbers of total (tlh) and pathogenic (tdh and trh) V. parahaemolyticus organisms and
unit of measure
Partial coefficient of
0.10 (0.079, 0.12)
0.0433 (0.023, 0.064)
MS oyster tlh/g
0.12 (0.082, 0.15)
0.0431 (?0.006, 0.092)
Turbidity 0.063 (0.023, 0.10)0.0034
0.11 (0.003, 0.22)
0.18 (0.03, 0.33)
?0.076 (?0.13, ?0.02)
0.060 (0.017, 0.10)
MS oyster trh/g
0.26 (0.07, 0.45)
0.38 (0.10, 0.66)
aMaximum likelihood estimates and parenthetical 95% confidence intervals of regression coefficients corresponding to environmental parameters identified to be
significant (P ? 0.05) or marginally significant (P ? 0.10) by backward stepwise identification in GLMM analyses.
bBased on Nagelkerke’s pseudo-R2statistic.
7594ZIMMERMAN ET AL.APPL. ENVIRON. MICROBIOL.
significant (P ? 0.05) except with total V. parahaemolyticus in
oysters, which association was only marginally significant (P ?
0.081). However, the effect sizes for turbidity on total V. para-
haemolyticus organisms were nearly identical for water and
oysters, and the lack of statistical significance per se is likely a
consequence of the smaller number of oyster samples ana-
lyzed. Due to the discrete nature of the data for pathogenic V.
parahaemolyticus, a pseudo-R2measure of association was
used to evaluate the proportion of variation explained. Based
on this statistic, turbidity was associated with up to 26% of the
variation in total V. parahaemolyticus densities and up to 23%
of the variation in pathogenic V. parahaemolyticus densities.
Although statistically significant associations between turbidity
and pathogenic V. parahaemolyticus densities were identified,
the estimates of effect sizes were not as robust as with total V.
parahaemolyticus densities. The adequacy of the fitted regres-
sion model was found to be more problematic for pathogenic
than for total V. parahaemolyticus strains. Increased turbidity
could potentially affect V. parahaemolyticus densities in water
and oysters in various ways. The most obvious is by resuspen-
sion of sediments, which have higher V. parahaemolyticus den-
sities than water, and subsequent uptake by oyster filtration.
Nutrients that were previously sediment bound may also be-
come more available in the water column, resulting in more V.
parahaemolyticus growth. On the other hand, high turbidity
resulting from the suspension of inorganic particles or other
particles that are not digestible by oysters may cause them to
shut down filter feeding and allow V. parahaemolyticus to ac-
cumulate in their tissues (2).
No significant associations with chlorophyll were observed at
either site. This is contradictory to our previous study (29),
which indicated a significant association (P ? 0.05) between
remotely sensed chlorophyll and total V. parahaemolyticus or-
ganisms, after correction for the effects of temperature and
salinity. However, the previous data were collected year round,
and the current data were collected only during warm weather,
when chlorophyll levels are typically higher. It is possible that
a relationship is evident in a comparison of the effects of
seasonal differences in chlorophyll levels that are not similarly
apparent in the effects of day-to-day variations within a single
season. It is also possible that the effects of day-to-day varia-
tions have a stronger temporal (i.e., lagging) relationship than
those of other environmental parameters and cannot be ap-
propriately identified based on contemporaneous statistical as-
sociations alone. However, the examination of potential tem-
poral associations based on the current data was found to be
problematic due to intermittent sampling dates over the course
of the study.
Other variables (chemical, biological, geological, and hydro-
lyticus densities at the two study sites. The Mississippi site is
within a highly eutrophic estuary known as the Mississippi
Sound. While this site is protected seaward from the Gulf of
Mexico by barrier islands, the ca. 1-km fetch between the site
and the innermost island allows some wave energy to impact
the site, especially during southerly wind flows, which prevailed
during the study period. A marine aquaculture facility housed
at the GCRL adjacent to the sampling site was considered a
potential source of pathogenic V. parahaemolyticus, but tests
did not detect V. parahaemolyticus in culture raceways or in the
effluent. The Alabama site is located along a bulkhead shore-
line in a small (2- to 3-km) bay on the north side of Dauphin
Island near the mouth of Mobile Bay. The low-wave energy,
especially during the study period, when prevailing winds were
from southerly directions, and relative isolation from major
freshwater sources (the Mobile River system is approximately
50 km to the north and had relatively low flow during the study
period) contributed to the lower turbidity and higher salinity at
Total V. parahaemolyticus densities in oysters paralleled
those in the overlying waters at both sites, as previously re-
ported (13). Regression analyses indicated a significant corre-
lation between total V. parahaemolyticus densities in oysters
and water collected on the same day at the Mississippi site but
not at the Alabama site. The correlation at the Mississippi site
was likely induced by the effects of salinity and turbidity, and
the lack of correlation at the Alabama site is consistent with
the absence of identifiable environmental effects at that site
and relatively constant V. parahaemolyticus densities. However,
oysters and their overlying waters were much greater in
Alabama than in Mississippi. It is possible that Alabama oys-
ters retain pathogenic V. parahaemolyticus, perhaps by coloni-
zation, even when their densities in the overlying water are
very low. These data suggest that V. parahaemolyticus popula-
tions in oysters and their overlying waters are controlled quan-
titatively and qualitatively by different factors. It is unlikely that
selective filtration of pathogenic V. parahaemolyticus alone can
account for the magnitude of different concentrations in oys-
ters at the two sites. Previous studies demonstrated the poten-
tial for V. parahaemolyticus growth (2), phagocytosis (16), and
phage lysis (3) in oysters, and these processes could affect
ratios of pathogenic to nonpathogenic V. parahaemolyticus or-
ganisms in oyster tissues.
The real-time PCR assay has two advantages over other
assays (28, 34): (i) the IAC is included to eliminate reporting of
false-negative results caused by sample matrix inhibition and
(ii) concentrations of tlh primers are limited, which improves
the detection of pathogenic targets (tdh and trh) in the pres-
ence of high densities of total V. parahaemolyticus (tlh) organ-
isms. By applying this assay to 10-fold serial dilutions of oysters
in APW in an MPN-PCR format, the LOD is lowered ?300-
fold (i.e., a three-tube MPN starting with a sample of 10 g of
homogenate has a theoretical LOD of 0.03 MPN g?1) over the
10-CFU g?1LOD for the colony hybridization method.
The IAC data indicated that matrix inhibition of the PCR in
the current study was primarily limited to oysters collected
from Mississippi. Partial matrix inhibition would likely affect
the detection of tdh and trh more than tlh, since they are
usually present at much lower copy numbers. The impact of
matrix inhibition on estimates of pathogenic V. parahaemo-
lyticus densities was reduced by the adjustments made with
improbable MPN data (excluding data at lower dilutions [10-g
portions] when more tubes at higher dilutions [1-g portions]
lyticus targets at the lower dilutions when higher dilutions were
positive appeared to be unrelated to matrix inhibition of the
PCR because the IAC was not inhibited. Failure to detect the
V. parahaemolyticus targets was likely due to their copy number
in APW enrichments being below the assay LOD of 500 ml?1.
VOL. 73, 2007DENSITY OF V. PARAHAEMOLYTICUS IN WATER AND OYSTERS 7595
The low frequency of partial IAC inhibition in Alabama oys- Download full-text
ters (ca. 1%) and all water samples (?1%) is similar to that
reported for this V. parahaemolyticus assay in a study of Alas-
kan oysters (28). While the IAC assists with the identification
of the few samples with matrix inhibition and prevents report-
ing of false-negative results, efficient methods to eliminate
these inhibitors need to be developed.
In conclusion, these results demonstrate greater temporal
and spatial variations in the densities of pathogenic V. para-
haemolyticus organisms in Gulf oysters during warm weather
than those observed for total-V. parahaemolyticus densities;
hence, there may be more uncertainty in the use of densities of
total V parahaemolyticus organisms as a surrogate for risk pre-
dictions than was previously recognized. While these results
underscore a continued poor understanding of the ecology and
distribution of pathogenic bacteria, including V. parahaemolyticus
in coastal environments (17), other important aspects of V.
parahaemolyticus risk assessment, such as seasonality and post-
harvest growth, are firmly established for risk predictions. The
site-to-site variability demonstrated in this study emphasizes
the need to use remote-sensing algorithms that are site specific
and to take niche-based signatures into account.
This study was funded by an Oceans and Human Health Initiative
grant (NA-04-OAR4600214) from the NOAA.
We are grateful to the following individuals and agencies for pro-
viding laboratory equipment, space, and aid during sample collection/
analyses: John Tennyson and Angela Ruple, NOAA, National Marine
Fisheries Service, National Seafood Inspection Laboratory; Scott Gor-
don, Mississippi Department of Marine Resources; Dawn Rebarchik,
GCRL; and George Blackstone, GCSL.
1. American Public Health Association. 1970. Recommended procedures for
the examination of seawater and shellfish, 4th ed. American Public Health
Association, Washington, DC.
2. Anonymous. 2005. Quantitative risk assessment on the public health impact
of pathogenic Vibrio parahaemolyticus in raw oysters. U.S. Food and Drug
Administration, Washington, DC.
3. Baross, J. A., P. A. Tester, and R. Y. Morita. 1984. Incidence, microscopy and
etiology of exoskeleton lesions in the tanner crab Chionoecetes tanneri. J.
Fish. Res. Board Can. 35:1141–1149.
4. Centers for Disease Control and Prevention. 1999. Outbreak of Vibrio para-
haemolyticus infection associated with eating raw oysters and clams har-
vested from Long Island Sound—Connecticut, New Jersey, and New York.
Morb. Mortal. Wkly. Rep. 48:48–51.
5. Centers for Disease Control and Prevention. 1998. Outbreak of Vibrio para-
haemolyticus infections associated with eating raw oysters—Pacific North-
west, 1997. Morb. Mortal. Wkly. Rep. 47:457–462.
6. Colwell, R. R., and D. J. Grimes. 1984. Vibrio diseases of marine fish pop-
ulations. Helgol. Meeresunters 37:265–287.
7. Colwell, R. R., and J. Kaper. 1977. Vibrio cholerae, Vibrio parahaemolyticus,
and other vibrios: occurrence and distribution in Chesapeake Bay. Science
8. Cook, D. W., J. D. Bowers, and A. DePaola. 2002. Density of total and
pathogenic (tdh?) Vibrio parahaemolyticus in Atlantic and Gulf Coast mol-
luscan shellfish at harvest. J. Food Prot. 65:1873–1880.
9. Daniels, N. A., L. MacKinnon, R. Bishop, S. Altekruse, B. Ray, R. M.
Hammond, S. Thompson, S. Wilson, N. H. Bean, P. M. Griffin, and L.
Slutsker. 2000. Vibrio parahaemolyticus infections in the United States, 1973–
1998. J. Infect. Dis. 181:1661–1666.
10. Daniels, N. A., B. Ray, A. Easton, N. Marana, E. Kahn, A. L. McShan, L. Del
Rosario, T. Baldwin, M. A. Kingsley, N. D. Puhr, J. G. Wells, and F. J.
Angulo. 2000. Emergence of a new Vibrio parahaemolyticus serotype in raw
oysters. JAMA 284:1541–1545.
11. Davidian, M., and D. M. Giltinan. 1995. Nonlinear models for repeated
measurement data. Chapman & Hall, New York, NY.
12. Deepanjali, A., H. S. Kumar, I. Karunasagar, and I. Karunasagar. 2005.
Seasonal variation in abundance of total and pathogenic Vibrio parahaemo-
lyticus bacteria in oysters along the southwest coast of India. Appl. Environ.
13. DePaola, A., L. H. Hopkins, J. T. Peeler, B. Wentz, and R. M. McPhearson.
1990. Incidence of Vibrio parahaemolyticus in U.S. coastal waters and oysters.
Appl. Environ. Microbiol. 56:2299–2302.
14. DePaola, A., C. A. Kaysner, J. Bowers, and D. W. Cook. 2000. Environmental
investigations of Vibrio parahaemolyticus in oysters after outbreaks in Wash-
ington, Texas, and New York (1997 and 1998). Appl. Environ. Microbiol.
15. DePaola, A., J. L. Nordstrom, J. C. Bowers, J. G. Wells, and D. W. Cook.
2003. Seasonal abundance of total and pathogenic Vibrio parahaemolyticus in
Alabama oysters. Appl. Environ. Microbiol. 69:1521–1526.
16. Genthner, F. J., A. K. Volety, L. M. Oliver, and W. S. Fisher. 1999. Factors
influencing in vitro killing of bacteria by hemocytes of the eastern oyster
(Crassostrea virginica). Appl. Environ. Microbiol. 65:3015–3020.
17. Grimes, D. J. 1991. Ecology of estuarine bacteria capable of causing human
disease: a review. Estuaries 14:345–360.
18. Grimes, D. J., P. Brayton, R. R. Colwell, and S. H. Gruber. 1985. Vibrios as
autochthonous flora of neritic sharks. Syst. Appl. Microbiol. 6:221–226.
19. Hervio-Heath, D., R. R. Colwell, A. Derrien, A. Robert-Pillot, J. M. Fournier,
and M. Pommepuy. 2002. Occurrence of pathogenic vibrios in the coastal
areas of France. J. Appl. Microbiol. 92:1123–1135.
20. Hlady, G. A., and K. C. Klontz. 1996. The epidemiology of vibrio infections
in Florida, 1981–1993. J. Infect. Dis. 173:1176–1183.
21. Huq, A., and R. R. Colwell. 1996. Vibrios in the marine and estuarine
environment: tracking Vibrio cholerae. Ecosyst. Health 2:198–214.
22. Huq, A., and R. R. Colwell. 1995. Vibrios in the marine and estuarine
environments. J. Mar. Biotechnol. 3:60–63.
23. Levine, W. C., and P. M. Griffin. 1993. Vibrio infections on the Gulf Coast:
results of first year of regional surveillance. J. Infect. Dis. 167:479–483.
24. Lipsitz, S. R., T. Leong, J. Ibrahim, and S. Lipschultz. 2001. A partial
correlation coefficient and coefficient of determination for multivariate nor-
mal repeated measures data. Statistician 50:87–95.
25. Lobitz, B., L. Beck, A. Huq, B. Wood, G. Fuchs, A. S. G. Faruque, and R. R.
Colwell. 2000. Climate and infectious disease: use of remote sensing for
detection of Vibrio cholerae by indirect measurement. Proc. Natl. Acad. Sci.
26. Nagelkerke, N. J. D. 1991. A note on a general definition of the coefficient
of determination. Biometrika 78:691–692.
27. Nordstrom, J. L., R. Rangdale, M. C. L. Vickery, A. M. B. Phillips, S. L.
Murray, and A. DePaola. 2006. Evaluation of an alkaline phosphatase-
labeled oligonucleotide probe for the detection and enumeration of the
thermostable-related hemolysin (trh) gene of Vibrio parahaemolyticus. J.
Food Prot. 69:2770–2772.
28. Nordstrom, J. L., M. C. L. Vickery, G. M. Blackstone, S. L. Murray, and A.
DePaola. 2007. Development of a multiplex real-time PCR assay with an
internal amplification control for the detection of total and pathogenic Vibrio
parahaemolyticus bacteria in oysters. Appl. Environ. Microbiol. 73:5840–
29. Phillips, A. M. B., A. DePaola, J. Bowers, S. Ladner, and D. J. Grimes. 2007.
An evaluation of the use of remotely sensed parameters for prediction of
incidence and risk associated with Vibrio parahaemolyticus in Gulf Coast
oysters (Crassostrea virginica). J. Food Prot. 70:879–884.
30. Pinheiro, J. C., and D. M. Bates. 1996. Unconstrained parametrizations for
variance-covariance matrices. Stat. Comput. 6:289–296.
31. U.S. Food and Drug Administration Center for Food Safety and Applied
Nutrition. 1986. Sanitation of the harvesting and processing of shellfish.
National Shellfish Sanitation Program manual of operations, part II. U.S.
Food and Drug Administration, Washington, DC.
32. U.S. Food and Drug Administration Center for Food Safety and Applied
Nutrition. 2004. Vibrio cholerae, V. parahaemolyticus, V. vulnificus, and other
Vibrio spp. Bacteriological analytical manual online, chapter 9. http://www
33. Vonesh, E. F., and V. M. Chinchilli. 1997. Linear and nonlinear models for
the analysis of repeated measurements. Marcel Dekker, Inc., New York, NY.
34. Ward, L. N., and A. K. Bej. 2006. Detection of Vibrio parahaemolyticus in
shellfish by use of multiplexed real-time PCR with TaqMan fluorescent
probes. Appl. Environ. Microbiol. 72:2031–2042.
35. Watkins, W. D., and V. J. Cabelli. 1985. Effect of fecal pollution on Vibrio
parahaemolyticus densities in an estuarine environment. Appl. Environ. Mi-
7596 ZIMMERMAN ET AL.APPL. ENVIRON. MICROBIOL.