Application of enterococci antibiotic resistance patterns for contamination source identification at...

8
Application of enterococci antibiotic resistance patterns for contamination source identification at Huntington Beach, California Samuel Choi a , Weiping Chu a , Jennifer Brown a , Sarah J. Becker a , Valerie J. Harwood b , Sunny C. Jiang a, * a Department of Environmental Analysis and Design, University of California, 1367 SE II, Irvine, CA 92697, USA b Department of Biology, University of South Florida, Tampa, FL 33620, USA Abstract Huntington Beach, California, one of the most popular surfing spots in the world, is plagued by sporadic, elevated levels of fecal bacteria. To assist with pollution source identification, we analyzed antibiotic resistance patterns (ARPs) of enterococci from four known sources (bird feces, urban runoff, coastal marsh sediment and sewage effluent from local sanitation district) and one un- known source (seawater) using seven antibiotics at four concentrations each. Of 2491 enteroccoci tested, all were resistant to at least one antibiotic at some level. Discriminant analysis indicated that the average correct classification rates for bird feces and urban runoff sources were above 80%. Sewage effluent contained mixed fecal sources. Sixty-four percent of the sewage isolates classified with the sewage category, while the other 35% of isolates were assigned evenly across the other three categories. When enterococci isolated from the seawater were classified using the known ARP database, it was evident that bird feces were the source of surf zone contamination on some days while the coastal salt marsh and sewage plume may have impacted the surf zone water quality to various degrees during other times. Ó 2003 Elsevier Science Ltd. All rights reserved. Keywords: Enterococcus; Antibiotic resistance pattern; Fecal bacteria; Pollution source identification; Bacterial source tracking; Huntington Beach; California 1. Introduction Huntington Beach, California, located approximately 40 miles south of Los Angeles and 60 miles north of San Diego, is one of the most popular surfing spots in the world, and is the site of the annual US Open Surfing Competition. However, a stretch of Huntington State beaches (between 6N and 9N, Fig. 1) is plagued by sporadic elevated levels of fecal indicator bacteria (FIB). To protect public safety, on July 1, 1999, the Orange County Health Care Agency closed a portion of Hun- tington State Beach to water-contact recreational ac- tivities. Closures continued throughout the summer, affecting up to six of the eight miles of state and city beaches for nearly three months (OCSD, 1999). A three-month risk-based source investigation led by the Orange County Sanitation District (OCSD) follow- ing the beach closures could not identify any significant source of sewage contamination due to infrastructure damage (OCSD, 1999), promoting speculation that urban runoff from non-point sources was the source of fecal pollution. Studies of the effect of urban runoff from nearby Talbert Marsh and the Santa Ana River, located approximately 1.7 miles down coast from the initial contamination site, suggest that these sources may contribute to the elevated levels of fecal indicator bac- teria but are not the sole contributor to surf zone con- tamination during the dry season (OCSD, 1999; Grant et al., 2001). A multi-disciplinary investigation of the impact of Talbert Marsh on surf zone water quality indicated that this coastal marsh can serve as a major source of * Corresponding author. Tel.: +1-949-824-5527; fax: +1-949-824- 2056. E-mail address: [email protected] (S.C. Jiang). 0025-326X/03/$ - see front matter Ó 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S0025-326X(03)00046-8 www.elsevier.com/locate/marpolbul Marine Pollution Bulletin 46 (2003) 748–755

Transcript of Application of enterococci antibiotic resistance patterns for contamination source identification at...

Application of enterococci antibiotic resistance patternsfor contamination source identification

at Huntington Beach, California

Samuel Choi a, Weiping Chu a, Jennifer Brown a, Sarah J. Becker a,Valerie J. Harwood b, Sunny C. Jiang a,*

a Department of Environmental Analysis and Design, University of California, 1367 SE II, Irvine, CA 92697, USAb Department of Biology, University of South Florida, Tampa, FL 33620, USA

Abstract

Huntington Beach, California, one of the most popular surfing spots in the world, is plagued by sporadic, elevated levels of fecal

bacteria. To assist with pollution source identification, we analyzed antibiotic resistance patterns (ARPs) of enterococci from four

known sources (bird feces, urban runoff, coastal marsh sediment and sewage effluent from local sanitation district) and one un-

known source (seawater) using seven antibiotics at four concentrations each. Of 2491 enteroccoci tested, all were resistant to at least

one antibiotic at some level. Discriminant analysis indicated that the average correct classification rates for bird feces and urban

runoff sources were above 80%. Sewage effluent contained mixed fecal sources. Sixty-four percent of the sewage isolates classified

with the sewage category, while the other 35% of isolates were assigned evenly across the other three categories. When enterococci

isolated from the seawater were classified using the known ARP database, it was evident that bird feces were the source of surf zone

contamination on some days while the coastal salt marsh and sewage plume may have impacted the surf zone water quality to

various degrees during other times.

� 2003 Elsevier Science Ltd. All rights reserved.

Keywords: Enterococcus; Antibiotic resistance pattern; Fecal bacteria; Pollution source identification; Bacterial source tracking; Huntington Beach;

California

1. Introduction

Huntington Beach, California, located approximately

40 miles south of Los Angeles and 60 miles north of San

Diego, is one of the most popular surfing spots in the

world, and is the site of the annual US Open Surfing

Competition. However, a stretch of Huntington Statebeaches (between 6N and 9N, Fig. 1) is plagued by

sporadic elevated levels of fecal indicator bacteria (FIB).

To protect public safety, on July 1, 1999, the Orange

County Health Care Agency closed a portion of Hun-

tington State Beach to water-contact recreational ac-

tivities. Closures continued throughout the summer,

affecting up to six of the eight miles of state and city

beaches for nearly three months (OCSD, 1999).

A three-month risk-based source investigation led by

the Orange County Sanitation District (OCSD) follow-

ing the beach closures could not identify any significant

source of sewage contamination due to infrastructure

damage (OCSD, 1999), promoting speculation thaturban runoff from non-point sources was the source of

fecal pollution. Studies of the effect of urban runoff from

nearby Talbert Marsh and the Santa Ana River, located

approximately 1.7 miles down coast from the initial

contamination site, suggest that these sources may

contribute to the elevated levels of fecal indicator bac-

teria but are not the sole contributor to surf zone con-

tamination during the dry season (OCSD, 1999; Grantet al., 2001).

A multi-disciplinary investigation of the impact of

Talbert Marsh on surf zone water quality indicated that

this coastal marsh can serve as a major source of

*Corresponding author. Tel.: +1-949-824-5527; fax: +1-949-824-

2056.

E-mail address: [email protected] (S.C. Jiang).

0025-326X/03/$ - see front matter � 2003 Elsevier Science Ltd. All rights reserved.

doi:10.1016/S0025-326X(03)00046-8

www.elsevier.com/locate/marpolbul

Marine Pollution Bulletin 46 (2003) 748–755

enterococci (Grant et al., 2001). High concentrations of

enterococci were detected in the marsh sediments and

bird feces. A dense seabird population made up largely

of seagulls resides at the marsh. Birds are frequently

observed bathing in the surf zone and near shore waterat Huntington Beach near the 9N location where ele-

vated levels of enterococci have been frequently detected

throughout the years. They are presumably attracted to

the location by the relatively warm cooling-water dis-

charged from a nearby power plant (AES Inc.). It is

hypothesized that bird feces can be an important source

of fecal bacterial contamination at the surf zone near the

9N beach monitoring station.Another hypothesis that is currently being investi-

gated suggests that OCSD�s submarine sewage outfall

field, located �5 miles offshore from the mouth of Santa

Ana River, may be transported back to the surf zone by

oceanographic processes (Boehm et al., 2002). The po-

tential transport of fecal bacteria from the sewage field

to the surf zone is highly dependent on oceanographic

conditions such as water column stratification, the scaleof internal waves, tidal range and wind-driven currents

(Boehm et al., 2002). The signature of FIB transport

may therefore not be easily detectable through routine

monitoring.

Given the geographic location of Huntington Beach

and its surrounding area, bird feces, the offshore sewage

outfall field, urban runoff and the coastal marsh are all

possible sources of FIB. To assist with fecal contami-nation source tracking, this study attempted to use an-

tibiotic resistance patterns (ARPs) of enterococci to

classify isolates from Huntington Beach into different

source categories in order to aid pollution source iden-

tification efforts. ARPs of FIB have proven very useful in

identification of contamination sources in several earlier

studies (Kaspar et al., 1990; Wiggins, 1996; Parveen

et al., 1997; Hagedorn et al., 1999; Wiggins et al., 1999;

Harwood et al., 2000). Using fecal streptococcus ARPs,Hagedorn et al. (1999) showed correct classification rates

of over 95% for separations between animal and human

sources. Harwood et al. (2000) demonstrated that ARPs

of both fecal coliform and fecal streptococcus were ca-

pable of identifying human fecal contamination in sur-

face waters impacted by septic tank discharges, and

found that variation in fecal coliform sources in an urban

Florida watershed correlated with seasonal rainfall(Whitlock et al., 2002). Enterococci, formerly included

within the fecal streptococci, were recently adopted by

the State of California as an additional standard indi-

cator for determining marine recreational water safety

(State Assembly Bill 411) because this group of bacteria

survive longer in marine environments and are more

resistant to wastewater treatment processes than fecal

coliforms (Cabelli, 1980; Miescier and Cabelli, 1982).The concentration of Enterococcus spp. in marine and

freshwaters also correlates with the increased risk of

acquiring an infection during water recreational activity

(Cabelli, 1980; Dufour, 1984; Kay et al., 1994). However,

like other currently recognized fecal indicators, entero-

cocci are found in feces of all warm-blooded animals and

therefore share the drawback of host non-specificity with

the fecal and total coliforms.The study presented here analyzed ARPs of entero-

cocci from known sources (bird feces, sewage, urban

runoff, and marsh sediments) and one unknown source

(seawater). Discriminant function analysis was used to

determine the predictive value of ARPs for classification

of Enterococcus isolates from known fecal and envi-

ronmental sources, and to categorize each seawater

isolate into the source category based upon its ARP.

2. Materials and methods

2.1. Sample collection

Samples for enterococcus isolation were obtained

from bird feces, sewage, urban runoff, marsh sedimentsand seawater. Twenty to 150 isolates were picked at each

time depends on the density of enterococci within each

sample. Bird fecal samples were collected from birds�excretion (discrete feces) on the sidewalk along the

beach at Huntington State Beach, CA on December 1,

11, and 12, 2000 and December 28, 2001. Seagulls are

the predominant bird at Huntington Beach. The fresh

fecal samples were picked using a sterilized spatula or asterile cotton swipe and suspended in 5 ml of saline

buffer (per liter contains 8.5 g NaCl, 0.3 g KH2PO4, 0.6

g Na2HPO4, pH 7.3) in sterile tubes. Isolation of en-

terococci was performed within 2 h of sample collection.

Fig. 1. Map of study area. Seawater samples for enterococcus isolation

were collected at station 9N. Orange County Sanitation District�ssewage outfall is located �5 miles offshore from the mouth of Santa

Ana River. AES Inc. power plant�s cooling water pipe is located �1.5

miles offshore from the station 9N.

S. Choi et al. / Marine Pollution Bulletin 46 (2003) 748–755 749

Urban runoff samples were collected three times be-

tween December 2001 and April 2002 from one of the

storm water catch basins (Atlanta Station) located in the

city of Huntington Beach. Urban runoff water from thisbasin is pumped periodically into the tributary con-

nected to Talbert Marsh, which eventually empties into

the Pacific Ocean at Huntington State Beach. Sediment

samples from Talbert Marsh were also collected on four

separate dates using sterile spatulas. Sediment samples

were collected during low tide from the freshly exposed

mud flat. Only the top 2–4 mm of sediment was scraped

into a sterile tube and transported back to the lab forimmediate isolation of enterococci. Primary treated

sewage effluent (sedimentation only) was received from

OCSD in Fountain Valley, CA on December 12, 2000,

and April 24, October 16, and December 21, 2001.

Seawater samples were collected in ankle-deep water

from the beach at station 9N (Fig. 1), approximately 1.7

miles (9000 feet) north of the Santa Ana River delta on

December 12, 2000, April 9 and 30, July 16 and De-cember 21, 2001. No significant storm events were re-

corded in southern California prior to each sample

collection. Seawater samples were stored on ice in sterile

sampling bags and were processed within 2 h of collec-

tion.

2.2. Isolation and confirmation of enterococci

The bird feces, sediment, urban runoff and sewage

samples were diluted in saline buffer. All samples were

filtered onto 0.45 lm membrane filters (millipore) and

incubated cell-side up on 55-mm-diameter plates (Fisher

Scientific) containing mE agar (Difco) at 41 �C for 48 h.

The filters were then transferred to EIA (esculin iron

agar) plates and incubated at 41 �C for 20 min. Red-

pigmented colonies that developed a black shadow(precipitates) were considered enterococci and were used

for further analysis. To verify that the mE/EIA agars

were selective for gram positive cocci, gram stain tests

were performed on a random sampling of isolates in-

cluding 10% of all bacteria used in the database. Further

confirmation of all enterococci was performed by culti-

vation on brain heart infusion agar (BHI) (Difco) at

44.5 �C for 48 h. Nearly all isolates were gram-positivecocci, and grew on BHI at elevated temperature con-

firming that the agar plates were selective for entero-

cocci.

2.3. Determination of antibiotic resistance pattern

The ARP of each isolate is comprised of 28 obser-

vations (seven antibiotics at four concentrations each).With sterile tooth picks, each isolate was inoculated into

a well of a 96-well microtiter plate (Fisher Scientific)

containing 80 ll of trypticase soy broth per well. Mic-

rotitre plates were incubated at 37 �C for 18–24 h. Using

a 48-prong replicator (Sigma Chemical Inc.), the isolates

were transferred onto 100-mm-diameter trypticase soy

agar (TSA) plates amended with the following concen-

trations of antibiotics: ampicillin, erythromycin, andtetracycline, 10, 15, 30, and 50 lg/ml; chlortetracycline,

oxytetracycline, and streptomycin, 20, 40, 60, and 80 lg/ml; and salinomycin, 1, 5, 10, and 15 lg/ml. Isolates

were also spotted onto a control TSA plate containing

no antibiotic. All plates were incubated at 37 �C for 24

h, and the growth of each isolate on each antibiotic was

determined. An isolate was considered resistant if

growth could be detected by the unaided eye. Isolatesthat did not grow on the control plates were excluded

from the analysis. The final database contained a total

of 2491 enterococci, 493 from bird feces, 504 from

sewage, 480 from urban runoff, 495 from sediment of

Talbert Marsh and 519 from beach water.

2.4. Discriminant analysis

Data for ARPs was analyzed using SAS software

(version 8.0; SAS Institute Inc.) essentially as described

by Wiggins et al. (1999). The DISCRIM procedure was

run to calculate the number and percent of isolates from

each known source that are classified in each source

category. Correct classification rates were calculated

using one set of ARPs both to establish the classification

rule and as test subjects. The number of isolates from agiven source that were placed in the correct source cat-

egory by the discriminant analysis was used to calculate

the average rate of correct classification for the database

by averaging the correct classification percentages for all

sources. Discriminant analysis was then used to assign

isolates from beach water to a source category based on

comparison of its ARPs to those of isolates from known

sources. To confirm the significance of discrimination,an F ratio, defined as the ratio of the unbiased estimates

of the variance from two population, was calculated for

all possible pairs of source groups using the formula

F ¼ ½n1n2ðn1 þ n2 � p� 1Þ=ðn1 þ n2ðn1 � n2 � 2Þp�D2 and

was then compared to the appropriate critical value of

the F distribution for df 1 ¼ p, and df 2 ¼ n1 þ n2 �p � 1 (Gardiner, 1997).

3. Results

3.1. Prevalence of multiple antibiotic resistant enterococci

Of the total 2491 enteroccoci isolated from different

sources, all were resistant to at least one antibiotic at

some level. Nearly 30% of the isolates were resistant tosome level of four different antibiotics and 8% of strains

were resistant to some level of all seven antibiotics used

for testing (Table 1). Bacteria resistant to some level of

six to seven antibiotics were more commonly found

750 S. Choi et al. / Marine Pollution Bulletin 46 (2003) 748–755

among isolates from sewage, urban runoff and marsh

sediments. While a lower percentage of isolates frombeach water was resistant to six to seven antibiotics, 37%

were resistant to some level of four different antibiotics

(Table 1).

Distinct ARPs were observed for isolates from dif-

ferent sources. A significantly higher percentage of ery-

thromycin resistance ðp < 0:01Þ at all concentrations

was observed amongst isolates from bird feces (Fig. 2B)

while little variation was observed when testing isolatesfrom different sources against different concentrations of

streptomycin (Fig. 2F). All isolates were highly resistant

to streptomycin at the highest concentration used (80

lg/ml) suggesting that higher concentrations of strep-

tomycin or a different antibiotic would be more useful

for developing distinct ARPs. All isolates were also

highly resistant to salinomycin (Fig. 2G). Isolates from

urban runoff were nearly 100% resistant to all concen-trations of salinomycin tested. However only concen-

trations up to 15 lg/ml were used for testing. An

increase in concentration may improve the discriminant

power of this antibiotic. Isolates from marsh sediments

and urban runoff showed greater resistance to chloro-

tetracycline at all concentrations used than did isolates

from other sources (Fig. 2C). Both sewage and urban

runoff isolates showed greater resistance to oxytetracy-clin than isolates from other sources, especially at the

high concentrations used for testing (Fig. 2E). Isolates

from beach waters also displayed high percentage of

resistance when tested against low concentrations of

ampicillin.

3.2. Discriminant analyses of ARP database

Table 2 shows the results of classification of entero-

cocci from known sources. Over 85% of the isolates

from bird feces were correctly classified as from bird

source, and 81% of the isolates from urban runoff were

correctly classified as urban runoff. Isolates from sedi-

ments have a correct classification rate of 73% with a

10% and 11% miss-match with sewage and urban runoff,

respectively. Sixty-four percent of the isolates fromprimary sewage effluent were correctly classified as from

sewage source. The rest of the isolates evenly distributed

across the other three categories in the database. These

results reflect the mixed sources of OCSD�s sewage.

Since 1999, urban runoff from the lower Santa Ana

watershed has been diverted to OCSD�s treatment fa-

cility for treatment.

The statistical significance of group separation indiscriminant analysis can be tested by the F ratio. The

statistic (Section 2) is calculated using the discriminant

functions ðD2Þ. Large D2 values indicate greater degrees

of group separation, and can be used to determine the

statistical significance of correct classification into host

source categories by calculating the F ratio (Gardiner,

1997). The F ratio is calculated for each pair of catego-

ries, and the F statistic is compared to the F distribution.For all pair-wise comparisons of categories, F exceeds

the critical value ðP < 0:001Þ, demonstrating that the

ARPs of the isolates comprising the four source groups

are significantly different from one another (between-

group variance is higher than within-group variance).

Antibiotic resistance analysis (ARA) of all entero-

cocci isolated from Huntington State Beach assigned

39% of the isolates to sewage source, 30% to bird sourceand 24% to marsh sediment (Table 3). Only 6% of the

beach isolates were assigned to urban runoff (Table 3).

When beach water isolates were broken down to dif-

ferent sampling dates and the analysis was rerun using

the same database, results showed that the bird feces

was the predominant source for enterococci in the

sample collected on April 30, 2001, with 67% of isolates

assigned to this category (Table 4). Both water samplescollected in December (either 2000 or 2001) had over

53% of isolates assigned to sewage source, while 54.8%

of the isolates obtained from the July 16, 2001 sampling

were classified with the category of marsh sediment

(Table 4), demonstrating temporal variability of the

dominant sources.

Table 1

Percent of multiple antibiotic resistant (MAR) enterococci isolated from each source

Source % of resistant strains

AR-Ia MAR-IIb MAR-IIIb MAR-IVb MAR-Vb MAR-VIb MAR-VIIb Total

Bird feces 0 3 17 28 19 17 16 100

Sewage 0 6 16 17 21 29 10 100

Urban runoff 0 3 4 17 41 31 3 100

Sediment 2 5 11 16 38 23 5 100

Seawater 3 5 15 37 21 14 6 100

All sources 1 4 13 23 28 23 8

An isolate was counted resistant if it grew at any level of antibiotic treatment.aAR-1: resistant to one antibiotic tested.bMAR-II to MAR-VII: multiple resistant to two to seven antibiotics tested.

S. Choi et al. / Marine Pollution Bulletin 46 (2003) 748–755 751

4. Discussion

Resistance to multiple antibiotics is not uncommon inenterococci isolated from animals and humans (Aarest-

rup et al., 1998, 2000). It is a common belief that the

selective pressure imposed on the commensal gastroin-

testinal flora of animals and humans by wide-spread

antibiotic usage results in patterns of antibiotic resis-

tance that reflect the microflora�s past exposure to anti-

biotics. However, the observation that 498 enterococci

isolated from wild bird feces are resistant to some level of

two or more antibiotics used in this study is interesting.Literature on antibiotic resistant bacteria from wild birds

is scarce. One recent report found tetracycline-resistant

(>32 lg/ml) enterococci in the GI tracts of 4 out of 38

(10.8%) magpies tested in rural west Wales (Livermore

et al., 2001). An investigation by Nakamura et al. (Na-

kamura et al., 1982) found resistant Escherichia coli

Fig. 2. Prevalence of antibiotic resistant enterococci from environmental sources. Antibiotics used for testing are: (A) Ampicillin, (B) Erythromycin,

(C) Chlortetracycline, (D) Tetracycline, (E) Oxytetracycline, (F) Streptomycin, (G) Salinomycin.

752 S. Choi et al. / Marine Pollution Bulletin 46 (2003) 748–755

in several Japanese avian species that have little or no

human contact, while resistant bacteria were hardly ever

isolated from wild ducks in Japan (Saito et al., 1979).

However, it is important to point out, that the majority

of bird feces tested in our study were from seagulls re-

siding in a urban city, having habits of eating out of

dumpsters and close human contact. Therefore, their diet

and possibly exposure to antibiotics is different thanwhat one might see in a true wild bird. Significantly

higher rates of resistance to erythromycin were observed

among isolates from gull feces than those from other

sources. However, concentrations and types of antibi-

otics used for testing, as well as the methods for deter-

mination of resistant characteristics were different from

previous reports for antibiotic resistant E. coli in Apal-

achicola Bay (Parveen et al., 1997) and urban and ruralwaters in the eastern US by Kaspar et al. (Kaspar et al.,

1990). Therefore, it is not possible to make a more

pr�eecise comparison regarding the prevalence of antibi-

otic resistant bacteria from different environments.

The ability of ARPs to distinguish sources of fecal

contamination from human sewage and non-human

feces is based on the theory that antibiotics are widely

used for treating human diseases, therefore selecting forantibiotic resistant commensal fecal flora in human

feces. However, the field study is complicated by the

wide distribution of antibiotic resistant bacteria in the

environment and the presence of intrinsically antibiotic

resistant bacteria. For example, isolates from all sources

were highly resistant to salinomycin and streptomycin at

the concentration used for testing (Fig. 2) making the

distinction between sources impossible by a simple

comparison of antibiotic resistant characteristics. The

use of multiple antibiotics at several concentrations toestablish the ARPs, coupled with statistical treatment of

the data by discriminant analysis, provides the predic-

tive power necessary for source identification (Wiggins,

1996; Harwood et al., 2000).

The mathematical calculations required for discrimi-

nant analysis are identical to those used for single factor

multiple analysis of variance (MANOVA), although the

typical uses of the two types of analysis differ (Quinn andKeough, 2002). Discriminant analysis is used to catego-

rize observations into pre-determined groups, while

MANOVA is more often used for hypothesis-testing

purposes in order to determine whether significant dif-

ferences between group centroids exist. Discriminant

analysis calculates a series of derived variables that are

linear combinations of the observed variables. The de-

rived variables maximize the distance between groups,thus maximizing the likelihood that observations will be

classified in the correct group (Quinn and Keough,

Table 2

Classification of known enterococcus isolates by source

Source No. (%) of isolates classified to each source category

Bird feces Sediments Urban runoff Sewage

Bird feces ðn ¼ 493Þ 431 (87.4) 7 (1.4) 23 (4.7) 32 (6.5)

Sediments ðn ¼ 495Þ 18 (3.7) 365 (73.7) 58 (11.7) 54 (10.9)

Urban runoff ðn ¼ 480Þ 6 (1.2) 56 (11.7) 391 (81.5) 27 (5.6)

Sewage ðn ¼ 504Þ 60 (11.9) 59 (11.7) 60 (11.9) 325 (64.5)

Table 3

Classification of total enterococci isolated from beach water at Huntington State Beach, California

Source No. (%) of isolates classified to each source category

Bird feces Sediments Urban runoff Sewage

Water ðn ¼ 519Þ 155 (29.9) 126 (24.3) 34 (6.5) 204 (39.3)

Table 4

Classification of dated enterococci isolated from beach water at Huntington State Beach, California

Source No. (%) of isolates classified to each source category

Bird feces Sediments Urban runoff Sewage

Water 12/12/00 ðn ¼ 88Þ 8 (9.1) 27 (30.7) 6 (6.8) 47 (53.4)

Water 4/09/01 ðn ¼ 74Þ 8 (10.8) 33 (44.6) 7 (9.5) 26 (35.1)

Water 4/30/01 ðn ¼ 88Þ 59 (67.1) 10 (11.3) 6 (6.8) 13 (14.8)

Water 7/16/01 ðn ¼ 93Þ 14 (15.1) 51 (54.8) 4 (4.3) 24 (25.8)

Water 12/21/01 ðn ¼ 176Þ 66 (37.5) 5 (2.8) 11 (6.3) 94 (53.4)

S. Choi et al. / Marine Pollution Bulletin 46 (2003) 748–755 753

2002). Combining a large number of isolates n with

multiple numbers and concentrations of antibiotic p will

allow the establishment of a powerful identification

system capable of source distinction (Wiggins et al.,1999; Harwood et al., 2000). Hagedorn et al. (1999) also

suggested that classification accuracy of isolates from

known sources in the database is insufficient evidence

that the database will accurately categorize isolates that

are not in the database. The database must also contain a

sufficient number of isolates that have been obtained

from wide enough temporal and geographic spans to be

representative of the population being classified. Theyrecommend the database of known sources contain a few

hundred isolates per source before the point of ‘‘repre-

sentativeness’’ is reached. In fact, it was recently noted

that small libraries, while frequently showing high cor-

rect classification rates, are also subject to stochastic

(random) clustering (Whitlock et al., 2002).

Nearly 500 isolates per source ðnÞ were used to test

against 28 different observations ðpÞ in this study. Thecorrect classification rates for the category of bird feces

and urban runoff were above 80% at the isolate-level.

This is comparable to previous studies using a similar

method. However, antibiotic resistance analysis of

sewage isolates suggests that OCSD�s sewage contains

heterogeneous sources of fecal material, since 35% of the

isolates were evenly distributed through the categories of

bird feces, sediment and urban runoff. When interpret-ing the field result, Harwood et al. (2000) suggested that

an expected frequency of misclassification for each

source is helpful in understanding the significance of the

assignment of unknown isolates to a particular source.

In our database, the misclassification rate for isolates

from other categories into the sewage source ranges

from 5.6% (urban runoff) to 10.9% (sediments). The

percentage of isolates from seawater classified as sewagesource exceeded the expected rates of misclassification

on each sample event, strongly suggesting that sewage is

an important source of pollution to beach waters. The

temporal variability noted in dominant source assign-

ments for the five beach sample events was also found in

a study conducted in Florida (Whitlock et al., 2002).

Bird feces may be an important source of fecal con-

tamination to beach water on the selected days as indi-cated by the resistance pattern analysis using dated

water isolates. Coastal salt marsh can also act as a

contributor for the enterococci contamination to the

surf zone on other sampling days. The concentrations of

enterococcus in sediment samples collected from the

marsh ranges from 12 to 250 CFU/g of sediment. A

greater sewage impact was observed on the two winter

days. Urban runoff showed minimal impact to thecoastal water quality. This is likely due to the urban

runoff diversion effort to allow a minimal volume of

urban runoff to enter coastal water directly. The Atlanta

Station is the only station discharging urban runoff into

the Talbert Marsh. The lack of rainfall near the sample

events probably also contributed to these results.

The results presented in this study verify previous

reports that ARPs are a useful tool for identification ofcontamination sources. However, our study differs from

previous reports by using mixed sources as known

source categories. For example, the category urban

runoff contains fecal sources from dog, rabbit, rodent,

and other domestic, wild and agriculture animals. The

results of this study demonstrate that, in spite of the

complex matrices of the fecal sources, urban runoff

isolates were correctly classified at a rate greater than80%, and were most frequently misclassified as sediment

isolates (11.7%). The advantage of using a mixed source

category is to construct the database based on the pro-

portional contribution from each fecal source in the

watershed. Arbitrary selection of individual sources,

most often based on the accessibility of fecal materials

(i.e. domestic dog feces are easier to obtain than some

wild animal feces), may introduce bias to the database.Compared with other source identification methods

such as ribotyping and PCR-based techniques, the ARA

method is relatively simple and inexpensive, therefore

allowing examination of the large number of isolates

necessary to avoid extensive errors introduced by under-

sampling the true population. However, this method

may require the establishment of individual databases

that are most relevant to a specific geographical loca-tion. The establishment of such databases should be

designed to include sufficient representatives from the

most likely sources of contamination. The results of our

study are an important contribution to the under-

standing of the contamination problem at Huntington

Beach and to the development of coastal management

strategies.

Acknowledgements

We thank the Orange County Sanitation District for

providing primary sewage effluent samples. This study

was partially supported by a UCI Young Faculty De-

velopment Award to S.J. and a grant from the UC

Water Resource Center #P-00-38.

References

Aarestrup, F.M., Agerso, Y., Gerner-Smidt, P., Madsen, M., Jensen,

L.B., 2000. Comparison of antimicrobial resistance phenotypes and

resistance genes in Enterococcus faecalis and Enterococcus faecium

from humans in the community, broilers, and pigs in Denmark.

Diagn. Microbiol. Infect. Dis. 37, 127–137.

Aarestrup, F.M., Bager, F., Jensen, N.E., Madsen, M., Meyling, A.,

Wegener, H.C., 1998. Surveillance of antimicrobial resistance in

bacteria isolated from food animals to antimicrobial growth

promoters and related therapeutic agents in Denmark. APMIS

106, 606–622.

754 S. Choi et al. / Marine Pollution Bulletin 46 (2003) 748–755

Boehm, A.B., Sanders, B.F., Winant, C.D., 2002. Cross-shelf transport

at Huntington Beach. Implication for fate of sewage discharged

through an offshore ocean outfall. Environmental Science and

Technology 36, 1899–1906.

Cabelli, A.P., 1980. Health criteria for marine recreational waters.

Environmental Protection Agency, Washington DC.

Dufour, A.P., 1984. Health effects criteria for fresh recreational waters.

Environmental Protection Agency, Washington DC.

Gardiner, W.P., 1997. Statistics for the Biosciences. Prentice Hall,

London, p. 339.

Grant, S.B., Sanders, B.F., Boehm, A.B., Redman, J.A., Kim, J.H.,

Mrse, R.D., et al., 2001. Generation of enterococci bacteria in a

coastal saltwater marsh and its impact on surf zone water quality.

Environmental Science and Technology 35, 2407–2416.

Hagedorn, C., Robinson, S.L., Filtz, J.R., Grubbs, S.M., Angier, T.A.,

Reneau, R.B., 1999. Determining sources of fecal pollution in a

rural Virginia watershed with antibiotic resistance patterns in fecal

streptococci. Applied and Environmental Microbiology 65, 5522–

5531.

Harwood, V.J., Whitlock, J., Withington, V., 2000. Classification of

antibiotic resistance patterns of indicator bacteria by discriminant

analysis: Use in predicting the source of fecal contamination in

subtropical waters. Applied and Environmental Microbiology 66,

3698–3704.

Kaspar, C.W., Burgess, J.L., Knight, I.T., Colwell, R.R., 1990.

Antibiotic resistance indexing of Escherichia coli to identify sources

of fecal contamination in water. Canadian Journal of Microbiology

36, 891–894.

Kay, D., Fleisher, J.M., Salmon, R.L., Jones, F., Wyer, M.D.,

Godfree, A.F., et al., 1994. Predicting likelihood of gastroenteritis

from sea bathing––results from randomised exposure. Lancet 344,

905–909.

Livermore, D.M., Warner, M., Hall, L.M.C., Enne, V., Projan, S.J.,

Dunman, P.M., et al., 2001. Antibiotic resistance in bacteria from

magpies (Pica Pica) and rabbits (Orycytolagus cuniculus) from west

Wales. Environmental Microbiology 3, 658–661.

Miescier, J.J., Cabelli, V.J., 1982. Enterococci and other microbial

indicators in municipal wastewater effluents. J. Water Pollut.

Control Fed. 54, 1599–1606.

Nakamura, M., Yoshimura, H., Koeda, T., 1982. Durg resistance and

R plasmids of Escherichia coli strains isolated from six species of

wild birds. Nippon Juigaku Zasshi 44, 465–471.

OCSD, 1999. Huntington Beach Closure Investigation. Phase I. Final

Report. URL: http://www.ocsd.com/main.htm.

Parveen, S., Murphree, R.L., Edmiston, L., Kaspar, C.W., Portier,

K.M., Tamplin, M.L., 1997. Association of multiple-antibiotic-

resistance profiles with point and nonpoint sources of Escherichia

coli in Apalachicola Bay. Applied and Environmental Microbiol-

ogy 63, 2607–2612.

Quinn, G.P., Keough, M.J., 2002. Experimental design and data

analysis for biologists. Cambridge University Press, Cambridge,

UK, New York.

Saito, G., Asagi, M., Yamamoto, H., 1979. Low incidence of

Enterobacteriaceae in wild ducks (Aythya spp.) and antibiotic

resistance of the isolates. Japanese Journal of Veterinary Science

41, 181–183.

Whitlock, J.E., Jones, D.T., Harwood, V.J., 2002. Identification of

the Sources of Fecal Coliforms in an Urban Watershed

Using Antibiotic Resistance Analysis. Water Research 36, 4265–

4274.

Wiggins, B.A., 1996. Discriminant analysis of antibiotic resistance

patterns in fecal streptococci, a method to differentiate human and

animal sources of fecal pollution in natural waters. Applied and

Environmental Microbiology 62, 3997–4002.

Wiggins, B.A., Andrews, R.W., Conway, R.A., Corr, C.L., Dobratz,

E.J., Dougherty, D.P., et al., 1999. Use of antibiotic resistance

analysis to identify nonpoint sources of fecal pollution. Applied

and Environmental Microbiology 65, 3483–3486.

S. Choi et al. / Marine Pollution Bulletin 46 (2003) 748–755 755