characteristics and dynamics of salmonella …from six sites between july 2004 and may 2008. the...

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Dec. 2009, p. 7700–7709 Vol. 75, No. 240099-2240/09/$12.00 doi:10.1128/AEM.01852-09Copyright © 2009, American Society for Microbiology. All Rights Reserved.

Characteristics and Dynamics of Salmonella Contaminationalong the Coast of Agadir, Morocco�

Ibtissame Setti,1 Alba Rodriguez-Castro,2 Maria P. Pata,3 Carmen Cadarso-Suarez,4 Bouchra Yacoubi,5Laila Bensmael,6 Abdellatif Moukrim,5 and Jaime Martinez-Urtaza2*

Institut National de Recheche Halieutique, Agadir, Morocco1; Instituto de Acuicultura,2 Departamento de Zoología y Antropología Física,Facultad de Biología,3 and Departamento de Estadistica e Investigacion Operativa, Facultad de Medicina,4 Universidad de

Santiago de Compostela, 15782 Santiago de Compostela, Spain; Universite Ibn Zohr, Agadir, Morocco5; andInstitut National de Recherche Halieutique, Casablanca, Morocco6

Received 31 July 2009/Accepted 1 October 2009

The occurrence of Salmonella enterica in the environment of tropical and desert regions has remained largelyuninvestigated in many areas of the world, including Africa. In the present study, we investigated the presenceof Salmonella spp. along 122 km of the coastline of Agadir (southern Morocco) in relation to environmentalparameters. A total of 801 samples of seawater (243), marine sediment (279), and mussels (279) were collectedfrom six sites between July 2004 and May 2008. The overall prevalence of Salmonella spp. was 7.1%, with thehighest occurrence in mussels (10%), followed by sediment (6.8%) and seawater (4.1%). Only three serotypeswere identified among the 57 Salmonella sp. strains isolated. S. enterica serotype Blockley represented 43.8% ofall Salmonella strains and was identified in mussel and sediment samples. S. enterica serotype Kentucky (29.8%)was found almost exclusively in mussels, whereas S. enterica serotype Senftenberg (26.3%) was detected insediment and seawater. Statistical analysis using generalized additive models identified seawater temperature,environmental temperature, rainfall, and solar radiation as significant factors associated with the presence ofSalmonella. Rainfall was the only variable showing a linear positive effect on the presence of Salmonella in thesea, whereas the remaining variables showed more complex nonlinear effects. Twenty-eight (49.1%) Salmonellaisolates displayed resistance to ampicillin (22 isolates), nalidixic acid (9 isolates), sulfonamide compounds (2isolates), and tetracycline (1 isolate), with six of these isolates displaying multiple resistance to two of theseantimicrobial agents. Pulsed-field gel electrophoresis analysis revealed homogenous restriction patterns withineach serotype that were uncorrelated with the resistance pattern profiles.

Salmonella enterica bacteria are one of the most frequentcauses of food-borne infections transmitted to humans, mainlyfrom animal products (9). In addition to health concerns, thepresence of Salmonella contamination in the food chain hasserious economic consequences related to the costs of medicalcare and lost productivity (36). Thus, studies aimed at exam-ining the capacity of Salmonella spp. to survive in differenthabitats are critical for controlling contamination and under-standing the routes of colonization of new hosts (40).

Salmonella bacteria display a high degree of resistance to alarge variety of stress factors, which provides them with anenhanced capacity to persist in changing environments (40).However, the persistence of the organism outside of the host isnot uniform among the different serogroups (3, 4, 13, 24, 32,34). About 50 of the more than 2,500 different serotypes ofSalmonella included in the current classification scheme (29)are dominantly identified in human and animal sources (34).Information about groups that predominate in a given envi-ronment and their relationship to potential human or animalorigins remains scarce. In recent years, some studies carriedout in aquatic environments have provided new insights intothe ecological preferences of different Salmonella serogroups

in these environments. These studies have revealed the pres-ence of specific patterns of contamination in different geo-graphical areas in association with environmental and ocean-ographic variables (13, 24, 32) and have identified majorfactors and conditions that favor the presence of the contam-inating bacteria. Identification of the different Salmonellastrains present in the environment at serotype level is an es-sential preliminary step in discriminating potentially clinicallyimportant strains among the Salmonella present in contami-nated areas, providing invaluable information about the natureof the contamination and allowing the inference of potentialroutes of dissemination through microbial source tracking(31). All of this information is critical for an improved assess-ment of the potential risks to public health associated withSalmonella and for the evaluation of the ecological preferencesof the diverse and heterogeneous group of organisms whichcomprise the genus Salmonella (26).

The lack of information regarding the epidemiology, con-tamination, and potential routes of transmission of Salmonellais of particular concern in many regions of the world, such asAfrica and Central America, where gastrointestinal diseasescontinue to be a major cause of illness, primarily due to poorsanitary conditions and nutritional deficiencies. In the presentstudy, we investigated the dynamics of Salmonella contamina-tion in the coastal areas of Agadir, a populous region of south-ern Morocco where shellfish production and maritime tourismare important to the local economy. Information concerningthe biological characteristics of the isolates was correlated with

* Corresponding author. Mailing address: Instituto de Acuicultura,Universidad de Santiago de Compostela, 15782 Santiago de Compos-tela, Spain. Phone: 34 981 528024. Fax: 34 981 547165. E-mail: [email protected].

� Published ahead of print on 9 October 2009.

7700

on March 21, 2020 by guest

http://aem.asm

.org/D

ownloaded from

http://aem.asm.org/

environmental data in order to evaluate the climatic conditionsthat favor contamination of this region by this pathogen and toidentify the potential sources of contamination.

MATERIALS AND METHODS

Study area. Agadir is located on the southwestern Atlantic coast of Moroccoand covers an area of approximately 170 km2 (Fig. 1). The central area has morethan 500,000 inhabitants although the population may sometimes grow to1,000,000 due to the presence of tourists. The Agadir coast is one of the mostimportant areas of shellfish production in Morocco but is subject to frequentcontamination from urban and port wastewater discharges. The climate of Aga-dir is semiarid, with low annual rainfall, and is influenced by the Sahara. Averageannual temperatures range from 14°C to 16°C in January and 19°C to 25°C inJuly although temperatures may rise to over 40°C under the influence of Saharanwinds. The rainy season occurs in autumn and the first weeks of winter, andannual rainfall is highly variable, ranging from 100 mm to 443.20 mm. Coastalwaters are further influenced by currents from the Canary Islands, which travelfrom the southwest to the north. The influx of cold waters into the area isespecially important in summer months, resulting in upwelling along the coast.Samples were collected from six sites extending over 122 kilometers along theAgadir coast: Tamri and Cap Ghir located in the northern area, Anza and Tifnitin the central region, and Sidi Rabat and Douira in the southern zones (Fig. 1).

Sampling program. The presence of Salmonella spp. was investigated in 801samples, including seawater (243), marine sediment (279), and mussels (279),between July 2004 and May 2008. Samples were collected every month from eachsite over the entire study period, with the exception of Cap Ghir, where sample

collection started in September 2005 and included only samples of seawater andmussels.

Analysis of Salmonella isolates. Samples were collected in sterile containersand transported to the laboratory under refrigeration. The presence of Salmo-nella spp. was determined according to the ISO 6579:1993 standard method (18).For liquid samples, 100 ml of sample was filtered through a 0.45-�m-pore-sizesterile filter (Millipore Corporation, Bedford, MA), and the filter was thenplaced in 225 ml of buffered peptone water (Merck, Darmstadt, Germany). Forsolid samples, 25 g of the sample was added to 225 ml of buffered peptone water(Merck) and incubated at 37°C for 20 h. Ten milliliters of preenriched cultureswas then transferred to 100 ml of selenite cystine broth (Oxoid, Basingstoke,England), and 0.1 ml was transferred to 10 ml of RV10 Rappaport-Vassiliadisbroth (Difco, Sparks, MD). The cultures were then incubated at 37°C and 42°C,respectively, for 24 h. Aliquots of enriched broth were streaked onto Hektoenenteric agar (Oxoid), phenol red-brilliant green agar (Oxoid), and bismuth sulfiteagar (Oxoid) and incubated at 37°C for 24 h (if only slight growth was observed,the plates were reincubated for an additional 24 h). Typical colonies wereselected and streaked onto nutrient agar and subjected to initial biochemicalscreening in triple-sugar iron agar (Oxoid). Cultures displaying a reaction typicalof Salmonella (an alkaline slant and acid butt, with or without production of H2S)were confirmed by biochemical tests using an API-20E strip (bioMerieux, Marcy-l’Etoile, France) and PCR analysis involving the amplification of a 284-bp frag-ment of the invA gene, according to the protocol described by Malorny et al. (20).

Salmonella serotyping. All Salmonella isolates were serotyped by seroaggluti-nation with commercial antiserum (Statens Serum Institut, Copenhagen, Den-mark). Polyvalent Salmonella O and H antisera were used to obtain a presump-tive diagnosis, and the definitive antigenic designation was then assigned by usingmonovalent antisera.

FIG. 1. Study area in Agadir with locations of the sampling stations and spatial distribution of the presence of Salmonella bacteria at thedifferent sites throughout the study period, as determined using the GIS data.

VOL. 75, 2009 SALMONELLA CONTAMINATION ON THE MOROCCAN COAST 7701


http://aem.asm

.org/D

ownloaded from

http://aem.asm.org/

PFGE. Pulsed-field gel electrophoresis (PFGE) was performed according tothe 1-day (24 to 28 h) standardized laboratory protocol for molecular subtypingof nontyphoidal Salmonella by PFGE (6). Chromosomal DNA was digested with50 U of XbaI (Promega, Southampton, United Kingdom). PFGE was performedon a CHEF DRIII system (Bio-Rad, Hercules, CA) in 0.5� Tris-Borate-EDTAextended-range buffer (Bio-Rad) with recirculation at 14°C. DNA macrorestric-tion fragments were resolved on 1% SeaKem Gold agarose (Cambrex) in 0.5�Tris-Borate-EDTA buffer. DNA from S. enterica serotype Braenderup H9812restricted with XbaI was used as a size marker. Pulse times were ramped from 2.2to 63.8 s during an 18-h run at 6.0 V/cm. Macrorestriction patterns were com-pared using BioNumerics software (Applied Maths, Sint-Martens-Latem, Bel-gium).

Antimicrobial susceptibility testing. Isolates were screened for susceptibilityto 16 antibiotics on Mueller-Hinton agar (Oxoid, Basingstoke, Hampshire,United Kingdom) by disk diffusion, as described in the Clinical LaboratoryStandard Institute (formerly NCCLS) guidelines (7). The following disks (Oxoid)were used: amikacin (30 �g), apramycin (15 �g), amoxicillin-clavulanic acid (30�g), ampicillin (10 �g), chloramphenicol (10 �g), cefoperazone (30 �g), cefta-zidime (30 �g), colistin (25 �g), furazolidone (15 �g), gentamicin (10 �g),nalidixic acid (30 �g), neomycin (10 �g), streptomycin (25 �g), sulfamethox-azole-trimethoprim (25 �g), sulfonamide compound (300 �g), and tetracycline(10 �g).

Environmental parameters. The environmental parameters considered in thestudy were air temperature, wind, hours of sunshine per day, rainfall, solarradiation, salinity, and seawater temperature. The minimum, maximum, andaverage daily air temperatures were recorded each day. Wind direction wasmeasured as the time in hours that the wind blew in each of the four prevailingquadrants (northwest, northeast, southwest, and southeast) or was measured asno wind (calm). Wind speed was expressed as kilometers per day. Rainfall wasmeasured as millimeters of precipitation per day, and solar radiation was mea-sured as watts per square meter (W/m2). All the climatic data were provided byLe Service Climatologique de Sud de la Direction Regional de la MeterologieNationale from the Agadir-Inezguane station (9°34�58.5�W, 30°25�10.18�N). Sa-linity and seawater temperature were recorded on each sampling day at eachsampling site, with a conductivity meter (WTW, Weilheim, Germany).

Spatial analysis. The results of the analyses were processed by using thegeographical information system (GIS) ArcGIS software (ArcView, version 3.3),produced by the Environmental Systems Research Institute. The data formatswere Shapefile (vector data) and were provided by the Agence Nationale de laConservation Fonciere du Cadastre et de la Cartographie, Direction du Cadastreet de la Cartographie.

Statistical analysis. The associations between environmental factors and thepresence of Salmonella spp. were analyzed by generalized additive models(GAMs) (43). Independent models were initially constructed for each environ-mental variable to select the data set of the variable (data from the same day ofsample collection to 4 days prior to sample collection, and the mean values forthe 5 days) with the best explanatory values based on the Akaike informationcriterion (AIC) (1). The selected data sets from each variable were used toconstruct different models incorporating correlations among binary responsedata (presence/absence of Salmonella spp.). The final multivariate logistic re-gression model selected according to both the AIC value and highest proportionof explained deviance was the following:

logit � �0 � s1�month� � s2�sea temperature�

� s3�environmental temperature� � s4�rain� � s5�solar radiation�

� sites � origin

where �0 is an unknown constant, si are unknown partial smooth functions, andlogit � log{p[Y � 1/X]/[1 (Y � 1/X)]}, where X is the vector of the variables

considered in the model (i.e., month, solar radiation, etc.), and Y represents thebinary outcome of interest (Y � 1 when Salmonella spp. are present and Y � 0when Salmonella spp. are absent).

To ensure identification of the above logistic GAM, the smooth functions si areall centered on the mean (i.e., zero-mean functions).

Additionally, GAMs for continuous response were used to obtain the pre-dicted values of the environmental variables (seawater temperature, salinity,environmental temperature, rainfall, and solar radiation) over the course of thestudy.

In all the GAMs considered, thin-plate regression splines (41) were used assmoothers of s, with optimal smoothing parameters (or, equivalently, with opti-mal effective degrees of freedom [edf]) chosen automatically by use of (i) theunbiased risk estimator (42, 43) criterion for GAMs with a binary response or (ii)the generalized cross-validation (43) criterion for GAMs with a continuousresponse.

Relationships between the presence of each serotype and the environmentalvariables were analyzed with the use of the following logistic generalized linearmodel (25):

logit � �0 � �1�sea temperature� � �2�environmental temperature�

� �3�salinity� � �4�rain� � �5�solar radiation�

where, � � (�0, �1, �2, �3, �4, �5, �5) is the vector of the parameters. The strengthof the associations between the binary response and each of the explanatoryvariables was evaluated through the calculation of the corresponding adjustedodds ratio ([OR] e�) with a 95% confidence interval.

All regression models were fitted with the gam function of the mgcv library(43) in the free R software (30), version 2.9.1.

RESULTS

Salmonella spp. were detected in 57 of 801 (7.1%) samplesanalyzed over the study period (July 2004 to May 2008) (Table1). The highest prevalence of Salmonella bacteria was detectedin mussels, with 28 positive samples occurring in the 279 ana-lyzed (10%), followed by sediment, with 19 positive samples(6.8%), and seawater, with 10 positive samples (4.1%) (Table1). The prevalence of Salmonella spp. was significantly higher(P 0.05) in mussel and sediment samples than in seawatersamples (Table 2).

The presence of Salmonella at the six sampling sites (Table2 and Fig. 1) was significantly higher (P 0.05) in Tifnit andAnza (southern and central areas), where 20 (13.8%) and 15(11%) positive samples, respectively, were detected (Table 3).The site with the lowest occurrence of Salmonella spp. wasDouira, with 5 positive samples occurring in the 144 studied(3.5%). Finally, no positive samples were detected in Cap Ghir(northern area).

Salmonella bacteria were detected throughout the period ofstudy. The presence of Salmonella spp. was higher in 2006, with22 positive samples occurring in a total of 201 samples(10.9%), and lower in 2008, when the bacteria were found inonly 1.7% of the 116 samples analyzed. The highest incidencecoincided with autumn (13%) and the winter months (12%)

TABLE 1. Number of samples collected and prevalence of Salmonella bacteria in the different sample types

Sampletype

No. ofsamples

No. of samples by year (% Salmonella positive) Total no. of Salmonella-positive samples (%)2004 2005 2006 2007 2008

Mussels 279 33 (9.1) 62 (9.7) 71 (16.9) 72 (7.0) 41 (4.9) 28 (10.0)Sediment 279 33 (3.0) 62 (9.7) 71 (8.5) 72 (8.3) 41 (0.0) 19 (6.8)Seawater 243 33 (3.0) 57 (3.5) 59 (6.7) 60 (5.0) 34 (0.0) 10 (4.1)

Total 801 99 (5.1) 181 (7.7) 201 (10.9) 204 (6.9) 116 (1.7) 57 (7.1)

7702 SETTI ET AL. APPL. ENVIRON. MICROBIOL.


http://aem.asm

.org/D

ownloaded from

http://aem.asm.org/

and reached a maximum in November 2006, when 47.1% pos-itive samples were detected (Fig. 2). In contrast, Salmonellaspp. were rarely detected in spring and summer during the 4years of the study, with two (1%) and three (1.5%) positivesamples, respectively.

The seasonal trends in the environmental variables (salinity,seawater temperature, environmental temperature, rainfall,and solar radiation) over the period of study were initiallyevaluated separately using independent GAMs with continu-ous responses for each variable (Fig. 3). The predicted salinity(data not shown), seawater temperature, and environmentaltemperature values formed a clear seasonal pattern, with val-ues peaking during the summer months. In contrast, the high-est predicted rainfall occurred during the last months of 2004,with a second, less significant, peak toward the end of 2005 andat the beginning of 2006, with no rainfall detected thereafter.Predicted solar radiation showed a variable pattern over thecourse of the study, with the highest value observed in June2005, followed by a drastic reduction in November 2005, withhigh variability being observed throughout the remainder ofthe study period.

Associations between the environmental and oceanographicparameters considered in the study (salinity, seawater temper-ature, environmental temperature, rainfall, and solar radia-tion) and the presence of Salmonella were analyzed by logisticGAM for binary responses. Independent models were con-structed for each variable using data from the same day of

sample collection to 4 days prior to sample collection and fromthe mean values for the 5 days. The best explanatory modelsfor each variable, selected on the basis of the AIC, were thoseconstructed using the mean values of the variables, with theexception of rainfall, which showed the lowest AIC value in thedata from the 4 days prior to sample collection. Selected datafrom each variable were used to construct different regressionmodels. The best explanatory model, with AIC and unbiasedrisk estimator values of 283.07 and 0.601, respectively, ac-counted for 50% of the deviance explained. The selectedmodel included the environmental variables seawater temper-ature, environmental temperature, rainfall, and solar radia-tion, as well as sampling site and sample source, as the vari-ables significantly (P 0.01) associated with the response(Table 4 and Fig. 4). Rainfall was the only explanatory envi-ronmental variable showing a linear relationship with the pres-ence of Salmonella (Fig. 4) and positively influenced the prob-ability of Salmonella detection. The presence of Salmonellawas also positively influenced by solar radiation although thiseffect was restricted to low values since no effect was observedfor values exceeding 5 W/m2. The remaining variables showedmore complex nonlinear relationships with the presence ofSalmonella. The overall probability of Salmonella detectionover the course of our study, as obtained from the final GAM,which integrated the four significant environmental variables(seawater temperature, environmental temperature, rainfall,and solar radiation) in addition to variables of sampling siteand the sample source, showed a pattern that was distinctivefor each period (Fig. 4). The probability of the presence ofSalmonella showed a similar pattern for the years 2005 and2006, with the presence of Salmonella peaking in May of bothyears. However, a different pattern was observed in 2007 and2008, with the highest predicted presence of Salmonella occur-ring from September of 2007 to January of 2008, followed by adrastic decline after this point.

Serotyping of Salmonella sp. strains revealed three differentserotypes among the 57 isolates (Table 5). The dominant wasS. enterica serotype Blockley, which occurred in 25 strains(43.8%), 12 of which were detected in mussels (48%), 12 insediment samples (48%), and only 1 strain in seawater (4%).PFGE analysis revealed a homogenous pattern of restrictionfragments among the strains of this serotype (Fig. 5). Nearly85% of the strains exhibited indistinguishable pulsotypes, andonly four isolates presented slight differences in restrictionprofiles. S. enterica serotype Kentucky was identified in 17isolates (29.8%) occurring almost exclusively in mussels

TABLE 2. Estimated coefficients of the categorical predictors forthe logistic GAM with SE, z-statistic, and P values

Parametric term(s)a Estimated coefficient(SE) z statistic P value

Intercept 4.788 (0.4213) 11.362 0.001Site 1 vs site 3 1.258 (0.3935) 3.197 0.001Site 1 vs site 4 2.288 (0.3639) 6.287 0.001Site 1 vs site 5 0.379 (0.4220) 0.899 0.369Site 1 vs site 6 0.970 (0.3912) 2.481 0.013Site 3 vs site 4 1.030 (0.3064) 3.361 0.001Site 3 vs site 5 0.879 (0.3889) 2.260 0.024Site 3 vs site 6 0.288 (0.3507) 0.821 0.412Site 4 vs site 5 1.909 (0.3480) 5.486 0.001Site 4 vs site 6 1.318 (0.3029) 4.350 0.001Site 5 vs site 6 0.591 (0.3827) 1.545 0.122Mussel vs water 1.424 (0.2663) 5.348 0.001Mussel vs sediment 0.545 (0.2236) 2.437 0.015Water vs sediment 0.879 (0.2755) 3.191 0.001

a Sites are as follows: 1, Tamri; 2, Cap Ghir; 3, Anza; 4, Tifnit; 5, Douira; 6,Sidi Rabat.

TABLE 3. Incidence of Salmonella spp. at each sampling site over the course of the study period

Samplingsite

No. ofsamples

No. of samples by year (% Salmonella positive) Total no. of Salmonella-positive samples (%)2004 2005 2006 2007 2008

Tamri 159 27 (3.7) 39 (5.1) 36 (5.6) 36 (8.3) 21 (5.0) 9 (5.7)Cap Ghir 74 0 (0.0) 12 (0.0) 24 (0.0) 24 (0.0) 14 (0.0) 0 (0.0)Anza 136 18 (22.2) 31 (9.7) 33 (15.1) 36 (5.6) 18 (5.6) 15 (11.0)Tifnit 144 18 (0.0) 33 (12.9) 36 (25.0) 36 (19.4) 21 (0.0) 20 (13.8)Douira 144 18 (0.0) 33 (9.1) 36 (5.6) 36 (0.0) 21 (0.0) 5 (3.5)Sidi Rabat 144 18 (0.0) 33 (6.1) 36 (11.1) 36 (5.6) 21 (0.0) 8 (5.5)

Total 801 99 (5.1) 181 (7.7) 201 (10.9) 204 (6.8) 116 (1.7) 57 (7.1)



http://aem.asm

.org/D

ownloaded from

http://aem.asm.org/

(94.1%) and in 1 isolate from sediment (5.9%). All serotypeKentucky isolates showed indistinguishable restriction patternsand grouped into a single cluster. Finally, S. enterica serotypeSenftenberg was identified in 15 strains (26.3%), primarilyoccurring in seawater (60%) but also in sediment, with sixisolates (40%). Serotype Senftenberg was entirely absent frommussels. PFGE analysis of serotype Senftenberg isolatesshowed four different related pulsotypes included in a single

cluster with 77.5% similarity. The three serotypes were de-tected throughout the study area. No significant associationbetween the different serotypes and the environmental vari-ables was obtained by applying the generalized linear models.

Almost 50% of the Salmonella isolates obtained in this studyshowed resistance to at least one antibiotic (Fig. 5). Among thethree serotypes identified, serotype Senftenberg displayed thehighest degree of resistance, with 60% of the isolates being

FIG. 2. Overall incidence of Salmonella bacteria along the Agadir coast throughout the study period (A) and variations in rainfall andtemperature during the same period (B).



http://aem.asm

.org/D

ownloaded from

http://aem.asm.org/

resistant, followed by serotypes Kentucky and Blockley, withresistances of 47.1% and 44%, respectively. Overall, 22 isolateswere resistant to ampicillin (38.6%), 9 to nalixidic acid(15.8%), 2 to sulfonamide compound (3.5%), and 1 to tetra-

cycline (1.7%). Among the 28 resistant isolates, 22 displayedresistance to one antimicrobial agent, and the remaining 6isolates displayed resistance to two antimicrobial agents. Atotal of six different resistance patterns were observed amongthe Salmonella isolates; these were predominantly ampicillin(17 isolates), nalidixic acid (4 isolates), and ampicillin-nalidixicacid (4 isolates) No associations between the resistance pat-terns and the different genetic profiles identified by PFGEanalysis were found (Fig. 5).

DISCUSSION

Data on the presence of Salmonella in human and nonhu-man sources in Africa are extremely limited. This is true formost developing countries worldwide. Almost 90% of all re-

FIG. 3. Predicted values for the principal environmental variables estimated through the smooth effect (s) obtained from the GAM. Predictedvalues for seawater temperature (A), environmental temperature (B), rainfall (C), and solar radiation (D) over the course of the study are shown.Shading indicates the 95% point-wise confidence bands for the predictions.

TABLE 4. Estimated nonparametric components of the logisticGAM, with corresponding edf, chi-square statistic, and P values

Smooth effect (s)variable edf Chi-square P value

Month 8.32 55.556 0.001Seawater temp 8.77 54.083 0.001Environmental temp 3.18 45.866 0.001Rainfall 1.00 23.238 0.001Solar radiation 4.59 14.119 0.001



http://aem.asm

.org/D

ownloaded from

http://aem.asm.org/

ported Salmonella isolates submitted to the WHO GlobalSalm-Surv country databank (www.who.int/salmsurv) are sub-mitted from North America and Europe (11) even though theseregions account for only 16% of the world population (U.S. Cen-sus Bureau, International Data Base, Population Division [http://www.census.gov/ipc/www/idb/worldpopinfo.html]). The limitedinformation available for other countries seriously restricts theunderstanding and interpretation of the data obtained in stud-ies carried out in these areas, which cannot be compared withreports from similar areas or with general trends. According tothe information included in the WHO Global Salm-Surv coun-try databank provided by the National Institute of Hygiene,Morocco, the dominant serotype isolated from human sourcesin Morocco between 2000 and 2003 was S. enterica serotypeEnteritidis, which accounted for 76.3% of the reported strainsin 2003. Animal-related data provided by the same institu-tion showed a well-differentiated pattern of serotype domi-nance, with S. enterica serotype Gallinarum representing52.5% of total Salmonella isolates for 2003 and serotypeEnteritis representing 12% of strains. Although this infor-mation is based on only a small number of strains and is not

likely representative of the overall situation in this largecountry with diverse habitats, the dominance of serotypeEnteritidis appears to be a constant feature of most re-ported human infections. The serotype Enteritidis was prev-alent in two studies carried out in Rabat (17, 19), and S.enterica serotypes Wien and Infantis were the second-mostfrequently isolated serotypes.

The serotype dominance observed in the coastal areas ofAgadir throughout the present study exhibited a pattern clearlydifferent from the serotypes reported in human isolates inMorocco. Only three serotypes—Blockley, Kentucky, and Sen-ftenberg—were identified among the 57 Salmonella strains iso-lated along 122 km of coastline. Serotypes Kentucky, Blockey,and Senftenberg, however, are not included in the top 10 mostfrequently reported serotypes from human infection and ani-mal sources in Morocco, according to the WHO Global Salm-Surv country databank. The restricted number of serotypesoccurring in the coastal regions of Morocco notably contrastswith the high degree of diversity of serotypes previously re-ported for aquatic and marine environments in many otherparts of the world. Typically, environmental studies of riversand coastal areas have identified between 10 and 20 differentserotypes (4, 5, 13, 24, 32, 35, 39). The variety of serotypes hasbeen associated with the presence of highly diverse sources ofSalmonella contamination along the shores of rivers and mar-itime zones (3, 13, 23, 24). Although most contamination isexpected to be of animal or human origin, serotypes that pre-vail in the environment often do not coincide with the mostcommon zoonotic or human serotypes identified in these areas(5, 16, 21, 24, 28, 35, 39). The discrepancy between the dom-inant serotypes in human and animal hosts and in the environ-ment may be ascribed to the diverse rates of survival of thedifferent Salmonella serotypes in the presence of adverse or

FIG. 4. Estimated centered smooth effect (s) of the presence of Salmonella obtained from the GAM. Temporal variation in the probability ofdetection of Salmonella throughout the period of study (A) and association between the presence of Salmonella and the environmental variables:rainfall (B), environmental temperature (C), seawater temperature (D), and solar radiation (E). Numbers along the y axis indicate the edf for eacheffect; values along the x axis represent the values of the environmental variables recorded over the sampling period. Shading indicates the 95%point-wise confidence bands for the variables.

TABLE 5. Distribution of distinct serotypes among S. entericaisolates obtained throughout the study period

per sample type

SerotypeNo. of isolates per serotype

Seawater Mussels Sediment Total (%)

Senftenberg 9 0 6 15 (26.3)Blockley 1 12 12 25 (43.8)Kentucky 0 16 1 17 (29.8)

Total 10 28 19 57 (100.0)



http://aem.asm

.org/D

ownloaded from

http://aem.asm.org/

stressful conditions (24) although differences in the pathogenicpotential and host range of the different serotypes and clonescould also influence in the rates of detection (34). S. entericaserotype Typhimurium is the most frequently reported sero-type of clinical importance and predominates over other sero-types in most environmental studies (3, 5, 24, 28, 32, 39), afinding that has been attributed to its high survival rate outsidethe host (3, 10, 24). Serotype Senftenberg is the most resistantto extreme environments (22) and is the most widespread

serotype in tropical and temperate marine regions of the world(3, 5, 15, 16, 22, 24, 33). The identification of only three sero-types along the Agadir coast throughout the entire period ofthe study and among all the sample sites investigated may bedirectly related to two factors: the particular pattern of Salmo-nella contamination in this region linked to sources of contam-ination and the ability of the prevailing serotypes to survive inthe environment outside the host. Most of the Salmonellaisolates in this study were detected in the central sampling

FIG. 5. Dendrogram generated by Bionumerics software showing the relationship between the different PFGE fingerprints and antibioticresistance patterns of the strains. Antimicrobial resistance, indicated by a black box, was present for amikacin (AK), apramycin (APR),amoxicillin-clavulate (AMC), chloramphenicol (C), cefoperazone (CAZ), colistin sulfate (CT), ampicillin (AMP), sulfonamides (S3), tetracycline(TE), furazolidone (FR), gentamicin (CN), nalidixic acid (NA), neomycin (N), streptomycin (S), and sulfamethoxazole trimethoprim (SXT). Thenumbers at the top of the PFGE profiles indicate molecular sizes in kbp.



http://aem.asm

.org/D

ownloaded from

http://aem.asm.org/

sites, surrounding the mouth of the Souss River, which is theprincipal river in the region and passes through the populatedareas surrounding Agadir.

The present study identified a high rate of antimicrobialresistance which was similarly distributed among the threeserotypes identified; almost 50% of the strains displayed resis-tance to at least one of the antibiotics tested. The antibioticresistance observed was restricted to four antibiotics whichhave been reported to be commonly used in veterinary prac-tices in the region (2). Although antimicrobial-resistant bacte-ria has been identified in areas without any evidence of selec-tive pressure, the presence of resistant and multidrug-resistantstrains linked to an enteric bacteria like Salmonella occurringin the marine environment may be considered a very prelimi-nary evidence suggesting the extensive use of antibiotics inveterinary and medical practices for the control of bacterialdiseases in the region, as has been reported for other regions(8, 14, 27, 37, 38).

The presence of Salmonella contamination in coastal watersof Agadir was detected throughout the entire period of study,with most of the contamination events coinciding with theperiods of rainfall. The statistical analysis identified the pres-ence of rain as a direct factor favoring the occurrence ofSalmonella in coastal areas, especially when rainfall occurredin the days immediately prior to sample collection. As found inthe present study, the influence of rainfall as the primary factorin determining Salmonella contamination events in rivers andcoastal areas has been reported in various other studies carriedout in diverse geographical areas worldwide (3, 12, 13, 24, 32).Rainfall has been identified as the universal environmentaldriver for the presence of Salmonella in the environment (32),with storm waters participating in the transport of Salmonellafrom its source points to marine environments by aquifers (13,24, 32). Recognition of the decisive role that rainfall plays inthe process of environmental contamination by Salmonellaprovides a practical tool for the development of simple sur-veillance programs to prevent contamination events in naturalhabitats and food-producing areas and, consequently, to fur-ther advance the methods of controlling diseases caused by thisorganism.

ACKNOWLEDGMENTS

Ibtissame Setti received a travel grant to carry out research at theUniversity of Santiago de Compostela, funded by project numberA/012228/07 from the Agencia Espanola de Cooperacion Internacio-nal para el Desarrollo. The work of M.P.P. and C.C.-S. was supportedby grants MTM2008-001603 from the Ministerio de Ciencia e Innova-cion and INCITE08PXIB208113PR from the Xunta de Galicia.

We are grateful to Isabel Mayan Barreiro and Silvia Carles Gonza-lez (University of Santiago de Compostela) and A. Hanoune (InstitutNational de Recheche Halieutique) for technical assistance and toMorjani Zine for help with the GIS analysis. We also thank A. Ber-raho, A. Chafik, A. Bernoussi, and A. Lahnin for their support in thedevelopment of the present study.

REFERENCES

1. Akaike, H. 1970. Statistical predictor identification. Ann. Inst. Stat. Math.22:203–217.

2. Bada-Alambedji, R., A. Fofana, M. Seydi, and A. J. Akakpo. 2006. Antimi-crobial resistance of Salmonella isolated from poultry carcasses in Dakar(Senegal). Braz. J. Microbiol. 37:510–515.

3. Baudart, J., K. Lemarchand, A. Brisabois, and P. Lebaron. 2000. Diversityof Salmonella strains isolated from the aquatic environment as determinedby serotyping and amplification of the ribosomal DNA spacer regions. Appl.Environ. Microbiol. 66:1544–1552.

4. Brands, D. A., A. E. Inman, C. P. Gerba, C. J. Mare, S. J. Billington, L. A.Saif, J. F. Levine, and L. A. Joens. 2005. Prevalence of Salmonella spp. inoysters in the United States. Appl. Environ. Microbiol. 71:893–897.

5. Catalao Dionisio, L. P., M. Joao, V. S. Ferreiro, M. L. Hidalgo, M. E. GarcíaRosado, and J. J. Borrego. 2000. Occurrence of Salmonella spp. in estuarineand coastal waters of Portugal. Antonie van Leeuwenhoek 78:99–106.

6. Centers for Disease Control and Prevention. 2002. Standardized molec-ular subtyping of foodborne bacterial pathogens by pulsed-field gel elec-trophoresis: training manual. Centers for Diseases Control and Preven-tion, Atlanta, GA.

7. Clinical and Laboratory Standards Institute (CLSI). 2008. Performancestandards for antimicrobial disk and diffusion susceptibility testing for bac-teria isolated from animals, 3rd ed. Approved standard M31-A3. CLSI,Wayne, PA.

8. Cooke, M. D. 1976. Antibiotic resistance among coliform and fecal coliformbacteria isolated from sewage, seawater, and marine shellfish. Antimicrob.Agents Chemother. 9:879–884.

9. Desenclos, J. C., P. Bouvet, V. Pierre, A. Brisabois, S. Fremy, C. Lahellec, F.Grimont, and P. A. D. Grimont. 1996. Epidemeologie des infection a Sal-monella: tendances recentes en France et en Europe. Bull. Soc. Fr. Micro-biol. 11:209–215.

10. Dupray, E., and A. Derrien. 1995. Influence of the previous stay of Esche-richia coli and Salmonella spp. in waste water on their survival in seawater.Water Res. 29:1005–1011.

11. Galanis, E., D. M. Lo Fo Wong, M. E. Patrick, N. Binsztein, A. Cieslik, T.Chalermchikit, A. Aidara-Kane, A. Ellis, F. J. Angulo, and H. C. Wegener.2006. Web-based surveillance and global Salmonella distribution, 2000–2002.Emerg. Infect. Dis. 12:381–388.

12. Goyal, S. M., C. P. Gerba, and J. L. Melnick. 1977. Occurrence and distri-bution of bacterial indicators and pathogens in canal communities along theTexas coast. Appl. Environ. Microbiol. 34:139–149.

13. Haley, B. J., D. J. Cole, and E. K. Lipp. 2009. Distribution, diversity, andseasonality of waterborne salmonellae in a rural watershed. Appl. Environ.Microbiol. 75:1248–1255.

14. Harwood, V. J., J. Whitlock, and V. H. Withington. 2000. Classification of theantibiotic resistance patterns of indicator bacteria by discriminant analysis:use in predicting the source of fecal contamination in subtropical Floridawaters. Appl. Environ. Microbiol. 66:3698–3704.

15. Hatha, A. A. M., and P. Lakshmanaperumalsamy. 1997. Prevalence of Sal-monella in fish and crustaceans from markets in Coimbatore, South India.Food. Microbiol. 14:111–116.

16. Heinitz, M. L., R. D. Ruble, D. E. Wagner, and S. R. Tatini. 2000. Incidenceof Salmonella in fish and seafood. J. Food. Prot. 63:579–592.

17. Hessissen, L., and N. Benseffaj. 2004. Les Salmonelles au Maroc, le guide dela medecine et de la sante au Maroc-CHU Rabat-Sale-8mars2004. www.santemaghreb.com.

18. International Organization for Standardization. 1993. Microbiology: gen-eral guidance on methods for the detection of Salmonella, 3rd ed. Interna-tional standard method ISO 6579:1993. International Organization for Stan-dardization, Geneva, Switzerland.

19. Kennedy, M., and F. J. Angulo. 1999. Incidence of foodborne illnesses: datafrom FoodNet. Irish J. Agric. Food Res. 39:295–300.

20. Malorny, B., J. Hoorfar, C. Bunge, and R. Helmuth. 2003. Multicentervalidation of the analytical accuracy of Salmonella PCR: towards an inter-national standard. Appl. Environ. Microbiol. 69:290–296.

21. Martinez-Urtaza, J., E. Liebana, L. Garcia-Migura, P. Perez-Pineiro, andM. Saco. 2004. Characterization of Salmonella enterica serovar Typhi-murium from marine environments in coastal waters of Galicia (Spain).Appl. Environ. Microbiol. 70:4030–4034.

22. Martinez-Urtaza, J., J. Peiteado, A. Lozano-Leon, and O. Garcia-Martin.2004. Detection of Salmonella Senftenberg associated with high saline envi-ronments in mussel processing facilities. J. Food. Prot. 67:256–263.

23. Martinez-Urtaza, J., M. Saco, G. Hernandez-Cordova, A. Lozano, O. Garcia-Martin, and J. Espinosa. 2003. Identification of Salmonella serovars isolatedfrom live molluscan shellfish and their significance in the marine environ-ment. J. Food. Prot. 66:226–232.

24. Martinez-Urtaza, J., M. Saco, J. de Novoa, P. Perez-Pineiro, J. Peiteado, A.Lozano-Leon, and O. Garcia-Martin. 2004. Influence of environmental fac-tors and human activity on the presence of Salmonella serovars in a marineenvironment. Appl. Environ. Microbiol. 70:2089–2097.

25. McCullagh, P., and J. A. Nelder. 1989. Generalized linear models. Chapmanand Hall, London, United Kingdom.

26. McQuiston, J. R., S. Herrera-Leon, B. C. Wertheim, J. Doyle, P. I. Fields,R. V. Tauxe, and J. M. Logsdon, Jr. 2008. Molecular phylogeny of thesalmonellae: relationships among Salmonella species and subspecies deter-mined from four housekeeping genes and evidence of lateral gene transferevents. J. Bacteriol. 190:70607.

27. Parveen, S., R. L. Murphree, L. Edminston, C. W. Kaspar, K. M. Portier,and M. L. Tamplin. 1997. Association of multiple-antibiotic-resistance pro-files with point and nonpoint sources of Escherichia coli in Apalachicola Bay.Appl. Environ. Microbiol. 63:2607–2612.

28. Polo, F., M. J. Figueras, I. Inza, J. Sala, J. M. Fleisher, and J. Guarro. 1999.



http://aem.asm

.org/D

ownloaded from

http://aem.asm.org/

Prevalence of Salmonella serotypes in environmental waters and their rela-tionships with indicator organisms. Antonie van Leeuwenhoek 75:285–292.

29. Popoff, M. Y., J. Bockemuhl, and L. L. Gheesling. 2004. Supplement 2002(no. 46) to the Kauffmann-White scheme. Res. Microbiol. 155:568–570.

30. R Development Core Team. 2009. R: a language and environment for sta-tistical computing. R Foundation for Statistical Computing, Vienna, Austria.http://www.R-project.org/.

31. Scott, T. M., J. B. Rose, T. M. Jenkins, S. R. Farrah, and J. Lukasik. 2002.Microbial source tracking: current methodology and future directions. Appl.Environ. Microbiol. 68:5796–5803.

32. Simental, L., and J. Martinez-Urtaza. 2008. Climate patterns governing thepresence and permanence of salmonellae in coastal areas of Bahia de TodosSantos, Mexico. Appl. Environ. Microbiol. 74:5918–5924.

33. Tavechio, A. T., A. C. Ghilardi, J. T. Peresi, T. O. Fuzihara, E. K. Yonamine,M. Jakabi, and S. A. Fernandes. 2002. Salmonella serotypes isolated fromnonhuman sources in Sao Paulo, Brazil, from 1996 through 2000. J. Food.Prot. 65:1041–1044.

34. Uzzau, S., D. J. Brown, T. Wallis, S. Rubino, G. Leori, S. Bernard, J.Casadesu�s, D. J. Platt, and J. E. Olsen. 2000. Host adapted serotypes ofSalmonella enterica. Epidemiol. Infect. 125:229–255.

35. Venkateswaran, K., T. Takai, I. M. Navarro, H. Nakano, H. Hashimoto, andR. J. Siebeling. 1989. Ecology of Vibrio cholerae non-O1 and Salmonella spp.and role of zooplankton in their seasonal distribution in Fukuyama coastalwaters, Japan. Appl. Environ. Microbiol. 55:1591–1598.

36. Voetsch, A. C., T. J. Van Gilder, F. J. Angulo, M. M. Farley, S. Shallow, R.Marcus, P. R. Cieslak, V. C. Deneen, R. V. Tauxe, and Emerging InfectionsProgram FoodNet Working Group. 2004. FoodNet estimate of the burden ofillness caused by nontyphoidal Salmonella infections in the United States.Clin. Infect. Dis. 38:S127–134.

37. Wiggins, B. A. 1996. Discriminant analysis of antibiotic resistance patterns infecal streptococci, a method to differentiate human and animal sources offecal pollution in natural waters. Appl. Environ. Microbiol. 62:3997–4002.

38. Wiggins, B. A., R. W. Andrews, R. A. Conway, C. L. Corr, E. J. Dobratz, D. P.Dougherty, J. R. Eppard, S. R. Knupp, M. C. Limjoco, J. M. Mettenburg,J. M. Rinehardt, J. Sonsino, R. L. Torrijos, and M. E. Zimmerman. 1999.Use of antibiotic resistance analysis to identify nonpoint sources of fecalpollution. Appl. Environ. Microbiol. 65:3484–3486.

39. Willson, I. G., and J. E. Moore. 1996. Presence of Salmonella spp. andCampylobacter spp. in shellfish. Epidemiol. Infect. 116:147–153.

40. Winfield, M. D., and E. A. Groisman. 2003. Role of nonhost environments inthe lifestyles of Salmonella and Escherichia coli. Appl. Environ. Microbiol.69:3687–3694.

41. Wood, S. N. 2003. Thin plate regression splines. J. R. Statist. Soc. B 65:95–114.

42. Wood, S. N. 2004. Stable and efficient multiple smoothing parameter esti-mation for generalized additive models. J. Am. Stat. Assoc. 99:673–686.

43. Wood, S. N. 2006. Generalized additive models: an introduction with R.CRC Press, Boca Raton, FL.



http://aem.asm

.org/D

ownloaded from

http://aem.asm.org/

characteristics and dynamics of salmonella …from six sites between july 2004 and may 2008. the...

Documents