International Journal of Food Microbiology 184 (2014) 92–97

Contents lists available at ScienceDirect

International Journal of Food Microbiology

j ourna l homepage: www.e lsev ie r .com/ locate / i j foodmicro

Study of the cross-contamination and survival of Salmonella in fresh apples

F. Perez-Rodriguez a,b,⁎, M. Begum c, G.S. Johannessen c

a Department of Food Science and Technology, University of Córdoba, Campus de Rabanales, C-1, 14014 Córdoba, Spainb International Campus of Excellence in the AgriFood Sector (CeIA3), Spainc Norwegian Veterinary Institute, Section for Bacteriology—Food and GMO, P.O. Box 750 Sentrum, 0106 Oslo, Norway

⁎ Corresponding author at: Department of Food Scienof Córdoba, Campus de Rabanales, C-1, 14014 Córdoba,fax: +34 957 212000.

E-mail address: b42perof@uco.es (F. Perez-Rodriguez)

http://dx.doi.org/10.1016/j.ijfoodmicro.2014.03.0260168-1605/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Available online 29 March 2014

Keywords:Cross contaminationSalmonellaPredictive microbiologySampling planFruitWeibull model

The present work aimed at studying the cross contamination of apples by Salmonella during the processing ofcommercial fresh apples and its survival capacity on apple at room temperature. For the first study, the typicalprocess of fresh apples was simulated at laboratory scale in which an apple that was artificially contaminatedby Salmonella at different concentration levels (8, 6 and 5 log cfu/apple) was introduced in one batch and proc-essed including a simulated transport/washing step and drying step using sponges to simulate the porousmate-rial used in the industry. Results indicated that at 8 log cfu/apple, 50% fresh apples were contaminated afterprocessing, with all analysed environmental samples being positive for the pathogen, consisting of washingwater and sponges. However, at lower inoculum levels (5–6 log cfu/apple) no cross contamination was detectedin apples, and only environmental samples showed contamination by Salmonella after processing including bothwater and sponges. Experiments on the survival of Salmonella on apple showed that the pathogenwas capable tosurvive for 12 days, only showing a significant drop at the end of the experiment. Finally, two-class attributesampling plans were assessed as tool to detect Salmonella in different contamination scenarios in fresh apple.This analysis indicated that with the highest inoculum level, a total of 16 apples would be needed to reach 95%of detecting Salmonella (i.e. lot rejection). In turn, when low levels were assessed (5–6 log cfu/apple), a largenumber of apples (n= 1021) would have to be sampled to obtain the same confidence level (95%). If the envi-ronment is sampled (i.e. water and sponges), a lower number of samples would be needed to detect the patho-gen. However, the feasibility of environmental sampling has not been assessed from a practical point of view.Overall, the results in this study evidenced that cross contamination by Salmonellamight occur during processingof fresh apples and subsequently, the pathogen might survive for a noticeable period of time.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

In recent years, foodborne illness associated with fresh produce hasbecomemore common (EFSA, 2013). Themain challengeswith the foodsafety of fresh produce are that these products are often grown outsidein an uncontrollable environment where they are exposed to contami-nation from different sources and are often consumed raw and withoutany forms of risk reducing treatments, such as heat treatment. The con-tamination of fresh producewith hazardousmicroorganisms is also het-erogeneous, unevenly spread and most likely present in low numbers.This makes it very difficult to detect the presence of foodborne patho-gens such as Salmonella spp. or others during routine sampling.

Until now, leafy greens and sprouted seeds have caused the mostnumbers of outbreaks due to many outbreaks associated with leafy

ce and Technology, UniversitySpain. Tel.: +34 957 212057;

.

greens and large outbreaks (N1000 persons ill) associated with sprouts(EFSA, 2013;Michino et al., 1999; Buchholz et al., 2011). However, othertypes of fresh produce, such as tomatoes and melons have also causedoutbreaks. There is little information about microbial hazards oftree fruits, such as apples, although there have been a few outbreaksof foodborne disease associated with consumption of apple cider. Inthese cases, E. coli O157:H7 and Cryptosporidium were the culprits(Millard et al., 1994; Anonymous, 1997; Blackburn et al., 2006). Thereis sufficient evidence, through experiments, that such products maybe contaminated by enteric pathogens and that the pathogens maysurvive for some time onand in the products, especiallyminimally proc-essed fruits and also in apple cider (Alegre et al., 2009; Zhao et al., 1993).A study by Abadias et al. (2006) indicated that only a few apple samplesfrom orchards or packhouses harbored E. coli, suggesting that contami-nation is sporadic and consequently will be difficult to detect. In thissense, the development of new real-time, non-destructive online in-spectionmethods could help to improve detection of pathogens in fruitsas already demonstrated in previous work (Yang et al., 2012).

Given the low frequency of contamination, and the great lack of data,it is unknown if and how a sporadic contamination of tree fruits could

93F. Perez-Rodriguez et al. / International Journal of Food Microbiology 184 (2014) 92–97

be spread during processing and thus further contaminate uncontami-nated products. In this sense, it is assumed that the potential for cross-contamination is present if one (or more) units are contaminated. Ifcross contamination occurs and no lethal or inactivation treatmentsare applied afterward, sampling plans may become important toolsintended to detect the possible contamination (Jongenburger et al.,2011a, 2011b), although their effectiveness will depend on the levelsof contamination and their distribution in the final product to be tested.In this study apples were chosen as a model product for tree fruits forinvestigating the potential for cross contamination and survival of anenteric pathogen such as Salmonella spp. during industrial processingand subsequent storage, respectively.

The aim of this work was i) to study the potential for cross contam-ination of apples simulating industrial processing, ii) to assess the sur-vival Salmonella on the apples after a cross contamination event, andiii) based on these data, to assess suitable sampling plans to detectSalmonella in apples.

2. Material and methods

For the experiments, fresh, unprocessedNorwegian apples of the va-riety Summerred (apples grown in Norway) were collected directly at alocal packing house. In addition, organically produced apples importedfrom Italy (Süd-Tirol) of the variety Braeburn were purchased at alocal supermarket for the final rounds of the experiments. The applesused presented no injuries (cuts) and were not treated prior to the ex-periment. The organic apples may have been washed, but had not re-ceived any further treatment.

2.1. Preparation of inoculum and contamination of apples

Salmonella Reading (VI 51763), previously isolated from spent irri-gation water from sprout production (Robertson et al., 2002), was plat-ed onto Blood agar from glycerol stock kept at−80 °C and incubated at37 °C overnight. One single colony was transferred to 9 ml of bufferedpeptonewater (BPW) (OXOID, Basingstoke, United Kingdom) and incu-bated overnight at 37 °C. The culture was serially diluted in BPW and100 μl of the appropriate dilution was carefully spot-inoculated on theapple surface to obtain the initial contamination on apple. The contam-inated apples were dried for 1–2 h in a safety cabinet at room tempera-ture. A total of 3 apples were contaminated in each round for thewashing experiments; two were used as inoculation controls, whilethe third was stored in a plastic bag with 29 uninoculated applesin the refrigerator at 3 ± 2 °C overnight before processing. To quantifySalmonella in the inoculum, 100 μl of the appropriate dilutionswas plat-ed in parallel on blood agar (bovine blood) and incubated at 37 °C over-night. The cross-contamination experiment was performed at threeinitial contamination levels corresponding to 5, 6 and 8 log cfu/apple.

For the storage experiment, single apples were drop inoculated andair-dried as described above before further storage. The initial inocu-lums on the apples corresponded to 6 log cfu/apple.

2.2. Experimental design

2.2.1. Simulation of processing lineA total of 30 apples including one contaminated apple were proc-

essed through a simulated apple processing line. A vat with 12 l potablewater was used to simulate transport/washing bath and two roundswith 15 apples in each round were left in the bath for approximately5 min with gentle shaking. After the transport/wash bath the appleswere rolled over sponges (to simulate the porous material that areused to dry and drain off water) before they were left to air-dry for1 h. After drying the apples were packed in zip-lock plastic bags withseven or eight apples in each bag simulating commercial packaging.The apples were stored at 3 ± 2 °C overnight prior to analysis.

2.2.2. Storage experimentTo investigate the survival of Salmonella on the surface of apples

after a simulated cross contamination event, a storage experimentwas carried out at room temperature. A total of 30 apples were used,where 28 were spot inoculated as described above (2.1.1), and twoleft uninoculated as negative controls. Three apples were analysedimmediately as inoculation control, while the remaining 25 inoculatedapples were stored in zip-lock bags with five apples in each bag. Thenegative apples were also stored together. The apples were stored at22 °C and ~70% RH for a total of 12 days. One bag with five apples wasremoved and analysed after 1, 2, 5, 6 and 12 days, respectively.

2.3. Bacterial analysis

The apples, sponges and water were analysed for the presence ofSalmonella using a modified version of NMKL no. 71, 5th ed. 1999(NMKL, 1999). Briefly, for qualitative analysis in the processing experi-ment, each apple was analysed separately, adding 225 ml BPW to eachapple. The appleswere gently rubbed in theBPW. The spongeswere dis-tributed into Stomacher bags, 1–2 per bag depending on the size and225 ml of BPW was added. A volume of 500 ml of the processingwater was retrieved from the vat and filtered through a 0.45 μm filter(Millipore S-PakTM Membrane Filters, Millipore, Billerica, MA, USA)and 100 ml BPW was added to the filter prior to enrichment. Enrich-ments were incubated 37 °C for 24± 3 h. A total of 100 μl of the enrich-ment cultures was transferred to 10 ml of Rappaport-Vassiliadis-SoyaPeptone Broth (OXOID) and incubated for 24 ± 3 h at 41.5 °C, followedby plating on XLD (OXOID) and Brilliance™Salmonella agar (OXOID).The plates were incubated at 37 °C for 24 h and examined for typicaland suspicious colonies. The presence of Salmonella was confirmed bytesting the colonies on Triple Sugar Iron agar (Difco, MD, USA) andUrea agar (Agar base: OXOID, with 40% Urea (Sigma-Aldrich, St. Louis,MO, USA) followed by agglutinationwith omnivalent Salmonella antise-rum (Enteroclon Anti-Salmonella A-67, omnivalent, Sifin, Berlin,Germany).

For quantification of Salmonella in the storage experiment, eachapple was gently rubbed in 100 ml of BPW. Serial dilution series wereprepared in BPW and 0.1 ml of the appropriate dilutions was platedon XLD agar. An aliquot of 1 ml of the primary dilution was plated onthree plates to achieve a detection limit of 100 cfu/apple. The apple–BPW mix was further incubated and analysed as described above forqualitative detection in case there were less than 100 cfu of Salmonellaper apple.

2.3.1. Data treatment and mathematical modellingExperiments were repeated three times in different days in order to

capture biological variability. For cross contamination experiments,probability of cross contamination was described as percentage (perone) of positive samples obtained in the different scenarios. Survivaldata of Salmonella spp. on fresh apples were tabulated and standardizedto represent ΔN= N0 − Nt expressed in log cfu/g with respect to timein days. These standardized data were submitted to regression analysisto fit different mathematical functions describing survival or log de-crease along time (Table 1). The regression procedure was performedby the curve fitting tool implemented in the MATLAB 7.7.0 Software(The MathWorks Inc. 2008).

3. Results

3.1. Cross contamination at processing line

Concentration levels transferred to environment and apples werenot quantified in the experiment since levels were below the limit ofquantification (b100 cfu/apple). Hence, results were expressed as thenumber or the percentage of positive samples. Results indicated thatcross contamination did not take place at low initial inoculum levels,

Table 1Description of the survival kinetic models used to analyse the experimental data obtained in this study.

Model name Model Model parameters References

Log linear log10 N = log10 N0 − (kmax × t / Ln(10)) kmax Bigelow and Esty, 1920Log linear + shoulder log10 N = log10 [(10^log10 N0 − 10^log10 N_res) × exp(−kmax × t) + 10^log10 N_res] kmax, N_res Geeraerd et al., 2000Weibull log10 N = log10 N0 − ((t / delta)^ p)) delta, p Mafart et al., 2002Biphasic model log10 N = log10 N0 + log10 [f × exp(−kmax1 × t) + (1 − f) × exp(−kmax2 × t)] f, kmax1, kmax2 Cerf, 1997

kmax: the specific inactivation rate (h−1);N: themicrobial population at time t (cfu/cm2);N0: themicrobial population at time zero (cfu/cm2);N_res: the residual populationdensity (cfu/cm2);delta: the scale parameter; p: the shape parameter; kmax1 and kmax2: the specific inactivation rates of the two subpopulations (h−1); f: the fraction of amajor less-resistant subpopulation in thetotal initial population.

94 F. Perez-Rodriguez et al. / International Journal of Food Microbiology 184 (2014) 92–97

i.e. 5–6 log/apple since the only detected positive sample correspondedto the initially inoculated one. Thiswas only observed for one of the rep-etitions for 5 log cfu/apple. In turn, at a high inoculum level, 8 log cfu/apple, 50% apples were contaminated after processing (Table 2). Albeitlow levels did not yield cross contamination, environmental samples,sponges and water, became contaminated during processing. Percent-ages of positive environmental samples are given in Table 2. Lookingat the average values of the three repetitions, 50 and 37% spongeswere positive at initial inoculum of 5 and 6 log cfu/apple, respectively.Importantly, in one of the repetitions for 5 log cfu/apple, all spongeswere positive for Salmonella spp. Forwater samples, nopositive sampleswere found for 6 log cfu/apple; however for 5 log cfu/apple, only onesample showed contamination by Salmonella. This higher contamina-tion rate in environmental samples for 5 log cfu/apple could be derivedfrom the experimental variability, which becomes more importantwhen contamination levels are close to the detection limit of the appliedmethod. So, slight variations in the contamination levels between ex-periments could result in either positive or negative samples. In thecase of high inoculum levels, all environmental samples were positive(Table 2). These differences between the different spiking levels suggestthat contamination distribution is enhanced at high inoculum levels.Furthermore, although the processing water was expected to be themain contamination vehicle, sponges were more likely contaminated.

0

0.5

1

1.5

og(N

/N0)

3.2. Salmonella survival during storage

The results of the Salmonella survival on apple during storage atroom temperature are shown in Fig. 1. Themicroorganismshowed a no-ticeable ability to survive at the assayed storage conditions, particularlyin the first six days in which themicroorganism did not exhibit any sig-nificant decrease. Then, from days 6 to 12, levels of Salmonella signifi-cantly decreased around 1 log. This decrease means a reduction rate of0.16 log cfu/apple day in the last 6 days.

The surface of an apple is a harsh environment for enteric pathogens,such as Salmonella, and it is expected that the numbers decline overtime. The fact that the apples were stored at room temperature andhigh RH may have assisted the survival up till six days. After six days,the shortage of available nutrients and competition from resident mi-croorganisms may have led to the decrease we observe from six to12 days. Although significant, the decrease is still rather low (approxi-mately 1 log cfu), indicating that if Salmonella is present on applesstored at ambient room temperature and humidity (RH), they may sur-vive for a noticeable time.

Table 2Mean and standard deviation of the percentages (%) of positive samples (apples andenvironment) at the different inoculum levels.

Experiment Positive apple Positive sponge Positive water

5 log/apple 2.2 ± 1.9 50.0 ± 50.0 33.36 log/apple 0.0 ± 0.0 37.5 ± 53.0 0.08 log/apple 50.0 ± 8.8 100.0 ± 0.0 100.0

In order to assess the suitability of the use of predictive models todescribe Salmonella survival on apples during storage, different math-ematical functions representing inactivation were fitted to survivaldata. The analysis indicated that given the great variability of data inthe initial shoulder, no model might be adequately fitted to data. How-ever, theWeibull model with parameters a= 12.03 and b= 22.63wasat least capable to account for the phases observed in the survival data(Fig. 1), the initial shoulder and the final decline.

3.3. Using data from lab-scale experiments to assess sampling schemes

Outcomes from the lab-scale experiments (i.e. cross contamination)were used to assess potential sampling schemes for Salmonella spp. inapple. For this, sampling schemes were derived based on the followingassumptions: 1) contamination levels in apples corresponded to thevalues found in our experiments, 2) the analytical method appliedwas used as reference method, and sensitivity and specificity are notconsidered in this study, 3) two different scenarios were considered,i.e., a first scenario considering no cross contamination (5–6 log) and asecond scenario including cross contamination (8 log), and 4) a two-class attribute sampling was assessed based on parameters n and c,which represents the number of samples to be tested and the numberof positive sample determining the acceptability of a lot, respectively(Valero et al., 2013). In the case of pathogens, c is usually establishedto 0, which means the lot is accepted if no samples are positive for thepathogen. So, the performance of the sampling plan was assessed fordifferent values of n.

Sampling the lot is assumed to follow a binomial process “Binomial(n;p)” and uncertainty about the prevalence detected after processing(p) is modelled following a Bayesian approach with a beta distributiondefinedwith parameters s+1 and n− s+1, s and n being the numberof positive and total samples, respectively, obtained from experimentswith inoculated apples. Based on this uncertainty distribution, 5th,

-1.5

-1

-0.5

0 2 4 6 8 10 12 14

Time (days)

l

Fig. 1. Survival data (mean and standard deviation) obtained for Salmonella spp. on freshapples and the fitted Weibull model (——————).

00.10.20.30.40.50.60.70.80.9

1

0 2 4 6 8 10 12

Number of samples (n)P

roba

bilit

y of

rej

ecti

on

00.10.20.30.40.50.60.70.80.9

1

0 2 4 6 8 10 12

Number of samples (n)

Pro

babi

lity

of r

ejec

tion

0

0.2

0.4

0.6

0.8

1

0 2 4 6 8 10 12

Number of samples (n)

Pro

babi

lity

of r

ejec

tion

A

B

C

Fig. 2. Curves representing probability of rejection of a lot when one contaminated apple enters the processing line with low (A) and high (B) levels of Salmonella spp. and curverepresenting probability of detection of Salmonella spp. in sponges when one contaminated apple enters the processing line with low levels (C) as function of the number of samples(n) at different prevalence levels after processing corresponding with 5th, 50th and 95th percentiles of the uncertainty distribution of prevalence.

95F. Perez-Rodriguez et al. / International Journal of Food Microbiology 184 (2014) 92–97

50th and 95th percentiles of prevalence (p) were used to simulate dif-ferent contamination scenarios as worst and best cases.

In Fig. 2, the probability of rejection is represented versus the num-ber of samples for apples at both initial levels (Fig. 2A and B) andsponges at low levels (Fig. 2C), respectively, considering the three prev-alence uncertainty levels explained above. These graphs show in onesnapshot how feasible the application of sampling schemes can be ineach case or scenario. On the other hand, if contamination data, thatis, prevalence (p) uncertainty levels are applied to Eq. (1), the numberof samples (n) needed to reject a contaminated lot can be estimatedfor a specific significance level α (i.e. confidence level) according toEq. (1):

n ¼ log αð Þ.log 1−pð Þ ð1Þ

In the case of one apple entering the processing linewith a relativelylow contamination level (non-cross-contamination), the sample sizeneeded to detect one contaminated sample (to reject the lot)with a sig-nificance level of 0.05 corresponded to n = 1021, 85 and 22 for 5th,50th and 95th percentiles of prevalence. For one apple entering theprocessing line with high contamination level, the sample size neededto detect one contaminated sample (to reject the lot)with a significancelevel of 0.05 corresponded to n = 16, 10, and 7 for 5th, 50th and 95thpercentiles of prevalence. If sponges are taken as contamination indica-tors in the scenario of one contaminated apple entering processing lineat the low level, the number of sponge samples needed to detect thepresence of the pathogen with a significance level of 0.05 corresponded

to n = 71, 7 and 3 samples for 5th, 50th and 95th percentiles of preva-lence. Results indicated that low levels required a very high number ofsamples to detect Salmonella spp.

3.4. Discussion

The results from the described experiments indicate that cross con-tamination of apples with Salmonellamay take place during processing.However, this was only observed at high levels. In spite of this result, itcannot be discarded that in the scenario simulating initial low levels,processed apples could be contaminated at so low levels that the recov-ery method was not sufficiently efficient in removing the bacteria fromthe apple surface. Interestingly, contamination was detected in spongesand water in the scenario simulating low levels when no cross contam-ination was detected in apples. As a consequence of this result, it couldbe suggested that the processing water and sponges or porous materialused to remove excess water could act as cross-contamination vehicleor reservoir for cross contamination affecting different productionsand days. In such cases, the application of more stringent cleaning andsanitation procedures can be suitable control measures to reduce theappearance of contamination reservoirs and hence the probabilityof cross contamination between work shifts, productions and days(Carrasco et al., 2012). Besides that, sampling plans could be used as ad-ditional tools to check the existence of cross contamination during pro-duction. Although scarce, someworks have assessed the performance ofdifferent sampling plan strategies for the detection of pathogenic con-tamination in different foods such as milk powder (Jongenburger et al2011ab). In our case, the analysis of the performance of the two-class

96 F. Perez-Rodriguez et al. / International Journal of Food Microbiology 184 (2014) 92–97

attribute sampling plans to detect the pathogen in apples was not effec-tive in detecting cross contamination due to the relatively high numberof apples to be tested (i.e. 16 samples at high levels). However, environ-mental sampling showed a better performance, especially in the case ofsponges (or porous material) which could be used as a contaminationindicator. So far, no assessment has beenmade regarding if this environ-mental sampling plan is affordable from a practical point of view. In ourstudy, the whole sponge was removed and tested. However, accordingto feedback from the industry, the removal and subsequent analysis ofthe sponges/porous material may prove difficult in real life. Therefore,another alternative could be to use the excess/drainwater from the dry-ing process, though this was not tested in the experiment describedhere. Although testing of the processing water was not as successful asexpected, the sensitivity of the analysis could be improved by analyzinga larger volume of water. Here, 500 ml was filtered, and the filter wasfurther enriched and analysed according to the described method. Byincreasing the volume of water filtered, the sensitivity of the methodwould be improved.

Data from the storage experiment showed that Salmonellamay sur-vive on the surface of whole, unblemished apples for at least 12 dayswhen the apples are stored at room temperature (22 °C). Other studieshave shown that the number of pathogens, such as SalmonellaTyphimurium and E. coli O157:H7 on the surface of whole apples de-creases significantly after one to two days, but these apples were storedat lower temperatures and at other relative humidity values (Tian et al.,2013) than used in our study. The levels of reduction were higher forwhole apples, than for bruised and cut apples. There were also varia-tions in reduction levels depending on the type of microorganism,i.e. E. coli O157 and Salmonella. Collignon and Korsten (2010) investi-gated the survival of Salmonella Typhimurium and E. coli O157:H7 onpeaches and plums through a simulated commercial export chain.Their results indicated that Salmonella decreased quite rapidly duringthe period of cold storage, and that the numbers stayed low after a 24hour period at 21 °C. However, Collignon and Korsten (2010) did notextend the storage at room temperature as we have done in thepresented study. The same authors showed that the numbers ofE. coli O157:H7 and Salmonella Typhimurium on peaches did not over-all decrease significantly during a storage period of six days underfluctuating temperature conditions (0–1 day at 21 °C, 1–3 days at0.5 °C, 3–5 days at 21 °C, 5–6 days at 4 °C). However, the numbersdecreased significantly on plums stored under the same temperatureregime (Collignon and Korsten, 2012). The results from Tian et al.(2013) and Collignon and Korsten (2010, 2012) indicate that pathogen-ic bacteria are able to survive on the surface of fruits, albeit the resultsvary depending on fruit type, temperatures, relative humidity andtype of bacteria. These results support our results that demonstratethat if cross contamination occurs during primary processing, thenSalmonella is capable to survive during a significant period of time onwhole apples. The results indicate thus that the risk by Salmonellacould remain along the food chain of fresh apples.

Although the prevalence of pathogenic bacteria in whole fresh ap-ples is seemingly low, experiments indicate that pathogens do surviveon whole apples, bruised apples and also cut surfaces. Taking into ac-count the increase in convenience foods, such as minimally processed,pre-cut fruit where there is a potential of cross contamination of patho-gens from the whole fruit to the ready-to-eat product, it is important tostart considering sampling strategies for such products. It is believedthat the results from the described experiments also can be transferred,probably not directly, to other fruit types, similar to apples, wherewateris used for transport or washing.

4. Conclusion

Results showed that cross contamination could occur during wash-ing and preparation of fresh apples; however, high initial levels wereneeded to result in a significant increase of positive apples at the end

of the process. Moreover, the environment (water and sponges)seems to be a relevant transmission vehicle during apple processing.In this sense, the theoretical assessment of sampling plans suggeststhat environmental sampling might enable better detection of crosscontamination than sampling apples as the latter required a highernumber of samples. The long survival showed by Salmonella on applesalso evidences the potential risk associated to this hazard/foodcombination. In spite of the conclusions, scaled up experiments shouldbe performed in order to both confirm these results and assessSalmonella cross contamination in different tree fruits and food process-ing scenarios.

Acknowledgements

This work was carried out in the BASELINE project. BASELINE isfunded by The Seventh Framework Program of the European Commis-sion (Grant agreement number 222738).

References

Abadias, M., Cañamas, T.P., Asensio, A., Anguera, M., Viñas, I., 2006. Microbial quality ofcommercial “Golden Delicious” apples throughout production and shelf-life in Lleida(Catalonia, Spain). Int. J. Food Microbiol. 108, 404–409.

Alegre, I., Abadias, M., Anguera, M., Oliveira, M., Viñas, I., 2009. Factors affecting growth offoodborne pathogens on minimally processed apples. Food Microbiol. 27, 70–76.

Anonymous, 1997. Outbreaks of Escherichia coli O157:H7 infection and cryptosporidiosisassociated with drinking unpasteurized apple cider, Connecticut and New York,October 1996. Morb. Mortal. Wkly MMWR 46, 4–8.

Bigelow, W.D., Esty, J.R., 1920. The thermal death point in relation to typical thermophylicorganisms. J. Infect. Dis. 27, 602–617.

Blackburn, B.G., Mazurek, J.M., Hlavsa, M., Park, J., Tillapaw, M., Parrish, M.K., Salehi, E.,Franks, W., Koch, E., Smith, F., Xiao, L., Arrowood, M., Hill, V., da Silva, A., Johnston,S., Jones, J.L., 2006. Cryptosporidiosis associated with ozonated apple cider. Emerg. In-fect. Dis. 12 (4), 684–686.

Buchholz, U., Bernard, H., Werber, D., Böhmer, M.M., et al., 2011. German outbreak ofEscherichia coliO104:H4 associatedwith sprouts. N. Engl. J.Med. 365 (19), 1763–1770.

Carrasco, E., Morales-Rueda, A., Garcia-Gimeno, R.M., 2012. Cross-contamination and re-contamination by Salmonella in foods: a review. Food Res. Int. 45, 545–556.

Cerf, O., 1997. Tailing of survival curves of bacterial spores. J. Appl. Bacteriol. 42, 1–19.Collignon, S., Korsten, L., 2010. Attachment and colonization by Escherichia coli O157:H7,

Listeria monocytogenes, Salmonella enterica subsp. Enterica serovar Typhimurium, andStaphylococcus aureus on stone fruit surfaces and survival through a simulated com-mercial export chain. J. Food Prot. 72 (7), 1247–1256.

Collignon, S., Korsten, L., 2012. Growth dynamics of Escherichia coli O157:H7, Listeriamonocytogenesm Salmonella enterica subst. enterica serovar Typhimurium and Staph-ylococcus aureus under different nutrient and temperature conditions. In: Collignon, S.(Ed.), Risk assessment of Escherichia coli O157:H7, Listeria monocytogenes, Salmonellaenterica subsp. enterica serovar Typhimurium and Staphylococcus aureus within astone fruit production environment. Faculty of Natural and Agricultural Sciences, De-partment of Microbiology and Plant Pathology, University of Pretoria, South Africa(PhD-thesis, July 2012).

EFSA, 2013. Scientific opinion on the risk posed by pathogens in food of non-animal ori-gin. Part 1 (outbreak data analysis and risk ranking of food/pathogen combinations).EFSA J. 11 (1), 3025 (Available online: www.efsa.europa.eu/efsajournal).

Geeraerd, A.H., Herremans, C.H., Van Impe, J.F., 2000. Structural model requirements todescribe microbial inactivation during a mild heat treatment. Int. J. Food Microbiol.59, 185–209.

Jongenburger, I., Reij, M.W., Boer, E.P., Gorris, L.G., Zwietering, M.H., 2011a. Actualdistribution of Cronobacter spp. in industrial batches of powdered infant formu-la and consequences for performance of sampling strategies. Int. J. Food Microbiol.151, 62–69.

Jongenburger, I., Reij, M.W., Boer, E.P.J., Gorris, L.G.M., Zwietering, M.H., 2011b. Random orsystematic sampling to detect a localised microbial contamination within a batch offood. Food Control 22, 1448–1455.

Mafart, P., Couvert, O., Gaillard, S., Leguerinel, I., 2002. On calculating sterility in thermalpreservation methods: application of the Weibull frequency distribution model. Int.J. Food Microbiol. 72, 107–113.

Michino, H., Araki, K., Minami, S., Takaya, S., Sakai, N., Miyazaki, M., Ono, A., Yanagawa, H.,1999. Massive outbreak of Escherichia coli O157:H7 infection in schoolchildren inSakai City, Japan, associated with consumption of white radish sprouts. Am. J.Epidemiol. 150, 787–796.

Millard, P.S., Gensheimer, K.F., Addiss, D.G., Sosin, D.M., Beckett, G.A., Houck-Jankoski, A.,Hudson, A., 1994. An outbreak of cryptosporidiosis from fresh-pressed apple cider.JAMA 272 (20), 1592–1596.

NMKL, 1999. Salmonella. Detection in foods, NMKL Method No. 715th ed. NMKL, Oslo,Norway.

Robertson, L.J., Johannessen, G.S., Gjerde, B.K., Loncarevic, S., 2002. Microbiological analy-sis of seed sprouts in Norway. Int. J. Food Microbiol. 75 (1–2), 119–126.

97F. Perez-Rodriguez et al. / International Journal of Food Microbiology 184 (2014) 92–97

Tian, J.Q., Bae, Y.M., Lee, S.Y., 2013. Survival of foodborne pathogens at different relativehumidities and temperatures and the effect of sanitizers on apples with different sur-face conditions. Food Microbiol. 35, 21–26.

Valero, A., Carrasco, E., Pérez-Rodríguez, F., García-Gimeno, R.M., Romero, M.P., Marín, S.,2013. Food contaminanants. In: Pawliszyn, J., Bayona, J.M. (Eds.), ComprehensiveSampling and Sample Preparation. Elsevier, Academic Press, Oxford, UK, pp. 381–394.

Yang, C.-C., Kim, M.S., Kang, S., Cho, B.-K., Chao, K., Lefcourt, L.M., Chan, D.E., 2012. Red tofar-red multispectral fluorescence image fusion for detection of fecal contaminationon apples. J. Food Eng. 108, 312–319.

Zhao, T., Doyle, M.P., Besser, R.E., 1993. Fate of enterohemorrhagic Escherichia coli O157:H7 in apple cider with and without preservatives. Appl. Environ. Microbiol. 59 (8),2526–2530.