Crescimento de Salmonella Enteritidis e Listeria Monocytogenes em
Transcrição
Crescimento de Salmonella Enteritidis e Listeria Monocytogenes em
Growth of Salmonella Enteriditidis and Listeria Monocytogenes in Melon Pulp: Predictive Modelling and Evaluation of Model Performance Crescimento de Salmonella Enteritidis e Listeria Monocytogenes em Polpa de Melão: Modelamento Preditivo e Avaliação do Desempenho de Modelos AUTORES AUTHORS Maria Fernanda Pontes Penteado Moretzsohn de CASTRO Instituto de Tecnologia de Alimentos (ITAL) Grupo Especial de Engenharia e Pós-Colheita (GEPC) Caixa Postal: 139 Av. Brasil 2880, Jd. Brasil CEP 13070-178 Campinas/SP - Brasil e-mail: [email protected] Flávio Luis SCHMIDT Faculdade de Engenharia de Alimentos (FEA) Departamento de Tecnologia de Alimentos e-mail: [email protected] Ana Lucia PENTEADO Empresa Brasileira de Pesquisa Agropecuária Centro Nacional de Pesquisa de Tecnologia Agroindustrial de Alimentos (CTAA) e-mail: [email protected] SUMMARY The growth of Salmonella Enteritidis and Listeria monocytogenes in melon pulp was verified at 10, 20 and 30 °C, and the observed growth parameters (generation time, lag phase duration and exponential growth rate) for these microorganisms compared. The experimental data were modelled using the Gompertz function and the DMFit programme and the kinetic values obtained were compared with those obtained from the observed growth and with the kinetic values for the growth of Salmonella spp. and L. monocytogenes as predicted by the U.S. Department of Agriculture’s Pathogen Modelling Program version 5.1. The results confirmed that low temperature is not a barrier to the growth of Listeria and S. Enteritidis; it can retard growth but not inhibit it. Tukey’s Paired Comparison Procedure, which compared the predicted and observed values for GT, LPD and EGR, showed that the values predicted by PMP were generally those that statistically differed most from the values observed, especially for the LPD values at 10 °C. The use of PMP in this situation could be critical, as the predictions were fail dangerous The DMFit and Gompertz values differed statistically from the observed ones in a few cases, but these cases were not critical since they predicted on the safe side. It was concluded that the Gompertz function, DMFit and PMP programmes can generally be used with a certain confidence, with the exception of PMP at low temperatures, which, in some cases can fail on the dangerous side. RESUMO PALAVRAS-CHAVE KEY WORDS Predictive microbiology; Salmonella Enteritidis; Listeria monocytogenes; Melon pulp; Modelling; Validation. Microbiologia preditiva; Salmonella Enteritidis; Listeria monocytogenes; Polpa de melão; Modelamento; Validação. O crescimento de Salmonella Enteritidis e Listeria monocytogenes na polpa de melão foi verificado a 10, 20 e 30 °C e os parâmetros de crescimento observados (tempo de geração, duração da fase lag (latência) e a taxa de crescimento exponencial para estes microrganismos foram comparados. Os dados experimentais foram modelados utilizando-se a função de Gompertz e o programa DMFit, e os valores cinéticos foram comparados com os valores cinéticos obtidos para Salmonella spp. e L. monocytogenes preditos pelo U.S. Department of Agriculture’s Pathogen Modelling Program (PMP) versão 5.1. Os resultados confirmaram que baixa temperatura não é uma barreira para o crescimento de L. monocytogenes e S. Enteritidis, pois ela pode retardar, mas não inibir o crescimento desses microrganismos. O procedimento de comparação pareada de Tukey, que compara os valores preditos e os observados para tempo de geração, duração da fase lag e taxa de crescimento exponencial indicaram que, de um modo geral, os valores preditos pelo PMP foram aqueles que mais diferiram estatisticamente dos observados. Especialmente para os valores de duração da fase lag a 10 °C, o uso do PMP, nessa situação, poderia ser crítico, uma vez que as predições falham perigosamente. Em poucas situações os valores do DMFit e Gompertz diferiram estatisticamente daqueles observados e, nesses casos, não foi crítico, uma vez que predisseram pelo lado seguro. Concluiu-se que a função de Gompertz e os programas DMFit e PMP, de um modo geral, podem ser usados com alguma confiança, com exceção do PMP para temperaturas baixas, que em alguns casos pode falhar perigosamente. Autor Correspondente Corresponding Author Braz. J. Food Technol., Campinas, v. 10, n. 3, p. 176-182, jul./set. 2007 176 Recebido/Received: 16/04/2007. Aprovado/Approved: 03/10/2007. CASTRO, M. F. P. P. M. et al. 1. INTRODUCTION Predictive microbiology is a tool that uses mathematical models to estimate or predict the responses of microorganisms under environmental conditions of interest (QUINTAVALLA and PAROLARI, 1993; WHITING and BUCHANAN, 1994). Amongst other uses, predictive microbiology can be of great value in quantitative microbial risk assessment by predicting the growth parameters for pathogens of interest under different conditions in a specific substrate, where little or no information is available. However, the extrapolation of predicted data to real-world scenarios may not provide completely accurate predictions. Consequently the validation of predictive models is necessary in order to use the models with confidence, including the ability to understand where key inaccuracies in the predication may exist (McELROY et al., 2000). In recent years, fresh produce has been detected as the vehicle of transmission in several food-borne outbreaks (De ROEVER, 1999). The potential for the microbial contamination of fruits and vegetables is high due to the wide variety of contamination conditions to which the produce is exposed during growth, harvest and distribution (MADEN, 1992). Melon is a highly popular fruit in Brazil. It is a low acid fruit with an average pH above 4.5 and is often served sliced in food establishments either as fresh pieces or in mixtures for salad bars and deli counters, or as a pulp for drinking as a juice. Salmonella spp. and Listeria monocytogenes can survive and grow in this fruit (PENTEADO and LEITÃO, 2004a, b; GOLDEN et al. 1993). There is little experimental information available regarding the growth of these pathogens in fruits, making it difficult to use predictive microbiological models such as the U.S. Department of Agriculture’s Pathogen Modelling Program (PMP, R. Whiting U.S. Department of Agriculture-Agricultural Research Service, Philadelphia, Pa). So this study was carried out (1) to compare the laboratory-based growth of Salmonella Enteritidis and Listeria monocytogenes in melon pulp, (2) to model the observed growth data using the Gompertz and DMFit functions, and (3) to compare the observed, Gompertz-derived and DMFit-derived kinetic values to those predicted by the PMP version 7.0 (USDA, 2005). 2. MATERIAL AND METHODS 2.1 Fruit Ripe, damage-free melons (Cucumis melo L. cv.) valenciano amarelo were obtained from supermarkets in Campinas, State of São Paulo, Brazil. 2.2 Bacterial cultures A strain of Salmonella Enteritidis (SE) and another of Listeria monocytogenes Scott A (serotype 4b), from the culture collection of the “Laboratório de Higiene e Legislação, FEA-UNICAMP”, Brazil were used in this study. The S. Enteritidis culture was maintained in tryptone soy agar slants (TSA; Oxoid, Oxoid Ltd. Basingstoke, Hampshire, England) and the L. monocytogenes in tryptone soy agar slants containing 0.6% yeast extract (TSA-YE); (TSA, Oxoid Ltd Basingstoke, Hampshire, England; YE, Difco) at 5 °C. The identities of both strains were previously confirmed by biochemical and serological tests. Growth of Salmonella Enteriditidis and Listeria Monocytogenes in Melon Pulp: Predictive Modelling and Evaluation of Model Performance use in the experiment, the inoculae of S. Enteritidis were transferred to TSA and those of L. monocytogenes to TSA-YE after three consecutive 24 h intervals. Cells were collected from the respective media and transferred to 5 mL saline solution (0.85% NaCl ) to adjust the suspension to a concentration of 2 x 108.mL–1 according to the MacFarland turbidity scale using the Densimat equipment (bioMerieux). The bacterial suspensions were serially diluted (1:10) in 0.1% peptone water, and 1mL aliquots of each dilution pour plated in TSA agar and TSA-YE, respectively, for S. Enteritidis e L. monocytogenes, followed by incubation at 35 °C for 24 h to determine the viable cell concentration. A dilution concentration of 104 CFU.mL–1 of each bacterium was used to inoculate the fruit pulp. 2.4 Sample pulp preparation After being washed and scrubbed the external surfaces of the fruit were cotton scrubbed with an alcoholic iodine solution (2%) (FDA, BACTERIOLOGICAL ANALYTICAL MANUAL 1995), and allowed to air dry inside a laminar airflow cabinet (VLFS-12, VECO). Defined areas (100 cm2) of the fruit skin, and the inner fruit pulp without seeds, were aseptically removed with sterilised spoons and the fruit portions transferred to a sterilised shaker. After mixing, 50 g portions of the pulp were carefully removed with a spoon and transferred to sterilised conical flasks (250 mL). Before the inoculation tests, the pulp was checked for sterility and then frozen. 2.5 Pulp inoculation and enumeration of S. Enteritidis and L. monocytogenes For each bacteria, triplicate test portions (50 g) of homogenized pulp were inoculated with 1 mL (104 CFU) suspension and incubated for 0, 24, 48, 72, 96, 120 and 144 h at 10 °C; 0, 12, 18, 24, 36, 42 and 48 h at 20 °C; and 0, 2, 4, 6, 8, 10, 12 and 24 h, at 30 °C. For sampling, 1 mL portions of fruit pulp were collected, serially diluted (1:10) in peptone water (0.1%) and pour plate dispersed in TSA-YE for L. monocytogenes (45 °C) and TSA for S. Enteritidis. The plates were incubated at 35 °C for 24 h and then counted using a colony counter (Phoenix, CP 600), the results being expressed in CFU.g-1. Uninoculated pulp controls were also prepared to assure the absence of any background microflora before and after the incubation time. 2.6 Chemical and physical-chemical analyses The pH of the uninoculated fruit pulp was determined using a calibrated pH meter (model B374, Micronal). The pH was not monitored during the incubation period. Brix was determined using a Carl Zeiss (Jena) refractometer, model 32-G 110d. Titratable acidity was determined using the method described by Anon (1997). Sugars (total and reducing) were analysed as described by Lara et al. (1976). 2.7 Growth parameters 2.3 Inoculum preparation S. Enteritidis and L. monocytogenes were cultured, respectively, in TSA and TSA-YE slants at 35 °C. Immediately before their Braz. J. Food Technol., Campinas, v. 10, n. 3, p. 176-182, jul./set. 2007 Parameters for comparison with the laboratory-based growth of Salmonella Enteritidis and Listeria monocytogenes in melon pulp, were obtained as follows: 177 Growth of Salmonella Enteriditidis and Listeria Monocytogenes in Melon Pulp: Predictive Modelling and Evaluation of Model Performance CASTRO, M. F. P. P. M. et al. 2.7.1 Generation time (GT) 2.7.2 Exponential growth rate (EGR) The EGR value was the slope of the line obtained in the calculation of GT as described in item 2.7.1 Log [population (CFU.g–1)] The generation time (GT) was calculated from the slope of the line obtained in the semi-logarithmic plot of exponential growth, using the mean of three repetitions for each temperature and time. The following equation was used: GT = 0.301/slope as described by Madigan et al. (1997). 10 Braz. J. Food Technol., Campinas, v. 10, n. 3, p. 176-182, jul./set. 2007 5 4 3 2 100 150 200 Time (h) PMP predicted growth DMFit modeled growth (r2 = 1.00) Observed growth Gompertz modeled growth (r2 = 0.99) 50 Log [population (CFU.g–1)] 10 9 20 °C 8 7 6 5 4 3 2 1 0 0 10 20 30 Time (h) 40 50 60 PMP predicted growth DMFit modeled growth (r2 = 0.97) Observed growth Gompertz modeled growth (r2 = 0.93) Log [population (CFU.g–1)] 3. RESULTS AND DISCUSSION As shown in Figures 1 and 2, the results confirmed that low temperature was not a barrier to the growth of Listeria and S. Enteritidis; it can retard growth but not inhibit it. Although the growth rate was reduced at 10 °C, both microorganisms were able to survive and grow in melon pulp stored at 10, 20 and 30 °C. The present results confirmed the observations of Ukuku and Fett (2002) who verified that L. monocytogenes could survive but not grow during 15 days of storage in cantaloupe at 4 °C, while growth was evident 6 0 The growth parameters were compared using Tukey’s paired comparison procedure (BOX et al., 1978). The procedure applied for the aseptic removal of the fruit pulp was adequate. All the analyses of uninoculated samples performed initially and during the incubation period revealed the absence of Listeria, Salmonellae and of any other endogenous micro flora in the internal tissues of the fruits. Samish et al. (1963) mentioned that in healthy fruits the bacterial flora is assumed to be limited to the surface, while the inner tissue remains sterile. 7 0 2.9 Statistical analyses The mean of three repetitions for each experiment plus the standard deviation for the following melon characteristics were: acidity (%) 1.99 ± 0.27; total sugar (%) 7.76 ± 1.40; reducing sugar (%) 4.63 ± 0.51; pH 5.87 ± 0.13 and Brix (°Brix) 10.25 ± 2.33. The carbohydrate contents of the fruit pulp showed it was an adequate substrate for the growth of both of the microorganisms under study. In addition the average pH values could not be considered inhibitory for either L. monocytogenes or S. Enteritidis, and thus the fruit pulp composition was not a barrier to the growth of these microorganisms. 10 °C 8 1 2.8 Growth modelling and evaluation of performance Data for growth modelling were compiled using Excel spreadsheets (Microsoft, Redmond, WA). The Statistica Program was used to fit the observed growth data to the Gompertz function. Kinetic values, including the exponential growth rate (EGR), generation time (GT) and lag-phase duration (LPD) were calculated from the Gompertz parameters generated (WHITING, 1995). The curves were also fitted to the Microsoft Excel Add-In “DMFit.xls”, downloadable from http://www.ifr.ac.uk/safety/DMFit/, based on the model of Baranyi and Roberts (1994). An evaluation of the performance of the models was accomplished by comparing the coefficient of determination values (R2), the use of a “fail-safe/faildangerous” graphical method and calculations of the bias and accuracy factors (ROSS, 1996). 9 10 9 8 7 6 5 4 3 2 1 0 30 °C 0 5 10 15 Time (h) 20 25 30 DMFit modeled growth (r2 = 099) PMP predicted growth Observed growth Gompertz modeled growth (r2 = 0.93) FIGURE 1. Observed, Gompertz-modelled, DMFit and PMPpredicted growth curves for Salmonella Enteritidis in melon pulp held at 10, 20 and 30 °C. after 4 h of storage at 8 and 20 °C. These observations emphasise the importance of strict low temperature control and good manufacturing practices to avoid contamination and growth. For both microorganisms there was an increase of about six log cycles after 24, 48 and 168 h, respectively, at 30, 20 and 10 °C (Figures 1 and 2). Leverentz et al. (2001) observed an increase of 178 CASTRO, M. F. P. P. M. et al. was provided about the characteristics of the cantaloupe in this work (mainly pH). 10 Log [population (CFU.g–1)] 9 Table 1 shows the growth parameters observed for S. Enteritidis and L. monocytogenes in the melon pulp. 10 °C 8 7 6 5 4 3 2 1 0 0 50 100 Time (h) DMFit modeled growth (r2 = 1.00) Gompertz modeled growth (r2 = 0.98) 150 200 PMP predicted growth Observed growth Log [population (CFU.g–1)] 9 20 °C 8 7 6 5 4 3 2 1 0 0 30 40 50 60 Time (h) DMFit modeld growth (r2 = 1.00) PMP predicted growth Gompertz modeled growth (r2 = 1.00) Observed growth 10 9 8 7 6 5 4 3 2 1 0 10 There was no significant difference in the growth parameters observed at 10 and 20 and 30 °C amongst the microorganisms under study with the exception of LPD at 30 °C, when Salmonella Enteritidis showed a two-hour shorter lag phase. Golden et al. (1993) also reported the ability of Salmonella ssp. to grow inside honeydew melons. The fruits were inoculated with a pool of five species of Salmonella (S. Anatum, S. Chester, S. Havana, S. Poona and S. Senftenberg) and the generation time detected at 23 °C was 1.1 h. The result obtained in the present work was in relative agreement with this published data, since at 20 °C the GT was 1.54 h. Table 2 shows the growth parameters obtained from the observed values, Gompertz function, DMFit and PMP programs. In general there was less than half an hour’s difference between the observed GT values and the Gompertz modelled, DMFit and PMP predicted GT values for both bacteria at 20 and 30 °C; the observed GT values always being higher than those modelled by either Gompertz or by the PMP and DMFit programs, with the exception of the GT at 20 °C for DMFit. On the other hand at 10 °C the observed GT value (7.17 h) for S. Enteritidis was about 6 h shorter than the PMP-predicted GT value (13.3 h), whilst for Listeria monocytogenes it was 1.5 h longer than the PMP-predicted value. 10 Log [population (CFU.g–1)] Growth of Salmonella Enteriditidis and Listeria Monocytogenes in Melon Pulp: Predictive Modelling and Evaluation of Model Performance 20 30 °C In general there was little difference (equal or inferior to 2 h) between the observed LPD values and those obtained from the modelled function or predicted by PMP, with the exception of L. monocytogenes at 20 °C and both microorganisms at 10 °C in relation to the values obtained with DMFit and PMP. At 10 °C the observed LPD value was 31 h and 14.5 h shorter than the LPD predicted by PMP for Salmonella and L. monocytogenes, respectively. In general the data modelled by DMFit showed the safest values with the exception of Salmonella Enteritidis at 20 °C. In these cases, the LPD values obtained using PMP were much higher than the values observed, representing a great risk for those making use of this program in this situation. For Salmonella at 30 °C, none of the predicted or modelled values were safe. With respect to EGR, the observed values were, in general, equal or inferior to those obtained from Gompertz and predicted by PMP. So for this parameter, the Gompertz function and PMP programme made safe predictions. 0 5 10 15 Time (h) DMFit modeled growth (r2 = 1.00) Gompertz modeled growth (r2 = 0.98) 20 25 30 PMP predicted growth Observed growth FIGURE 2. Observed, Gompertz-modelled, DMFit and PMPpredicted growth curves for Listeria monocytogenes in melon pulp held at 10, 20 and 30 °C. about 2 log units in cut honeydew melon slices inoculated with S. Enteritidis at 10 °C, and of up to 5 log units at 20 °C after 168 h incubation. Ukuku and Fett (2002) observed that populations of L. monocytogenes in fresh cut cantaloupe pieces stored at 8 or 20 °C increased by 1 log unit by the end of storage. These results are in disagreement with those of the present study, but no information Braz. J. Food Technol., Campinas, v. 10, n. 3, p. 176-182, jul./set. 2007 The Tukey’s paired comparison procedure, which compared the predicted and observed values for GT, LPD and EGR showed that, in general, the values predicted by PMP were the ones showing the greatest statistical difference from the observed values at the 5% level, especially at 10 and 20 °C. The use of PMP in these situations could be critical, since the predictions were in the “fail dangerous” zone. In a few cases the DMFit and Gompertz values differed statistically from the observed ones, but in these cases it was not critical, since they predicted on the safe side. Figures 1 and 2 show the observed, Gompertz-modelled, DMFit and PMP predicted growth curves for S. Enteritidis and L. monocytogenes in melon pulp. Duh and Schaffner (1993) used the r2 values as an indication of the reliability of the models when applied to foods. In the present study the values for the determination coefficients (r2) of the Gompertz and DMFit functions were fitted to both 179 CASTRO, M. F. P. P. M. et al. of the data sets and also used to compare these equations. As can be seen, all the determination coefficient values were either equal or higher for the DMFit program than for the Gompertz function, showing that the first had a better fit to the observed data. The predicted and observed values were also compared using a graphical method. Figures 3 and 4 show the “fail safe” and “fail dangerous” plots for GT, LPD and EGR for both microorganisms. With the exception of the PMP predicted value at 10 °C and the DMFit predicted value at 20 °C for Salmonella, in general the GT values for both microorganisms provided a margin of safety, since all the GT points were below the line of equivalence, indicating that in general all the observed GT values were longer than the predicted ones. The DMFit programme provided safer predictions for the LPD values for Listeria monocytogenes than for Salmonella. Safe values for LPD were only obtained for Salmonella at 10 °C with DMFit and at 20 °C with PMP. Growth of Salmonella Enteriditidis and Listeria Monocytogenes in Melon Pulp: Predictive Modelling and Evaluation of Model Performance All the points for EGR were either on or above the line of equivalence (with the exception of Salmonella at 20 °C for DMFit), so the prediction can be considered safe in these cases. Ross (1993) introduced the bias and accuracy factors for the quantitative measurement of the plots shown in Figures 3 and 4. The bias factor checks for systematic over or under prediction by the model, and the accuracy (or precision) factor provides a measure of the average difference between the observed and predicted values (McMEEKIN and ROSS, 1996). Bias and accuracy factors of 1 indicate perfect agreement between the observed and predicted values (Mac ELROY et al., 2000), while, for example, factors of 1.1 and 0.9, respectively, indicate over prediction and under prediction at an average of 10% (McMEEKIN and ROSS, 1996). As the calculation is an average of all the factors obtained at each temperature, the result should be cautiously evaluated, because distortions can occur at individual temperatures. Table 3 shows the results for the bias and accuracy factors. TABLE 1. Growth parameters1 observed for S. Enteritidis and L. monocytogenes in melon pulp stored at 10, 20 and 30 °C. Microorganisms S. Enteritidis L. monocytogenes S. Enteritidis L. monocytogenes S. Enteritidis L. monocytogenes Temperature (°C) G.T (h)2 7.17a ± 0.20 7.16a ± 0.77 1.54a ± 0.14 1.74a ± 0.05 0.79a ± 0.30 0.86a ± 0.12 10 20 30 L.P.D. (h)2 24.00a ± 0 24.00a ± 0 8.00a ± 3 6.00a ± 0 2.00a ± 0 4.00b ± 0 E.G.R. [log(CFU.g–1.h–1]2 0.04a ± 0.00 0.04a ± 0.01 0.20a ± 0.02 0.17a ± 0.01 0.41a ± 0.16 0.38a ± 0.01 Results expressed as the mean of three repetitions plus the standard deviation for each parameter. 1 Results between microorganisms followed by different letters differ significantly at the 5% level. 2 TABLE 2. Observed, Gompertz function, DMFit and PMP growth parameters for S. Enteritidis and L. monocytogenes in melon pulp. Growth parameters1 10 °C 20 °C 30 °C GT (h) Observed Gompertz DMFit PMP LPD (h) Observed Gompertz DMFit PMP Observed EGR -1 Log [(CFU.g ) /h] Gompertz DMFit PMP S. L. Enteritidis2 monocytogenes2 7.17 ± 0.20a.d 7.16 ± 0.77a 6.02 ± 0.00b, d 5.69 ± 0.58a d 6.84 ± 0.00 6.84 ± 1.18a c 13.30 ± 0.00 5.70 ± 0.00a a 24.00 ± 0.00 24.00 ± 0.00a a 24.51 ± 1.10 24.13 ± 1.82a a 17.83 ± 5.19 16.62 ± 6.41a 55.00 ± 0.00b 38.50 ± 0.00b 0.04 ± 0.00a 0.04 ± 0.01a c 0.05 ± 0.00 0.05 ± 0.01a a, b, c 0.04 ± 0.00 0.05 ± 0.01a a, b 0.04 ± 0.00 0.05 ± 0.00a S. L. Enteritidis2 monocytogenes2 1.54 ± 0.14a 1.74 ± 0.05a, c 1.24 ± 0.18a 1.46 ± 0.11b, d a 1.70 ± 0.28 1.68 ± 0.9c b 1.20 ± 0.00 1.40 ± 0.00d a 8.00 ± 3.46 6.00 ± 0.00a a 9.61 ± 4.08 5.84 ± 0.32a a 12.74 ± 0.00 4.43 ± 0.72b 6.00 ± 0.00a 10.20 ± 0.00c 0.20 ± 0.02a, b 0.17 ± 0.01a a 0.25 ± 0.04 0.21 ± 0.02a, b b 0.18 ± 0.03 0.18 ± 0.01a, b a, b 0.20 ± 0.00 0.21 ± 0.00b 1 Results expressed as the mean of three repetitions for each parameter. 2 Results for the different models of the same microorganism followed by different letters (a, b, c or d) differ significantly at the 5% level. S. Enteritidis2 0.79 ± 0.30a 0.61 ± 0.20a 0.76 ± 0.36a 0.40 ± 0.00a 2.00 ± 0.00a 3.39 ± 0.68a 2.87 ± 0.00a 3.30 ± 0.00a 0.41 ± 0.16a 0.52 ± 0.17a 0.45 ± 0.21a 0.54 ± 0.00a L. monocytogenes2 0.86 ± 0.12a 0.73 ± 0.07a 0.83 ± 0.18a 0.60 ± 0.00a 4.00 ± 0.00a 3.23 ± 0.16a, c 2.29 ± 0.96b, c 4.00 ± 0.00a 0.38 ± 0.01a 0.41 ± 0.04a, c 0.37 ± 0.07a 0.51 ± 0.00b, c TABLE 3. Bias and accuracy factors for the GT, LPD and EGR of Salmonella Enteritidis and Listeria monocytogenes in melon pulp. Bacteria Factors Salmonella sp. Listeria monocytogenes Gompertz Bias 0.81 Accuracy 1.24 Bias 0.84 Accuracy 1.20 GT DMFit 1.01 1.06 0.97 1.03 PMP 0.90 1.70 0.77 1.30 Braz. J. Food Technol., Campinas, v. 10, n. 3, p. 176-182, jul./set. 2007 Gompertz 1.28 1.31 0.92 1.09 180 LPD DMFit 1.19 1.46 0.66 1.52 PMP 1.42 1.76 1.40 1.43 Gompertz 1.24 1.24 1.19 1.19 EGR DMFit 1.01 1.08 1.04 1.04 PMP 1.11 1.12 1.26 1.27 Growth of Salmonella Enteriditidis and Listeria Monocytogenes in Melon Pulp: Predictive Modelling and Evaluation of Model Performance CASTRO, M. F. P. P. M. et al. 1.2 1.2 1.0 1.0 "Fail dangerous" Log (GT (h)) Log (GT (h)) 0.6 0.4 0.2 0.0 –0.2 0.0 0.2 0.4 0.6 0.8 Log (observed GT (h)) 1.0 "Fail safe" 0.2 0.4 0.6 0.8 1.0 1.2 Log (observed LPD (h)) 1.4 1.6 1.8 "Fail safe" –0.2 0.0 0.2 0.4 0.6 0.8 Log (observed GT (h)) 2.0 1.8 "Fail dangerous" 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Log (observed LPD (h)) 1.0 1.2 "Fail safe" 1.6 1.8 2.0 –0.2 –0.2 "Fail safe" –0.4 Log (EGR[log(CFU.g–1)/h]) Log (EGR[log(CFU.g–1)/h]) 0.2 –0.4 –0.4 1.2 "Fail dangerous" 0.0 0.4 –0.2 Log (LPD (h)) Log (LPD (h)) –0.4 –0.4 0.6 0.0 "Fail safe" –0.2 2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 "Fail dangerous" 0.8 0.8 –0.6 –0.8 –1.0 –1.2 "Fail dangerous" –1.4 –1.4 –1.2 –1.0 –0.8 –0.6 –0.4 –0.2 –0.6 –0.8 –1.0 DMfit PMP "Fail dangerous" –1.2 –1.4 –1.4 Log (observed EGR[log(CFU.g–1)/h]) Gompertz "Fail safe" –0.4 –1.2 –1.0 –0.8 –0.6 –0.4 Log (observed EGR[log(CFU.g–1)/h]) Gompertz Line of equivalence DMfit PMP –0.2 Line of equivalence FIGURE 3. Graphical comparison of observed versus predicted kinetic values for the growth of Salmonella sp. in melon pulp. FIGURE 4. Graphical comparison of observed versus predicted kinetic values for the growth of Listeria sp. in melon pulp. For instance, for both microorganisms the bias factors for GT obtained with Gompertz, DMFit and PMP were below 1 (with the exception of DMFit with a bias factor of 1.01) indicating a safe prediction for this parameter. For the accuracy factors all values for GT were above 1 for both bacteria, indicating a percentage of variation between the predicted and observed data of 24% (Gompertz), 6% (DMFit) and 70% (PMP) in the case of Salmonella, and of 20% (Gompertz), 3% (DMFit) and 30% (PMP) for L. monocytogenes. (PMP), and for L. monocytogenes of 9% (Gompertz), 52% (DMFit) and 43% (PMP). The bias factor for LPD only showed safe values for Gompertz and DMFit for Listeria monocytogenes. For the accuracy factors, all the values for LPD were above 1 indicating a percentage of variation for Salmonella of 31% (Gompertz), 46% (DMFit) and 76% Dalgaard and Jorgensen (1998) calculated accuracy factors in seafood ranging from 1.4 to 4.0 for the growth rates of L. monocytogenes, while in the present study the EGR values ranged from 1.04 to1.27. Braz. J. Food Technol., Campinas, v. 10, n. 3, p. 176-182, jul./set. 2007 For the bias factor, all values for EGR were above 1 indicating safe values for Listeria monocytogenes and Salmonella. For the accuracy factor, all values for EGR were above 1 indicating a percentage of variation for Salmonella of 24% (Gompertz), 8% (DMFit) and 12% (PMP), and for L. monocytogenes of 19% (Gompertz), 4% (DMFit) and 27% (PMP). 181 CASTRO, M. F. P. P. M. et al. Growth of Salmonella Enteriditidis and Listeria Monocytogenes in Melon Pulp: Predictive Modelling and Evaluation of Model Performance The PMP has been validated for some food borne pathogens in various food systems (HUDSOM and MOTT, 1993; WALLS and SCOTT, 1997; WALLS et al., 1996) but there are no studies for S almonellae in any food substrate so far, and for Listeria monocytogenes the few studies available focussed on other foods rather than on fresh fruits. Validations of the PMP for L. monocytogenes in baby food, ham salad, and pâté demonstrated good agreement between the model and observed values (HUDSON and MOTT, 1993; WALLS and SCOTT, 1997). This fact was not verified in all the situations of the present study, as, for instance the lag phase duration at 10 °C (fail dangerous). Mc Elroy et al. (2000) in a validation study for B. cereus spores in boiled rice also observed that the model was “fail safe” for lag-phase duration at 20 and 30 °C but not at 15 °C. These authors verified that as the temperature decreased below 15 °C, the PMP may become fail dangerous for LPD when evaluating the growth of psychotropic B. cereus strains, and they suggested additional validation experiments to better evaluate this phenomenon. In the present study we also verified that PMP was fail dangerous for LPD for both microorganisms, as well as for GT for Salmonella. LARA, A. B. W. H. et al., Métodos químicos e físicos para análise de alimentos. In: REBOCHO, D. D. E. (Ed.). Normas analíticas do Instituto Adolfo Lutz, v. 1, Instituto Adolfo Lutz, São Paulo, p. 42-44, 1976. It was concluded that in general the Gompertz, DMFit and PMP models could be used with reasonable confidence, with the exception of the PMP at low temperatures, when it can become fail dangerous in some cases. PENTEADO, A. L.; LEITÃO, M. F. F. Growth of Salmonella Enteritidis in Melon, Watermelon and Papaya Pulp Stored at Different Times and Temperatures. Food Control, London, v. 15, n. 5, p. 369-373, 2004a. REFERENCES ______. Growth of Listeria monocytogenes in melon, watermelon and papaya pulps. International Journal of Food Microbiology, London, v. 92, n. 1, p. 89-94, 2004b. ANON. Official methods of analysis of the association of official analytical chemist. Titratable acidity method 37.1. 37 (16th ed.) Gaithersburg, Maryland, USA: Association of Official analytical Chemists. 1997. BARANYI, J.; ROBERT, T. A. A dynamic approach to predicting bacterial growth in food. International Journal of Food Microbiology, London, v. 23, n. 3, 4, p. 277-294, 1994. BOX, G. E. P.; HUNTER, W. G.; HUNTER, J. S. Statistics for experimenters. An introduction to design data analysis and model building. New York, John Wiley., 1978. DALGAARD, P.; JØRGENSEN, L. V. Predicted and observed growth of Listeria monocytogenes in seafood challenge tests and in naturally contaminated cold-smoked salmon. International Journal of Food Microbiology, London, v. 40, n. 1, 2, p. 105-115, 1998. De ROEVER, C. Microbiological safety evaluations and recommendations on fresh produce. National Advisory Committee on Microbiological Criteria for Foods. Food Control, London, v. 10, n. 2, p. 117-143, 1999. DUH Y. H.; SCHAFFNER, D. W. Modelling the effect of temperature on the growth rate and lag time of Listeria innocua and Listeria monocytogenes. Journal of Food Protection, Des Moines, v. 56, n. 3, p. 205-210, 1993. FDA. Bacteriological Analytical Manual – BAM, . 8 th ed. Gaithersburg. USA, 1995. GOLDEN, D. A.; RHODEHAMEL, E. J.; KAUTTER D. A. Growth of Salmonella spp. in cantaloupe, watermelon, and honeydew melons. Journal of Food Protection, Des Moines, v. 56, n. 33, p. 194-196, 1993. HUDSON, J. A.; MOTT, S. J. Growth of Listeria monocytogenes, Aeromonas hydrophila and Yersinia enterocolitica in pâté and a comparison with predictive models. International Journal of Food Microbiology, London, v. 20, n. 1, p. 1-11, 1993. Braz. J. Food Technol., Campinas, v. 10, n. 3, p. 176-182, jul./set. 2007 LEVERENTZ, B. et al. Examination of bacteriophage as a biocontrol method for Salmonella on fresh-cut fruit: A Model study. Journal of Food Protection, Des Moines, v. 64, n. 8, p. 1116-1121, 2001. MADEN, J. M. Microbial pathogens in fresh produce-the regulatory perspective. Journal of Food Protection, Des Moines, v. 55, n. 10, p. 821-823, 1992. MADIGAN, M. T.; MARTINKO, J. M.; PARKER, J. Biology of Microorganisms. Brock (Ed.), 8th ed. Prentice-Hall, New Jersey, p. 153, 1997. McELROY, D. M.; JAYKUS, L. A.; FOEGEDING, P. M. Validation and Analysis of Modelled Prediction of Growth of Bacillus cereus Spores in Boiled Rice. Journal of Food Protection, Des Moines, v. 63, n. 2, p. 268-272, 2000. McMEEKIN, T. A.; ROSS, T. Modelling Applications. Journal of Food Protection; Des Moines, Supp. p. 37-42; 1996 QUINTAVALLA, S.; PAROLARI, G. Effects of temperature, aw, and pH on the growth of Bacillus cells and spores: a response surface methodology study. International Journal of Food Microbiology, London, v. 19, n. 3, p. 207-216, 1993. ROSS, T. Indices for performance evaluation of predictive models in food microbiology. Journal of Applied Bacteriology, London, v. 81, n. 5, p. 501-508, 1996. SAMISH, Z.; TULCZYNSKA, R. E.; BICK, M, The micro flora within the tissue of fruits and vegetables. Journal of Food Science, Chicago, v. 28, n. 3, p. 259-266, 1963. UKUKU, D. O.; FETT, W. Behaviour of Listeria monocytogenes inoculated on cantaloupe surfaces and efficacy of washing treatments to reduce transfer from rind to fresh-cut pieces. Journal of Food Protection, Des Moines, v. 65, n. 6, p. 924-930, 2002. USDA-UNITED STATES DEPARTMENT OF AGRICULTURE, 2005. Pathogen Modelling Program version 7.0 - Current models. Available in: http://ars.usda.gov/Services/docs.htm?docid=6795 . USA. Acessed in: oct/2006. WALLS, I.; SCOTT, V. N. Validation of predictive mathematical models describing the growth of Listeria monocytogenes. Journal of Food Protection, Des Moines, v. 60, n. 9, p. 1142-1145, 1997. WALLS, I.; SCOTT, V. N.; BERNARD, D. T. Validation of predictive mathematical models describing growth of Staphylococcus aureus. Journal of Food Protection, Des Moines, v. 59, n. 1, p. 11-15, 1996. WHITING, R. C.; BUCHNAN, R. L. Microbial modelling. Food Technology, Chicago, v. 48, n. 6, p.112-120, 1994. WHITING, R. C. Microbial Modelling in Foods. Critical Reviews in Food Science and Nutrition, New York, v. 35, n. 6, p. 467-494, 1995. 182