| Roast shrimp with high moisture is a kind of shrimp products which has highnutrition value. Some heat-resistance bacteria such as Bacillus spp. may survive infoods and contaminate the foods due to mild processes, affecting the shelf-life of thefoods. At the same time microbial safety problems of foods is always the mostconcerned issue to manufacturers. This problem was solved by the development ofpredictive microbiology. Up to now many microbial growth kinetic models areestablished. Though these models make possible the calculation of the food shelf-life, itis more important to describe the growth/no growth information than growth rate forpathogens with a low infective dose.In this paper, some studies were made on the roast shrimp with high moisture. Alsosome dominant spoilage bacteria were isolated from the products during the storagetime. It is showed that all the three strains were Bacillus cereus. At the same time thegrowth/no growth model of Bacillus cereus was established. The results indicated thatthe model has a high goodness fit for the growth/no growth data and the predictedvalues can well quantify the combinations of environmental factors on Bacilluscereus.The results were as follows:(1) Some dominant spoilage bacteria were isolated from the high-moisture roastshrimp which had gone bad. Three typical strains were selected and were named CCL1,CCL2and CCL3.Then morphology observation, hemolytic test, physiological andbiochemical tests, microbial fluorescence identification,16S rDNA gene sequenceanalysis were done to identify them. In addition, growth characteristics of strain CCL1were also studied. The experimental results showed that three strains were identified asBacillus cereus. The lag phase of strain CCL1was0～2h, log phase2～10h on thecondition of37℃Meat Infusion Broth. Bacillus cereus CCL1could grow well in5g/L～45g/L NaCl (w/v) included in Meat Infusion Broth while was inhibited or even died in more than105g/L NaCl. Also B.cereus CCL1grew slowly in Meat InfusionBroth (5g/L NaCl) with pH3～5.5and rapidly with pH4.5～9while it was inhibitedwithpH≥9.(2) A cocktail of the five strains of B.cereus CCL1,CCL2and CCL3were studiedand a logistic regression model was chosen to study the interaction of differenttemperature (10℃,15℃，20℃，25℃，30℃，35℃), pH(4.5,5,5.5,6,6.5,7,7.5), Aw(14levels from0.992to0.932) on the growth probability of B.cereus. This papermade afractional factorial design and theexperimental data was divided to two parts:80%ofdata was chosen as model data and20%of data was chosen as validation data. At lastcomparison was made between the predicted values and observed value by choosing30experimental data from the published studies to test the application of the model. Theresults showed the concordance index of model data was0.991while the validation datawas0.976, indicating that model had a high predictive accuracy for homogeneous data.Also The performance statistics obtained indicate a reasonable goodness of ft of themodel obtained, mainly due to the high values of R~2-Nagelkerke (0.954) and χ~2=0.0119,P=1of the Hosmer–Lemeshow statistic. With the decreasing temperature, the domain ofthe growth/no growth became narrow. As an example, at T=10℃, it can strongly affectthe growth probability of B.cereus while small changes on pH values and water activityhappen. The growth of B.cereus was only observed at optimum temperature andpH7.0-7.5when at lower water activity. While at pH=4.5、T≤20℃, B.cereus couldn’tgrow at any temperature. A high predictive accuracy was obtained (83.3%) with testdata, showing a wide range of application.(3) A cocktail of the strain CCL1, CCL2and CCL3were studied and a logisticregression model was chosen to model the growth/no growth data of B.cereus on theroast shrimp products under different water activity (0.960,0.956,0.952,0.948,0.942,0.938,0.935,0.925) and pH values (5.33,5.67,5.9,6.21,6.62,6.8). At the same timesamples were vacuum packaged and then stored at different temperatur（e10℃，15℃，20℃，25℃，30℃，35℃）for60days. Finally a cocktail of B.cereusDSMZ4312,DSMZ4313, ATCC49064and ATCC33019was chosen as test strains to validate andestimate the fitness and predictive ability of the model. The results showed theconcordance rate of the model was97.86%; false positive was0.855%and false negative was1.28%, indicating the model had a high predictive accuracy.Also Theperformance statistics obtained indicate a reasonable goodness of ft and a goodpredictive ability of the model obtained, mainly due to the high values of R~2-Nagelkerke(0.947) and χ~2=0.012, P=1of the Hosmer–Lemeshow statistic. According to thegrowth/no growth contour graphs, the interaction of each factor had strong effects onthe growth probabilities of B.cereus.The concordance rate for test data by using thismodel was93.16%and c value was0.926, indicating the model had a good performancefor other strains. This growth/no growth interface model can predict the growthprobability of the B.cereus along the shelf-life of the products and has an importantvalue for well optimizing the manufacturing process. |
PDF Full Text Request