• Challenge test
• Lactic acid bacteria
• Lightly preserved seafood
• Listeria monocytogenes
• Organic acids
• Predictive model

Changes in consumers’ habits have led to an increase in the production of ready-to-eat (RTE) food, including lightly preserved seafood. In lightly preserved seafood, lactic acid bacteria (LAB) has often been detected as the dominant spoilage microbiota and Listeria monocytogenes has been identified as the most relevant pathogen. Recently, a mathematical model that predicts the simultaneous growth of L. monocytogenes and psychrotolerant LAB in processed seafood and mayonnaise-based shrimp salads was developed. The LmLAB model takes into account the effect of microbial interaction and predicts the individual, as well as the interactive effects, of 12 preserving parameters on microbial growth, including acetic, benzoic, citric, lactic and sorbic acids (Mejlholm and Dalgaard, 2015).
The aim of this thesis was to study the influence of the composition of the food matrix on the antimicrobial effect of organic acids. Furthermore the performance of the LmLAB model was evaluated in order to examine if additional parameters should be included in the model in order to obtain better predictions and to increase its range of applicability.
Two challenge test were performed on (i) minced salmon and cod filets, and (ii) mayonnaise-based shrimp salad stored under modified atmosphere packaging (MAP), including products with different concentrations of lipid and organic acids. Benzoic and sorbic acids, with relatively high solubility in the lipid phase, were examined together with acetic, citric and/or lactic acid, having a low solubility in the lipid phase. L. monocytogenes and Lb. sakei were inoculated in all products. The growth kinetics of the microbiota and the chemical characteristics of the products were determined in order to be able to compare observed and predicted growth of L. monocytogenes and LAB. Growth of L. monocytogenes was observed only in minced salmon and cod filets with added lactic acid. Hence, the presence of benzoic and sorbic acids, in combination with other organic acids and chilled storage, seems to control the growth of L. monocytogenes in mayonnaise-based shrimp salad.
As expected, LAB became predominant within the microbiota of all examined products.
The LmLAB model overestimated the growth of L. monocytogenes (i.e. fail-safe predictions) in salmon and cod products, whereas no-growth of L. monocytogenes was correctly predicted in mayonnaise-based shrimp salad. The growth of LAB was underestimated in salmon products (Bias factor= 0.82) and mayonnaise-based shrimp salads (Bias factor= 0.67), and overestimated in cod products (Bias factor= 1.20).
The performance of the model was extensively improved for Lactobacillus spp. with respect to prediction of µmax values when partitioning of organic acids between the lipid and the water phase of salmon products and mayonnaise-based shrimp salad (including benzoic and sorbic acids) was considered. Bias factors of 0.91 and 1.01 were obtained, respectively. No improvements were observed for treatments based on cod (Bias factor= 1.24).
In conclusion the partitioning of organic acids between the lipid and the water phase should be included in the model in order to improve the model performance.