On the selection of relevant environmental factors to predict microbial dynamics in solidified media

Several studies have shown that food structure causes slower growth rates and narrower growth boundaries of bacteria compared to laboratory media. In predictive microbiology, both a_w or corresponding solute concentration (mainly NaCl) have been used as a growth influencing factor for kinetic models or growth/no growth interface models. The majority of these models have been based on data generated in liquid broth media with NaCl as the predominant a_w influencing solute. However, in complex food systems, other a_w influencing components might be present, next to NaCl. In this study, the growth rate of Salmonella typhimurium was studied in the growth region and the growth/no growth response was tested in Tryptic Soy Broth at 20 °C at varying gelatin concentration (0, 10, 50 g L⁻¹ gelatin), pH (3.25–5.5) and water activity (a_w) (0.929–0.996). From the viewpoint of water activity, the results suggest that NaCl is the main a_w affecting compound. However, gelatin seemed to have an effect on medium a_w too. Moreover, there is also an interaction effect between NaCl and gelatin. From the microbial viewpoint, the results confirmed that the a_w decreasing effect of gelatin is less harmful to cells than the effect of Na⁺ ions. The unexpected shift of the growth/no growth interface to more severe conditions when going from a liquid medium to a medium with 10 g L⁻¹ gelatin is more pronounced when formulating the models in terms of a_w than in terms of NaCl concentrations. At 50 g L⁻¹ gelatin, the model factored with NaCl concentration shifts to milder conditions (concordant to literature results) while the model with a_w indicates a further shift to more severe conditions, which is due to the water activity lowering effect of gelatin and the interaction between gelatin and NaCl. The results suggest that solute concentration should be used instead of a_w, both for kinetic models in the growth region and for growth/no growth interface models, if the transferability of models to solid foods is to be increased.