We demonstrate that the commonly assumed matrix-independence of the cardinal (minimum, optimum and maximum) temperatures for bacterial growth is not necessarily valid for Bacillus licheniformis growing in plant-based milk. If confirmed, a consequence is that the ratio (called correction factor) between the maximum specific growth rate in a specific food matrix and in culture medium is not temperature-independent for every food matrix, opposed to general expectations. We found that, while the cardinal temperatures of B. licheniformis growing in either white almond or coconut beverages did not significantly differ from those in culture medium, this was not the case for another almond-based beverage, where we observed a smaller growth range and lower optimum temperature. A possible reason for this is that the food composition affects the cardinal temperatures. Our investigation is an example of "tertiary modelling" inasmuch it studies the effect of food matrix (a category variable) on the parameters of secondary models.