Purpose: We investigated the validity of the above assumption in a sytematic way, using Bacillus cereus as a model organism.
Methods: Three strains of Bacillus cereus were grown in RIF (Reconstituted Infant Formulae) and the maximum specific growth rates were estimated using viable count measurements at temperatures ranging from 12 to 25oC. Both the (i) square-root and the (ii) natural logarithm transformations, as link functions, were applied to the rates, in order to model the effect of temperature on them by (i) Ratkowsky-type square root and (ii) quadratic polynomial models, respectively. The results were compared with each other and with predictions from ComBase Predictor, using an appropriate bias factor.
Results: The bias factor was fairly constant, 1.3 in our case, providing a convenient simplification that can be used, with confidence, between 12 and 25˚C, as a generic rule for the three strains. We also showed that the answer to the question, which link function should be preferred, depends on various other considerations; notably, how generic the model should be in terms of applicability to more strains and bigger range of environmental factors.
Significance: These findings strengthen confidence in using culture medium-based predictive models for food scenarios, adjusted by constant (but food-dependent) bias factors, which can bring significant saving to the food industry.