Validation of predictive growth modeling in food during shelf-life: a case study on Bacillus amyloliquefaciens ropy bread spoilage

Dehydrated raw materials used to produce bread are known to be highly contaminated with spore-forming bacteria. Once incremented in the bread, spores will germinate and exponential growth of amylase producing species such as Bacillus amyloliquefaciens may cause ropy bread spoilage in Mediterranean countries and huge economical costs. The aim of this study was to perform challenge test studies to compare growth experimental counts in bread and in silico growth predictions for various bread recipes. B. amyloliquefaciens ISPA-S109.3 was selected after previous characterization underlining its resistance to the baking process and high spoilage potential. Challenge tests were performed according to standardized methods with ISPA-S109.3 spore inoculation (4log spore/g) in the ingredient mixture. Home bread-making machine was used to bake the 4 bread recipes, i.e. wheat bran bread (WBB) and white wheat bread +/- bioingredient yielding various intrinsic parameters (aw: 0.93-0.96 and pH: 4.67-5.87). Bread incubation was performed at 20, 25 and 30°C +/- 0.01. Artificially contaminated pan bread and controls were prepared in three replicates. Distinction between ISPA-S109.3 and naturally contaminated Bacillus was performed using Rep PCR fingerprinting. The enumeration of ISPA-S109.3 was performed to determine growth kinetics and mathematical modeling using Sym'Previus. Based on fitted experimental growth kinetics, growth rate in WBB at 30°C was determined (µmax= 0.415+/-0.013h-1) and optimal growth rate calculated (µopt=1.61+/-0.048h-1) to further predict growth for the 5 tested conditions. Comparisons between experimental kinetics and growth predictions underlined that challenge test data were comprised into the 90% confidence interval of the predictions. These results highlight that major impact on growth was due to pH, aw and temperature while the impact of food matrix quantified by the µopt was sufficient, even in the case of sourdough. This study received funding from FP7-222-654-2 DREAM project and data related to strain and behavior diversity were incremented in Sym'Previus.