Abstract :
[en] Food waste is currently a major problem since it is estimated that about one third of the food produced in the world is discarded before it is consumed. The reasons for these food losses and waste are varied and one cause is the bacterial spoilage, rendering foods unacceptable for consumption. The study of the dynamics of bacterial spoilage populations and the prediction of their dynamics would therefore be interesting to better understand and anticipate this phenomenon. This research focused on the study of predictive models for spoilage bacteria of fresh meat and meat products, considered as highly perishable foodstuffs. The two working matrices were pork minced meat and white pudding, considering variations in storage conditions (temperature and packaging).
The first chapter of this thesis provides a general overview of bacterial spoilage of meat and meat products, as well as factors that may promote or limit its development. The different techniques used in this study to characterize and modelize the dynamics of spoilage microbiota are also described.
This research was then divided into four main areas that are discussed in the other chapters: (1) describing the spoilage bacterial microbiota naturally present in the matrices studied; (2) characterizing the spoilage bacteria of interest for these products; (3) developing and validating predictive models with one or more bacteria; (4) and studying the metabolome of minced meat inoculated by spoilage microorganisms of interest.
These studies have demonstrated the interest of combining results from classical microbiology and 16S rDNA-based metagenetic to monitor and predict the dynamics of spoilage microbiota. For the white pudding, the bacteria of interest were Brochothrix thermosphacta, Carnobacterium maltaromaticum, Lactobacillus spp. (Lb. fuchuensis, Lb. graminis, Lb. oligofermentans), Lactococcus lactis, Leuconostoc mesenteroides, Pseudomonas psychrophila, Pseudomonas sp., Psychrobacter spp. (Psy. okhotskensis, Psy. urativorans), Raoultella terrigena and Serratia sp. For minced pork samples they were B. thermosphacta, Lb. algidus, Lc. piscium, Leuconostoc spp. (Ln. inhae, Ln. gelidum), Photobacterium spp. (Ph. kishitanii, Ph. phosphoreum) and Pseudomonas spp. (Ps. fragi, Ps. fluorescens, Ps. psychrophila). The type of packaging and storage temperature have a significant effect on the different dynamics, as well as the food companies and the production batches analyzed.
Some of these bacteria of interest were then inoculated on sterile and non-sterile matrices, stored at different temperatures and packaging. The growth parameters to each bacterium were collected: maximum growth rate, lag time, minimum and maximum bacterial populations, time to reach the stationary phase, time to reach the spoilage threshold, minimum growth temperature, etc. Packaging seems to have the greatest impact on the maximum growth rate, itself having the greatest influence on the microbiological shelf life of the foods studied.
Based on these data, good adjustments were obtained for the growth simulations, but overestimations were often observed. The same observations could be made by comparing the simulations performed on the white pudding with those available from software (ComBase and Sym'Previus). For minced pork, the data obtained allowed the development of three species interaction models based on the Lotka-Volterra (prey-predator model) and the modified Jameson models. The simulations obtained were validated by monitoring the spoilage microbiota of naturally contaminated pork minced meat matrices. The modified Jameson model obtained the best adjustments, although the prey-predator approach seems to be an interesting interaction model for complex microbiota. However, these proposals for models with three or more spoilage bacteria need to be validated by more experimental repetitions.
Finally, metabolomic analyses (1H-NMR), in collaboration with CIRM-CHU, were performed in order to monitor the metabolites produced by inoculated bacteria in sterile minced pork samples. The dynamics of the metabolome for sterile non-inoculated matrices was also monitored. The different metabolomic patterns and metabolites produced were highlighted according to the inoculated bacteria and the food packaging. Moreover, the storage temperature seems to have the lowest impact on the metabolome.
Development of predictive models based on data obtained by multi-omics analyses, combined with classical microbiology, provide an interesting approach. Further research on the development of complex models integrating the dynamics of two or more spoilage bacteria, interacting with each other and with the natural microbiota of foodstuffs, will be also an important step for better understanding and anticipating the bacterial spoilage of perishable foodstuffs.
