Dynamics of bacterial communities and interaction networks in thawed fish fillets during chilled storage in air

Thawed hake (Merluccius capensis and M. paradoxus) and plaice (Pleuronectes platessa) fillets were used as a model to evaluate the effect of storage temperature (0 or 10 °C) and biological variability (fish species, lot to lot) on bacterial growth kinetics and microbial successions. Both culture dependent methods (plate counts on non-selective and selective media) and culture independent methods (qPCR and 16S rRNA gene metabarcoding) were used. Bacterial counts exceeded 107 cfu/g within 2-3 days at 10 °C and 7-8 days at 0 °C. Plate counts on three media (Plate Count Agar +0.5% NaCl, Iron Agar Lyngby and Pseudomonas Selective medium) and 16S rRNA gene counts estimated by qPCR were highly correlated. Growth was modelled using the D-model and specific growth rate ranged between 0.97 and 1.24 d-1 and 3.54 and 5.90 d-1 at 0 and 10 °C, respectively. The initial composition of the microbiota showed lot-to-lot variation, but significant differences between the two fish species were detected. Alpha diversity significantly decreased during storage. When bacterial counts exceeded 107 cfu/g, the microbiota was dominated by members of the genera Pseudomonas, Psychrobacter, Acinetobacter, Serratia, Flavobacterium, Acinetobacter, Carnobacterium, Brochothrix and Vagococcus. However, Photobacterium and Shewanella, two genera frequently associated with fish spoilage, were either absent or minor components of the microbiota. As expected, storage temperature significantly affected the abundance of several species. The inference of microbial association networks with three different approaches (an ensemble approach using the CoNet app, Sparse Correlations for Compositional data, and SParse InversE Covariance Estimation for Ecological Association Inference) allowed the detection of both a core microbiota, which was present throughout storage, and a number of taxa, which became dominant at the end of spoilage and were characterized by a disproportionate amount of negative interactions.