Abstract
Fresh-cut produce satisfies the ever-growing consumers’ need for healthy and convenient foods. However, the naturally occurring microbial populations along with further microbial contamination during the processing tend to limit the shelf-life of these products. Given that there is limited knowledge of the microbial communities associated with spoilage of these products, the finer characterization of the fresh-cut produce microbiota, and monitoring of the impact of important environmental factors on its composition during storage is needed. On the other hand, the establishment of rapid and effective methods for microbiological spoilage evaluation also is of great importance to the fresh-cut food industry. The two main objectives of the present thesis were: i) to characterize the microbial communities associated with spoilage of these commodities during storage at different temperatures, using a metagenetic approach, (Section A: Chapters 2 and 3), and ii) to evaluate the microbiologica ...
Fresh-cut produce satisfies the ever-growing consumers’ need for healthy and convenient foods. However, the naturally occurring microbial populations along with further microbial contamination during the processing tend to limit the shelf-life of these products. Given that there is limited knowledge of the microbial communities associated with spoilage of these products, the finer characterization of the fresh-cut produce microbiota, and monitoring of the impact of important environmental factors on its composition during storage is needed. On the other hand, the establishment of rapid and effective methods for microbiological spoilage evaluation also is of great importance to the fresh-cut food industry. The two main objectives of the present thesis were: i) to characterize the microbial communities associated with spoilage of these commodities during storage at different temperatures, using a metagenetic approach, (Section A: Chapters 2 and 3), and ii) to evaluate the microbiological spoilage of different fresh-cut produce commodities using rapid spectroscopy-based technologies (Section B: Chapters 4, 5 and 6).Concerning Chapter 2, the fungal and bacterial communities associated with the spoilage of ready-to-eat (RTE) pineapple during storage at different temperatures were characterized using the ITS2 and gyrB metagenetic sequencing, respectively. Significant variability in fungal species composition was observed among the different batches of RTE pineapple. The initial microbiota composition was the main influencing factor determining the progress of spoilage. Depending on the initial prevalent fungal species, the temperature’s and storage time’s impact varied. With reference to the subdominant bacterial communities, neither temperature nor batch significantly influenced the bacterial diversity and composition. In Chapter 3, a metagenetic amplicon approach, based on gyrB sequencing, was applied for deciphering the bacterial communities associated with the spoilage of RTE rocket and baby spinach stored at different temperature conditions. The vegetable communities differed in the dominance of specific bacterial species. Specifically, Pseudomonas viridiflava was dominant in most samples of rocket, while a new Pseudomonas species, as well as P. fluorescens and/or P. fragi were highly abundant or even dominant in baby spinach. Significant variability in bacterial species composition among the different batches of each vegetable also was observed. At batch-level, the impact of temperature and/or storage time on bacterial microbiota was not apparent for baby spinach. Concerning rocket, the storage time was the most influencing factor for some batches resulting in reduction and increase of Pseudomonas species and lactic acid bacteria abundance, respectively. To conclude for both Chapters 2 and 3, a large-scale sampling of fresh-cut RTE produce should be conducted in order to assess the full biodiversity of its spoilage microbiota and unravel the impact of important environmental factors on microbial diversity and composition.In Chapter 4, the suitability of four analytical technologies (sensors) coupled with different machine learning algorithms for the evaluation of RTE pineapple’s quality was assessed, while the utilization potential of selected analytics tools, namely the Unscrambler software and the SorfML platform, also was explored. Pineapple samples stored at different temperature conditions were subjected to microbiological (total viable count, TVC) and sensory analyses, with parallel acquisition of Fourier transform infrared (FTIR), near infrared (NIR), fluorescence (FLUO), visible (VIS), and multispectral imaging (MSI) spectroscopy data. Similar trends concerning the applicability of the different sensors and algorithms were observed for both analytics tools. For TVC, almost all the combinations of sensors and the partial least squares regression (PLSR) algorithm showed relatively satisfactory performances. Moreover, most of the studied sensors in conjunction with linear support vector machine (SVM Linear) resulted in similar performances. Overall, the tested sensors apart from the NIR constitute promising tools for the assessment of pineapple’s microbiological spoilage. Concerning sensory features, FLUO spectral data and MSI sensor seem to be also promising for the evaluation of pineapple odour. Chapter 5 provided a comparative assessment of sensors and machine learning approaches for evaluating the microbiological spoilage of RTE rocket and baby spinach stored at different temperature conditions under modified atmosphere packaging (MAP). The samples were subjected to microbiological analysis (TVC and Pseudomonas spp.), and parallel acquisition of FTIR, NIR, VIS and MSI data. Two data partitioning approaches, namely random and dynamic data partitioning, and two machine learning algorithms, namely PLSR and radial support vector regression (SVR), were comparatively evaluated. Concerning baby spinach stored under passive MAP, the two algorithms yielded similar performances for most of the developed models. The random data partitioning resulted in better (for MSI and NIR) or similar (for FTIR and VIS) performances to the ones attained with the dynamic approach. Regarding rocket, the SVR algorithm resulted in better prediction for most of sensors, while random data partitioning yielded also considerably or slightly better results compared to the dynamic test set. The microbiological spoilage of baby spinach stored under passive MAP was better assessed by models derived mainly from the VIS sensor, while FTIR and MSI models were more suitable in rocket, with the selected best models exhibiting satisfactory performances. Contrarily, FTIR and MSI were more appropriate for TVC prediction of baby spinach stored under active MAP, exhibiting an even higher predictive power. The results indicated that personalised (i.e. product-specific) sensor applications and data analysis workflows are needed, while the applied storage conditions should be also taken into account. In Chapter 6, the assessment of microbiological spoilage of oyster mushrooms (Pleurotus ostreatus) using FTIR and MSI technologies was studied. Fresh-cut mushrooms were stored at different temperature conditions and were subjected to microbiological analysis (TVC) as well as to FTIR and MSI measurements. The FTIR and MSI data were collected for both the cap and the gills sides of mushrooms. Developed PLSR models for both mushroom sides exhibited poor prediction performances, regardless of the applied data partitioning, the reduction of the observed high spectral variability and the selection of the most informative independent variables. The results indicated that, under the conditions of this study and the applied computational analysis, the application of FTIR and MSI do not appear to be promising for the evaluation of the microbiological spoilage of oyster mushrooms.
show more
