Volatilome of Chill-Stored European Seabass (Dicentrarchus labrax) Fillets and Atlantic Salmon (Salmo salar) Slices under Modified Atmosphere Packaging.

Fish spoilage occurs due to production of metabolites during storage, from bacterial action and chemical reactions, which leads to sensory rejection. Investigating the volatilome profile can reveal the potential spoilage markers. The evolution of volatile organic molecules during storage of European seabass (Dicentrarchus labrax) fillets and Atlantic salmon (Salmo salar) slices under modified atmosphere packaging at 2 °C was recorded by solid-phase microextraction combined with gas chromatography-mass spectrometry. Total volatile basic nitrogen (TVB-N), microbiological, and sensory changes were also monitored. The shelf life of seabass fillets and salmon slices was 10.5 days. Pseudomonas and H2S-producing bacteria were the dominant microorganisms in both fish. TVB-N increased from the middle of storage, but never reached concentrations higher than the regulatory limit of 30–35 mg N/100 g. The volatilome consisted of a number of aldehydes, ketones, alcohols and esters, common to both fish species. However, different evolution patterns were observed, indicating the effect of fish substrate on microbial growth and eventually the generation of volatiles. The compounds 3-hydroxy-2-butanone, 2,3-butanediol, 2,3-butanedione and acetic acid could be proposed as potential spoilage markers. The identification and quantification of the volatilities of specific fish species via the development of a database with the fingerprint of fish species stored under certain storage conditions can help towards rapid spoilage assessment.


Introduction
Fish spoilage is a well-documented process resulting from chemical reactions, autolytic degradation by fish enzymes and microbial metabolic activity. However, in the chilled seafood supply chain, fresh fish quality is diminished mainly by microbial mediated changes [1,2]. Specific spoilage organisms (SSOs) is a small part of the initial total microbiota which dominate against others under particular

Sensory Acceptance Evaluation and Shelf-Life Determination
The overall sensory acceptance of cooked fish samples was evaluated, and the results are presented in Figure 1. Initially, the fish acceptance was excellent and remained at those levels for the first 4 days of storage at 2 • C under MAP (p > 0.05). Afterwards, the overall acceptance score diminished linearly as expressed by the regression equations. A score of 3 for overall impression was judged as the lower limit of acceptability. The time after that coincided with slight off flavor and off taste development. The shelf life of both fish products studied was estimated to be no longer than 11 days, where at least one of the panelists scored the product below 3. The shelf life of fisheries and aquaculture products depends on various factors such as the applied storage (temperature, atmosphere and packaging, e.g., gas concentration, film permeability, and headspace), transportation (storage requirements) and distribution (storage facilities, temperature) conditions and the composition and population level of the initial total microbiota (including indigenous and exogenous microbiota).
Molecules 2020, 25, x 3 of 16 aquaculture products depends on various factors such as the applied storage (temperature, atmosphere and packaging, e.g., gas concentration, film permeability, and headspace), transportation (storage requirements) and distribution (storage facilities, temperature) conditions and the composition and population level of the initial total microbiota (including indigenous and exogenous microbiota). Thus, shelf life varies between fish species or even among the products (whole, gutted, fillets) [16,17]. In a recent work by our team [18], it was found that whole gutted seabass had a shelf life of 13 days, under the same conditions of temperature and atmosphere as herein. According to Kostaki et al. [19], the shelf life of sea bass fillets was 12 days when stored at 4 °C under identical gaseous composition to ours. In another study, Poli et al. [20] reported that the shelf life of seabass fillets stored at 2 °C under MAP was 8 days. However, they used a different gas atmosphere. Similarly, variable results have been reported for salmon fillets stored under MAP [21][22][23].

Microbiological Changes
After 2 days of storage, the total microbial population of seabass fillets expressed by total viable count (TVC) was at the level of 4.5 log cfu/g. At the time of sensory rejection, the total microbial population reached 6.8 log cfu/g ( Figure 2a). Initial (day 2) microbial populations of spoilage bacteria were at the level of 3.3, 4.6, 4.5, 3.4 and 2.3 log cfu/g, for lactic acid bacteria (LAB), Pseudomonas spp., H2S producing bacteria (presumable Shewanella spp.), Enterobacteriaceae and B. thermosphacta, respectively. The dominant microorganisms were Pseudomonas spp., reaching at the end of shelf life populations densities as high as 6.7 log cfu/g (p < 0.05), followed by H2S producing bacteria and LAB with 5.9 log cfu/g, approximately. B. thermosphacta and Enterobacteriaceae populations were not higher than 4.6 log cfu/g. These results are in agreement with our previous work [18] conducted under the same conditions with whole gutted seabass. In all cases, our previous and present study, it was noticed that Pseudomonas was the most dominant microorganism, possibly due to the use of a high O2 concentration level (10%). On the other hand, B. thermosphacta and LAB populations usually predominate under reduced O2 and elevated CO2 of MAP by outcompeting the strictly aerobic Pseudomonas spp. [24]. However, this was not observed in the present work, probably due to different gas composition. Indeed, the study of Parlapani et al. [25], using molecular techniques found that Pseudomonas spp. was still a great part of spoilage microbiota together with Carnobacterium spp. and other LAB in gilt-head seabream fillets stored in the same packaging conditions.
Regarding salmon slices, the total microbial load did not exceed 3 log cfu/g after 2 days of storage, whereas at the end of shelf life reached the level of 5.6 log cfu/g (Figure 2b). In contrast to seabass fillets, the dominant microorganisms were H2S producing bacteria and LAB that reached a Thus, shelf life varies between fish species or even among the products (whole, gutted, fillets) [16,17]. In a recent work by our team [18], it was found that whole gutted seabass had a shelf life of 13 days, under the same conditions of temperature and atmosphere as herein. According to Kostaki et al. [19], the shelf life of sea bass fillets was 12 days when stored at 4 • C under identical gaseous composition to ours. In another study, Poli et al. [20] reported that the shelf life of seabass fillets stored at 2 • C under MAP was 8 days. However, they used a different gas atmosphere. Similarly, variable results have been reported for salmon fillets stored under MAP [21][22][23].

Microbiological Changes
After 2 days of storage, the total microbial population of seabass fillets expressed by total viable count (TVC) was at the level of 4.5 log cfu/g. At the time of sensory rejection, the total microbial population reached 6.8 log cfu/g ( Figure 2a). Initial (day 2) microbial populations of spoilage bacteria were at the level of 3.3, 4.6, 4.5, 3.4 and 2.3 log cfu/g, for lactic acid bacteria (LAB), Pseudomonas spp., H 2 S producing bacteria (presumable Shewanella spp.), Enterobacteriaceae and B. thermosphacta, respectively. The dominant microorganisms were Pseudomonas spp., reaching at the end of shelf life populations densities as high as 6.7 log cfu/g (p < 0.05), followed by H 2 S producing bacteria and LAB with 5.9 log cfu/g, approximately. B. thermosphacta and Enterobacteriaceae populations were not higher than 4.6 log cfu/g. These results are in agreement with our previous work [18] conducted under the same conditions with whole gutted seabass. In all cases, our previous and present study, it was noticed that Pseudomonas was the most dominant microorganism, possibly due to the use of a high O 2 concentration level (10%). On the other hand, B. thermosphacta and LAB populations usually predominate under reduced O 2 and elevated CO 2 of MAP by outcompeting the strictly aerobic Pseudomonas spp. [24]. However, this was not observed in the present work, probably due to different gas composition. Indeed, the study of Parlapani et al. [25], using molecular techniques found that Pseudomonas spp. was still a great part of spoilage microbiota together with Carnobacterium spp. and other LAB in gilt-head seabream fillets stored in the same packaging conditions. and Tamplin [26], who highlighted the importance of LAB in fresh Atlantic salmon stored under MAP. Using culture-independent methods, they found that the microbial communities were dominated by Shewanella spp. and Carnobacterium spp., after 15 days of storage. The low spoilage level of 5.8 log cfu/g might be due to the fact that other mechanisms, such as lipid oxidation of salmon, which is quite fatty fish compared to seabass, occurred taken into account the 10% of oxygen used in this package. Between the two fish species, LAB and B. thermosphacta counts were the same (p > 0.05) within each sampling day. Enterobacteriaceae and Pseudomonas spp. counts were always higher in seabass fillets. H2S bacteria counts were significantly higher in seabass during the 7 days of storage, but afterwards, their levels did not differ greatly (p > 0.05) between salmon and sea bass.  [27,28] or salmon [29].

TVB-N Determination
At the end of shelf life, TVB-N values never reached the legislated regulatory limit, which is at 30-35 mg N/100 g, [30]. It has already been shown that this parameter-often used as a spoilage quality indicator for seafood kept on ice-displays lower values for fish stored in a CO2 atmosphere and considered spoiled by sensory analysis [17]. Therefore, TVB-N should be considered as a poor indicator of fish freshness, as also proposed by others [27,28]. Regarding salmon slices, the total microbial load did not exceed 3 log cfu/g after 2 days of storage, whereas at the end of shelf life reached the level of 5.6 log cfu/g ( Figure 2b). In contrast to seabass fillets, the dominant microorganisms were H 2 S producing bacteria and LAB that reached a value of 5.3 log cfu/g, followed by Pseudomonas spp. with 4.9 log cfu/g (p > 0.05). B. thermosphacta and Enterobacteriaceae counts were significantly lower (p < 0.05) at the sensory rejection point, reaching the values of 4.1 and 2.8 log cfu/g, respectively. Similar observations have been reported by Powell and Tamplin [26], who highlighted the importance of LAB in fresh Atlantic salmon stored under MAP. Using culture-independent methods, they found that the microbial communities were dominated by Shewanella spp. and Carnobacterium spp., after 15 days of storage. The low spoilage level of 5.8 log cfu/g might be due to the fact that other mechanisms, such as lipid oxidation of salmon, which is quite fatty fish compared to seabass, occurred taken into account the 10% of oxygen used in this package.
Between the two fish species, LAB and B. thermosphacta counts were the same (p > 0.05) within each sampling day. Enterobacteriaceae and Pseudomonas spp. counts were always higher in seabass fillets. H 2 S bacteria counts were significantly higher in seabass during the 7 days of storage, but afterwards, their levels did not differ greatly (p > 0.05) between salmon and sea bass.  [27,28] or salmon [29].

TVB-N Determination
At the end of shelf life, TVB-N values never reached the legislated regulatory limit, which is at 30-35 mg N/100 g, [30]. It has already been shown that this parameter-often used as a spoilage quality indicator for seafood kept on ice-displays lower values for fish stored in a CO 2 atmosphere and considered spoiled by sensory analysis [17]. Therefore, TVB-N should be considered as a poor indicator of fish freshness, as also proposed by others [27,28].

Production of Volatile Compounds During Fish Storage
The analysis by SPME-GC/MS of the salmon slices and seabass fillets at different storage stages under MAP at 2 °C identified 54 volatile compounds (excluding hydrocarbons, terpenoids and miscellaneous compounds) that were classified by their characteristic functional group ( Table 1).
The majority of them were mainly aldehydes (22), followed by alcohols (13), ketones (12), esters (6), and one acid (acetic acid), which were all found in the two species studied except 2,3-butanediol, 3-hydroxy-2-butanone (acetoin) and ethyl lactate. The first two compounds were detected only in seabass fillets whereas ethyl lactate was found only in salmon. By comparing the relative concentrations of the compounds at different stages of storage (day 2, 7, 11, 14), it was possible to identify compounds whose levels increased, decreased or fluctuated during spoilage. Interestingly, the behavior during storage was similar within certain classes of compounds but different between fish species, as it will be further discussed.
Among the numerous aldehydes detected in seabass fillets, the amounts of the higher members of saturated aldehydes (nonanal, decanal, undecanal), unsaturated aldehydes with 8, 10, 11 carbon atoms (2-octenal, 2-decenal, 2-undecenal), aromatic aldehydes (benzaldehyde, phenylacetaldehyde) as well as 2,4-decadienal increased by 66% from the middle stages of storage (day 7) until the last sampling point (day 14). Due to their similar evolution profile, they are referred hereafter as group Ald-1 (Table 1). A different trend was observed for the rest aldehydes, such as the saturated homologues with 2-8 carbon atoms and the unsaturated ones with 5-7 carbon atoms (group Ald-2). Their levels remained approximately constant during the first seven days of storage, then reached a maximum at the rejection point (66% increase) and subsequently, they declined by a factor of 50%.

Production of Volatile Compounds During Fish Storage
The analysis by SPME-GC/MS of the salmon slices and seabass fillets at different storage stages under MAP at 2 • C identified 54 volatile compounds (excluding hydrocarbons, terpenoids and miscellaneous compounds) that were classified by their characteristic functional group ( Table 1).
The majority of them were mainly aldehydes (22), followed by alcohols (13), ketones (12), esters (6), and one acid (acetic acid), which were all found in the two species studied except 2,3-butanediol, 3-hydroxy-2-butanone (acetoin) and ethyl lactate. The first two compounds were detected only in seabass fillets whereas ethyl lactate was found only in salmon. By comparing the relative concentrations of the compounds at different stages of storage (day 2, 7, 11, 14), it was possible to identify compounds whose levels increased, decreased or fluctuated during spoilage. Interestingly, the behavior during storage was similar within certain classes of compounds but different between fish species, as it will be further discussed.
Among the numerous aldehydes detected in seabass fillets, the amounts of the higher members of saturated aldehydes (nonanal, decanal, undecanal), unsaturated aldehydes with 8, 10, 11 carbon atoms (2-octenal, 2-decenal, 2-undecenal), aromatic aldehydes (benzaldehyde, phenylacetaldehyde) as well as 2,4-decadienal increased by 66% from the middle stages of storage (day 7) until the last sampling point (day 14). Due to their similar evolution profile, they are referred hereafter as group Ald-1 (Table 1). A different trend was observed for the rest aldehydes, such as the saturated homologues with 2-8 carbon atoms and the unsaturated ones with 5-7 carbon atoms (group Ald-2). Their levels remained approximately constant during the first seven days of storage, then reached a maximum at the rejection point (66% increase) and subsequently, they declined by a factor of 50%.
Molecules 2020, 25, x 1 of 16 Since the focus of the present study was to find volatiles suitable as spoilage markers, the relationships of the aforementioned evolution profiles with the remaining shelf life were depicted schematically (Figure 4a,b). (c) Group Ket-3 and 2,3-butanediol; (d) Acetic acid. The compounds included in each group are described in Table 1.
As it can be seen, when the fish was considered fresh (remaining shelf life over 7 days), the amounts of aldehydes, ketones, alcohols and esters groups remained relatively low. However, at the rejection point (0 days of remaining shelf life), their respective amounts increased substantially. The most remarkable evolution patterns were observed for 2,3-butanedione, 3-hydroxy-2-butanone (group Ket-3) and acetic acid (Table 1, Figure 4c, 4d). Their amounts were negligible initially, when the fish remaining shelf life was over 5 days. Their relative concentration started to increased and subsequently reached a maximum at the end of shelf life (0 remaining days), corresponding to an almost 100% increase. 2,3-Butanediol was the only compound that was not detected until day 11, which is the end of shelf life, but reached a high concentration at the end of storage. The formation of the aforementioned compounds has been associated with microbial activity occurring during storage. 2,3-Butanedione is reduced to acetoin, which is in turn reduced to 2,3-butanediol through enzymatic mediated reaction [31]. Acetoin formation in seafood has been associated mainly with LAB [32] and in some cases to Photobacterium phosphoreum [33] and Shewanella baltica [34], whereas in meat has been reported for Pseudomonas spp. [35]. The production of acetic acid has been associated with the metabolic activity of B. thermosphacta, some heterofermentative LAB and Shewanella spp. [34,[36][37][38]. In a previous study [39], these compounds were attributed exclusively to microbial activity, as they were detected only in inoculated sterile fish juice. According to our results, these four compounds (diacetyl, acetoin, 2,3-butanediol, acetic acid) may be suggested as spoilage markers of seabass stored under MAP. (c) Group Ket-3 and 2,3-butanediol; (d) Acetic acid. The compounds included in each group are described in Table 1.
As it can be seen, when the fish was considered fresh (remaining shelf life over 7 days), the amounts of aldehydes, ketones, alcohols and esters groups remained relatively low. However, at the rejection point (0 days of remaining shelf life), their respective amounts increased substantially. The most remarkable evolution patterns were observed for 2,3-butanedione, 3-hydroxy-2-butanone (group Ket-3) and acetic acid (Table 1, Figure 4c,d). Their amounts were negligible initially, when the fish remaining shelf life was over 5 days. Their relative concentration started to increased and subsequently reached a maximum at the end of shelf life (0 remaining days), corresponding to an almost 100% increase. 2,3-Butanediol was the only compound that was not detected until day 11, which is the end of shelf life, but reached a high concentration at the end of storage. The formation of the aforementioned compounds has been associated with microbial activity occurring during storage. 2,3-Butanedione is reduced to acetoin, which is in turn reduced to 2,3-butanediol through enzymatic mediated reaction [31]. Acetoin formation in seafood has been associated mainly with LAB [32] and in some cases to Photobacterium phosphoreum [33] and Shewanella baltica [34], whereas in meat has been reported for Pseudomonas spp. [35]. The production of acetic acid has been associated with the metabolic activity of B. thermosphacta, some heterofermentative LAB and Shewanella spp. [34,[36][37][38]. In a previous study [39], these compounds were attributed exclusively to microbial activity, as they were detected only in inoculated sterile fish juice. According to our results, these four compounds (diacetyl, acetoin, 2,3-butanediol, acetic acid) may be suggested as spoilage markers of seabass stored under MAP.
As with seabass, aldehydes dominated the volatile fraction of salmon slices stored under MAP at 2 • C. The most abundant compounds were hexanal and nonanal followed by acetaldehyde (Table 1). Among aldehydes, two groups having different evolution patterns can be distinguished. The first group (Ald-1) includes the saturated homologues with 5-10 carbon atoms (pentanal to decanal) and three unsaturated members, namely 2-pentenal, 2-hexenal and 4-heptenal. Their amounts decreased from day 2 to day 11 and afterwards they increased or remained constant. This was more pronounced with the unsaturated members, whose relative amounts were found 50-60% lower at day 11. On the contrary, the Ald-2 group, which comprised of 2-alkenals with 7-11 carbon atoms (2-heptenal, 2-octenal, 2-decenal, 2-undecenal) as well as 2,4-decadienal, followed an almost linear declining trend throughout storage. This is depicted clearly in Figure 5a,b relatively to the remaining shelf life. This linear trend can be exploited as a potential freshness index for salmon stored under MAP.  Table 1.

Fish Provision, Handling and Storage
Seabass fillets and salmon slices were obtained directly from a Greek fish processing plant (Selonda Aquaculture SA, Magoula, Attica, Greece). Seabass was farmed in the geographical area designated as FAO 37, 3.1 (Aegean Sea, Greece), while salmon was farmed in the European Union (Norway) and was imported to Greece. The products were packed under MAP in polysterene trays (Sirap Gema S.p.A., Verolanuova, Italy) covered with a MAP film (BDF 8050F, Cryovac-Sealed Air Ltd., Athens, Greece). Each tray contained either two seabass fillets (approximately 120 g each fillet) or one salmon slice (approximately 200 g). The concentrations of gases were CO2: 60%, O2: 10%, N2: 30% as recommended by the Hellenic seafood industry for this type of products. The products were delivered to the laboratory within 5 h after packaging (day 0) in insulated boxes with ice flakes. Subsequently, the products were stored in incubators (Panasonic MIR-254 cooled incubator, PHC Europe B.V., Etten-Leur, The Netherlands) operating at 2 °C. The two batches (lots) of each product were provided with 15 days difference in June 2016. The sampling started the day after receiving the products (day 1). At each sampling point, a suitable quantity of product (sample) was taken from 2  Table 1. Among the detected ketones, the 2-alkanones (2-butanone, 2-pentanone, 2-heptanone, 2-nonanone) as well as 2,3-octanedione and the two 3,5-octadien-2-one isomers presented a distinct evolution pattern (Ket-1 group, Table 1). Their levels remained constant or decreased slightly during the first 11 days of storage. After the sensory rejection point (0 days of remaining shelf life), the relative amount of Ket-1 group increased by 150%, (Figure 5b). This characteristic profile was also observed for Alc-1 group, which comprised mainly of 1-octen-3-ol, (5Z)-octa-1,5-dien-3-ol and 1-penten-3-ol (Figure 5a). The latter one was the main alcohol produced during spoilage besides ethanol, whose amount fluctuated. Hexanol, 3-methyl-1-butanol (Alc-2 group) and ethyl esters of C6-C10 fatty acids (Est-1 group) followed a pattern (Figure 5c) similar to that in seabass (Figure 4b). The levels of acetic acid and group Alc-2 (propanol and dodecanol) presented a maximum before the rejection point (0 days of remaining shelf life), and then they declined rapidly (Figure 5d). However, the observed profile of acetic acid in salmon was different from the respective one in seabass. Furthermore, it is in contrast to other researchers who suggested acetic acid as a spoilage marker in fresh king salmon [12] and salmon fillets [29].
A notable difference between the examined fish species was the absence of acetoin and 2,3-butanediol in the salmon samples. This is contradictory to the results reported by Wierda et al. [12] albeit the storage conditions were different from ours. Furthermore, a small amount of ethyl lactate was detected only in salmon, but its levels fluctuated during storage ( Table 1).
Most of the VOCs detected in this study have also been reported for other fish and seafood by other researchers as well [11,[13][14][15]32,40,41]. It is generally known that most of the saturated or unsaturated aldehydes, alcohols and carbonyls in fish flesh come from the autoxidation of the polyunsaturated fatty acids resulting in the formation of hydroperoxides. Although, autoxidation of fatty acids can be initiated by a catalyst such as light or oxygen or by enzymes coming from the fish flesh, it can be also initiated by the enzymes of the microorganisms present in fish [42]. Thus, we can infer that the different evolution patterns observed between the two fish species could be attributed both to their characteristic microbiota growth as discussed earlier and the lipid composition of each fish. It is known that different fish species from different geographical areas are spoiled by different SSOs, even if the fish are stored under identical conditions (temperature and atmosphere), which means that different metabolites might be produced (9). This can explain the difference on VOCs profile between the two different fish. Various alcohols, aldehydes and esters, such as 3-methyl-1-butanol, 3-and 2-methylbutanal and ethyl esters of short chain fatty acids (C4-C10) have been suggested as potential spoilage indicators in fish and meat products [10,13,32,39,[43][44][45]. However, our results from both fish species studied, indicate that they were produced (if any) at very low levels during storage. Thus, their usefulness as biomarkers may be limited under MAP in combination with low temperatures.

Fish Provision, Handling and Storage
Seabass fillets and salmon slices were obtained directly from a Greek fish processing plant (Selonda Aquaculture SA, Magoula, Attica, Greece). Seabass was farmed in the geographical area designated as FAO 37, 3.1 (Aegean Sea, Greece), while salmon was farmed in the European Union (Norway) and was imported to Greece. The products were packed under MAP in polysterene trays (Sirap Gema S.p.A., Verolanuova, Italy) covered with a MAP film (BDF 8050F, Cryovac-Sealed Air Ltd., Athens, Greece). Each tray contained either two seabass fillets (approximately 120 g each fillet) or one salmon slice (approximately 200 g). The concentrations of gases were CO 2 : 60%, O 2 : 10%, N 2 : 30% as recommended by the Hellenic seafood industry for this type of products. The products were delivered to the laboratory within 5 h after packaging (day 0) in insulated boxes with ice flakes. Subsequently, the products were stored in incubators (Panasonic MIR-254 cooled incubator, PHC Europe B.V., Etten-Leur, The Netherlands) operating at 2 • C. The two batches (lots) of each product were provided with 15 days difference in June 2016. The sampling started the day after receiving the products (day 1). At each sampling point, a suitable quantity of product (sample) was taken from 2 different packages for each batch (lot) of product (2 replicate samples per batch) and analyzed as described below. Thus, four replicate determinations were accomplished at each sampling point (n = 4 = 1 sample/package × 2 packages/batch × 2 batches), except for volatiles, where the samples were pooled as described in VOCs analyses section.

Sensory Acceptance Evaluation
The attributes of cooked fish (flavor and taste) were evaluated by five trained members of our Department (Dept. Ichthyology & Aquatic Environment, University of Thessaly, Greece). Approximately 20 g of fish flesh were removed, wrapped in aluminum foil and cooked in an oven preheated at 180 • C for 20 min. After that, the cooked samples were left to cool and evaluated.
The panel was asked to evaluate overall impression and acceptability. Rating was assigned on a 1-5 descriptive hedonic scale (5 = like extremely, 4 = like, 3 = neutral, 2 = dislike and 1 = dislike extremely). A score of 3 was considered as the score for minimum acceptability, hence the time point corresponded to a score below 3 was set as end of shelf life. Only cooked samples were chosen to be evaluated since the aim of the sensory acceptance evaluation was to determine the shelf life of fish products and correlated with the production of various potential freshness/spoilage markers and for this reason an in-depth analysis of sensory descriptors changes of raw samples was not performed.

Microbiological Analysis
A sample of 25 g fish product (representing 1 replicate as described in Section 3.1) was placed into a stomacher bag containing 225 mL sterile MRD (Maximum Recovery Diluent, 8.5 g/L NaCl, 1.0 g/L bacteriological peptone) and homogenized for 1 min using a Stomacher (Bug Mixer, Interscience, London, UK). Then, 0.1 mL of 10-fold serial dilutions in MRD were used for the spread plate technique for the enumeration of the following microorganisms: (a) Pseudomonas spp. on cetrimide-fucidin-cephaloridine agar (CFC, LAB M, Lancashire, UK), and (b) Brochothrix thermosphacta on streptomycin sulphate, thallus acetate, cycloheximide (actidione) agar (STAA, Biolife Italiana srl, Milano, Italy), after incubation at 25 • C for 48 h. Additionally, 1 mL of the serial dilution were used for the pour plate with overlay technique for the enumeration of (a) the Total viable counts (TVC) on Iron Agar (IA, prepared according to Gram et al. 1987

Determination of TVB-N
A sample of 10 g fish product (representing 1 replicate as described in Section 3.1) was homogenized in trichloroacetic acid solution (TCA 60 g/L), filtered through Whatman No.1 paper in a 100 mL volumetric flask and made up to volume with TCA solution. An aliquot of 50 mL extract was analyzed for TVB-N using the steam-distillation procedure [46]. Each extract was analyzed twice, and the mean value was used for further calculations. The results were expressed as mean mg N/100 g ± standard deviation of 4 replicate samples (2 replicates per batch of fish product). All chemical reagents were supplied by Sigma-Aldrich (Steinheim, Germany).

VOCs Determination by Headspace SPME-GC/MS
At each sampling point, a total amount of 50 g fish product was obtained from 4 different packages (2 packages from each batch) and pooled. Then, 5 g of the pooled fish sample, 5 mL of 30% NaCl solution and 100 µL of internal standard (4-methyl-1-pentanol, in-vial concentration 1000 µg/L) were transferred into a 20 mL glass vial and homogenized with a glass rod for 1 min. The vial was hermetically closed with a Mininert valve (Sigma Aldrich, St. Louis, MO, USA) and headspace SPME-GC/MS analysis was performed according to Parlapani et al. [18]. Identification of the compounds was performed by comparing: (i) the retention indices (RI) based on an homologous series of even numbered n-alkanes (C8-C24, Polyscience, Niles, IL, USA) with those of authentic compounds and by comparison with literature data, and (ii) MS data with those of reference compounds and by MS data obtained from NIST 14 (NIST/EPA/NIH Mass Spectral Library with Search Program, software version 2.0d, Gaithersburg, MD, U.S.) and WILEY 7 libraries. AMDIS software (version 2.62, http://chemdata.nist.gov/mass-spc/amdis/, Gaithersburg, MD, U.S.) was used for the deconvolution of mass spectra and identification of target components. The volatile compounds were quantified by dividing the peak area of the compound of interest by the peak area of internal standard (IS) and multiplying this ratio by the concentration of the IS (expressed as µg/L). The peak areas were measured by selecting single ions (Table 1). Each pooled fish sample was extracted and analysed twice and the mean values were used. All authentic compounds used were of analytical grade and purchased from Sigma Aldrich (Steinheim, Germany).

Statistical Analysis
The t-test of means or Analysis of Variance followed by Tukey s significant difference test, using STATISTICA 6.0 (Stat Soft Inc., Tulsa, OK, U.S.), were used to compare the means in viable counts, TVB-N and overall sensory acceptance score. A probability level of p ≤ 0.05 was considered statistically significant. Statistical analysis of VOCs was not performed, for the reason that the measurements were conducted in duplicates from a pooled sample, since the aim of the study was to monitor the profile of VOCs evolution during storage and distribution and reveal any potential spoilage marker.

Conclusions
The same storage conditions, MAP and temperature, affected the growth of microbial populations in a completely different way between seabass and salmon, thus affecting the evolution of fish volatiles. This reflects the difficulty of defining common markers of fish spoilage or freshness among fish species. It is known that different spoilage microorganisms are grown on fish from different geographical areas, even if these fish are stored under the same conditions, hence different microbial origin VOCs are produced. Therefore, in seeking potential spoilage markers using the volatilome, the research should be focused on the following directions: a) identification and quantification of species specific volatile markers under specified storage conditions or b) exploration of the overall trend of volatilome through multivariate data analysis and development of a large database with the volatile fingerprint of each fish species produced in certain storage conditions.