1. Introduction
Animal derived products, such as fresh meat and fish products, are amongst the most resource demanding yet perishable food product categories. Examining fish products specifically, 35% of the fisheries and aquaculture production is wasted globally. This underlines the importance of novel tools and processes that can assist in minimizing these losses and thereby reduce the pressure and enhance the sustainability of the current production systems [
1].
When examining the losses at a retail level, limited product shelf life can be a major contributor [
2]. This is especially applicable to fresh fish fillets, a preferred product choice in many countries, yet with a limited shelf life [
3,
4,
5]. To address this, a range of processing and preservation methods, such as salting, curing, drying, marinating, canning, smoking etc., have been developed and applied traditionally post-harvest to extend the shelf life of different fish products [
6,
7]. Other common approaches applied to extend the shelf life of fish is refrigeration and freezing after harvesting, which delays or inhibits microbial growth and physicochemical reactions, resulting in quality deterioration during storage.
Besides the processing effects on shelf life, the application of different technologies can alter the sensory and nutritional characteristics of the product, affecting its organoleptic quality (appearance, aroma, taste-flavor, texture) and having in some cases detrimental effects on its nutritional value [
7,
8]. Due to the importance of the appearance of the product in retail (choice) and its flavor and texture during consumption (eating experience), treatment effects on the sensory aspects of both raw and cooked fish products are important to consider, besides the nutritional alterations and shelf life extension achieved via processing [
9,
10].
Moreover, whereas processing can result in high sensory and nutritional quality, the perceived freshness is an established part of fish product quality [
7,
9,
10]. This is shown by the higher perceived quality associations of consumers to fresh versus processed fish products and a preference for food products that retain their natural appearance [
5,
11]. Indeed, nowadays the consumer demands fresh or minimally processed foods of high quality and nutritional value, natural and safe, with the minimum addition of preservatives [
12]. Within these frames, preservation and processing methods that increase the shelf life of fish products, while not compromising on the quality of sensory characteristics, are sought by the industry [
13].
Considering the role of water activity (
aw) in microbial growth and thereby perceived freshness, the current work examined a minimal processing method consisting of a mild tissue dehydration (ranging 5–10%, corresponding to final a
w values between 0.93–0.95), for its ability to extend shelf life, while not affecting the raw quality of gilthead seabream (
Sparus aurata Linnaeus, 1758) fillets [
14,
15,
16]. Osmotic dehydration has been reported as an effective non-thermal processing method, for delaying growth of spoilage bacteria and extending the shelf life of fresh fish. Several osmotic media have been investigated, such as sucrose, maltodextrin, trehalose and glycerol at different processing times and temperatures [
16,
17,
18]. Towards the current trend for fresh-like food products with improved quality and extended shelf life, the systematic evaluation of consumer perception for minimally processed fish is a prerequisite.
In the present study, the shelf life extension was benchmarked by the Quality Index Method (QIM) and supported by microbial analyses of the tissue [
19]. To ensure that the sensory characteristics of the raw product were not altered, sensory discrimination testing was performed [
20]. Moreover, due to the importance of the eating quality and the sensory characteristics upon consumers’ acceptance (and re-purchase) the treatment was evaluated regarding its effects on cooked tissue throughout its commercial shelf life via generic descriptive analyses (DA) [
20]. Gilthead seabream was chosen as a species of importance due to its production numbers and contribution to the growing aquaculture sector [
21]. Gilthead seabream (
Sparus aurata) is one of the most cultured species in the Mediterranean area, with a total production of almost 200,000 tons in 2019 [
22]. Besides the desired organoleptic characteristics (aroma, taste) of the species, gilthead seabream has a high nutritional value, reflected in flesh which contains a high concentration of proteins, essential amino acids and polyunsaturated fatty acids [
23,
24]. Although usually sold as whole fish, gilthead seabream filleted products have high commercial potential but suffer from short shelf life [
16,
25,
26].
The specific aims of the current work were to evaluate:
the shelf-life extension achieved by the treatment and raw quality of the Treated gilthead seabream fillets versus the Control ones via sensory (QIM, Tetrad discrimination) and microbial methodologies
the cooked quality and the sensory profile of the Treated gilthead seabream fillets versus the Control ones throughout shelf life via sensory descriptive analysis (DA).
2. Materials and Methods
2.1. Fish Samples, Processing and Storage Conditions
The fresh gilthead seabream (Sparus aurata, Linnaeus, 1758) fillets used for the pre-trials and the main experiment were commercially obtained from Select Fish SA, Athens, Greece. Fish destined for the main experiment were harvested on 1 March 2022. Pre-processing of fish into fillets was conducted on the following day at the Select Fish SA facilities. The weight of the obtained fish fillets ranged from 90–120 g. Directly after pre-processing 400 fillets were divided into seven polystyrene boxes (~6 kg each), with the addition of an appropriate amount of flaked ice. Fillets were coated with high density polyethylene films to avoid direct contact of fish flesh with ice or fish skin. After packaging, the boxes were transported directly to the Agricultural University of Athens (AUA), Department of Food Science and Human nutrition, Laboratory of Food Process Engineering. Upon arrival, the ice was removed from the polystyrene boxes, which were then placed in refrigerated storage at 2 °C (±0.2). The packaging date and expiry date set on the package were the second and ninth of March 2022, respectively.
The experimental trials at AUA were initiated the following day with the processing of the fillets, which was assigned as Day 0 (D0) of the experiments. For this purpose, fish fillets were divided randomly into two groups. Half of the fillets remained untreated at refrigerated storage (Control fillets). The other half of the fillets were mildly processed via osmotic dehydration (Treated-SUSEA fillets). The Osmotic dehydration was applied on fish fillets using a hypertonic aquatic solution optimized by SuSea BV (SuSea BV, Eindhoven, The Netherlands). Fish was processed in an aquatic solution consisting of 30% glycerol, 10% dextrose and 5% NaCl for 40 min at 15 °C [
27,
28]. Directly after processing, the fillets were dried onto filter paper and placed back in the polystyrene boxes for storage at 2 °C.
2.2. Design of Freshness and Shelf Life Experiments of Raw Products
The shelf-life trials of the raw fillets were initiated on D1 of the experiments with subsequent analyses and sampling being conducted on D1, D5, D7, D8, D11 and D13. This included up to D11 for both the Control and Treated fillets, whereas at D13 only the Treated fillets were evaluated due to the expiry of the Control ones. The storage time and sampling frequency were defined according to the Control fillets’ expected shelf life (expiry date) in order to (i) capture three freshness stages (High: D1; Medium: D5; Low: D7), (ii) monitor the transition of the fillets from acceptable to unacceptable for consumption (D7–D8) and (iii) capture the expected shelf-life extension (up to 50%) for the Treated fillets.
The freshness and corresponding shelf life of the gilthead seabream fillets undergoing storage at 2 °C were assessed by the Quality Index Method (QIM) and through microbial evaluations. Five fillet replicates were used per treatment at each shelf-life date examined. In addition to the shelf life trials, a Tetrad test (ISO 6658:2017) with 36 assessors was conducted on D1 to determine whether any perceivable differences between the raw fillets of the two treatments existed.
2.2.1. Discrimination Testing (Tetrad)
Discrimination testing was conducted at D1 of storage, following the Tetrad method. For this purpose, N = 36 assessors with no previous experience in sensory science were recruited. Assessors were presented with four fillets (two of each treatment) on 3-digit coded paper plates that did not reflect light, and they were asked to group the samples into two pairs taking into account their appearance, odor and texture by hand. The Tetrad evaluation was conducted at room temperature on tablets using the 2020 EyeQuestion® Software (Version: 5.2.3, Logic8. BV, Elst, The Netherlands) and followed a fully randomised sample presentation among panellists.
The statistical power of the test and the data analysis were calculated according to the V power program written by Virginie Jesionka and based on the Discrimination Test Planning and Analysis Tools developed by Tom Carr (Carr Consulting, 1215 Washington Ave, Suite 203, Wilmette, IL 60091, USA). Due to the natural high variability of the within batch variation of the samples, the following parameters were taken into account when setting up the test: d’ = 1.2, a = 0.1 and Power = 0.9. As a result, the minimum number of assessors required to participate in the test was 32 and our actual sample size was N = 36. For the data analysis and the setting of parameters before the test, the Thurstonian model for a Similarity test was followed.
2.2.2. Quality Index Method (QIM)
Development and Validation of the QIM Scheme for Gilthead Seabream Fillets
The QIM scheme development was based on the methodology described by Martinsdóttir, Sveinsdottir, Luten, Schelvis-Smit and Hyldig [
19]. Previous work describing QIM schemes and sensory changes of whole fish and fillets of fresh gilthead seabream was used for its development and parameter definition [
29,
30,
31]. This scheme was then validated through pre-trials by examining and recording the changes occurring at fresh gilthead seabream fillets stored at 2 °C for the defined appearance, odor and texture parameters until spoilage. During the pre-trials, fillets corresponding to different freshness stages were stored (−40 °C, storage duration: 1–2 weeks) in order to create reference material for the introductory QIM evaluation session.
The validated scheme included six appearance, odor and texture parameters (
Table 1). For each of the parameters a demerit point scale of up to 3 points, accompanied by appropriate descriptions reflecting the quality changes with storage, was used. Zero (0) corresponded to very fresh gilthead seabream fillet, whereas the demerit points increased according to the spoilage degree of the fillet with two (2) and three (3) points corresponding to spoiled fish appearance/texture and odor parameters, respectively. The Quality Index (QI) of the scheme ranged from 0 to 13 points.
QIM Evaluation
A panel consisting of eight (8) judges was employed for the QIM. The panel participated in an initial training session during which the panel leader introduced the process, preliminary QIM scheme and point scale used to evaluate each parameter, by presenting reference samples corresponding to different freshness stages stored during the pre-trials.
Five to eight judges participated at each of the QIM freshness evaluation sessions conducted on D1, D5, D7, D8, D11 and D13. The fillets of each treatment were presented to the panelists in triplicates on 3-digit coded white paper plates that did not reflect light at room temperature. The QIM scheme evaluation was performed on tablets using the Compusense20 software (Version 22, Compusense Inc., Guelph, ON, Canada). For each of the fillets the judge had to evaluate the parameters using the QIM demerit point scheme and at the end indicate whether the fillet was acceptable for consumption. A panel discussion was conducted after the initial evaluation session to ensure panel consensus on the definition of the individual parameters and scale points.
The QIM results of the Control and Treated fillets are presented as the total QI (sum of demerit points), averaged across assessor and replica, per storage day of the freshness trials. The fitted linear model equation and R2 of the fit were calculated in Microsoft Excel 2016.
Microbial Evaluation
For microbiological enumeration, a representative sample (10 g) was collected from fish fillets (including skin and flesh), transferred to a sterile stomacher bag with 90 mL sterilized Ringer solution (Merck, Darmstadt, Germany) and homogenized for 60 s with a Stomacher (BagMixer®, Interscience, Saint-Nom-la-Breteche, France). Zero point one milliliters of tenfold serial dilutions of fish homogenates were spread on the surface of appropriate media in Petri dishes for microbial enumeration. Total aerobic viable count (TVC) was enumerated on Plate Count Agar (PCA; Neogen, Lansing, MI, USA) after incubation at 25 °C for 72 h. Pseudomonas spp. were enumerated on Cetrimide Agar (CFC; Condalab, Torrejon De Ardoz, Spain) after incubation at 25 °C for 48 h.
Two replicates of at least three appropriate dilutions were enumerated. The microbial growth was modelled using the Baranyi growth model (Baranyi and Roberts, 1995). For curve fitting the program DMFit (IFR, Institute of Food Research, Reading, UK) was used (available at
http://www.combase.cc/index.php/en/, accessed at 1 June 2022). Kinetic parameters of microbial growth such as the rate (k), lag phase (λ) and final population levels (N
max) were estimated. Furthermore, the linear correlations between different QI and microbial measurements across shelf-life sampling dates were studied by calculating simple linear regression models in Microsoft Excel software, 2016. The relationships were evaluated by an F-test with a significance level of 5%, which are included as correlation coefficients (r) accompanied with their respective
p-values (
pr).
2.3. Sensory Evaluation of Cooked Products
2.3.1. Samples Storage, Preparation, Cooking Trials and Presentation
Control and Treated fillets destined for sensory evaluation via generic Descriptive Analysis (DA) from D1, D5 and D7 were vacuum packed and stored at −40 °C at the AUA (Athens, GR), Department of Food Science and Human Nutrition, Laboratory of Food Process Engineering. The samples were then transported frozen to Aarhus University, Department of Food Science (AU FOOD, Aarhus, DK) where they were stored at −20 °C, until analysis. The thawing process involved placing the vacuum-packed fillets in a thermal chamber (Termaks, series 6000) overnight at 1 ± 0.1 °C to allow for gradual defrosting of the tissue. The time of frozen storage from sampling at the AUA facilities till thawing of tissue for sensory analysis at the AU facilities did not exceed a month.
Prior to the initiation of the sensory evaluation process, cooking trials were performed. The sample preparation for the cooking trials was conducted in like manner as for the sensory evaluation. This involved removing the dorsal line and ventral part of the fillet, due to the their higher fat content and contact with the internal organs, which can result in deviations in the volatile composition (and thereby sensory profile) [
32]. Thereafter, the dorsal part of the fillet remaining was cut into approximately 4 cm × 3 cm rectangles (15–20 g), and placed in ceramic containers that were covered by aluminum foil [
33]. For the culinary preparation of the samples, a series of different time (7–10 min) and temperature (80–100 °C) combinations were examined. The final cooking temperature was set at 100 °C for 7 min (relative humidity of the oven chamber was kept at 60%) and was performed in iCombi
® Pro RATIONAL ovens. The cooking process as well as temperature (60 °C) and resting time (maximum 10 min) of samples in the thermal chamber were validated with the sensory panel, to avoid drying of the tissue.
Six samples were profiled by the DA method corresponding to the Control fillets from D1 (CD1), D5 (CD5) and D7 (CD7) and Treated SUSEA fillets from D1 (SD1), D5 (SD5) and D7 (SD7).
2.3.2. Descriptive Analysis and Profiling of Cooked Products
All sensory training and evaluation sessions were conducted at the AU-FOOD iSense lab facilities (adhering to ISO standards ISO 8589:2007). A screened and trained panel (according to ISO 8586:2012) consisting of 10 assessors (5 male and 5 female, age: 23–48) with previous experience in the sensory evaluation of food and fish products in specific was employed. Due to COVID protocols the panel was divided during the whole process in morning and afternoon shifts to adhere to social distancing protocols established at AU-FOOD.
The DA process was conducted within two weeks and occupied a total of seven days. This included: (i) three 2 h generic training sessions placed on different days, (ii) one vocabulary generation and two training sessions with the Control and Treated samples accounting to a total of 5 h, placed on three separate days and (iii) three evaluation sessions accounting to 3 h and placed on two days.
The three generic training sessions were dedicated to training and reference development for attributes found in fish products. A variety of different products were presented to the panel as references, including fish fillets, seafood products and algae products as well as other relevant references (e.g., boiled potatoes, mushrooms, lactic acid). The aim of these sessions was to clarify the vocabulary and achieve alignment and consensus amongst the panel.
The vocabulary session involved the tasting of extreme samples (CD1, SD1, CD7 and SD7). During this session, assessors were asked to generate terms describing all the characteristics of appearance, odor, taste, flavor, texture, aftertaste and mouthfeel (after swallowing) perceived in the samples. This resulted to an initial vocabulary list containing 48 attributes. The remaining training involved three 1 h sessions: (a) sample pair comparisons, (b) booth sample training, including two blind duplicates to evaluate repeatability, and (c) sample pair comparisons. For booth sample training all generated attributes in the initial vocabulary list were included and evaluated for their intensity on a 150 mm scale, anchored at 0 and 150 mm with “None” to “Very high”, respectively (Exceptions for specific attributes can be found in
Table 2). The booth training was conducted on tablets using the Compusense20 software. After this process the initial attribute list was reduced to 37 attributes, which were used in the final DA evaluation and can be found along with their definitions and references in
Table 2. Panel Check V1.4.2 was used to provide panel feedback for improving the assessors’ performance throughout the training process. Prior to the initiation of the DA evaluation sessions, specific references were presented to the assessors to ensure clarification of unclear attributes.
The DA evaluation was conducted at the AU-FOOD iSense lab sensory booths in three sessions over two days. Data were registered on tablets using the Compusense20 software. The assessors evaluated three replications of each sample following a block randomised design (one replication per session). Samples were kept in a thermal chamber (60 °C) and time between culinary preparation and serving did not exceed 10 min. They were served in aluminum foil-covered white ceramic containers, blind coded with 3-digit numbers, and in randomized (William Squared Design) monadic order, accounting for first order and carry over effects [
34]. The attributes line scale and anchors used were the same as in booth training described above. Modalities were evaluated in the following order: odor for which assessors were instructed to lift the aluminum (and place it back after each sniffing), appearance, taste, flavor, texture and after-taste/-mouthfeel. Aftertaste and mouthfeel were evaluated 20 sec after sample swallowing and a break of 1 min was included in-between samples. A combination of two palate cleansers (green apple, cucumber, sparkling water and crackers) were used and assessors had access to tap water.
2.3.3. Statistical Analyses
A 2-way ANOVA model with interaction (factors: treatment, day, treatment × day) was performed to identify significant (p < 0.05) effects of the design factors on the perception of individual attributes. A sequential elimination of insignificant (p < 0.1) interaction effects was performed. Fisher’s LSD was used as a post-hoc test to identify sample groupings according to the design factors. Attributes showing a tendency of significance were considered in combination with the results of the post-hoc test, due to the inherent inter fillet variability.
Principal Component Analysis (PCA) was furthermore employed for the visualization of the sensory maps. Two PCA (co-variance) analyses were conducted; one including the design factors of the experiment as observations and the samples as supplementary observations, and one including only the samples as observations. In all cases, only attributes that varied significantly or showed a tendency of variation (
p < 0.1) according to the 2-way factorial ANOVA model were included in the calculation of the PCA models as variables. The PCA bi-plot (after Varimax rotation) with the design factors of the experiment as observations is included in the main body of the results; the PCA bi-plot with the samples as observations is included in the Supplementary material. In the latter, Convex hulls were calculated for the samples according to the partial bootstrap method [
35] based on the most extreme observations.
Agglomerative Hierarchical Clustering (AHC) was furthermore used to classify the samples according to their similarity and dissimilarity based on the raw sensory data. AHC on similarity was calculated using Pearson’s correlation coefficient, as agglomeration method, complete linkage and as truncation entropy. AHC on dissimilarity was calculated using Euclidian distance, as agglomeration method, Wards method and as truncation entropy.
All statistical analyses (ANOVA, PCA, AHC) were performed in XLSTAT® software, 2016 (Addinsoft™, New York, NY, USA), whereas Panel Check V1.4.2 was used to assess the assessor performance and data obtained from the DA evaluation prior to the initiation of statistical analyses.
5. Conclusions and Directions for Future Research
The treatment achieved a shelf-life extension of 29%, as validated by both the Quality Index Method and microbial evaluation of gilthead seabream fillets. No deviations in the raw quality of the fillets were caused by the treatment in the beginning of shelf life as indicated by the results of the discrimination test. The cooked quality of the fillets profiled by sensory descriptive analysis varied with the treatment, exhibiting a slower development (lower intensities) for sensory attributes related to spoilage (bitter and sour taste and aftertaste, lactic acid and metallic—mussels flavor) for samples in the middle (D5) and end (D7) of shelf life. Moreover, a range of attributes were perceived as having higher perceived intensities in the Treated fillets when compared to the Control ones’, either independently of, or consistently during, shelf life.
The results of the treatment are promising for the shelf-life extension of gilthead seabream fillets, however the authors argue that a more prolonged extension could have been achieved given a lower initial microbial load of the raw material (fillets) for processing. Furthermore, whereas the sensory attributes enhanced by the treatment in the cooked tissue can be evaluated in total as positive, further work examining consumers’ preferences for the Treated fillets and corresponding attributes versus the Control ones are needed for validation.