The Shelf Life of Farmed Turbot (Scophthalmus maximus) Bjorn Roth, Lene Kramer, Aase Vorre Skuland, Trond Løvdal, Sigurd Øines, Atle Foss, and Albert Kjartansson Imsland
Abstract: A total of 18 farmed turbot (Scophthalmus maximus) were slaughtered over 4 successive weeks in November 2012 and stored in polystyrene boxes with ice until analyzed. The fish were stored between 1 and 22 d and presented to a taste panel and further analyzed for quality index method (QIM), microbiological analysis by real-time quantitative PCR (qPCR), taste, pH, color by computer imaging, protein denaturation with differential scanner calorimeter (DSC), texture hardness, and shear force. Results show small, but significant changes in physical and visual attributes such as texture and color. No gaping was observed. Only small changes in texture were observed explained by lack of myosin denaturation. The fillets became more white and yellow during storage, whereas the major changes occurred during the 1st week. A panel evaluating QIM and taste could not distinguish major differences in appearance and taste and over 15 d storage period, but were able to quantify the age by smell. Analysis of microorganisms on the epidermis displayed growth of Carnobacterium maltaromaticum, potentially inhibiting growth of other spoilage bacteria. Fish stored for 22 d were rejected by the taste panel caused by a stale smell and taste, but not bitter or rancid. It is concluded that turbot has a shelf life of at least 16 d. Keywords: fish, proteolysis, sensory analysis, shelf life, texture
This research gives a better understanding on why flatfish do have a long shelf life and the spoilage criteria involved. This understanding can be used to improve the shelf life of turbot even more, but more importantly increase the market potentials for this and other fish species.
Practical Application:
Introduction
S: Sensory & Food Quality
Farmed turbot (Scophthalmus maximus) is a fish species that has been reported to have an exceptional long shelf life as compared to other fish species lasting over 22 d (Rodriguez and others 2006). Turbot is regarded as an intermediate fat flatfish species, containing approximately 12% fat, with 1.5% to 2% of the fat is distributed within the muscle tissue and the rest is stored as rims along the fillet. During ice storage, small changes are observed for turbot in total volatile basic nitrogen (TVB-N) and free fatty acids (FFAs), whereas TMA-N and K-values will increase (Pineiro and others 2005; Rodriguez and others 2006). The texture of turbot is characterized as firm, remaining relatively stable during the entire postmortem storage, with no changes of the actin and myosin breakdown over 3 wk storage occurring (Abugoch and others 2011). Similarly, stress has minor effects on texture hardness and shear force regardless of a fast pH drop and earlier onset of rigor mortis (Morzel and others 2003; Roth and others 2007; Knowles and others 2008). The color of the fillets is reported to display minor changes, only displaying a shift from white to yellow during storage (Ruff and others 2002; Santos and others 2013). Despite the fact that turbot has been reported to have a long shelf life ranging everything from 12 to 22 d (Ozogul and others MS 20131884 Submitted 12/18/2013, Accepted 5/2/2014. Authors Roth, Kramer, Skuland, Løvdal, and Øines are with Nofima, Dept. of Processing Technology, P.O. Box 8034, N-4068 Stavanger, Norway. Author Kramer is also with Inst. of Biochemistry, Univ. of Stavanger, Stavanger, Norway. Author Foss is with Akvaplan niva Bergen, Thormøhlensgate 53 D, N-5006 Bergen, Norway. Author Imsland is with Akvaplan niva, Iceland Office, Akralind 4, 201, Kopavogur. Iceland: and also with Dept. of Biology, Univ. of Bergen, High Technology Centre, 5020, Bergen, Norway. Direct inquiries to author Roth (E-mail: [email protected]).
S1568
Journal of Food Science r Vol. 79, Nr. 8, 2014
2006; Rodriguez and others 2006; Nuin and others 2008; Santos and others 2013), the industry and the turbot market is operating with shelf life up to 12 d. One reason for this could be the discrepancies in storage temperature and sensory attributes in, for example, microbiological growth and smell. Only a few studies on the microflora development on the dermis of turbot exists, and even less on specific bacteria species that thrives, only is identified as Pseudomonas spp. as the dominating bacteria (Nuin and others 2008). The aim of this study was to test the shelf life of turbot combining physical, microbiological, and sensory analysis during a 3 wk storage period.
Material and Methods A total of 18 market sized farmed turbot (2.3 to 3.0 kg) were slaughtered over 4 successive weeks at the farm of Stolt Seafarm, Kvinesdal, Norway in November 2012 and stored in polystyrene boxes with ice until analyzed. The fish were slaughtered according to commercial procedures, that is, the fish were gill cut and placed into ice slurry, before gutted and stored into polystyrene boxes with approximately 7 kg of ice. In each box, a temperature logger was placed to log the temperature each 5 min during the entire storage period. The fish were then shipped to the laboratories of Nofima, Stavanger, Norway, and held in 0.5 to 1 °C cooling room for the entire storage until the experiment. At analyzing day, all fish that were stored for 1 (n = 5), 8 (n = 5), 15 (n = 5), and 22 (n = 3) d were removed from the boxes and swabbed for bacterial counts before they were evaluated using quality index method (QIM). The fish were then filleted into 2 fillets, for measuring muscle pH and presented for a taste panel. The next day the rest of the fish were filleted and analyzed R C 2014 Institute of Food Technologists
doi: 10.1111/1750-3841.12541 Further reproduction without permission is prohibited
for color, texture hardness/shear force, and protein denaturation Falcon tubes, and stored frozen (−80 °C) until use. The frozen samples were left to thaw overnight at 4 °C, and the material was (differential scanner calorimeter [DSC]). pelleted by centrifugation at 4500 × g for 20 min. The supernatant was removed, and the pellet resuspended in 400 μL TE buffer Muscle pH Muscle pH was measured according to Roth and others (2007), supplemented with 1.2% (v/v) Tween 80 (Sigma, St. Louis, Mo., in the anterior part of the dorsal fillet using a Mettler Toledo U.S.A.) and 20 mg Lysozyme/mL (Sigma). The whole content Seven Go ProTM with an Inlab 489 pH probe (Mettler Toledo was transferred to 1.5 mL Axygen microcentrifuge tubes (Corning B.V. Life Sciences, Amsterdam, The Netherlands), vortexed, INC, Greifensee, Switzerland). and incubated for 30 min at 37 °C with rotary agitation (225 rpm). Twenty-five μL of a 25 mg/mL proteinase K (Sigma) stock Texture analysis solution and 100 μL GeneMole lysis buffer from the GeneMole For texture measurements, both fillets from the eye side of the Bacteria Convenience kit (Mole Genetics AS, Lysaker, Norway) fish were removed and used to measure shear force (blade) and were added, and the tubes vortexed before incubation for 30 min R hardness (puncture test) with a TA-XT2 -Pro Texture Analyser at 56 °C with rotary agitation (225 rpm). Total genomic DNA was (Stable Micro Systems, Surrey, U.K.) with a 50 kg load cell (Roth extracted by transferring the content to FastPrep tubes containing and others 2010). For shear force measurements, a standard muscle Lysis matrix B (0.1 mm silica glass beads (MP Biomedicals, Solon, sample (70 × 30 mm) was sampled, starting 1 cm from the visceral Ohio, U.S.A.) supplemented with one ¼ inch ceramic sphere (MP cavity and along the loin from the blind side of the fish. The muscle Biomedicals). The samples were homogenized 3 rounds at 6.0 m/s samples were sliced with a 3 × 70 mm flat bedded blade having for 30 s in a FastPrep24 machine (MP Biomedicals). Samples were a 60° knife edge. The samples were cut at a constant speed at then centrifuged at 4 °C for 10 min at 17000 × g. DNA was ex0.8 mm/s (Sigurgisladottir and others 1999). To measure texture tracted from the supernatant using strips from the GeneMole Bachardness, a flat cylinder with a diameter of 20 mm was used as teria Convenience kit in a GeneMole instrument (Mole Genetics the test probe. The penetration depth for the probe was 80% of the fillet height holding a constant speed equal of 1 mm/s. The AS) according to the manufacturer’s instructions. The concentratexture profile was sampled directly on the epaxial fillet on the eye tion of isolated DNA was measured spectrophotometrically at 260 side at 2 locations. The breaking force was defined as the force nm, and quality assessed by the ratio A260 /A280 . Spectrophotometry was performed with an Implen NanoPhotometer version 1.6.1 required to penetrating the cylinder through the fillet surface. (AH Diagnostics, Aarhus, Denmark). DNA samples were diluted to 10 ng/μL in nuclease-free water and stored at −20 °C until Color measurements further use. Fillet color measured as CIE L∗ a∗ b∗ was by image analysis Quantitative real-time PCR (qPCR). q-PCR were assayed DigiEyeTM (VeriVide Ltd., Leicester, U.K.). Each fillet was placed in an ABI StepOnePlus thermocycler (Applied Biosystems, Foster into an illumination cabinet that ensures a uniform lighting, stan- City, Calif., U.S.A.). Specific TaqMan primers and probes for Phodard daylight (6400K) and photographed with a Nikon camera tobacterium phosphoreum, Carnobacterium maltaromaticum, Shewanella D80 with a Nikkor lens. The color of the whole fillet, except putrefaciens, Pseudomonas aeruginosa, and Lactococcus lactis includthe belly flaps was measured using DigiPix (VeriVide Ltd.) color ing all 3 subspecies lactis, cremoris, and hordniae, and SybrGreen measurement software. primers for Vibrio spp., were designed, synthesized, and premixed by PrimerDesign Ltd., Southampton, U.K. (Table 1). The primers Differential scanning calorimeter (DSC) for Lactobacillus/Leuconostoc spp. were adapted from Rinttil¨a and One of the methods used for characterization of protein de- others (2004). The PCR consisted of 10 μL 2 × PCR Precinaturation in fish is DSC. Sampling from the turbot was done at sion Mastermix (PrimerDesign Ltd.) with or without SybrGreen specific days by dissecting the same myotome on each fish, and according to the assay, 1 μL primer mix or primer/probe mix consequently storing it at −80 °C. The sample was thawed before (PrimerDesign Ltd.), 5 μL of DNA template, and nuclease free DSC was performed at a heating rate of 5 °C/min over the range water in a total volume of 20 μL. Thermocycling was performed from 0 to 110 °C on a Mettler Toledo DSC1/200 (Star System, according to the specifications for the kits using the following Schwerzenbach, Switzerland). Crucibles of the size 7 mm were conditions: 10 min at 95 °C followed by 40 cycles of 10 s at used as sample containers for weighted samples of about 60 mg ± 95 °C and 60 s at 60 °C. Melting curve analysis (60 to 95 ˚C) 1. A leveler was used to close the crucible airtight. An empty pan was performed for SybrGreen assays. Real-time PCR results were containing H2 O at equal weight as a sample was used as reference recorded as the mean of 2 analytical replicates analyzed for each and 2 min equilibration at 0 °C was done before each run. The sample against respective copy no standard curves (PrimerDesign instrument was calibrated for temperature and heat transfer using Ltd.) with dynamic ranges from Log 1 to Log 8 supplied with each indium (Mettler Toledo, Schwerzenbach) as standard. After each kit. Negative controls were included in each run. scan a rescan was performed with the same heating rate as the sample. The residual denaturation enthalpy (H) was defined as Sensory analysis the area under the denaturation peak using a spline baseline. The The sensory analysis was performed in 2 steps: (i) QIM on raw specific enthalpy was found by dividing the residual enthalpy by whole turbot and (ii) quantitative descriptive analysis (QDA) on the weight of the sample. cooked turbot fillets. The sensory panel consisted of 6 persons trained according to ISO 8586 (ISO 2012) including selection, Microflora detection, and recognition of characteristic attributes of farmed Sampling and DNA isolation. Approximately 1 cm2 of the turbot. For QIM, 10 attributes were measured at a scale from fish surface was swabbed with sterile Q-tips swabs from 2 fish at 0 to 3 (Eurofish 2001, QIM Eurofins), while for QDA, 16 each day. The swabbed material was suspended in 5 mL Tris-HCl attributes were defined and described for cooked turbot fillets EDTA buffer (TE; 10 mM Tris-HCl, 1 mM EDTA) in 15 mL using a 9-point scale from 1 (low intensity) to 9 (high intensity). Vol. 79, Nr. 8, 2014 r Journal of Food Science S1569
S: Sensory & Food Quality
Shelf life of farmed turbot . . .
Rinttila and others (2004) NA
Commercially available (PrimerDesign) This study
This study
This study
Lactobacillus spp.
RNA polymeraseα-subunit (AJ842676) 16S rDNA (AB859011) Vibrio spp.
RegA (NA) P. aeruginosa
FerE (AF188713)
LuxA (AY345888) P. phosphoreum
S. putrefaciens
16S rDNA (HM241921) L. lactis subsp.
Reference
This study
ACCAAGCCCAACCTGAACCAAAACGAGT TTGCTTAGGTTTATGACTAAACCCCTCAAAACT AACGTTGATCACAGTATGACCCTCATCTGTT AGCCGCAAGCAATACTGAAATCACACCA Trade secret NA
GCTAATTTGCCACCATCTAAAACT GGCCCCACAATCAAGAATTTG AAACTCACGGCAYACATCTTCAG AGTGGTAGCGGGTAACTCTG Trade secret GTCATGGGTGATGTCACCTGC CACCGCTACACATGGAG GTGAGGTTAAAGACGCTATTGAC ATACAAAGACGTGAGCATTCAAC TAGAGATAGYGGTTACAGTGAAGAG CAGGCAATGATTTATCCGATAGTG Trade secret GCTGAAGGCAAAGATGAAGTGTT AGCAGTAGGGAATCTTCCA SodA (AM490329)
Assay
C. maltaromaticum
Probe (sequence 5 -3 ) Reverse primer (sequence 5 -3 ) Target (accession no)
Forward primer (sequence 5 -3 )
S: Sensory & Food Quality Table 1–Primer assays for qPCR.
This study
Shelf life of farmed turbot . . .
S1570 Journal of Food Science r Vol. 79, Nr. 8, 2014
Table 2–Average muscle pH and weight of gutted turbot stored 1–22 d at −0.7 ± 0.1 °C. Significant differences are represented by ∗ P < 0.05 and ∗∗ P < 0.005 from piecewise regression on pH and one-way ANOVA on weight. In each column, different letters a and b represent significant difference of ∗ P < 0.05 and ∗∗ P < 0.005. Muscle pH Days postmortem
Mean
1 8 15 22
6.54a∗ 6.33b 6.45a,b 6.60a∗
∗
Whole weight (kg)
S.E.
Mean
S.E.
n
0.029 0.065 0.014 0.062
2.82a
0.053 0.072 0.037 0.062
5 5 5 3
2.48b∗∗ 2.57b∗ 2.36b∗∗
∗∗
The sensory attributes used in the QDA were based on a corresponding score sheet for cooked turbot fillets (Psetta maxima) (Regost and others 2001). Sensorial parameters of the filets were odor, appearance including discoloration, flavor, and texture. The samples for the QDA were assessed in duplicate, while the QIM was conducted using all specimens. Sample presentation order was randomized between the panel lists and coded with a 3-digit code so that the identities of the samples were unknown to the panel. Before QDA, the fillets were cut in pieces of 2 × 4 cm. Each panellist got the same definite part of the filet from all samples throughout the evaluation. The samples were vacuum-packed (99%) in sous vide plastic pouches (PA/PE 80 μm from Lietpac, Vilniaus, Lithuania) and cooked in a convection oven (Rational, SelfCooking Center from Rational, Landsberg, Germany) at 80 °C for 5 min. The samples were presented to the panel immediately after cooking. A computerized system EyeQuestion Software version 3.5 (Logic8 BV, Wageningen, the Netherlands) was used for data recording. Average score of attributes was calculated.
Statistics All statistical analyses were performed in STATISTICA 12.0 (Statsoft, Inc., Tulsa, Okla., U.S.A.). To test the single variables such as color and calorimetric DSC against the continuous independent variable “time,” linear regression was used as the statistical model. To obtain linearity, square root transformation was used. For the nonlinear distributed variable pH against time, piecewise linear regression with breakpoint was used. Wald Chi square was used to test nonlinear estimation of the microbial load. ANCOVA was used to test changes on texture hardness and shear force against time using height as a sample height as a covariate. One-way ANOVA was used to test difference between weight, QIM, and results from the taste panel. Homogeneity of variances was evaluated by Levine’s as homogeneity of variance.
Results There was a significant difference in size (P < 0.005, oneway ANOVA, Table 2) during the analyzing period, where the fish analyzed on the last day was significant larger than all other groups (P < 0.05; Tukey U). The muscle pH significantly declined over the 1st week, reaching the breaking point at pH 6.34 before gradually increasing over the next 2 wk (P < 0.05, R = 0.60, piecewise linear regression).
Texture measurements In the puncture test (Table 3), no significant differences were observed in breaking force (P > 0.86, GLM) and at 80% compression (P > 0.89, ANCOVA). There was a significant drop hardness
Microflora There was an exponential growth (P < 0.0005, Wald chi square, Figure 1) of C. maltaromaticum during the storage period with a log-3 increase over the last week. The P. phosphoreum was below the detection limit (0.5 log/cm2 ) until the final sampling at day 23, and neither S. putrefaciens, nor P. aeruginosa were observed throughout the storage period. However, Vibro spp. remained stable during the entire storage period (P > 0.63, Wald chi square); hence, it is likely to constitute a natural part of the flora on the dermis and mucus. Lactococcus lactis subsp. also remained stable throughout the storage period (P > 0.55, Wald chi square), however, in low numbers. Subspecies of the spoilage bacteria related to Lactobacillus were absent at slaughter, but did occur after 1 wk of storage. The numbers remained constant during the entire storage period, likely due to the inhibiting competition from C. maltaromaticum.
∗
-2
ns 72.4 74.9 81.1 82.7
Mean SE
DSC No significant change of myosin denaturation was observed during the postmortem storage (P > 0.75, linear regression, Table 5), where the peak temperature for denaturation remained stable (P > 0.16, linear regression). Actin did, however, denature at a slow rate during postmortem storage (P < 0.05, R = 0.51, linear regression, Table 5), where the peak temperature decreased over time (P < 0.005, R = 0.80, linear regression).
Vibrio spp. Lactobacillus spp. L. lactis subsp. C. maltaromaticum P. phosphoreum
5
0.46 0.42 0.38 0.31 18.2b 18.3b 18.4b
20.5a
4 3 2 1 0 1
8
15
22
Day
2 9 16 23
Mean Days postmortem
Color measurements During storage, there was a significant increase of lightness (P < 0.0005, R = 0.88, linear regression, Table 4) and yellowness (P < 0.0005, R = 0.77, linear regression, Table 4), while no changes were observed in redness (P > 0.25, linear regression, Table 4).
Log bacteria cm
117.9 134.2 136.7 139.4 9.80 11.21 4.87 4.25
ns Mean SE
Breaking force (N) Sample height (mm)
measured at 60% compression (P < 0.05, ANCOVA), depending on the fillet thickness (P < 0.05, ANCOVA). There was a significant decrease in both breaking force (P < 0.005, GLM) and shear force (P < 0.05, ANCOVA) during storage. The shear force was also dependent on the sample height (P < 0.05, ANCOVA).
Figure 1–Cell numbers estimated by qPCR for skin microflora of gutted turbot stored 1 to 22 d at −0.7 ± 0.1 °C. Swab samples were taken from 2 fish at each sampling date and processed prior to q-PCR. The error bars represent standard deviation (SD) of 2 biological replicates run in parallel. The dashed line denotes the detection limit of log 0.5 bacteria/cm2 . Vol. 79, Nr. 8, 2014 r Journal of Food Science S1571
S: Sensory & Food Quality
86.6a,b 75.8b 83.8a,b
5.11 8.42 10.07 8.12
Hardness 80% (N)
SE
∗∗
4.84 4.67 4.33 6.48 94.1a
ns
0.49 0.36 0.45 0.33 6.4 5.1 4.4 5.6
73.1b 57.4a 48.3a
∗∗
85.8a 68.9a 68.3a
∗
3.74 2.96 5.28 4.36
10 10 10 6 4.64 3.91 6.93 4.96 70.9a
n SE Mean SE
61.3a
Mean SE Mean SE Mean
Shear force (N) Breaking force (N) Hardness 60% (N)
Sample height (mm)
Shear test Puncture test
Table 3–Mean (SE) texture hardness and shear force (N) from gutted turbot stored from 2 to 23 d at −0.7± 0.1 °C. Significant differences from ANCOVA are represented by ∗ P < 0.05 and ∗∗ P < 0.005 and ns = not significant. In each column, different superscript letter (a, b) represents a significant difference of P < 0.05 using post hoc test.
Shelf life of farmed turbot . . .
Shelf life of farmed turbot . . . Table 4–Mean (SE) values from color measurements as CIE L∗ a∗ b∗ from turbot stored from 2 to 23 d at −0.7 ± 0.1 °C. Significant differences using linear regression are represented by ∗∗∗ P < 0.0005 and ns = not significant. In each column, different superscript letter (a,b,c) represents a significant difference of P < 0.05 using post hoc test. Lightness (L∗ ) Days postmortem
Mean
2 9 16 23
75.2a 81.1b 83.8b,c 84.7c
Redness (a∗ )
∗∗∗
SE
Mean
0.89 0.51 0.53 0.90
−0.2 0.7 0.8 0.7
Yellowness (b∗ ) SE
Mean
SE
n
0.76 0.31 0.27 0.98
10.6a
0.61 0.40 0.44 0.10
5 5 5 3
11.5a,b 12.3a,b 14.0b
∗∗∗
ns
Table 5–Mean (SE) protein denaturation of actin and myosin for in turbot stored for 2 to 23 d, measured by enthalpy (H) and temperature (°C) from a differential scanning calorimetry. Significant differences by linear regression are represented by ∗ P < 0.05, ∗∗ P < 0.005, and ns = not significant. In each column, different superscript letters (a,b) represent a significant difference of P < 0.05 using post hoc test. Myosin H Days postmortem 2 9 16 23
Mean
SE
0.291 0.294 0.299 0.304
Actin H
°C
0.0304 0.0516 0.0290 0.0039
Mean
SE
48.42 48.29 48.56 49.37
ns
0.349 0.662 0.275 0.248 ns
S: Sensory & Food Quality
Sensory analysis The total QIM score kept increasing during the entire storage period (P < 0.0005, linear regression, Figure 2). The sensory panel was able to distinguish differences between newly slaughtered and stored fish (P < 0.05, ANOVA) in all attributes. However, the panel was not able to quantify differences between 1 and 3 wk old fish in all those attributes (P > 0.05, Tukey U). On cooked material (Table 6), no significant differences were detected between 1 d and 1 wk old fish, neither in smell or taste (P > 0.05, one-way ANOVA). After 2 wk of storage, the panel was able to detect a slight difference in texture hardness and elasticity, but no other changes were detected (P > 0.05, one-way ANOVA). After 3 wk storage, 2 of the judges rejected the blind material without knowing the background, and the remaining 3 judges scored negative (P < 0.005, post hoc), except on color. The panel reported a stale smell and taste, rather than rancid taste indicating a bacterial off-flavor, rather than oxidation. Also, a slight bitter aftertaste was reported after 3 wk of storage.
Discussion Compared to other farmed fish, turbot and other flatfish generally have a remarkable capacity for a long shelf life (Huss 1995). Considering the changes in postmortem softening, color, and QIM that occur over 3 wk gives an understanding why this species has been reported with shelf life lasting over 22 d before rejection (Rodriguez and others 2006). Despite the fact that the physical quality of the flesh in turbot did not change very much, the changing smell and flavor of the flesh resulted in rejection in a taste panel at 22 d postmortem (Table 6). The reasons for inconsistence between the physical quality of the flesh and sensory attributes are many. Like most other fish species, turbot displays a reverse modal change of pH followed by increased drip loss containing soluble proteins (Rodriguez and others 2006; Santos and others 2013). For turbot, results indicate that proteolysis is of less importance in the postmortem degradation (Carrera and others 2008). In correspondence with Abugoch and others (2011), actin was the S1572 Journal of Food Science r Vol. 79, Nr. 8, 2014
Mean 0.145a 0.159a 0.119a,b 0.126b
∗
°C SE
Mean
SE
n
0.0099 0.0083 0.0046 0.0123
76.99a 76.47a
0.103 0.209 0.247 0.254
8 6 8 6
76.42a 75.45a
∗∗
main muscle protein that was denaturated at such a slow rate that changes could only be observed on shear force. The absence of gaping combined with small changes of textural properties makes it difficult to predict the age of the meat by QIM only. Cooking the meat, however, and presenting it to a taste panel show that hardness decreased with time along with elasticity, but still in the category where the meat can be defined as firm and juicy even after 23 d (Figure 3). Rejection of the turbot fillets was based on stale smell and taste only. Previous reports on ice chilled turbot demonstrate that TVB-N values remain almost stable during storage, whereas the TMA values increase (Pinereo and others 2005; Rodriguez and others 2006). As with QIM, smell was the main cause for determining the shelf life followed by taste (Figure 2 and Table 6). Since the yellowness of the fillet in this study remained more or less the same, followed by the absence of a rancid flavor, one can expect lipid oxidation not to be the main cause for shortening the shelf life. As demonstrated by Rodriguez and others (2006), TMA-N levels in turbot appeared after 21 d of ice storage. Whether TMA levels are the reason for rejection is unclear, but this can partially be explained by the bacterial flora. The formation of TMA is often caused by bacteria, especially Lactobacillus, which was not in this case, the dominating spoilage bacteria. The question arises whether the shelf life is solely dependent on the microflora rather than lipid oxidation. Rodriguez and others (2006) showed that there might be an interaction between them as they observed an increase of TMA formation along with bacterial growth. Apparently, the dominating bacterium in our study was C. maltaromaticum, but it is unclear whether this is a bacterium that thrives in mucus of flatfish or not. As shown in Figure 1, C. maltaromaticum was a part of the natural flora of the skin. The spoilage potential of C. maltaromaticum is reported to be very strain-dependent (Laursen and others 2006), and the understanding of this bacteria and possible impact on the shelf life is still poorly understood (Leroi 2010). However, it is well known that strains of C. maltaromaticum are capable of bacteriocin production under specific environmental
Shelf life of farmed turbot . . . Table 6–Sensory score of cooked turbot stored for 1, 8, 15, and 22 d at −0.7 ± 0.1 °C. In each row, a difference of superscript letters (a, b) represents a significant difference of P < 0.05. Days postmortem Attributes
Odor
Appearance
Flavor
Texture
Fresh Stale Rancid Off-odor Denatured protein Color intensity Discoloration Fresh Stale Rancid Off-flavor Bitter aftertaste Hardness Elasticity Juiciness Stickiness
1
8
15
22
Mean
SD
Mean
SD
Mean
SD
Mean
SD
ANOVA (P-value)
6.40a 1.20a 1.10 1.00 1.50 7.30 1.10 6.00a 1.50a 1.00 1.00 1.60 6.10ab 5.50ab 5.40 4.40
1.35 0.63 0.32 0.00 0.71 1.06 0.32 2.45 0.85 0.00 0.00 0.70 1.60 0.97 1.08 1.27
6.50a 1.30a 1.00 1.00 2.20 7.50 1.10 6.30a 1.80a 1.00 1.00 1.50 6.30a 5.60a 5.20 4.20
1.78 0.68 0.00 0.00 0.63 0.97 0.32 2.31 2.20 0.00 0.00 0.53 0.82 0.70 0.79 1.48
5.90a 1.70a 1.00 1.00 1.90 7.60 1.00 6.30a 1.70a 1.00 1.00 1.70 4.90ab 4.70ab 5.40 4.30
2.38 1.34 0.00 0.00 0.74 0.70 0.00 1.95 1.25 0.00 0.00 0.82 1.20 0.68 0.97 1.77
2.50b 5.40b 1.00 1.00 2.10 6.90 1.10 2.30b 5.40b 1.00 2.00 1.50 4.60b 4.40b 4.40 3.10
1.65 2.91 0.00 0.00 0.88 1.10 0.32 1.64 2.84 0.00 2.11 0.71 1.58 1.26 1.35 2.23
Abstract: A total of 18 farmed turbot (Scophthalmus maximus) were slaughtered over 4 successive weeks in November 2012 and stored in polystyrene boxes with ice until analyzed. The fish were stored between 1 and 22 d and presented to a taste panel and further analyzed for quality index method (QIM), microbiological analysis by real-time quantitative PCR (qPCR), taste, pH, color by computer imaging, protein denaturation with differential scanner calorimeter (DSC), texture hardness, and shear force. Results show small, but significant changes in physical and visual attributes such as texture and color. No gaping was observed. Only small changes in texture were observed explained by lack of myosin denaturation. The fillets became more white and yellow during storage, whereas the major changes occurred during the 1st week. A panel evaluating QIM and taste could not distinguish major differences in appearance and taste and over 15 d storage period, but were able to quantify the age by smell. Analysis of microorganisms on the epidermis displayed growth of Carnobacterium maltaromaticum, potentially inhibiting growth of other spoilage bacteria. Fish stored for 22 d were rejected by the taste panel caused by a stale smell and taste, but not bitter or rancid. It is concluded that turbot has a shelf life of at least 16 d. Keywords: fish, proteolysis, sensory analysis, shelf life, texture
This research gives a better understanding on why flatfish do have a long shelf life and the spoilage criteria involved. This understanding can be used to improve the shelf life of turbot even more, but more importantly increase the market potentials for this and other fish species.
Practical Application:
Introduction
S: Sensory & Food Quality
Farmed turbot (Scophthalmus maximus) is a fish species that has been reported to have an exceptional long shelf life as compared to other fish species lasting over 22 d (Rodriguez and others 2006). Turbot is regarded as an intermediate fat flatfish species, containing approximately 12% fat, with 1.5% to 2% of the fat is distributed within the muscle tissue and the rest is stored as rims along the fillet. During ice storage, small changes are observed for turbot in total volatile basic nitrogen (TVB-N) and free fatty acids (FFAs), whereas TMA-N and K-values will increase (Pineiro and others 2005; Rodriguez and others 2006). The texture of turbot is characterized as firm, remaining relatively stable during the entire postmortem storage, with no changes of the actin and myosin breakdown over 3 wk storage occurring (Abugoch and others 2011). Similarly, stress has minor effects on texture hardness and shear force regardless of a fast pH drop and earlier onset of rigor mortis (Morzel and others 2003; Roth and others 2007; Knowles and others 2008). The color of the fillets is reported to display minor changes, only displaying a shift from white to yellow during storage (Ruff and others 2002; Santos and others 2013). Despite the fact that turbot has been reported to have a long shelf life ranging everything from 12 to 22 d (Ozogul and others MS 20131884 Submitted 12/18/2013, Accepted 5/2/2014. Authors Roth, Kramer, Skuland, Løvdal, and Øines are with Nofima, Dept. of Processing Technology, P.O. Box 8034, N-4068 Stavanger, Norway. Author Kramer is also with Inst. of Biochemistry, Univ. of Stavanger, Stavanger, Norway. Author Foss is with Akvaplan niva Bergen, Thormøhlensgate 53 D, N-5006 Bergen, Norway. Author Imsland is with Akvaplan niva, Iceland Office, Akralind 4, 201, Kopavogur. Iceland: and also with Dept. of Biology, Univ. of Bergen, High Technology Centre, 5020, Bergen, Norway. Direct inquiries to author Roth (E-mail: [email protected]).
S1568
Journal of Food Science r Vol. 79, Nr. 8, 2014
2006; Rodriguez and others 2006; Nuin and others 2008; Santos and others 2013), the industry and the turbot market is operating with shelf life up to 12 d. One reason for this could be the discrepancies in storage temperature and sensory attributes in, for example, microbiological growth and smell. Only a few studies on the microflora development on the dermis of turbot exists, and even less on specific bacteria species that thrives, only is identified as Pseudomonas spp. as the dominating bacteria (Nuin and others 2008). The aim of this study was to test the shelf life of turbot combining physical, microbiological, and sensory analysis during a 3 wk storage period.
Material and Methods A total of 18 market sized farmed turbot (2.3 to 3.0 kg) were slaughtered over 4 successive weeks at the farm of Stolt Seafarm, Kvinesdal, Norway in November 2012 and stored in polystyrene boxes with ice until analyzed. The fish were slaughtered according to commercial procedures, that is, the fish were gill cut and placed into ice slurry, before gutted and stored into polystyrene boxes with approximately 7 kg of ice. In each box, a temperature logger was placed to log the temperature each 5 min during the entire storage period. The fish were then shipped to the laboratories of Nofima, Stavanger, Norway, and held in 0.5 to 1 °C cooling room for the entire storage until the experiment. At analyzing day, all fish that were stored for 1 (n = 5), 8 (n = 5), 15 (n = 5), and 22 (n = 3) d were removed from the boxes and swabbed for bacterial counts before they were evaluated using quality index method (QIM). The fish were then filleted into 2 fillets, for measuring muscle pH and presented for a taste panel. The next day the rest of the fish were filleted and analyzed R C 2014 Institute of Food Technologists
doi: 10.1111/1750-3841.12541 Further reproduction without permission is prohibited
for color, texture hardness/shear force, and protein denaturation Falcon tubes, and stored frozen (−80 °C) until use. The frozen samples were left to thaw overnight at 4 °C, and the material was (differential scanner calorimeter [DSC]). pelleted by centrifugation at 4500 × g for 20 min. The supernatant was removed, and the pellet resuspended in 400 μL TE buffer Muscle pH Muscle pH was measured according to Roth and others (2007), supplemented with 1.2% (v/v) Tween 80 (Sigma, St. Louis, Mo., in the anterior part of the dorsal fillet using a Mettler Toledo U.S.A.) and 20 mg Lysozyme/mL (Sigma). The whole content Seven Go ProTM with an Inlab 489 pH probe (Mettler Toledo was transferred to 1.5 mL Axygen microcentrifuge tubes (Corning B.V. Life Sciences, Amsterdam, The Netherlands), vortexed, INC, Greifensee, Switzerland). and incubated for 30 min at 37 °C with rotary agitation (225 rpm). Twenty-five μL of a 25 mg/mL proteinase K (Sigma) stock Texture analysis solution and 100 μL GeneMole lysis buffer from the GeneMole For texture measurements, both fillets from the eye side of the Bacteria Convenience kit (Mole Genetics AS, Lysaker, Norway) fish were removed and used to measure shear force (blade) and were added, and the tubes vortexed before incubation for 30 min R hardness (puncture test) with a TA-XT2 -Pro Texture Analyser at 56 °C with rotary agitation (225 rpm). Total genomic DNA was (Stable Micro Systems, Surrey, U.K.) with a 50 kg load cell (Roth extracted by transferring the content to FastPrep tubes containing and others 2010). For shear force measurements, a standard muscle Lysis matrix B (0.1 mm silica glass beads (MP Biomedicals, Solon, sample (70 × 30 mm) was sampled, starting 1 cm from the visceral Ohio, U.S.A.) supplemented with one ¼ inch ceramic sphere (MP cavity and along the loin from the blind side of the fish. The muscle Biomedicals). The samples were homogenized 3 rounds at 6.0 m/s samples were sliced with a 3 × 70 mm flat bedded blade having for 30 s in a FastPrep24 machine (MP Biomedicals). Samples were a 60° knife edge. The samples were cut at a constant speed at then centrifuged at 4 °C for 10 min at 17000 × g. DNA was ex0.8 mm/s (Sigurgisladottir and others 1999). To measure texture tracted from the supernatant using strips from the GeneMole Bachardness, a flat cylinder with a diameter of 20 mm was used as teria Convenience kit in a GeneMole instrument (Mole Genetics the test probe. The penetration depth for the probe was 80% of the fillet height holding a constant speed equal of 1 mm/s. The AS) according to the manufacturer’s instructions. The concentratexture profile was sampled directly on the epaxial fillet on the eye tion of isolated DNA was measured spectrophotometrically at 260 side at 2 locations. The breaking force was defined as the force nm, and quality assessed by the ratio A260 /A280 . Spectrophotometry was performed with an Implen NanoPhotometer version 1.6.1 required to penetrating the cylinder through the fillet surface. (AH Diagnostics, Aarhus, Denmark). DNA samples were diluted to 10 ng/μL in nuclease-free water and stored at −20 °C until Color measurements further use. Fillet color measured as CIE L∗ a∗ b∗ was by image analysis Quantitative real-time PCR (qPCR). q-PCR were assayed DigiEyeTM (VeriVide Ltd., Leicester, U.K.). Each fillet was placed in an ABI StepOnePlus thermocycler (Applied Biosystems, Foster into an illumination cabinet that ensures a uniform lighting, stan- City, Calif., U.S.A.). Specific TaqMan primers and probes for Phodard daylight (6400K) and photographed with a Nikon camera tobacterium phosphoreum, Carnobacterium maltaromaticum, Shewanella D80 with a Nikkor lens. The color of the whole fillet, except putrefaciens, Pseudomonas aeruginosa, and Lactococcus lactis includthe belly flaps was measured using DigiPix (VeriVide Ltd.) color ing all 3 subspecies lactis, cremoris, and hordniae, and SybrGreen measurement software. primers for Vibrio spp., were designed, synthesized, and premixed by PrimerDesign Ltd., Southampton, U.K. (Table 1). The primers Differential scanning calorimeter (DSC) for Lactobacillus/Leuconostoc spp. were adapted from Rinttil¨a and One of the methods used for characterization of protein de- others (2004). The PCR consisted of 10 μL 2 × PCR Precinaturation in fish is DSC. Sampling from the turbot was done at sion Mastermix (PrimerDesign Ltd.) with or without SybrGreen specific days by dissecting the same myotome on each fish, and according to the assay, 1 μL primer mix or primer/probe mix consequently storing it at −80 °C. The sample was thawed before (PrimerDesign Ltd.), 5 μL of DNA template, and nuclease free DSC was performed at a heating rate of 5 °C/min over the range water in a total volume of 20 μL. Thermocycling was performed from 0 to 110 °C on a Mettler Toledo DSC1/200 (Star System, according to the specifications for the kits using the following Schwerzenbach, Switzerland). Crucibles of the size 7 mm were conditions: 10 min at 95 °C followed by 40 cycles of 10 s at used as sample containers for weighted samples of about 60 mg ± 95 °C and 60 s at 60 °C. Melting curve analysis (60 to 95 ˚C) 1. A leveler was used to close the crucible airtight. An empty pan was performed for SybrGreen assays. Real-time PCR results were containing H2 O at equal weight as a sample was used as reference recorded as the mean of 2 analytical replicates analyzed for each and 2 min equilibration at 0 °C was done before each run. The sample against respective copy no standard curves (PrimerDesign instrument was calibrated for temperature and heat transfer using Ltd.) with dynamic ranges from Log 1 to Log 8 supplied with each indium (Mettler Toledo, Schwerzenbach) as standard. After each kit. Negative controls were included in each run. scan a rescan was performed with the same heating rate as the sample. The residual denaturation enthalpy (H) was defined as Sensory analysis the area under the denaturation peak using a spline baseline. The The sensory analysis was performed in 2 steps: (i) QIM on raw specific enthalpy was found by dividing the residual enthalpy by whole turbot and (ii) quantitative descriptive analysis (QDA) on the weight of the sample. cooked turbot fillets. The sensory panel consisted of 6 persons trained according to ISO 8586 (ISO 2012) including selection, Microflora detection, and recognition of characteristic attributes of farmed Sampling and DNA isolation. Approximately 1 cm2 of the turbot. For QIM, 10 attributes were measured at a scale from fish surface was swabbed with sterile Q-tips swabs from 2 fish at 0 to 3 (Eurofish 2001, QIM Eurofins), while for QDA, 16 each day. The swabbed material was suspended in 5 mL Tris-HCl attributes were defined and described for cooked turbot fillets EDTA buffer (TE; 10 mM Tris-HCl, 1 mM EDTA) in 15 mL using a 9-point scale from 1 (low intensity) to 9 (high intensity). Vol. 79, Nr. 8, 2014 r Journal of Food Science S1569
S: Sensory & Food Quality
Shelf life of farmed turbot . . .
Rinttila and others (2004) NA
Commercially available (PrimerDesign) This study
This study
This study
Lactobacillus spp.
RNA polymeraseα-subunit (AJ842676) 16S rDNA (AB859011) Vibrio spp.
RegA (NA) P. aeruginosa
FerE (AF188713)
LuxA (AY345888) P. phosphoreum
S. putrefaciens
16S rDNA (HM241921) L. lactis subsp.
Reference
This study
ACCAAGCCCAACCTGAACCAAAACGAGT TTGCTTAGGTTTATGACTAAACCCCTCAAAACT AACGTTGATCACAGTATGACCCTCATCTGTT AGCCGCAAGCAATACTGAAATCACACCA Trade secret NA
GCTAATTTGCCACCATCTAAAACT GGCCCCACAATCAAGAATTTG AAACTCACGGCAYACATCTTCAG AGTGGTAGCGGGTAACTCTG Trade secret GTCATGGGTGATGTCACCTGC CACCGCTACACATGGAG GTGAGGTTAAAGACGCTATTGAC ATACAAAGACGTGAGCATTCAAC TAGAGATAGYGGTTACAGTGAAGAG CAGGCAATGATTTATCCGATAGTG Trade secret GCTGAAGGCAAAGATGAAGTGTT AGCAGTAGGGAATCTTCCA SodA (AM490329)
Assay
C. maltaromaticum
Probe (sequence 5 -3 ) Reverse primer (sequence 5 -3 ) Target (accession no)
Forward primer (sequence 5 -3 )
S: Sensory & Food Quality Table 1–Primer assays for qPCR.
This study
Shelf life of farmed turbot . . .
S1570 Journal of Food Science r Vol. 79, Nr. 8, 2014
Table 2–Average muscle pH and weight of gutted turbot stored 1–22 d at −0.7 ± 0.1 °C. Significant differences are represented by ∗ P < 0.05 and ∗∗ P < 0.005 from piecewise regression on pH and one-way ANOVA on weight. In each column, different letters a and b represent significant difference of ∗ P < 0.05 and ∗∗ P < 0.005. Muscle pH Days postmortem
Mean
1 8 15 22
6.54a∗ 6.33b 6.45a,b 6.60a∗
∗
Whole weight (kg)
S.E.
Mean
S.E.
n
0.029 0.065 0.014 0.062
2.82a
0.053 0.072 0.037 0.062
5 5 5 3
2.48b∗∗ 2.57b∗ 2.36b∗∗
∗∗
The sensory attributes used in the QDA were based on a corresponding score sheet for cooked turbot fillets (Psetta maxima) (Regost and others 2001). Sensorial parameters of the filets were odor, appearance including discoloration, flavor, and texture. The samples for the QDA were assessed in duplicate, while the QIM was conducted using all specimens. Sample presentation order was randomized between the panel lists and coded with a 3-digit code so that the identities of the samples were unknown to the panel. Before QDA, the fillets were cut in pieces of 2 × 4 cm. Each panellist got the same definite part of the filet from all samples throughout the evaluation. The samples were vacuum-packed (99%) in sous vide plastic pouches (PA/PE 80 μm from Lietpac, Vilniaus, Lithuania) and cooked in a convection oven (Rational, SelfCooking Center from Rational, Landsberg, Germany) at 80 °C for 5 min. The samples were presented to the panel immediately after cooking. A computerized system EyeQuestion Software version 3.5 (Logic8 BV, Wageningen, the Netherlands) was used for data recording. Average score of attributes was calculated.
Statistics All statistical analyses were performed in STATISTICA 12.0 (Statsoft, Inc., Tulsa, Okla., U.S.A.). To test the single variables such as color and calorimetric DSC against the continuous independent variable “time,” linear regression was used as the statistical model. To obtain linearity, square root transformation was used. For the nonlinear distributed variable pH against time, piecewise linear regression with breakpoint was used. Wald Chi square was used to test nonlinear estimation of the microbial load. ANCOVA was used to test changes on texture hardness and shear force against time using height as a sample height as a covariate. One-way ANOVA was used to test difference between weight, QIM, and results from the taste panel. Homogeneity of variances was evaluated by Levine’s as homogeneity of variance.
Results There was a significant difference in size (P < 0.005, oneway ANOVA, Table 2) during the analyzing period, where the fish analyzed on the last day was significant larger than all other groups (P < 0.05; Tukey U). The muscle pH significantly declined over the 1st week, reaching the breaking point at pH 6.34 before gradually increasing over the next 2 wk (P < 0.05, R = 0.60, piecewise linear regression).
Texture measurements In the puncture test (Table 3), no significant differences were observed in breaking force (P > 0.86, GLM) and at 80% compression (P > 0.89, ANCOVA). There was a significant drop hardness
Microflora There was an exponential growth (P < 0.0005, Wald chi square, Figure 1) of C. maltaromaticum during the storage period with a log-3 increase over the last week. The P. phosphoreum was below the detection limit (0.5 log/cm2 ) until the final sampling at day 23, and neither S. putrefaciens, nor P. aeruginosa were observed throughout the storage period. However, Vibro spp. remained stable during the entire storage period (P > 0.63, Wald chi square); hence, it is likely to constitute a natural part of the flora on the dermis and mucus. Lactococcus lactis subsp. also remained stable throughout the storage period (P > 0.55, Wald chi square), however, in low numbers. Subspecies of the spoilage bacteria related to Lactobacillus were absent at slaughter, but did occur after 1 wk of storage. The numbers remained constant during the entire storage period, likely due to the inhibiting competition from C. maltaromaticum.
∗
-2
ns 72.4 74.9 81.1 82.7
Mean SE
DSC No significant change of myosin denaturation was observed during the postmortem storage (P > 0.75, linear regression, Table 5), where the peak temperature for denaturation remained stable (P > 0.16, linear regression). Actin did, however, denature at a slow rate during postmortem storage (P < 0.05, R = 0.51, linear regression, Table 5), where the peak temperature decreased over time (P < 0.005, R = 0.80, linear regression).
Vibrio spp. Lactobacillus spp. L. lactis subsp. C. maltaromaticum P. phosphoreum
5
0.46 0.42 0.38 0.31 18.2b 18.3b 18.4b
20.5a
4 3 2 1 0 1
8
15
22
Day
2 9 16 23
Mean Days postmortem
Color measurements During storage, there was a significant increase of lightness (P < 0.0005, R = 0.88, linear regression, Table 4) and yellowness (P < 0.0005, R = 0.77, linear regression, Table 4), while no changes were observed in redness (P > 0.25, linear regression, Table 4).
Log bacteria cm
117.9 134.2 136.7 139.4 9.80 11.21 4.87 4.25
ns Mean SE
Breaking force (N) Sample height (mm)
measured at 60% compression (P < 0.05, ANCOVA), depending on the fillet thickness (P < 0.05, ANCOVA). There was a significant decrease in both breaking force (P < 0.005, GLM) and shear force (P < 0.05, ANCOVA) during storage. The shear force was also dependent on the sample height (P < 0.05, ANCOVA).
Figure 1–Cell numbers estimated by qPCR for skin microflora of gutted turbot stored 1 to 22 d at −0.7 ± 0.1 °C. Swab samples were taken from 2 fish at each sampling date and processed prior to q-PCR. The error bars represent standard deviation (SD) of 2 biological replicates run in parallel. The dashed line denotes the detection limit of log 0.5 bacteria/cm2 . Vol. 79, Nr. 8, 2014 r Journal of Food Science S1571
S: Sensory & Food Quality
86.6a,b 75.8b 83.8a,b
5.11 8.42 10.07 8.12
Hardness 80% (N)
SE
∗∗
4.84 4.67 4.33 6.48 94.1a
ns
0.49 0.36 0.45 0.33 6.4 5.1 4.4 5.6
73.1b 57.4a 48.3a
∗∗
85.8a 68.9a 68.3a
∗
3.74 2.96 5.28 4.36
10 10 10 6 4.64 3.91 6.93 4.96 70.9a
n SE Mean SE
61.3a
Mean SE Mean SE Mean
Shear force (N) Breaking force (N) Hardness 60% (N)
Sample height (mm)
Shear test Puncture test
Table 3–Mean (SE) texture hardness and shear force (N) from gutted turbot stored from 2 to 23 d at −0.7± 0.1 °C. Significant differences from ANCOVA are represented by ∗ P < 0.05 and ∗∗ P < 0.005 and ns = not significant. In each column, different superscript letter (a, b) represents a significant difference of P < 0.05 using post hoc test.
Shelf life of farmed turbot . . .
Shelf life of farmed turbot . . . Table 4–Mean (SE) values from color measurements as CIE L∗ a∗ b∗ from turbot stored from 2 to 23 d at −0.7 ± 0.1 °C. Significant differences using linear regression are represented by ∗∗∗ P < 0.0005 and ns = not significant. In each column, different superscript letter (a,b,c) represents a significant difference of P < 0.05 using post hoc test. Lightness (L∗ ) Days postmortem
Mean
2 9 16 23
75.2a 81.1b 83.8b,c 84.7c
Redness (a∗ )
∗∗∗
SE
Mean
0.89 0.51 0.53 0.90
−0.2 0.7 0.8 0.7
Yellowness (b∗ ) SE
Mean
SE
n
0.76 0.31 0.27 0.98
10.6a
0.61 0.40 0.44 0.10
5 5 5 3
11.5a,b 12.3a,b 14.0b
∗∗∗
ns
Table 5–Mean (SE) protein denaturation of actin and myosin for in turbot stored for 2 to 23 d, measured by enthalpy (H) and temperature (°C) from a differential scanning calorimetry. Significant differences by linear regression are represented by ∗ P < 0.05, ∗∗ P < 0.005, and ns = not significant. In each column, different superscript letters (a,b) represent a significant difference of P < 0.05 using post hoc test. Myosin H Days postmortem 2 9 16 23
Mean
SE
0.291 0.294 0.299 0.304
Actin H
°C
0.0304 0.0516 0.0290 0.0039
Mean
SE
48.42 48.29 48.56 49.37
ns
0.349 0.662 0.275 0.248 ns
S: Sensory & Food Quality
Sensory analysis The total QIM score kept increasing during the entire storage period (P < 0.0005, linear regression, Figure 2). The sensory panel was able to distinguish differences between newly slaughtered and stored fish (P < 0.05, ANOVA) in all attributes. However, the panel was not able to quantify differences between 1 and 3 wk old fish in all those attributes (P > 0.05, Tukey U). On cooked material (Table 6), no significant differences were detected between 1 d and 1 wk old fish, neither in smell or taste (P > 0.05, one-way ANOVA). After 2 wk of storage, the panel was able to detect a slight difference in texture hardness and elasticity, but no other changes were detected (P > 0.05, one-way ANOVA). After 3 wk storage, 2 of the judges rejected the blind material without knowing the background, and the remaining 3 judges scored negative (P < 0.005, post hoc), except on color. The panel reported a stale smell and taste, rather than rancid taste indicating a bacterial off-flavor, rather than oxidation. Also, a slight bitter aftertaste was reported after 3 wk of storage.
Discussion Compared to other farmed fish, turbot and other flatfish generally have a remarkable capacity for a long shelf life (Huss 1995). Considering the changes in postmortem softening, color, and QIM that occur over 3 wk gives an understanding why this species has been reported with shelf life lasting over 22 d before rejection (Rodriguez and others 2006). Despite the fact that the physical quality of the flesh in turbot did not change very much, the changing smell and flavor of the flesh resulted in rejection in a taste panel at 22 d postmortem (Table 6). The reasons for inconsistence between the physical quality of the flesh and sensory attributes are many. Like most other fish species, turbot displays a reverse modal change of pH followed by increased drip loss containing soluble proteins (Rodriguez and others 2006; Santos and others 2013). For turbot, results indicate that proteolysis is of less importance in the postmortem degradation (Carrera and others 2008). In correspondence with Abugoch and others (2011), actin was the S1572 Journal of Food Science r Vol. 79, Nr. 8, 2014
Mean 0.145a 0.159a 0.119a,b 0.126b
∗
°C SE
Mean
SE
n
0.0099 0.0083 0.0046 0.0123
76.99a 76.47a
0.103 0.209 0.247 0.254
8 6 8 6
76.42a 75.45a
∗∗
main muscle protein that was denaturated at such a slow rate that changes could only be observed on shear force. The absence of gaping combined with small changes of textural properties makes it difficult to predict the age of the meat by QIM only. Cooking the meat, however, and presenting it to a taste panel show that hardness decreased with time along with elasticity, but still in the category where the meat can be defined as firm and juicy even after 23 d (Figure 3). Rejection of the turbot fillets was based on stale smell and taste only. Previous reports on ice chilled turbot demonstrate that TVB-N values remain almost stable during storage, whereas the TMA values increase (Pinereo and others 2005; Rodriguez and others 2006). As with QIM, smell was the main cause for determining the shelf life followed by taste (Figure 2 and Table 6). Since the yellowness of the fillet in this study remained more or less the same, followed by the absence of a rancid flavor, one can expect lipid oxidation not to be the main cause for shortening the shelf life. As demonstrated by Rodriguez and others (2006), TMA-N levels in turbot appeared after 21 d of ice storage. Whether TMA levels are the reason for rejection is unclear, but this can partially be explained by the bacterial flora. The formation of TMA is often caused by bacteria, especially Lactobacillus, which was not in this case, the dominating spoilage bacteria. The question arises whether the shelf life is solely dependent on the microflora rather than lipid oxidation. Rodriguez and others (2006) showed that there might be an interaction between them as they observed an increase of TMA formation along with bacterial growth. Apparently, the dominating bacterium in our study was C. maltaromaticum, but it is unclear whether this is a bacterium that thrives in mucus of flatfish or not. As shown in Figure 1, C. maltaromaticum was a part of the natural flora of the skin. The spoilage potential of C. maltaromaticum is reported to be very strain-dependent (Laursen and others 2006), and the understanding of this bacteria and possible impact on the shelf life is still poorly understood (Leroi 2010). However, it is well known that strains of C. maltaromaticum are capable of bacteriocin production under specific environmental
Shelf life of farmed turbot . . . Table 6–Sensory score of cooked turbot stored for 1, 8, 15, and 22 d at −0.7 ± 0.1 °C. In each row, a difference of superscript letters (a, b) represents a significant difference of P < 0.05. Days postmortem Attributes
Odor
Appearance
Flavor
Texture
Fresh Stale Rancid Off-odor Denatured protein Color intensity Discoloration Fresh Stale Rancid Off-flavor Bitter aftertaste Hardness Elasticity Juiciness Stickiness
1
8
15
22
Mean
SD
Mean
SD
Mean
SD
Mean
SD
ANOVA (P-value)
6.40a 1.20a 1.10 1.00 1.50 7.30 1.10 6.00a 1.50a 1.00 1.00 1.60 6.10ab 5.50ab 5.40 4.40
1.35 0.63 0.32 0.00 0.71 1.06 0.32 2.45 0.85 0.00 0.00 0.70 1.60 0.97 1.08 1.27
6.50a 1.30a 1.00 1.00 2.20 7.50 1.10 6.30a 1.80a 1.00 1.00 1.50 6.30a 5.60a 5.20 4.20
1.78 0.68 0.00 0.00 0.63 0.97 0.32 2.31 2.20 0.00 0.00 0.53 0.82 0.70 0.79 1.48
5.90a 1.70a 1.00 1.00 1.90 7.60 1.00 6.30a 1.70a 1.00 1.00 1.70 4.90ab 4.70ab 5.40 4.30
2.38 1.34 0.00 0.00 0.74 0.70 0.00 1.95 1.25 0.00 0.00 0.82 1.20 0.68 0.97 1.77
2.50b 5.40b 1.00 1.00 2.10 6.90 1.10 2.30b 5.40b 1.00 2.00 1.50 4.60b 4.40b 4.40 3.10
1.65 2.91 0.00 0.00 0.88 1.10 0.32 1.64 2.84 0.00 2.11 0.71 1.58 1.26 1.35 2.23
