Spoilage of king salmon oncorhynchus tsh

JFS: Sensory and Nutritive Qualities of Food Spoilage of King Salmon (Oncorhynchus tshawytscha) Fillets Stored Under Different Atmospheres G.C FLETCHER, G SUMMERS, V CORRIGAN, S CUMARASAMY, AND J.P DUFOUR ABSTRACT: King salmon (Oncorhynchus tshawytscha) packaged in air (AIR), nitrogen (N2), or 40:60 carbon dioxide:nitrogen (CO2N2) was stored (0 8C) for 18, 25, and 54 d, respectively Air packs (AIR9) were also stored at 8C for d A quality index (QI) method was developed to monitor sensory quality of cooked salmon First detection of spoilage was 1.5, 15, 15, and 21 d for AIR9, AIR, N2, and CO2N2 treatments respectively Total aerobic and sulfideproducing bacteria, pH, drip loss, Eh, color, texture, ATP derivatives, trimethylamine, total volatile base nitrogen, thiobarbituric acid numbers, and peroxide values were determined Only total aerobic counts and hypoxanthine were indicators of sensory deterioration across treatments and times Keywords: king salmon (Oncorhynchus tshawytscha), modified atmosphere packaging (MAP), quality index method, spoilage indicators Introduction Sensory and Nutritive Qualities of Food M OST OF N EW Z EALAND ’ S FINFISH CATCH IS SOLD OVERSEAS 2362 JOURNAL OF FOOD SCIENCE—Vol 67, Nr 6, 2002 (261108 t in 1999), and the bulk of this is exported frozen with only 4.6% of the catch exported chilled The market, however, pays a premium for chilled fish with returns being on average 2.4 times as much per kg for chilled fillets compared with frozen fillets ( The New Zealand Seafood Industry Council Ltd 2000) One of the principal factors limiting the export of chilled seafood is its short shelf life Frozen seafood can be held in storage until required by the markets, ensuring continuity of supply while chilled product must be marketed and sold soon after it becomes available New Zealand’s distance from its major markets means that currently all chilled seafood must be transported by air, which adds to production costs Various methods have been investigated for extending the shelf life of seafood One of the more widely accepted is modified atmosphere packaging (MAP) in which the ambient atmosphere around the product is replaced by one that is more inhibitory to the growth of spoilage bacteria Carbon dioxide (CO2), nitrogen (N2), and oxygen (O2), either alone or in combination are the gases most commonly used CO2 provides the bacteriostatic effect, oxygen is sometimes included to inhibit the growth of anaerobic bacteria while N2 is used as a filler to prevent packs collapsing when CO2 is absorbed by the product Numerous studies have been published on the effects of MAP on seafood, and shelf life extensions of 50 to 300% have been reported (Bremer and Fletcher 1999) However, most studies have been empirical, with results only useful for a specific application, and although some gas mixtures are commonly recommended for use with seafood (for example, Cann 1990) there is limited published information on work defining the optimum gas mixture or fish:gas ratio A key factor determining the shelf life of a packaged product is the quality of the raw material In a wild fishery there can be considerable variability in the condition of fish within a single catch so our aquaculture species (salmon) was selected for this work to increase the ability to have control of the raw material jfsv67n6p2362-2374ms20010445-AF-TO RRD.P65 2362 Most salmon aquaculture is based on the Atlantic salmon (Salmo salar), and there is a considerable body of literature on the spoilage of this species (for example, Blokhus 1987; Pastoriza and others 1996; Sigurgisladottir and others 1997; Randell and others 1999; Sivertsvik and others 1999) and of other wild-caught salmon and other Oncorhynchus species (for example, Brown and others 1980; Bilinski and others 1981; Bilinski and others 1984; Himelbloom and others 1994; Crapo and Himelbloom 1999; Rodriguez and others 1999) However, the New Zealand salmon industry is entirely based on king (chinook) salmon (Oncorhynchus tshawytscha) and there has been very little research published on the spoilage of this species (Fey and Regenstein 1981) A range of methods has been used for sensory evaluation of salmon, with most involving an evaluation by trained panelists of the odor, flavor, and texture of the cooked product (Fey and Regenstein 1982; Cann and others 1984; Himelbloom and others 1994; Sivertsvik and others 1999) To grade raw whole fish the use of the quality index (QI) method is gaining popularity (Botta 1995a; Anonymous 2000) In the QI method, a wide range of individual spoilage indicators (eye shape, gill odor, and so forth) is graded (usually to a maximum of demerit points) based on precise, objective, independent, and primary descriptions The total score across all indicators is summed to give a QI, which, by adjusting the size of the descriptors allocated to the individual indicators, can be manipulated to give linearly increasing scores with increasing storage time This has the benefit that no undue emphasis is placed on a single criterion and minor differences in judgments in any one criterion not unduly influence the total score In the current work, we attempt to apply similar principles to evaluating cooked salmon The aims of the current study were: to identify suitable indices (including sensory characteristics) for evaluating spoilage of king salmon, to compare the spoilage characteristics of king salmon with those reported for other salmonid species, and to obtain an indication of the potential of MAP to extend the shelf life of king salmon produced in New Zealand 8/22/2002, 2:03 PM Spoilage of king salmon Materials and Methods of 100% nitrogen and stored at –70 ЊC for sensory evaluation and chemical testing, respectively Fish King salmon (n = 90) were supplied by the New Zealand King Salmon Co Limited, Nelson, New Zealand The salmon were starved for d then commercially harvested by ‘rested harvesting’ procedures using the anesthetic AQUI-S™ (AQUI-S New Zealand, Lower Hutt) and iced for overnight transfer to the company’s processing facility Eighty-five fish were filleted under commercial conditions on 10/29/98 while the other were supplied whole The 170 paired fillets and representative whole fish were transferred overnight to the Crop & Food Research laboratory in Auckland (mean temperature on arrival = ЊC) and placed in a chiller (0.5 ± ЊC) until packaging that day The fillets and whole fish were weighed and the fork length of the whole fish was measured The belly cavity of the whole fish was opened to determine sex and the gonads and liver were removed and weighed These fish were then filleted and the amounts of protein, lipid, moisture, and ash in the right hand (skin-off ) fillets were determined by methods 968.05, 922.06, 950.46, and 900.02 respectively of the Association of Official Analytical Chemists (1998) Carbohydrate was estimated by difference Treatments The 170 fillets were trimmed to loins of at least cm thick (Figure 1), weighed (mean weight 429 g, standard deviation 65 g), placed on water absorbent pads (SecureFresh Pacific Ltd, Auckland) and packaged in pouches (40 ϫ 29 cm) of double metallized laminate with an oxygen transmission rate of less than 0.4 cc/m2/24 h at atmospheric pressure, 23 ЊC and 90% relative humidity (SecureFresh Pacific Limited, Auckland) All fillets, except those sampled at time zero, were packed with a SecurePack machine (SecureFresh Pacific Ltd, Auckland, NZ) A fish:gas ratio of approximately 1:4.2 was achieved by setting the packaging machine timers to deliver a nominal gas volume of 1800 cc to each pack Four packaging/storage treatments were applied: ambient air (AIR), 100% nitrogen (N2), and a 40:60 carbon dioxide:nitrogen mix (CO2N2), all stored in melting ice (0 ± 0.1 ЊC), and the fourth treatment of ambient air packs stored at ± 0.5 ЊC (AIR9) Gases were supplied by BOC Gases New Zealand Limited, Auckland, and certified as containing less than 10 ppm oxygen Sensory evaluation A 14 member sensory panel was selected from a group of consumers who said they enjoy eating fresh salmon During training sessions, a 34 attribute QI score sheet for cooked salmon was developed (Table 1) using a selection of sensory terms used to describe the quality of salmonid fish (trout and salmon) derived from the literature (Fey and Regenstein 1981; Hansen and others 1995; Pastoriza and others 1996; Nair and others 1997; Reddy and others 1997b; Sigurgisladottir and others 1997; Einen and Skrede 1998; Einen and Thomassen 1998; Leroi and others 1998; Paludan-Müller and others 1998; Rodriguez and others 1999) These were augmented as required in the training sessions during which panelists discussed terms until panel consensus was reached on terminology and scale usage Panelists also provided an evaluation of overall quality where zero was a pass and a fail This overall evaluation was not included in the QI Fish samples for testing were taken from frozen ventral strips (containing the fat line but no pin bones), which were thawed overnight at ЊC and cut into pieces (about cm ϫ cm ϫ cm) from positions 11 to 15 (Figure 1) These pieces were individually wrapped in aluminum foil and baked at 180 ЊC for approximately 14 to give an internal temperature of 65 to 70 ЊC Fish from replicates of the different treatment and time combinations were used in training sessions and score sheet development, while the third was used for the formal evaluation of the samples These evaluations were run on consecutive days with samples randomly distributed among panelists Presentation order was balanced to ensure that no sample followed one more often than another, and the sampling design ensured that each panelist tasted sample from each treatment and time Panelists in separate booths received up to warm foil-wrapped pieces at each session and evaluated them according to the QI score sheet Samples were coded with 3-digit random numbers, and rinsing water was provided for use between samples Panelist consistency was monitored by including a control sample in each session The control was purchased from a local wholesaler the day after harvest, cut into strips and held at -18 ЊC for d before the first session It was sampled and cooked as for the other samples Chemical and physical analyses Sampling occurred at time zero before packaging (FRESH) and on the following days for the different treatments: AIR9:1 and d; AIR: 1, 4, 8, 11, 15, and 18 d; N2: 8, 18, and 25 d; CO2N2: 1, 8, 18, 40, and 54 d On each occasion, 10 packs containing loins from fish were sampled for the given treatment The oxygen content of the gas in the packs was determined using a Gaspace oxygen analyzer (Systech Instruments, Thame, Oxfordshire, United Kingdom) The right hand loins were weighed, and pH and redox readings recorded A sample (about 10 g) was aseptically removed from position (Figure 1) of the left hand loins for microbiological analysis The 10 loins were then split longitudinally, ventral to the pin bones, and the strips were trimmed to a width of cm and a thickness of cm Three squares of cm2 were cut from the anterior end of the dorsal strips (positions 2, 3, and 4) for chemical analysis Sample 16 was removed from the ventral strips for gas chromatographic analysis of volatiles produced on cooking (to be reported elsewhere) The ventral strip (positions 11 to 15) and the separate squares (positions 2, 3, 4, and 16) were individually packaged in foil bags in an atmosphere Five samples from each treatment were analyzed for each test at each sampling time Drip loss, expressed as a percentage of the initial loin weight, was calculated as the weight loss between packaging and sam- Figure 1—Sampling positions on a king salmon fillet Vol 67, Nr 6, 2002—JOURNAL OF FOOD SCIENCE jfsv67n6p2362-2374ms20010445-AF-TO RRD.P65 2363 8/22/2002, 2:03 PM 2363 Sensory and Nutritive Qualities of Food Sampling Spoilage of king salmon Table 1—Sensory attributes and scores used in panel score sheet Attribute score1 Attribute group Attribute Appearance Odor Off odor Flavor Off flavor Mouth feel Aftertaste Texture Oily Slight Moderate Very Curds Small amount Moderate Large amount Gaping No Slight Some Severe Characteristic salmon Characteristic Weak No Fishiness No Slight Strong Freshness Fresh, slight sea smelling Neutral Stale Sweetness Fresh sweet Not Sickly sweet Meatiness Not present Slight Moderate Strong Sourness None Slight Moderate Severe Rancidity None Slight Moderate Severe Yeasty Not present Slight Moderate Strong Vinegar Not present Slight Moderate Strong Metallic Not present Slight Moderate Strong Characteristic salmon Characteristic Weak No Fishiness None Slight Strong Cardboard Characteristic salmon flavor Bland Slight Strong Freshness Fresh salmon Neutral Slightly stale Stale Sweetness Sweet Not sweet Sickly sweet Sourness Fresh tang Bland Slight Moderate Strong persistent Rancidity None Slight Moderate Strong Yeasty None Slight Moderate Strong Vinegar None Slight Moderate Strong Bitter None Slight Moderate Strong Metallic None Slight Moderate Strong Oiliness Normal Moderate Excessive Salmon Salmon Neutral Fishy No Slight Strong Oily Slight Moderate Strong, lingering Sour None Slight Moderate Strong Cardboard No Slight Marked Stickiness Slight Moderate Strong Flakiness Flaky Not flaky Pastiness Not Slight Moderate Very pasty Moistness Moist Slightly dry or slightly wet Dry or wet Chewiness Soft, slightly chewy Chewy Very chewy The quality index (QI) was taken as the sum of the attribute scores Sensory and Nutritive Qualities of Food pling (measured before portioning and freezing) Redox measurements were determined at position (Figure 1) of the right hand loins using a platinum combination electrode (MC241Pt, red rod in saturated KCl, Radiometer, Copenhagen, Denmark) inserted into the anterior dorsal portion of the loin Loins stored without oxygen (N2 and CO2N2 treatments) were measured in a chamber flushed with oxygen-free nitrogen while AIR treatments were measured in air The electrode was left in contact with the fish for 30 to ensure a stable reading and the flesh temperature at the time of reading was recorded (mean = 9.7 ЊC) E h was calculated following the method published by George and others (1998): Eh = Eobs + Eref + EN(pH – 7) where E obs is the observed redox potential, E ref is the temperature dependent redox potential of the internal electrolyte of the electrode and EN is the temperature dependent Nernst potential Using information supplied by the electrode manufacturer, Eref was calculated as: Eref = -0.98357 T + 223.3393 and using the information in Ingold (1980), the Nernst potential was calculated as: EN = 0.198 T + 54.2 where T was the temperature (°C) of the sample when the redox was measured pH was measured on a fresh cut surface at the anterior end of the right hand loins using a surface electrode (Sensorex 450C; 2364 Garden Grove, Calif., U.S.A.) Nucleotides and derivatives (adenosine 5’-triphosphate derivatives (ATP), adenosine 5’-diphosphate (ADP), adenosine 5’monophosphate (AMP), inosine 5’-monophosphate (IMP), inosine (INO), and hypoxanthine (HX)) were determined on samples from the left hand loin (position 2) Approximately 0.4 g of partially thawed muscle tissue was homogenized with mL chilled (0 ЊC) perchloric acid (0.6 M), allowed to stand (30 min, ЊC) then centrifuged (20 min, 10000 ϫ g) Supernatant (400 ␮L) was mixed (1 vortex) with 800 µL of a 3:1 mixture of Volasil 244 (BDH) and Tri-n-octylamine (Sigma, St Louis, Mo., U.S.A.) then centrifuged (2 min, 11000 ϫ g) The filtered aqueous phase was stored at –80 ЊC before HPLC analysis (Ryder 1985) K-values (Saito and others 1959) (the percentage of HX + INO over the total pool of ATP derivatives) and H-values (Botta 1995b) (the percentage of HX over the sum of IMP, INO and HX) were calculated Peroxide values were determined by the method in Firestone (1995) on samples from position Thiobarbituric acid number (TBA) (Vyncke 1970) and Trimethylamine Nitrogen ( TMA-N) (Dyer 1945) were determined on samples from position Total volatile base nitrogen (TVBN) was determined by the micro-diffusion method of Pearson (1973) on samples from position Color was measured on samples from right hand loins (position 2) at a 90Њ angle to both the frame and skin side of samples using a Minolta Chromameter (model CR-200b; Osaka, Japan) Color was JOURNAL OF FOOD SCIENCE—Vol 67, Nr 6, 2002 jfsv67n6p2362-2374ms20010445-AF-TO RRD.P65 2364 8/22/2002, 2:03 PM Spoilage of king salmon The sample from position was homogenized (2 min, Stomacher 400, Seward Limited, London, United Kingdo) in salt-peptone water (1% NaCl, 0.1% peptone (Difco, Detroit, MI 482327058, U.S.A.) at a ratio of 1:10 Further decimal dilutions were made in salt-peptone water and the drop plate method using 10 µL drops from each dilution (Barbosa and others 1995) was used for total aerobic counts (plate count agar (Difco) with 1% NaCl) and sulfide producers (peptone iron agar (Difco) with a peptone iron agar overlay) Plates were incubated aerobically for d at 20 ЊC and colonies counted and recorded as colony forming units (c.f.u.) Sulfide producers were black colonies on the peptone iron agar Six colonies were randomly picked from the plate count agar from each sample on the last sampling day of each treatment (30 colonies per treatment) These were stored on glass beads ( Jones and others 1984) for subsequent taxonomic identification and use in a full study on the spoilage microflora of king salmon and will be reported elsewhere (personal communication, Johanson 2002) Statistical analyses The GENSTAT statistical package (Lawes Agricultural Trust 1998) was used for all statistical analyses In order to compare treatments, analysis of variance (ANOVA) was applied to log 10 transformations of the microbiological counts and oxygen readings and to the untransformed results from the remaining physical and chemical tests Where the residual plots showed a random pattern with constant variance, the calculated least significant difference (l.s.d.) at p = 0.05 was used to compare mean results, otherwise (for drip loss) the ANOVA was repeated and comparisons were made using the l.s.d after a natural log transformation Where results appeared to give a linear response with time, linear regression analysis was applied over that range The sensory scores for all attributes (excluding overall quality) were summed for each panelist for a given sample to give a total QI with a maximum (worst score) of 86 These scores were then standardized so that each panelist’s scores had a mean of zero and standard deviation of This was done to remove the effect of panelist and to standardize their use of the attribute scales Samples were then compared using the resultant scores To establish the relationship between the panelists’ evaluation of overall quality and QI, mean pass-fail scores of overall quality were calculated for samples with the same QI, regardless of treatment and time, and these were plotted against QI Regression analysis was carried out on the linear portion of the plot Principal component analysis (PCA) of the sensory attribute data was carried out using the covariance matrix to identify relationships between treatments and sensory attributes To determine differences in individual spoilage attributes, tests of independence were performed on tables of counts of the numbers of panelists scoring a given attribute absent (score = 0), or present (score = 1, 2, 3, or 4) for each treatment and time Chi-squared statistics were calculated, with a large statistic indicating that the proportion of detections of an attribute differed among the treatments and times Results Fish The 170 fillets had mean weights of 687 g, (standard deviation 76 g) The whole fish consisted of females and male Oxygen There was a linear decline in mean oxygen levels in the AIR packs (0 ЊC) from the ambient of 20.8% to 15.1% during the first 15 d storage (0.39 %/d, standard error = 0.06%/d, r2 = 74.5) Between days 15 and 18 there was a rapid decline (p < 0.01) to 8.1% Mean residual oxygen levels of 26.4 ppm were achieved in the N2 and CO2N2 packs on packaging After 18 d storage, levels had declined (p < 0.05) to 3.1 and 5.7 ppm for the N2 and CO2N2 packs, respectively On subsequent testing days, oxygen levels in all packs had dropped from the 18 d level to below the detection limit of 0.5 ppb, except for CO2N2 pack that had an oxygen content of 41.8 ppb after 40 d Sensory evaluation The mean QI results are presented as percentages of the total possible score of 86 (Figure 2) Increases in mean QI tended to be linear with storage time except for the AIR9 and AIR, which showed sudden increases after and 15 d, respectively Although regression analysis of the linear portions of the QI plots only accounted for 25.2% of the variance (Table 2), the analyses were still highly significant (p < 0.001), indicating QI increases with time over these portions The slopes from the treatments stored at ЊC were not significantly (p < 0.05) different (mean = 0.276%/d) The results of the first principal components of the PCA results are presented in Figure 3a to 3d The sensory attributes and the treatment days are presented in separate plots for clarity but they represent the same analysis and need to be interpreted together For example, in Figure 3c the day result for AIR9 is in the upper left quadrant of the graph which can be interpreted as meaning that the sensory terms in the upper left quadrant of Figure 2—Sensory quality index (QI, expressed as a percentage of the maximum score of 86) on cooked king salmon held in different atmospheres All packs were stored at ЊC except AIR9 which was at ЊC Each point represents the mean of 12 assessments Points marked with the same letter (a, b, c, and so on) were not significantly different at the 5% level Vol 67, Nr 6, 2002—JOURNAL OF FOOD SCIENCE jfsv67n6p2362-2374ms20010445-AF-TO RRD.P65 2365 8/22/2002, 2:03 PM 2365 Sensory and Nutritive Qualities of Food Microbiological analyses with mean biological status as follows (standard deviations in parentheses): weight 3242 (260) g, length 59.9 (1.8) cm, liver 23.5 (5.0) g, and gonad 29.6 (5.8) g The proximate compositions of the right hand fillets had the following ranges: moisture 63.0 to 66.6%, protein 18.6 to 19.0%, lipid 12.7 to 20.4%, ash 1.1 to 1.2%, and carbohydrate (by difference) 0.0 to 2.8% Quality index (%) expressed in the Lab color space and the hue (H0ab) and chroma (C*ab) calculated according to Choubert and others (1992) Spoilage of king salmon IMP (␮mol/g) 86.1 Inosine (␮mol/g) 84.6 90.8 Hypoxanthine 74.0 (␮mol/g) H value (%) 79.2 QI (%) 25.3 Overall 80.0 quality score4 AIR9a AIRb 3.32 (0.26) 1.339 (0.089) 2.11 (0.21) 0.346 (0.017) AIR9a,b 0-4 - 18 0-4 AIRa - 18 CO2N2b - 54 N2a - 25 AIR9a 0-4 3.84 AIRa,b - 11 5.97 CO2N2b 0-8 6.57 N2b 0-8 5.09 AIR9a 0-4 3.27 AIRb - 11 2.70 CO2N2b 0-8 2.76 N2b 0-8 2.50 All °C - 11 2.65 AIR9 a 0-4 0.40 AIRb - 18 0.35 CO2N2c,d - 54 0.35 N2b,d - 25 0.30 AIR9a 0-4 4.24 AIRb - 18 3.54 CO2N2c - 54 3.43 N2b - 25 3.83 AIR9a 0-1 9.72 AIRa - 15 6.20 CO2N2a - 54 8.50 N2a - 25 9.53 Alla Not AIR9 8.67 d or AIR d 18 All - 32% -0.35 0.003 0.003 0.003 0.003 (0.58) (0.40) (0.49) (0.81) (0.28) (0.24) (0.26) (0.34) (0.11) (0.18) (0.14) (0.13) (0.19) (1.77) (1.34) (1.26) (1.84) (1.97) (1.62) (1.16) (1.65) (0.62) Slope (standard error) 0.272 (0.077) 0.137 (0.012) 0.088 (0.005) 0.153 (0.011) -0.939 (0.200) -0.546 (0.056) -0.489 (0.086) -0.489 (0.086) 0.842 (0.117) 0.466 (0.037) 0.484 (0.056) 0.480 (0.061) 0.475 (0.021) 0.354 (0.076) 0.105 (0.013) 0.055 (0.005) 0.085 (0.012) 4.435 (0.744) 1.170 (0.129) 0.620 (0.044) 0.887 (0.120) -0.330 (1.970) 0.525 (0.176) 0.273 (0.040) 0.257 (0.103) 0.276 (0.032) (0.10) 0.0446 (0.005) 1Treatments with different superscript letters have significantly different slopes (p < 0.05) 2Most of the regression analyses were executed on the whole data set for the particular test for example drip loss However, where results clearly gave a linear response for only a limited range (for example IMP–see Figure 7a), only this range was included in the regression analysis 3Constant taken as zero by definition 4Quality against QI between and 32% (See Figure 4) Sensory and Nutritive Qualities of Food Figure 3a (sour aftertaste, sour odor, bitter flavor, and so forth) were the dominant sensory characteristics Most of the variance (80.1%) was accounted for by the first principal component (PC1) with components and (PC2, PC3) accounting for 4.88 and 2.95% of the total variance (24.5 and 14.8% of the remaining variance) respectively All panelists always scored samples with a mean QI of less than (8.1%) as a pass in the overall quality evaluation and always scored samples with a mean QI of more than 27 (31.4%) as a fail (Figure 4) A linear response was observed for the mean overall quality ratings of samples with mean scores between and 27 (8 and 32%) Regression analysis (Table 2) suggested that a sample with a QI of 16.3 (19% of the maximum score of 86) would be expected to have a mean overall quality score of 0.5 (that is, on average, 50% of panelists would rate it with a pass score and 50% with a fail score) while 25% and 75% of panelists would give a pass score for samples with a QI of 11.5 (13%) and 21.2 (25%), respectively Drip loss All treatments gave approximately linear increases in drip loss with time ANOVA did not show significant differences between 2366 Eh and pH Although normally distributed, the Eh results were characterized by large variances (l.s.d = 124 mV ) with results from identically treated samples differing by up to 272 mV The mean Eh of the unpackaged fish flesh was +126 mV The mean E h of the CO2N2 and N2 packs did not change significantly from this value for the duration of the trial while the Eh in the AIR packs increased slightly and in the AIR9 packs it decreased (p < 0.01) to a mean of -143 after d (range to -270) mV (Figure 6) All results from individual samples in AIR, N2, and CO2N2 treatments gave positive Eh values except for N2 sample on day 25 (-35 mV) The mean cut-surface pH was 6.12 on arrival (time zero) and the l.s.d for the trial was calculated as 0.13 pH units The mean pH of treatments AIR, AIR9, N2, and CO2N2 increased to maximum levels Principle component Aerobic 95.1 plate count (log10 cfu/g) Drip loss (%) 75.2 Constant (standard error) (accounts for 4.88% of variance R2 Test Range of determinTreat- ant data ment1 used2 (d) treatments with the same day of sampling (Figure 5) However, regression analysis (Table 2) suggested that the AIR9 treatment had a higher rate of drip loss (0.272%/d) than the other treatments and the rate of drip loss from the N2 treatment (0.153%/d) was significantly higher than the CO2N2 treatment (0.088%/d) although this difference was only evident after 18 d storage Principle component (accounts for 2.95% of variance Table 2—Linear regression analyses on results where plots suggested a linear response Figure 3a and 3b—Principal component (PC) analysis of sensory terms and treatment times: Sensory terms as listed in Table JOURNAL OF FOOD SCIENCE—Vol 67, Nr 6, 2002 jfsv67n6p2362-2374ms20010445-AF-TO RRD.P65 2366 8/22/2002, 2:03 PM Spoilage of king salmon Chemical indicators The sum total of the pool of ATP derivatives (ATP + ADP + AMP + inosine + IMP + hypoxanthine) did not change (p > 0.05) from the initial total of 9.48 ␮mol/g (l.s.d = 1.46) for the duration of the trial for any of the treatments There was a rapid loss of IMP to form inosine followed by linear increases in hypoxanthine (Figure 7a to 7c) All ЊC treatments produced inosine at a constant rate of 0.475 µmol/g for the first 11 d to a maximum of around ␮mol/g while the fish at ЊC produced inosine at a rate of 0.842 ␮mol/g (Table 2) Because of the rapid conversion of IMP, K-values quickly reached a maximum of around 96% (Figure 8a) with the remaining 4% accounted for by ADP levels which did not vary (p > 0.05) from the mean of 0.36 ␮mol/g throughout the trial The H-values, however, showed linear increases for each treatment over the duration of the trial (Figure 8b) Although mean peroxide values (data not shown) recorded increased from an initial value of 0.00 to 0.42, 0.29, 0.42, and 0.05 meq/ kg in AIR, AIR9, N2, and CO2N2 samples stored for 18, 4, 25, and 54 d respectively, none of these increases were statistically significant (p > 0.05, l.s.d = 0.56) The initial TBA number was 0.416 and only samples held under CO2N2 for 54 d (mean = 0.742) were significantly above this level (p > 0.05, l.s.d = 0.201, data not shown) Significant increases in TMA-N were only observed in AIR9 samples after d and CO2N2 samples after 54 d (Figure 9) TVBN (data not presented) in the AIR treatment did not increase significantly from the initial level of 8.87 mg/100 g during 18 d while the other treatments all increased slightly to 10.8, 10.4, and 10.9 mg/100 g for the AIR9, N2, and CO2N2 treatments after 4, 25, and 54 d, respectively (l.s.d = 1.02) Color ANOVA did not show any significant variations in color between treatments or with time The mean values for the different tissue surfaces are recorded in Table Microbiology Logarithmic increases in total aerobic plate counts (APC) were recorded for all treatments (Figure 10) with lag phases of at least Mean overall quality score = Pass, = Fail of 6.34, 6.37, 6.52, and 6.44 on days 18, 4, 18, and 8, respectively The pH of the anaerobic treatments, N2 and CO2N2, subsequently declined, dropping to 6.22 and 6.29 on days 25 and 54 respectively Figure 3c and 3d—Principal component (PC) analysis of sensory terms and treatment times: Treatment times Numbers represent the d storage for the particular treatment Figure 5—Drip loss from king salmon held in different atmospheres All packs were stored at ЊC except AIR9, which was at ЊC The error bar depicts the least significant difference (p = 0.05) Vol 67, Nr 6, 2002—JOURNAL OF FOOD SCIENCE jfsv67n6p2362-2374ms20010445-AF-TO RRD.P65 2367 8/22/2002, 2:03 PM 2367 Sensory and Nutritive Qualities of Food Drip loss (%) Figure 4—Relationship between pass-fail scores of overall quality assigned by the trained panel and the sensory quality index (QI, expressed as a percentage of the maximum score of 86) on cooked king salmon held in different atmospheres at or ЊC Each point represents the mean overall quality score for samples which panelists assigned the same QI, regardless of treatment Spoilage of king salmon phase at lower levels than the aerobic treatments No sulfide producing organisms (< 50 c.f.u./g) were recorded throughout the trial for any of the treatments However, the method used will not distinguish sulfide producing organisms if their numbers are log cycles lower than the total aerobic plate count because the background flora obscures colonies of sulfide producing organisms Eh (mV) 4, 8, and 18 d for AIR, N2, and CO2N2 treatments, respectively CO2N2 counts showed a significant decline (p < 0.05) following packaging, an extended lag phase and a stationary Figure 7—Changes in nucleotide derivatives in king salmon held in different atmospheres All packs were stored at ЊC except AIR9 which was at ЊC Each point represents the mean of samples The error bars depict the least significant difference (p = 0.05) Figure 9—Trimethylamine nitrogen (TMA-N) levels in king salmon held in different atmospheres All packs were stored at ЊC except AIR9 which was at ЊC Each point represents the mean of samples The error bar depicts the least significant difference (p = 0.05) nitrogen (mg/100g) Figure 8—K- and H-values of king salmon held in different atmospheres All packs were stored at ЊC except AIR9 which was at ЊC Each point represents the mean of samples The error bars depict the least significant difference (p = 0.05) Trimethylamine Sensory and Nutritive Qualities of Food Nucleotide derivates (mm ol/g) Figure 6—Eh of king salmon held in different atmospheres All packs were stored at ЊC except AIR9 which was at ЊC Each point represents the mean of samples The error bar depicts the least significant difference (p = 0.05) 2368 JOURNAL OF FOOD SCIENCE—Vol 67, Nr 6, 2002 jfsv67n6p2362-2374ms20010445-AF-TO RRD.P65 2368 8/22/2002, 2:03 PM Spoilage of king salmon Discussion Oxygen Oxygen levels in the packs showed similar trends of decline with storage to those reported elsewhere (Lindroth and Genigeorgis 1986; Garcia and Genigeorgis 1987; Reddy and others 1995; Reddy and others 1996; Reddy and others 1997a; Reddy and others 1997b; Reddy and others 1999) The largest declines corresponded with the occurrence of high bacterial numbers (> 106 c.f.u./g, Figure 10) in the samples, suggesting that the oxygen was being consumed during microbial metabolism Although the previously mentioned researchers recorded oxygen levels of “0%” immediately after gas packaging, their methods of analysis were not able to demonstrate the very low levels (< 100 ppm) achieved in this study Sensory evaluation and shelf life Aerobic plate count (log10 c.f.u./g) Sensory evaluation showed very clear differences in spoilage between treatments (Figure 2) The usual aim in developing a QI is to have a score that has a linear response with time under constant conditions, for example ice-time This was generally observed with our score system except for the air treatments Figure 10—Changes in bacterial levels in king salmon held in different atmospheres All packs were stored at ЊC except AIR9 which was at ЊC Each point represents the mean of samples The error bar depicts the least significant difference (p = 0.05) Table 3—Color values for salmon flesh during storage under various conditions1 Hunterlab color scale mean (standard error) L a b Sample Frame side of fillet Skin side of fillet Dark meat 53.15 (2.72) 60.23 (2.55) 63.20 (1.15) 19.72 (1.95) 20.07 (2.47) 34.31 (2.67) 28.56 (2.96) 30.79 (1.97) 24.41 (2.77) Chroma2 Hue2 tan-1(b/a) (a2 + b2)1/2 55.38 34.71 56.90 36.75 35.43 42.11 1ANOVA did not suggest any significant (p < 0.05) color differences with treatment and/or storage time 2Calculated according to Choubert and others (1992) (AIR9 and AIR), which gave a sudden increase in scores at the end of storage where a large number of the demerit attributes became very obvious in a short period of time Our aim was not to develop a linear scale but to be able to compare the degree of spoilage between samples in a more absolute sense Discussions with the panelists when designing the score sheet suggested that perceived spoilage did not occur linearly For example, samples from the AIR treatment underwent a rapid deterioration between days 15 and 18 as reflected in Figure The scores developed in Table are our best evaluation of the sensory experience of the panelists Regression analysis of the linear portions of the QI plots (Table 2) only accounted for a limited amount (23.3%) of the variance and did not show any difference in the rates of spoilage between the treatments However, ANOVA confirmed that the differences with time for individual treatments were significant, particularly for the CO2N2 treatment (Figure 2), so the observed slopes did represent real changes The results of the PCA are useful in further analyzing the spoilage profiles of the salmon PC1 appeared to represent a general spoilage trend with increasing spoilage giving more negative values Almost all sensory attributes gave negative values for PC1 (Figure 3a and 3b) as was expected since they were designed as demerit attributes On the basis of PC1, the most important spoilage attributes for cooked salmon are sour flavor and Figure 11—Relationship between bacterial numbers and H-values for individual king salmon loins stored under different conditions Vol 67, Nr 6, 2002—JOURNAL OF FOOD SCIENCE jfsv67n6p2362-2374ms20010445-AF-TO RRD.P65 2369 8/22/2002, 2:04 PM 2369 Sensory and Nutritive Qualities of Food There was a positive relationship between APC and hypoxanthine levels with APC below 500000 c.f.u./g invariably giving Hvalues of 20% or less, as shown in Figure 11 When bacteria numbers were between 103 and 10 c.f.u./g, samples that had been stored longer showed higher hypoxanthine levels than those which were stored for shorter periods Thus, CO2N2 samples had higher H-values than N2, which were higher than the air stored samples while time levels gave the lowest hypoxanthine levels (Figure 11) The sensory scores were also positively related to APC with the mean QI beginning to increase more rapidly after mean counts reached 107 c.f.u./g (Figure 12) There was generally a linear relationship between sensory scores and hypoxanthine levels (Figure 13) although rapid growth of bacteria toward the end of storage for samples stored in air caused a more rapid increase in QI H-value (%) Interrelationships Spoilage of king salmon Quality index (%) Sensory and Nutritive Qualities of Food Figure 12—Relationship between bacterial numbers and sensory quality index (QI) Each point represents the mean of microbiological analyses and the mean of evaluations by 14 panelists of samples stored under the same treatments for the period of time 2370 Figure 4) were taken as the product shelf life, then from Figure the AIR, AIR9, N2, and CO2N2 treatments gave shelf-lives of 16, 2.2, > 25, and 43 d respectively If a more conservative shelf life of 13% (when 25% of panelists gave it a fail score) was taken as the end of shelf life, then shelf-lives of 15, 1.5, 15, and 21 d would be obtained from the AIR, AIR9, N2, and CO2N2 treatments respectively The latter values approximate the times at which negative values for PC1 were first recorded (Figure 3c and 3d) and might, therefore, be taken as the times when any spoilage was consistently detected by the panel If the former measure were taken as indicating the shelf life, this would give a shelf life extension of 170% for the CO2N2 treatment over the AIR treatment while the more conservative measure would give an extension of only 40% Results for cooked odor and flavor of salmon from other workers gave the following shelf lives: Atlantic salmon fillets–11 d in an air permeable overwrap and > 17 d in 60:40 CO2:N 2, 40:60 CO2:N 2, or vacuum at ЊC (Randell and others 1999); Atlantic salmon steaks–25 d in 60:40 CO2:N2 and 21 d under vacuum at ЊC (Cann 1990); whole Atlantic salmon–16 d in air at ЊC or 21 d if the fish were starved (Blokhus 1987); Atlantic salmon slices–18 d in 100% CO2 but less than in air at ЊC (Pastoriza and others 1996); whole king salmon–1 mo in 60:35:5 CO2:N2:O2 at ЊC (Fey and Regenstein 1982) Drip loss The amounts of drip loss were similar to those recorded in other studies on salmon (Fey and Regenstein 1982; Pastoriza and others 1996; Randell and others 1999) Packs stored at ЊC would be expected to produce drip more rapidly than those at ЊC, but the slower rate of drip in CO2N2 compared with N2 was unexpected The losses of around 3% recorded at the end of shelf life of the CO2N2 fillets may represent a significant cost to producers (Connell 1975), and may be an important consideration in deciding whether to use MAP technology to extend the shelf life of salmon In our packs, this drip was effectively absorbed by the water-absorbent pads included in the packs so did not detract from the appearance of the product Quality index (%) loss of fresh flavor followed by sour aftertaste, sour odor, loss of fresh odor, loss of salmon flavor, and development of cardboard flavor In contrast to the sensory attributes, the treatments gave positive values for PC1 until days 4, 18, 18, and 40 for the AIR9, AIR, N2, and CO2N2 treatments, respectively (Figure 3c and 3d) suggesting that substantial deterioration was not detected in the cooked flesh until these times PC2 distinguished between the different sensory attributes with loss of flavor and texture deterioration generally giving negative values while odors and aftertaste gave positive values (Figure 3a) Exceptions to this were rancid, bitter, yeast, and to a lesser extent vinegar flavors, and fishy aftertaste PC2 also separated the treatments: when spoilage was evident (PC1 was negative) the air treatments (AIR and AIR9) gave positive values in PC2 while the oxygen-free treatments (CO2N2 and N2) gave negative values (Figure 3c) PC3 gave negative values for the CO2N2 treatment separating it from the other treatments (Figure 3d) The sensory attributes associated with these (Figure 3b) suggest that the moisture and oily attributes were more important in the deterioration of samples stored under CO2N2 Overall the CO2N2 samples were associated with attributes more readily attributed to autolytic spoilage (texture degradation, perceived oiliness) The Chi-squared test confirmed significant interactions between treatment times and oily aftertaste (p = 0.026), moist texture (p = 0.73), and cardboard texture (p < 0.001) Panelists gave these attributes a greater proportion of defect scores in the CO2N2 samples from day 54 than other treatments, which shows that these attributes were important contributors to spoilage for the CO2N2 treatment Shelf life of a product should be defined by a supplier for a specific market based on the sensory properties required for the product, taking into account the company brand image and the product end-use For example, where salmon is marketed to be eaten raw it is likely to be assigned a much shorter shelf life than if expected to be cooked Profiling panels are trained to detect defects and tend to be more critical in their evaluation of products than ordinary consumers If the QI of 19% (where 50% of the trained panel gave the ungarnished baked salmon a fail score– Figure 13—Relationship between hypoxanthine and sensory quality index (QI) Each point represents the mean of chemical analyses and the mean of evaluations by 14 panelists of samples stored under the same treatments for the same period of time JOURNAL OF FOOD SCIENCE—Vol 67, Nr 6, 2002 jfsv67n6p2362-2374ms20010445-AF-TO RRD.P65 2370 8/22/2002, 2:04 PM Spoilage of king salmon Eh is as an important factor in considering the hazard from psychrotrophic strains of the anaerobe Clostridium botulinum (Huss 1981) Maximum Eh values for growth of this organism have been reported between +30 and +250 mV Optimum Eh for growth of C botulinum type A was reported as -350 mV (Kim and Foegeding 1993) and rapid growth is reported to occur between -6 and -436 mV (Notermans 1993) There have been few studies where the Eh of fish has been monitored during storage (Ishida and others 1976; Huss and Larsen 1979; Huss and others 1979; Huss and Larsen 1980; Wiberg and Nilsson 1980) The large variance in Eh we observed in identically treated samples appears to be common in such studies Storing fish under conditions of very low oxygen might be expected to give low E h values but the N2 and CO2N2 treatments did not show a decrease in Eh Huss and Larsen (1979) recorded high Eh values in cod fillets until substantial microbial spoilage occurred At 10 ЊC, regardless of whether fillets were unwrapped or vacuum packed, Eh values were above mV for the first d but rapidly declined to levels of around -250 mV by d storage, which is the same pattern we observed (Figure 6) At ЊC, they recorded no change from the initial value of around 250 mV for the first 13 d after which there was a dramatic decline in E h with values approaching -400 mV by 20 d storage They noted that the change in Eh corresponded to the development of trimethylamine (TMA) and total volatile bases (TVB) in the cod while, at ЊC, we found minimal changes in TVBN and TMA-N and no drop in E h despite the very low levels of oxygen detected in the packs Studies on smoked fish (Huss and others 1979; Huss and Larsen 1980; Wiberg and Nilsson 1980) also indicate that reductions in Eh not occur until the onset of microbial spoilage, regardless of packaging A number of studies on C botulinum, such as those recently published by Reddy and others (1999) and Dufresne and others (2000), suggest that there is little difference in the safety of fish packaged in air or under MAP but there is an increased hazard when fish are stored above recommended refrigeration temperatures, regardless of packaging environment The results of Eh studies support the assertion that packaging technologies not negatively affect the safety of fish from the effects of hazardous anaerobic microorganisms pH Jerrett and others (1996), Jerrett and Holland (1998), Jerrett and others (1998), and Jerrett and others (2000) have reported the pH of rested and exhausted muscle from king (chinook) salmon They found rested fish to have a mean flesh pH of 7.2 to 7.5 dropping to means of between 6.2 and 6.3 with glycolytic production of lactic acid The low pH recorded in our salmon (6.1) on arrival at the laboratory suggests that although harvested using commercially applied rested harvesting techniques and held at refrigeration temperatures, glycolysis was already complete Our low pH fish might be expected to deteriorate faster than fish with a higher pH (Connell 1975) Increases in fish muscle pH during storage have normally been attributed to microbial spoilage producing volatile bases such as TMA (Connell 1975) In the current work, increases in pH in the AIR and AIR9 packs corresponded to bacterial growth although there was minimal production of volatile bases, while the more rapid increases in the N2 and CO2N2 treatments occurred despite minimal bacterial growth These increases are unexplained although pH increases in the absence of bacterial growth have been observed in salmon by other workers (Pastoriza and others 1996; Rodriguez and others 1999) The presence of elevated CO levels in the CO2N2 treatment could be expected to lower the flesh pH as has been recorded previously in salmon (Fey and Regenstein 1982; Pastoriza and others 1996) We observed the opposite with higher (p < 0.05) pHs in the CO2N2 fish than the AIR fish after d at ЊC Nucleotides and derivatives Of the chemical indicators of spoilage, only the development of hypoxanthine and its dependent H-value were consistently related to storage time This is in agreement with Norwegian workers who suggested that hypoxanthine was the only measure available as a quality measure for their Atlantic salmon (Blokhus 1987) The difference in rates of hypoxanthine production between the ЊC and ЊC treatments are compatible with the square root relationship commonly used for seafood spoilage (for example, Dalgaard 2000) Given the observed rate of 0.105 µmol/d at ЊC, this predicts a rate of 0.378 ␮mol/g/d at ЊC, well within the standard errors of the observed rate of 0.354 ␮mol/g/d (Table 2) There were no significant differences in IMP and inosine results for the three ЊC treatments despite significant differences in bacterial numbers and shelf life Thus, the breakdown of ATP derivatives in the biochemical pathway through to inosine was not affected by the storage atmosphere However, hypoxanthine production was related to the growth of bacteria (Figure 11), which was affected by atmosphere The increasing hypoxanthine levels observed for samples with similar bacterial counts but longer storage times shows that hypoxanthine is also produced by autolytic processes, so hypoxanthine may be a useful measure of salmon deterioration across treatments that affect both bacterial and autolytic spoilage However, severe bacterial spoilage did not result in proportionally high levels of hypoxanthine as shown by the points (AIR9 day and AIR day 18) that did not fit the generally linear relationship between QI and hypoxanthine level in Figure 13 Blokhus (1987) demonstrated that starved Atlantic salmon had lower hypoxanthine levels and better sensory quality than normally fed salmon during storage in ice Fed fish had approximately 0.22 ␮mol/g (3 µg/g) hypoxanthine present at harvest while hypoxanthine appeared to be absent in starved fish Our d starved fish had 0.31 µmol/g hypoxanthine at packaging which, using the regression equation in Table 2, also suggests a value of close to when harvested d earlier Blokhus (1987) observed increases in hypoxanthine with time in starved salmon at similar rates to those we observed Himelbloom and others (1994) recorded more rapid hypoxanthine production in ice stored pink salmon (Oncorhynchus gorbuscha) than we observed, with levels increasing from about 0.8 µmol/g (120 ␮g/g) after d to about ␮mol/g at 10 d Fish held in chilled seawater contained much higher levels and also had poorer sensory scores Rodriguez and others (1999) also demonstrated a linear relationship for the H-value in rainbow trout (Oncorhynchus mykiss) but their slope was 4.052%/d at to ЊC Given our observed rate of 1.170 %/d at ЊC, the square root relationship predicts a rate of only 2.459 %/d, considerably less than that observed in rainbow trout Brown and others (1980) recorded hypoxanthine levels during storage under air and MAP (40:60 CO2:Air and 40:1:59 CO2:CO:Air) at 4.5 ЊC for silver salmon (Oncorhynchus kisutch) with a relatively high initial hypoxanthine level (2.5 ␮mol/g) Regression analysis of their data suggests production rates of 0.331, 0.201, and 0.295 µmol/g/d respectively for air, CO2, and CO2:CO treatments The rate in air is similar to what the square root relationship would predict from our king salmon data but that in the 40% CO2 is higher, possibly because air rather than N2 was used in the gas mix Overall, studies suggest that the rate of hypoxanthine production varies for different species and/or for the different bacterial flora present Vol 67, Nr 6, 2002—JOURNAL OF FOOD SCIENCE jfsv67n6p2362-2374ms20010445-AF-TO RRD.P65 2371 8/22/2002, 2:04 PM 2371 Sensory and Nutritive Qualities of Food Eh Spoilage of king salmon in the different trials More work needs to be done on the relationship between hypoxanthine levels in salmon and sensory scores, but hypoxanthine (or H-value) does show promise of providing an objective measure of deterioration across a range of treatments for a given species TBA and peroxide values Despite the presence of O2 in the air packs, lipid oxidation was insignificant in any packs as measured by either TBA number or peroxide value Some other workers have also recorded low (< mg/ kg) TBA numbers in salmon (Brown and others 1980; Randell and others 1999) while other studies have found substantial increases in TBA number Pastoriza and others (1996) reported linear increases exceeding 10 mg/kg for Atlantic salmon after 20 d in 100% CO2 stored at ± ЊC and Fey and Regenstein (1982) recorded up to 14 mg/kg (3.4 µmol/100 g) for king salmon after 24.5 d stored in MAP (60:35:5 CO2:N2:O2) Interestingly, both of these studies recorded higher TBA numbers in MAP salmon than in air-stored controls Salmon naturally contains substantial levels of carotenoid pigments, including astaxanthine (Blokhus 1987), a known antioxidant The conflicting results reported may represent different astaxanthine levels Although we did not measure carotenoid concentrations in our samples, Choubert and others (1992) showed linear relationships between carotenoid concentrations and color in raw rainbow trout (Oncorhynchus mykiss) with chroma (C*ab) giving the best correlation (r2 = 0.90) If a similar relationship were true for king salmon the color values observed in our fish (Table 3) suggest that carotenoid levels were high, exceeding 10 mg/kg TMA-N and TVBN Researchers have invariably found only low levels of TVB or TMA in spoiling salmonid species and our maximum levels of only 1.4 and 11 mg/100 g for TMA-N and TVBN respectively were not unexpected Such levels are well below the levels considered unacceptable in fish products (for example, 10 to 15 mg TMA-N or 35 to 40 mg TVBN/100 g; Connell 1975) suggesting that these compounds are unlikely to make a major contribution to unpleasant flavors and odors in salmon Studies of these compounds as indicators of salmon spoilage have given different results for different species Pink and silver salmon (Oncorhynchus gorbuscha and O kisutch) gave approximately linear increases in both TVBN and TMA with time (Brown and others 1980; Crapo and Himelbloom 1999) while Atlantic salmon gave approximately linear increases in TVBN (Pastoriza and others 1996; Rodriguez and others 1999), but the TMA results were more like those we observed with minimal change for a period followed by a sudden increase In our trial, all treatment times with TMA-N levels above 0.5 mg/100 g were past the end of shelf life, which agrees with the results of Pastoriza and others (1996) although we found much lower TVBN levels at the end of shelf life That TMA-N and TVBN did not increase during the test period for our AIR treatments despite obvious sensory spoilage shows that these compounds are not suitable as indicators of spoilage in king salmon Sensory and Nutritive Qualities of Food Microbiology The substantial differences observed in bacterial counts for the treatments are closely related to the sensory scores (Figure 12), suggesting that shelf life in this trial was primarily limited by bacterial growth From Figure 11 it appears that changes in aerobic plate counts during the logarithmic growth phase were similar for the AIR, CO2N2, and N2 treatments This suggests that the shelf life extensions achieved in the CO2N2 and N2 packs (Figure 2) were, therefore, due to the effectiveness of these atmospheres 2372 in delaying the onset of bacterial growth rather than slowing the rate of growth The extensions in shelf life found for the N2 and CO2N2 packs over the AIR packs were closely related to the APC lag times of approximately 9, 18, and d respectively (Figure 10) QI only increased gradually with increasing APC when counts were below 107 c.f.u./g while above this QI increased rapidly with increasing counts (Figure 12) The latter is likely to be a causal relationship with bacteria contributing much of the spoilage At lower counts it is more likely an incidental relationship with increasing bacterial numbers only indicating storage time, and the observed deterioration (Figure 2) being due to autolytic spoilage such as the loss of the flavor enhancing compound IMP (Figure 7a) In CO2N2, APC declined significantly after packing with counts from days and particularly day being significantly lower than those from day 0, suggesting that CO2 had a bactericidal as well as a bacteriostatic effect Our microbiological results are very similar to those reported by Cann (1990) for Atlantic salmon steaks except that all their curves reached lower stationary phase maxima (about 7.5 log10 c.f.u./g) Other workers using Atlantic salmon at different temperatures and MAP configurations recorded different results (Pastoriza and others 1996; Randell and others 1999) and further research should be carried out on the relative effect of different temperatures and gas mixes on bacterial lag phase and product deterioration The rapid deterioration we observed at the end of sampling in the AIR treatments is characteristic of bacterial spoilage However, spoilage proceeded in a linear fashion for the entire monitoring periods in the N2 and CO2N2 treatments (Figure 2) suggesting that autolytic processes were more important for these treatments The AIR9 growth rate of 1.339 log10cfu/g/d was 3.874 times higher than that of AIR (see Table 2) which is within the standard errors of the 3.61 times increase predicted by the square root relationship The absence of sulfide-producing bacteria in the fish shows that one of the most common seafood spoilage bacteria (Shewanella putrefaciens) did not contribute to the spoilage in this trial Randell and others (1999) reported that sulfide-producing organisms only just exceeded the detection limit in MAP and overwrapped Atlantic salmon Amanatidou and others (2000), however, found substantial numbers of S putrefaciens in their vacuum- and MAP (50:50 CO2:O2) Atlantic salmon fillets The safety of MAP seafood has been questioned, particularly with regard to Clostridium botulinum (see Reddy and others 1992 and the more recent work of Skinner and Larkin 1998; Hyytia and others 1999; Reddy and others 1999; Gibson and others 2000) This spore-forming bacterium produces an extremely potent neurotoxin and, because some nonproteolytic strains are capable of growing at temperatures as low as 2.9 ЊC (Graham and others 1997), growth and subsequent toxin production can occur under normal refrigeration conditions These strains (types B, E, and F) are unknown in New Zealand but there has been limited published survey work (Gill and Penney 1982) For our experiments, safety was assured by cooking to a controlled temperature that would denature the polypeptide toxin, and by either controlling temperature below the minimum growth temperature or, in the case of the AIR9 treatment, a short storage time Despite any theoretical risk of botulism from MAP seafood, the process has been used for many years in northern Europe (particularly the UK and France) with no published incidents of food poisoning A code of practice used by the UK industry (Betts 1996) places no restrictions on pre-distribution storage life where temperatures are maintained at less than ЊC The code allows for a post-distribution shelf life (retail and consumer) of up to 10 d if temperatures can be expected to be below ЊC JOURNAL OF FOOD SCIENCE—Vol 67, Nr 6, 2002 jfsv67n6p2362-2374ms20010445-AF-TO RRD.P65 2372 8/22/2002, 2:04 PM Conclusions T HE QI METHOD WAS SUCCESSFULLY APPLIED TO THE SENSORY evaluation of cooked salmon samples and gave good separation of the effects of different treatments on quality Indications from the pass-fail quality scores provided by the trained panel suggested that a good relationship existed between these and the more objective QI This relationship should be tested by asking an untrained consumer panel to evaluate the product for acceptability Overall, the results point to a combination of gradual autolytic deterioration while counts were below 10 c.f.u./g This process was not affected by the storage atmosphere Once bacterial numbers reached high levels, spoilage was much more rapid Both total aerobic plate counts and hypoxanthine levels were useful spoilage indicators pH increased with storage but was not related to increases in bacterial numbers The cause(s) of this pH increase in salmon warrants further study The square root relationship was able to explain differences in rates of change in sensory, microbiological, and chemical indices at and ЊC Despite being a different genus, results from farmed king salmon (Oncorhychus tshawytscha) more closely resembled those of farmed Atlantic salmon (Salmo salar) than of other wild-caught Oncorhynchus species This study shows that MAP has good potential for extending the shelf life of king salmon References Amanatidou A, Schluter O, Lamkau K, Gorris LGM, Smid EJ, Knorr D 2000 Effect of combined application of high pressure treatment and modified atmospheres on shelf life of fresh Atlantic salmon Innovative Food Sci and Emerging Technol 1(2):87-98 Anonymous 2000 Development and implementation of a computerized sensory system for evaluating fish freshness [Web page] http://qimit.rfisk.is/ QimIT 2000 Accessed December 2000 Association of Official Analytical Chemists 1998 Official methods of analysis Helrich K, Editor 16th ed Arlington, Va.: Association of Official Analytical Chemists Methods 900.02, 922.06, 950.46, 968.05 Barbosa HR, Rodrigues MFA, Campos CC, Chaves ME, Nunes I, Juliano Y 1995 Counting viable cluster-forming and noncluster-forming bacteria: a comparison between the drop and spread methods J Microbiol Meth 22:39-50 Betts GD 1996 Code of practice for the manufacture of vacuum and modified atmosphere packaged chilled foods with particular regard to the risks of botulism Glocstershire: Chipping Campden 114 p Bilinski E, Gibbard GA, Lee F, Gibbard SE, Smith DT, Peters MD, Jonas REE 1981 Evaluation of the quality of sockeye salmon, Oncorhynchus nerka, stored in ice or partially frozen in salt fortified refrigerated sea water Vol 1055 Canadian Technical Report of Fisheries and Aquatic Sciences Vancouver: Dept of Fisheries and Oceans 19 p Bilinski E, Jonas REE, Peters MD, Choromanski EM 1984 Effects of sexual maturation on the quality of Coho salmon (Oncorhynchus Kisutch) flesh Can Inst Food Sci Technol J 17(4):271-273 Blokhus H 1987 Aspects related to quality of farmed Norwegian salmon (Salmo salar) In: Kramer DE, Liston J, editors Proceedings of the International Symposium on Seafood Quality Determination November 10-14, 1986, Anchorage, Alaska, U.S.A Amsterdam: Elsevier Publishing Co Inc p 615-628 Botta JR 1995a Evaluation of seafood freshness quality New York: VCH Publishers, Inc 180 p Botta JR 1995b Nucleotide degradation products In: Botta JR, editor Evaluation of seafood freshness quality New York: VCH Publishers, Inc P 16-25 Bremer PJ, Fletcher GC 1999 Applications of modified atmosphere packaging for New Zealand seafoods The NZ Food J 29(2):50-55 Brown WD, Albright M, Watts DA, Heyer B, Spruce B, Price RJ 1980 Modified atmosphere storage of rockfish (Sebastes miniatus) and silver salmon (Oncorhynchus kisutch) J Food Sci 45:93-96 Cann BC, Houston NC, Taylor LY, Smith GL, Thomson AB, Craig A 1984 Studies of salmonids packed and stored under a modified atmosphere Aberdeen: Torry Research Station, MAFF 53 p Cann DC 1990 An update on modified atmospheres Fish Farmer 7:13-14 Choubert B, Blanc J-M, Courvalin C 1992 Muscle carotenoid content and color of farmed rainbow trout fed astaxanthin or canthaxanthin as affected by cooking and smoke-curing procedures Int J Food Sci Technol 27:277-284 Connell JJ 1975 Control of fish quality 2nd ed Farnham: Fishing News Books Ltd 222 p Crapo C, Himelbloom B 1999 Spoilage and histamine in whole Pacific herring (Clupea harengus pailast) and pink salmon (Oncorhynchus gorbuscha) fillets J Food Safety 19(1):45-55 Dalgaard, P 2000 Seafood Spoilage Predictor 1.1 (Predictive microbiology) Danish Institute for Fisheries Research, Dept of Seafood Research, Technical Univ of Denmark, Lyngby, Denmark Accessed December 6, 2000 Dufresne I, Smith JP, Liu JN, Tarte I, Blanchfield B, Austin JW 2000 Effect of headspace oxygen and films of different oxygen transmission rate on toxin production by Clostridium botulinum type E in rainbow trout fillets stored under modified atmospheres J Food Safety 20(3):157-175 Dyer WJ 1945 Amines in fish muscle I Colorimetric determination of trimethylamine as picrate salt J Fish Res Board Can 6(5):351-358 Einen O, Skrede G 1998 Quality characteristics in raw and smoked fillets of Atlantic salmon, Salmo salar, fed high-energy diets Aquacult Nutri 4(2):99-108 Einen O, Thomassen MS 1998 Starvation prior to slaughter in Atlantic salmon (Salmo salar) II White muscle composition and evaluation of freshness, texture, and color characteristics in raw and cooked fillets Aquaculture 169(1/2):37-53 Fey MS, Regenstein JM 1981 Extending shelf life of fresh wet red hake and salmon using CO2-O2 modified atmosphere and potassium sorbate ice at ЊC In: Martin RE, editor Proceedings of the First National Conference on Modified and Controlled Atmosphere Packaging of Seafood Products January 12-14, 1981, San Antonio, Texas, U.S.A Washington, DC: National Marine Fisheries Institute P 186-194 Fey MS, Regenstein JM 1982 Extending shelf life of fresh wet red hake and salmon using CO2-O2 modified atmosphere and potassium sorbate ice at °C J Food Sci 47:1048-1454 Firestone D 1995 AOAC official method 965.33 Peroxide values of oils and fats Titration method In: Cunniff P, editor Official Methods of Analysis of AOAC International Gaithersburg: AOAC International P 41.41.16 Garcia G, Genigeorgis C 1987 Quantitative evaluation of Clostridium botulinum nonproteolytic Types B, E, and F growth risk in fresh salmon tissue homogenates stored under modified atmospheres J Food Prot 50(5):390-397 George SM, Richardson LCC, Pol IE, Peck MW 1998 Effect of oxygen concentration and redox potential on recovery of sublethally heat-damaged cells of Escherichia coli O157:H7, Salmonella enteritidis and Listeria monocytogenes J Appl Microbiol 84:903-909 Gibson AM, Ellis-Brownlee RCL, Cahill ME, Szabo EA, Fletcher GC, Bremer PJ 2000 The effect of 100% CO2 on the growth of nonproteolytic Clostridium botulinum at chill temperatures Int J Food Microbiol 54:39-48 Gill CO, Penney N 1982 The occurrence of Clostridium botulinum at aquatic sites in and around Auckland and other urban areas of the North Island NZ Vet J 30:110112 Graham AF, Mason DR, Maxwell FJ, Peck MW 1997 Effect of pH and NaCl on growth from spores of non-proteolytic Clostridium botulinum at chill temperature Letters in Appl Microbiol 24:95-100 Hansen LT, Gill T, Huss HH 1995 Effects of salt and storage temperature on chemical, microbiological, and sensory changes in cold-smoked salmon Food Res Int 28(2):123-130 Himelbloom BH, Crapo C, Brown EK 1994 Pink salmon (Oncorhynchus gorbuscha) quality during ice and chilled seawater storage J of Food Qual 17(3):197-210 Huss HH 1981 Clostridium botulinum Type E and botulism Copenhagen: Technological Laboratory, Ministry of Fisheries, Technical Univ 58 p Huss HH, Larsen A 1979 The post mortem changes in the oxidation-reduction potential of fish muscle and internal organs In: Sobolenska-Ceronik K, Ceronik E, Zaleski S, editors Food as an ecological environment for pathogenic and index microorganisms Poland: Ars Polona P 265-279 Huss HH, Larsen A 1980 Changes in the oxidation-reduction potential (Eh) of smoked and salted fish during storage Lebensm-Wiss u-Technol 13:40-43 Huss HH, Schaeffer I, Pedersen A, Jepsen A 1979 Toxin production by Clostridium botulinum Type E in smoked fish in relation to the measured oxidation-reduction potential (Eh), packaging method, and the associated microflora In: Connell JJ, editor Advances in fish science and technology England: Fishing News (Books) Ltd P 476-479 Hyytia E, Hielm S, Mokkila M, Kinnunen A, Korkeala H 1999 Predicted and observed growth and toxigenesis by Clostridium botulinum type E in vacuum-packaged fishery product challenge tests Int J Food Microbiol 47:161-169 Ingold W 1980 Redox measurement Principles and problems Urdorf: Ingold 24 p Ishida Y, Fujii T, Kadota H 1976 Microbiological studies on salted fish stored at low temperature—I chemical and microbial changes of salted fish during storage Bull Japan Soc Sci Fish 42(3):351-358 Jerrett AR, Holland AJ 1998 Rigor tension development in excised “rested,” “partially exercised,” and “exhausted” Chinook salmon white muscle J Food Sci 63(1):48-52 Jerrett AR, Holland AJ, Cleaver SE 1998 Rigor contractions in “rested” and “partially exercised” Chinook salmon white muscle as affected by temperature J Food Sci 63(1):53-56 Jerrett AR, Law RA, Holland AJ, Cleaver SE, Ford SC 2000 Optimum postmortem chilled storage temperature for summer and winter acclimated, rested, chinook salmon (Onchorhynchus tshawytscha) white muscle J Food Sci 65(5):750-755 Jerrett AR, Stevens J, Holland AJ 1996 Tensile properties of white muscle in rested and exhausted king salmon (Oncorhynchus tshawytscha) J Food Sci 61(3):527532 Johanson, MR 2001 Unpublished data New Zealand Institute for Crop and Food Research Limited, Private Bag 92169, Auckland, New Zealand Jones D, Pell PA, Sneath PHA 1984 Maintenance of bacteria on glass beads at 60 ЊC to -76 ЊC In: Kirsop BE, Snell JJS, editors Maintenance of microorganisms London: Academic Press P 35-40 Kim J, Foegeding PM 1993 Principles of control In: Hauschild AHW, Dodds KL, editors Clostridium botulinum ecology and control in foods: Marcel Dekker, Inc P 121-176 Lawes Agricultural Trust 1998 Genstat (Statistics) Rothamsted: Rothamsted Experimental Station [Software package] Leroi F, Joffruad J-J, Chevalier F, Cardinal M 1998 Study of the microbial ecology of cold-smoked salmon during storage at ЊC Int J Food Microbiol 39:111-121 Lindroth SE, Genigeorgis CA 1986 Probability of growth and toxin production by nonproteolytic Clostridium botulinum in rockfish stored under modified atmospheres Int J Food Microbiol 3:167-181 Nair PS, Surendran PK, Gopakumar K 1997 Occurrence and distribution of lactic acid bacteria in fish and fishery environments Fishery Technology 34(1):34- Vol 67, Nr 6, 2002—JOURNAL OF FOOD SCIENCE jfsv67n6p2362-2374ms20010445-AF-TO RRD.P65 2373 8/22/2002, 2:04 PM 2373 Sensory and Nutritive Qualities of Food Spoilage of king salmon Spoilage of king salmon 38 Notermans SHW 1993 Control in fruits and vegetables In: Hauschild AHW, Dodds KL, editors Clostridium botulinum ecology and control in foods New York, U.S.A Marcel Dekker, Inc P 233-260 Paludan-Müller C, Dalgaard P, Huss HH, Gram L 1998 Evaluation of the role of Carnobacterium piscicola in spoilage of vacuum- and modified- atmosphere-packed cold-smoked salmon stored at ЊC Int J Food Microbiol 39:155-166 Pastoriza L, Sampedro G, Herrera JJ, Cabo ML 1996 Effect of carbon dioxide atmosphere on microbial growth and quality of salmon slices J Sci Food Agric 72:348352 Pearson D 1973 Flesh foods–Meat and fish In: Pearson D, editor Laboratory techniques in food analysis London: Butterworths P 166-212 Randell K, Hattula T, Skytta E, Sivertsvik M, Bergslein H, Ahvenainen R 1999 Quality of filleted salmon in various retail packages J of Food Qual 22:483-497 Reddy NR, Armstrong DJ, Rhodehamel EJ, Kautter DA 1992 Shelf life extension and safety concerns about fresh fishery products packaged under modified atmospheres: A review J Food Safety 12:87-118 Reddy NR, Paradis A, Roman MG, Solomon HM, Rhodehamel EJ 1996 Toxin development by Clostridium botulinum in modified atmosphere-packaged fresh Tilapia fillets during storage J Food Sci 61(3):632-635 Reddy NR, Roman MG, Villanueva M, Solomon HM, Kautter DA, Rhodehamel EJ 1997a Shelf life and Clostridium botulinum toxin development during storage of modified atmosphere–packaged fresh catfish fillets J Food Sci 62(4):878-884 Reddy NR, Solomon HM, Rhodehamel EJ 1999 Comparison of margin of safety between sensory spoilage and onset of Clostridium botulinum toxin development during storage of modified atmosphere (MA)-packaged fresh marine cod with MApackaged aquaculture fish fillets J Food Safety 19:171-183 Reddy NR, Solomon HM, Yep H, Roman MG, Rhodehamel EJ 1997b Shelf life and toxin development by Clostridium botulinum during storage of modified- atmosphere-packaged fresh aquacultured salmon fillets Food Protect 60(9):10551063 Reddy NR, Villanueva M, Kautter DA 1995 Shelf life of modified-atmospherepackaged fresh Tilapia fillets stored under refrigeration and temperatureabuse conditions Food Protect 58(8):908-914 Rodriguez CJ, Besteiro I, Pascual C 1999 Biochemical changes in freshwater Rainbow Trout (Oncorhynchus mykiss) during chilled storage J Sci Food Agric 79:1473-1480 Ryder JM 1985 Determination of adenosine triphosphate and its breakdown Sensory and Nutritive Qualities of Food 2374 products in fish muscle by high performance liquid chromatography J Agric Food Chem 33:678-680 Saito T, Arai K, Matsuyoshi M 1959 A new method for estimating the freshness of fish Bull Japan Soc Sci Fish 24(9):749-750 Sigurgisladottir S, Ole T, Oyvind L, Thomassen M, Hafstein H 1997 Salmon quality: Methods to determine the quality parameters Rev Fish Sci 5(3):223252 Sivertsvik M, Rosnes JT, Vorre A, Randell K, Ahvenainen R, Bergslien H 1999 Quality of whole gutted salmon in various bulk packages J Food Qual 22:387-401 Skinner GE, Larkin JW 1998 Conservative prediction of time to Clostridium botulinum toxin formation for use with time-temperature indicators to ensure the safety of foods J Food Prot 61(9):1154-1160 The New Zealand Seafood Industry Council Ltd 2000 New Zealand Seafood Exports Calendar Year to December 1999 Wellington: The New Zealand Seafood Industry Council Ltd 90 p Vyncke W 1970 Direct determination of the thiobarbituric acid value in trichloroacetic acid extracts of fish as a measure of oxidative rancidity Fette Seifen Anstrichmittel 12:1084-1087 Wiberg C, Nilsson G 1980 Redoxpotentialvarden (Eh) hos vakuumforpackad bockling, rokt makrill och rokt sik Var Foda 32:89-95 MS20010445, Submitted 8/15/01, Accepted 2/18/02, Received 2/27/02 The authors thank Tricia O Lee and Maria J Leonard for their skillful technical assistance and John Koolaard for assistance with statistical analyses Thanks also to the members of the Palmerston North sensory evaluation panel for their skill and tolerance in tasting the fish Proximate analyses of the salmon were carried out by AgriQuality New Zealand Limited, Lynfield, Auckland This work was funded by the New Zealand Foundation for Research, Science, and Technology, Contract CO2816 The support of the New Zealand King Salmon Co and SecureFresh Pacific Limited for respectively supplying the salmon and packaging machine and materials used in the research is gratefully acknowledged Authors Corrigan and Cumarasamy are with the New Zealand Institute for Crop & Food Research, Palmerston North, New Zealand Author Dufour is with the Food Science Dept., Univ of Otago, Dunedin, New Zealand Authors Fletcher and Summers are with the New Zealand Institute for Crop & Food Research Limited, Private Bag 92169, Auckland, New Zealand Direct inquiries to author Fletcher (E-mail: fletcherg@crop.cri.nz) JOURNAL OF FOOD SCIENCE—Vol 67, Nr 6, 2002 jfsv67n6p2362-2374ms20010445-AF-TO RRD.P65 2374 8/22/2002, 2:04 PM ... RRD.P65 2373 8/22/2002, 2:04 PM 2373 Sensory and Nutritive Qualities of Food Spoilage of king salmon Spoilage of king salmon 38 Notermans SHW 1993 Control in fruits and vegetables In: Hauschild... Qualities of Food Sampling Spoilage of king salmon Table 1—Sensory attributes and scores used in panel score sheet Attribute score1 Attribute group Attribute Appearance Odor Off odor Flavor Off flavor... JOURNAL OF FOOD SCIENCE—Vol 67, Nr 6, 2002 jfsv67n6p2362-2374ms20010445-AF-TO RRD.P65 2366 8/22/2002, 2:03 PM Spoilage of king salmon Chemical indicators The sum total of the pool of ATP derivatives