Application of soluble gas stabilization technology on ready‐to‐eat pre‐rigor filleted Atlantic salmon (Salmo salar L.)

Abstract The demand for high‐quality, convenient, and sustainable salmon products represents a potential for value‐added product development and novel packaging solutions. Soluble gas stabilization (SGS) technology, which applies dissolved CO2 in the product before packaging, represents a novel approach to retain product quality and prevent microbiological deterioration during cold storage of pre‐rigor filleted salmon loins. The present study aimed to examine the solubility of CO2 in salmon loins as affected by rigor status. In addition, the effect of predissolved CO2 on the overall quality of pre‐rigor vacuum‐packed Atlantic salmon (Salmo salar L.) was investigated during storage at 4°C. The CO2 pretreatment was conducted, exposing loins to 100% CO2 for 18 h at 4°C (the control group was kept in air at 4°C) before repackaging and storage for 15 days. Dissolved CO2 in the muscle (equilibrium achieved four days post packaging) was slightly higher in pre‐rigor than post‐rigor salmon loins (p equilibrium = 0.006). Moreover, the overall spoilage (Hvalue) and microbiological stability of salmon fillets stored in SGS‐vacuum were significantly improved compared to vacuum‐packed loins (p < 0.05). The results demonstrate that SGS technology can maintain the overall quality of pre‐rigor vacuum‐packed salmon loins without introducing the high gas‐to‐product volume ratio recognized by modified atmosphere packaging. Thus, the application of SGS technology on pre‐rigor loins can lead to higher economic gain and environmental benefits due to the reduced amount of required packaging material and reduced food waste. Practical Application CO2 can be dissolved in pre‐rigor salmon loins before vacuum packaging to increase product shelf life during cold storage.


INTRODUCTION
Norwegian farmed Atlantic salmon (Salmo salar L.) is a highly successful product and has become a well-known trademark globally. In 2020, Norway, as the world's largest producer of Atlantic salmon, exported 1.1 million tons with a value of €6.8 billion (FishFarmingExpert, 2021). Seventy-seven percent of the Norwegian Atlantic salmon are exported as whole Head-on-Gutted (HOG) fish stored on ice. However, in Norwegian retail stores, most salmon are sold as convenient vacuum or modified atmosphere (MA) portion packages of filleted products of both postand pre-rigor salmon (Heide, 2020). Products of pre-rigor origin are processed before the salmon enters rigor mortis, which in a commercial process usually starts at 6-12 h post-mortem, depending on various factors, for example, temperature, pre-mortem handling, and slaughtering procedures (Chan et al., 2020;Lerfall et al., 2017;Mørkøre et al., 2008;Wang et al., 2000). Pre-rigor processed salmon are regarded as a ready-to-eat product (sashimi and sushi quality), meaning that the producer or the manufacturer intends it for direct human consumption without the need for cooking or other processing to eliminate or reduce to acceptable level microorganisms of concern (European Commission, 2007). The trend of consumer preferring high-quality, convenient, and sustainable salmon products represent a potential for value-added product development (VAPD) and increased processing before export (Dobrucka & Cierpiszewski, 2014;Nilsen, 2019). A significant advantage of pre-rigor filleting is that the product can reach the market 3-5 days earlier than post-rigor fillets (Skjervold et al., 2001), supplying the market with super-fresh products. Furthermore, from a sustainability perspective, transportation of pre-rigor filleted products instead of HOG will be favorable as the greenhouse gas emissions are estimated to be reduced by 20-50% (Madslien & Kwan Kwong, 2015;Rotabakk et al., 2020). Thus, effort should be put into VAPD and novel packaging techniques to establish a variety of high-quality salmon products.
Microbiological control during storage is one of the critical factors for achieving high-quality products. MA and vacuum packaging combined with low temperature have been used for decades to prolong the shelf life of seafood, counteracting deteriorative effects during storage due to microbiological and endogenous enzymatic activity (Bouletis et al., 2017). MA packaging utilizes the bacteriostatic effects of CO 2 to suppress the present spoilage microbiota (Devlieghere & Debevere, 2000). Dissolved CO 2 in the food matrix will reshape the product microbiota because of inter-and intraspecies variation in CO 2 tolerance (Kolbeck et al., 2021). Growth parameters (mmax and lag phase) of Gram-negative bacteria are more affected by CO 2 than Gram-positive bacteria (Devlieghere & Debevere, 2000), making gram-positive pathogens, such as Clostridium botulinum and Listeria monocytogenes of particular concern (Jami et al., 2014;Peck et al., 2020). Furthermore, the presence of no or little O 2 in vacuum and MA packages may promote the growth of these bacteria and the formation of botulinum neurotoxin by psychrotrophic anaerobic non-proteolytic C. botulinum. The current industry practices in European countries, keeping retail vacuum-and MA-packed food below 3-8 • C, are supported by challenge tests for several meat products (Peck et al., 2020). However, measures toward nonproteolytic C. botulinum types B and F that are associated with seafood and can produce toxins at low temperature (≥3.3 • C) (Peck, 2006) must be taken. Microbial growth in raw seafood is sensitive to deviations from optimal storage temperature (Hoel et al., 2017(Hoel et al., , 2018; thus, keeping an unbroken refrigerated chain during product distribution is of utmost importance.
The bacteriostatic effect of CO 2 is proportional to the concentration of dissolved CO 2 in the food matrix (Devlieghere et al., 1998a, b). Several factors affect the amounts of dissolved CO 2 in food, including temperature, the ratio of gas-to-product volume, the initial composition of the gas mixture used, and product characteristics such as pH, salt content, the content and composition of lipids, and the water content (Abel et al., 2018(Abel et al., , 2020Gill, 1988;Jakobsen & Bertelsen, 2004;Mendes et al., 2011;Rotabakk, 2013;. However, there is no available knowledge about how the rigor-mortis status of salmon affects the solubility of CO 2 in the muscle. To achieve the optimal effect of CO 2 in MA packaging, a gas-to-product volume ratio of 2:1 or 3:1 is typically used (Sivertsvik et al., 2002). From an environmental or financial point of view, this is one of the disadvantages of MA packaging. The degree of filling can be improved by combining MA-packaging with a CO 2 emitter (Hansen et al., 2009a) or soluble gas stabilization (SGS) technology . In SGS-technology, recently reviewed by Esmaeilian et al. (2021), CO 2 is dissolved in the product prior to packaging. Previous studies have mainly combined SGS-technology with MA packaging (Rotabakk & Sivertsvik, 2012;Rotabakk et al., 2008). Compared to MA packaging, vacuum packaging is more transport economical and requires less packaging material. However, vacuum-packed products are, in general, more rapidly deteriorated (Dalgaard et al., 1993;Hansen et al., 2009a;Lerfall et al., 2018a). Combining SGS-technology and vacuum packaging represents an innovative and sustainable approach to meet the consumer demand for high-quality fresh salmon products.
The present study aimed to (1) establish new knowledge of solubility of CO 2 in pre-rigor salmon loins and (2) evaluate the general quality and microbiological stability of prerigor salmon loins combining pre-dissolved CO 2 and vacuum packing during cold storage.

Raw material
Atlantic salmon was slaughtered at a nearby slaughterhouse in Mid-Norway before the fish were immediately transported on ice in polystyrene (EPS) boxes to the Norwegian University of Science and Technology (NTNU, Trondheim, Norway). At arrival NTNU (5 h postslaughter), eight fish with no signs of rigor-mortis was chosen and immediately scaled (average weight of 4.2 ± 0.4 kg) and filleted. The right fillets were portioned and processed directly (pre-rigor), whereas left fillets were stored on ice for 4 days before further handling (post-rigor samples). To ensure similar chemical composition of the experimental portions, the back-and mid-loin running from the gills to the Norwegian quality cut (NQC) was used to prepare uniform experimental samples (Katikou et al., 2001;Lerfall et al., 2011). Raw material (n = 5) composition was assessed by measuring fat (according to Bligh & Dyer, 1959) and water content (according to ISO.6496 (1983)), whereas the residual content was assumed to be protein and ash. Furthermore, degradation products of adenosine triphosphate (ATP) (n = 3), colorimetric properties (n = 9) and pH (n = 5) were analyzed as described in Section 2.4-2.6.

Experimental design
The experimental setup was divided into two experiments (Figure 1), where Experiment 1 was designed to study the solubility of CO 2 in portioned salmon loins as affected by the state of rigor mortis (pre-versus post-rigor). Experiment 2 followed the effect of predissolved CO 2 on quality attributes of pre-rigor vacuum-packed salmon.
2.2.1 Experiment 1: Solubility of CO 2 in preand post-rigor filleted salmon In the first experiment (Figure 1), pre-and post-rigor samples (n = 8 for each group) underwent a CO 2 pretreatment (described subsequently) for 18 h (100% CO 2 , 4 • C) followed by MA-packing (60% CO 2 and 40% N 2 ) (Group ID: SGS prerigor and SGS post-rigor respectively). The control groups were pretreated in air (18 h, 4 • C), and packed in MA with the same gas composition (Group ID: control pre-rigor and control post-rigor, respectively). All samples (80 ± 2 g) were stored at ≤4 • C for 4 days to obtain equilibrium between the headspace gas phase and product.
The CO 2 pretreatment was performed according to Abel et al. (2020) in batches (n = 8). The control samples were meanwhile stored in air. After 18 hours at 4 • C, pretreated samples (CO 2 or air) were repacked into 230 ml semi-rigid crystalline polyethylene terephthalate trays (C2125-1A, Faerch Plast, Denmark) using a tray sealing packaging machine (TL250, Webomatic, Germany). The air was evacuated (final vacuum pressure of 25 mbar) and the packages were flushed with the pre-set MA gas mixture. The cover film comprised of a 40 mm combination of polyethylene, ethylene vinyl alcohol, polyamide, and polyethylene terephthalate (Topaz B-440 AF, Plastopil, The Netherlands). Food-grade CO 2 (60%) and N 2 (40%) were mixed using a MAP Mix 900 gas mixer (Dansensor, Denmark). The oxygen transmission rate (OTR) was 66−78 cm 3 × 25 mm × m −2 × 24 h × bar at 23 • C for the tray, 2.5 cm 3 × 40 mm × m −2 × 24 h × atm at 23 • C for the cover film, and 50 cm 3 × m −2 × 24 h × bar at 23 • C for the high-barrier pouches. A sample filling degree of approximately 35% was achieved.
The headspace gas composition was measured using a Checkmate 9900 oxygen and carbon dioxide analyzer (PBI-Dansensor, Denmark) as described by Abel et al. (2018). The gas composition was measured in empty trays immediately after packaging and in all trays at the end of the storage period (95 h). The headspace gas volume was moreover measured 2, 7,11,23,35,47,59,71,83, and 95 h after packaging for the pre-rigor samples and 2, 7,11,23,35,47,60,71,83, and 95 h after packaging for the post-rigor samples, according to a method described by Rotabakk et al. (2007) modified by Abel et al. (2018). The concentration of absorbed CO 2 was calculated according to Rotabakk et al. (2007) related to changes in package volume as described by: where =∞ CO 2 is the total CO 2 (ppm) absorbed by the product, P is absolute pressure (Pa), V g is gas volume The experimental design showing the set up aiming to investigate the solubility of CO 2 in pre-versus post-rigor salmon portions (Experiment 1) and to evaluate the effect of the soluble gas stabilization (SGS) technology on microbiological, chemical, and physiochemical parameters of vacuum-packaged (25 mbar) pre-rigor filleted salmon loins (Experiment 2). Modified atmosphere (MA, 60% CO 2 , 40% N 2 ) packaging was used in the repacking step in experiment 1 due to the chosen methodology to measure the solubility of CO 2 in salmon samples (Rotabakk et al., 2007) (m 3 ) at start, and at equilibrium, MwCO 2 is the molecular weight of CO 2 , R is the gas constant, T is the absolute temperature (K), and Wf is the weight of the product (kg). According to Henry's law, a sample has reached its equilibrium with the surrounding atmosphere when the amount of CO 2 in the product headspace is proportional to the amount of CO 2 absorbed in the sample. According to Schumpe et al. (1982) this can be described as: is the partial equilibrium pressure of CO 2 in the headspace gas (Pa), CO 2, is the temperaturedependent Henry's constant for CO 2 in the sample (Pa × ppm −1 ).

2.2.2
Experiment 2: Effect of SGS-technology on quality of pre-rigor filleted vacuum-packed salmon loins In the second experiment (Figure 1), pre-rigor filleted salmon was cut to loin pieces of 80 ± 2 g and separated into two groups. The SGS-vacuum group (n = 27) was pretreated with CO 2 (18 h, 100% CO 2 , 4 • C) and the control-vacuum group (n = 27) were kept in air (18 h, 4 • C) as described for Experiment 1. The experiments were conducted simultaneously and with the same experimental conditions. After the pretreatment, all samples were repacked in vacuum (50 mbar), using 20-mm PA/70-mm PE pouches (120 × 80 mm, Star-Pack Productive, France) and a Webomatic Supermax s3000 chamber machine (Webomatic, Germany). The OTR value of the pouches was 50 cm 3 × m −2 × 24 h × bar at 23 • C. Three packages were opened and examined for microbiological-, chemical-and physiochemical parameters at day 1, 4, 6, 8 (microbiology and pH only), 11, 13, 15 during storage at ≤ 4 • C. Assessment of surface color (n = 3) was conducted at day 4, 11 and 15 after repacking. Samples to be analyzed for degradation products of ATP were immediately frozen at -80 • C until analysis.

Degradation products of ATP
Frozen samples were shredded using a kitchen grater, and approximately 1.0 g (exact weight listed) was homogenized with 7,5 ml of trichloroacetic acid (TCA, 7% w/v) for 1 min with an Ultra Turrax T25 Basic (Janke & Kunkel IKA R -Labortechnik, Staufen, Germany). Potassium hydroxide was then added to the sample solution (KOH, 1 M, 3.25 mL) to achieve a pH of 6.25. The mixing tubes were kept on ice during preparation thereafter centrifuged (18000 rpm, 4 • C, 15 min) in a Rotina 420R centrifuge (Hettich Zentrifugen, Germany) before the supernatant was filtered through a nylon filter (0.45 mm) and transferred to HPLC vials (Agilent, 862-09-16, 2 ml) for analysis. The samples were analyzed on a Poroshell 120 porous column (ECC18 3.0 × 100 mm, porous size 2.7 mm, with a Poroshell 120 Fast Guard (3.0 × 5 mm, Sub-2 mm), Agilent InfinityLab) after a modified method by Sellevold et al. (1986), as described by Lerfall et al. (2018a). The K value and H value were calculated based on the concentrations of ATP degradations products (Hong et al., 2017;Howgate, 2006 where Ino is inosine, Hx is hypoxanthine, ATP is adenosine triphosphate, ADP is adenosine diphosphate, AMP is adenosine monophosphate, and IMP is inosine monophosphate.

Muscle pH
pH was measured in the center of the salmon muscle at each sampling point using a Testo 206 pH2meter (Testo SE &Co, Germany). The pH meter was calibrated before use by a two-point calibration at pH 4 and 7.

Color
A digital color imaging system (DigiEye full system, Veri-Vide Ltd., Leicester, UK) was used to measure the surface color. The test area was manually selected, and measurements were performed above the lateral line of the fillet. Analysis was carried out in a standardized lightbox (6400 K) using a digital camera (Nikon D7000, 35 mm lens, Nikon Corp., Japan). The data obtained from the DigiEye system was processed using the software Digipix version 2.8.0.2 (VeriVideLtd.) to transform the red, green, and blue (RGB) values into the Commission Internationale de I'éclairage (CIE) values L*, a*, and b*.

Statistics
The data were analysed by a general linear model (GLM) with state of rigor and pretreatment (Air and CO 2 ) as fixed factors in Experiment 1, and pretreatment (air and CO 2 ) as fixed factor in Experiment 2. One-way ANOVA tests using Tukey HSD procedures to derive statistical differences (p < 0.05) were used to compare groups on their respective storage days. Statistical analysis on microbial growth was done at log-transformed data. Samples with no detected bacterial counts were scored as 1 CFU/g before log-transformation. The bacterial counts are presented as means ± standard error (SE), and other mean values are presented as ±1 standard deviation (SD). Pearson's correlation coefficient (r) was used to calculate the linearity dependence between quantitative variables. Statistical analyses were performed using an IBM Statistical Package for the Social Sciences statistics software (release 26, IBM Corporation, USA). The log-transformed average bacterial counts were fitted to the primary model of Baranyi and Roberts (1994) (available at www.combase.cc) to estimate and compare the effects of different packaging conditions on the maximum growth rates (mmax) and duration of the lag phases.
TA B L E 1 Solubility of CO 2 in salmon loins as affected by rigor mortis and pretreatment (air for the control group and CO 2 for the SGS-group

and chemical composition of the raw material
The muscle pH at the filleting time was 7.0 ± 0.3, indicating high-quality pre-rigor loins not exposed to pre-mortem stress (Lerfall et al., 2015). The average muscle fat-and water content were 12.1 ± 2.3% and 67.6 ± 3.0%, respectively, whereas the residual content was assumed to be protein and ash (not measured).

Experiment 1: Solubility of CO 2 in preand post-rigor filleted salmon
The solubility of CO 2 , and thus Henry's constant, depends on the product's composition (Abel et al., 2018;Jakobsen & Bertelsen, 2002;Schumpe et al., 1982;. The present study assumed that the water and lipid content was similar in the pre-and post-rigor samples  and did not affect the experimental results. Control pre-rigor packaged samples showed a significantly higher Henrys constant than the control postrigor equivalents (Table 1, 40.3±1.7 and 36.5±1.5, respectively). Moreover, the observed Henry's constants corresponded well with previously reported values for fish in general (Abel et al., 2018; and salmon specifically (Abel et al., 2020). A similar pattern was observed for the product-and headspace CO 2 concentration (Table 1), showing a coincidental reduction of headspace CO 2 with increased product concertation. However, the observed difference was only 46.3 ppm at equilibrium (p = 0.015), showing the practical effect of rigor mortis on the CO 2 solubility as minor. However, this result implies that pre-rigor CO 2 -treatment can benefit from the increased solubility of CO 2 and that pre-rigor filleting and packaging is a highly relevant option for the salmon indus-try. Moreover, the SGS-technology can be implemented as a part of the logistic chain from the slaughtering and filleting plant to a processing plant for further processing.
The CO 2 solubility observed in the SGS pre-rigor and SGS post-rigor samples are presented in Table 1 as the initial CO 2 product concentration at the point of repackaging and as solubilized CO 2 at equilibrium after storage in a MA consisting of 60% CO 2 and 40% N 2 . Similar to the headspace CO 2 concentration (Table 1), significantly higher CO 2 concentration was observed in the SGSgroups as compared to the controls (GLM, p < 0.001). Moreover, the effect of rigor mortis was significant, showing higher amounts of dissolved CO 2 in pre-rigor compared to post-rigor samples (Table 1, p initial < 0.001 and p equilibrium = 0.006).
Relative changes in product CO 2 concentration in salmon loins as a function of storage time are presented in Figure 2. The SGS pre-rigor and SGS post-rigor groups had a significantly lower change in dissolved CO 2 than the control samples (GLM, p < 0.001), as expected, as CO 2 has already been dissolved into the product during the CO 2 pretreatment step, bringing the system closer to the new equilibrium. Figure 2 shows that 50% of the equilibrium concentration is achieved between 4 and 7 h for the control samples, showing that a significant amount of CO 2 is dissolved quickly, enabling CO 2 as a pretreatment before repackaging in vacuum, MA, and further distribution to the market. The effect of rigor mortis on the relative change in product CO 2 concentration among the control samples was insignificant. However, among the SGS samples, an effect of rigor mortis was observed, showing significantly less dissolved CO 2 in post-rigor than pre-rigor samples (p < 0.05).
At equilibrium (t > 47 h), it was assumed that the prerigor treated samples were in a post-rigor state, which is supported by Mørkøre et al. (2008), who reported stressed Atlantic salmon to be in full rigor within 36 h. Moreover, Wang et al. (2000) reported from their study that rigor F I G U R E 2 Relative changes in dissolved CO 2 (ppm) as affected by time, rigor mortis (pre-versus post-rigor filleted salmon loins), and pretreatment (air for the control group or CO 2 for the SGS group). Presented values are based on the difference between the initial concentration and the dissolved concentration at the measure time (h). Legends: . Control pre-rigor, .
Control post-rigor . SGS pre-rigor, . SGS post-rigor mortis started 8 h after death and that it achieved full rigor after 24-60 h before rigor was wholly dissolved after 72 h. The results obtained in experiment 1 indicate better solubility in pre-rigor than post-rigor Atlantic salmon portions. As far as we know, this is the first study that has investigated the CO 2 solubility in pre-rigor Atlantic salmon. The findings highlight the potential of SGS-technology for an industrial application for both pre-and post-rigor salmon loins.

Experiment 2: Effect of SGS-technology on quality of pre-rigor filleted vacuum-packed salmon loins
A second experiment was designed to investigate further the SGS technology's potential to retain quality attributes of high-end cuts of pre-rigor filleted vacuumpacked salmon. Quality parameters were evaluated during 15 days of refrigerated storage and compared to vacuumpacked portions. For both groups, the muscle pH dropped significantly between filleting and storage day 1 (GLM, p <0.001), that is, from pH 7.0 ± 0.3 to 6.3 ± 0.08 for the SGSvacuum samples and 6.4 ± 0.04 for the control-vacuum samples. The mean pH of the samples remained stable through the storage period.
The CO 2 -pretreatment of the SGS-samples resulted in a significantly lower mean muscle pH of 6.3 ± 0.07 throughout the storage period than for the control-vacuum group with a mean pH value of 6.4 ± 0.06 (GLM, p < 0.005). CO 2 lowers the food matrix's pH due to dissociation into carbonic acid when reacting with water (Sivertsvik et al., 2002). Thus, similar pH-drops due to CO 2 dissolutions have been reported for different kinds of food products such as pork loins (Sørheim et al., 1996), Atlantic halibut (Rotabakk et al., 2008), and chicken breast (Rotabakk et al., 2006).

Microbiological stability
The initial concentration of APC in the pre-rigor salmon loin was below the method quantification limit of 2.4 log CFU × g −1 (Nordic Committee on Food Analysis, 2006), indicating that contamination during handling and processing was at a minimum. The low initial microbiological contamination level is in accordance with previous studies of pre-rigor salmon loin quality (Hanesen et al., 2009;Lerfall et al., 2018b). Several bacterial genera can evolve in MA-packed salmon during storage and contribute to spoilage, including genera of LAB such as Carnobacterium spp., Brochothrix thermosphacta, and gram-negative species capable of anaerobic respiration, including, Photobacterium phosphoreum, psychotropic Enterobacteriaceae, and Shewanella putrefaciens (Gram & Huss, 1996;Macé et al., 2012;Powell & Tamplin, 2012). The application of dissolved CO 2 in the food matrix is a hurdle in food preservation that reshape the microbiota initially present because of inter-and intraspecies variation in CO 2 tolerance (Kolbeck et al., 2021). The evolution of APC was significantly inhibited in the SGS vacuum-packed samples compared to the control group (Figure 3, GLM, p < 0.0001), reaching a maximum level of 2.6 ± 1.3 and 6.2 ± 1.1 log CFU × g −1 respectively, after 15 days of cold storage. There was no correlation between the APC and the measured muscle pH (r = 0.27 and r = 0.012 SGS-vacuum and control-vacuum samples, respectively). Although a randomized experimental design was conducted, high biological variation was observed (seen as high SE of microbiological counts), indicating a nonuniform distribution of microorganisms due to, for example, point contamination of the fillets during processing or atmosphere leakage in the packages during storage. The packages were controlled before analysis, so the uneven distribution of fillet microbiota (concentration and community structure) is likely the reason. Thus, attention should be given to the overall picture, not on specific sampling days.
There are no specific criteria for APC of RTE raw salmon loins, but several guidelines and studies state APC levels > 6 log CFU × g −1 as borderline or unsatisfactory quality (Food Safety Authority of Ireland, 2020; Gilbert et al., 2000). The estimated duration of the lag-phase by the primary model of Baranyi and Roberts was 0.9 ± 2.6 days for APC in the control-vacuum group, which increased to 9.7 ± F I G U R E 3 Evolution of total aerobic plate counts (APC) and H 2 S-producing bacteria of pre-rigor vacuum-packed Atlantic salmon loins pretreated with CO 2 (SGS-vacuum) or air (control-vacuum) during storage at 4 • C. Each sampling point represents the mean bacterial count (n = 3) ± SE. Legends: . APC, control-vacuum; . APC, SGS-vacuum; . H 2 S-producing bacteria, control-vacuum 1.1 days for SGS-vacuum. Thus, the CO 2 pretreatment can be an efficient hurdle to delay microbial growth, similar to previously reported for pre-rigor filleted Atlantic salmon packed in MA with or without a CO 2 emitter (Hansen et al., 2009a). The lag-phase extension depends on the initial microbiota present and the amount of dissolved CO 2 in the food matrix as Gram-negative spoilage bacteria, in general, are more influenced by the concentration of dissolved CO 2 than Gram-positive bacteria (Devlieghere & Debevere, 2000). Furthermore, Devlieghere and Debevere (2000) established an inverse relationship between the concentration of dissolved CO 2 and the lag phase for common spoilage microorganisms.
After microbiological growth was initiated in the present study, the estimated maximum growth rate of APC was similar under the two packaging conditions applied (m max 0.41 ± 0.07 day −1 and m max 0.43 ± 0.11 day −1 for controlvacuum and SGS-vacuum samples, respectively). Contradictory, SGS packaging in combination with conventional heat treatment (Abel et al., 2019) or microwave heat treatment (Lerfall et al., 2018b) approximately halved the maximum growth rate of APC compared to heat treatment of vacuum-packed salmon. However, the latter study reported no effect of CO 2 emitters on the growth rate of APC compared to vacuum. Devlieghere and Debevere (2000) established a linear relationship between dissolved CO 2 in the food matrix and mmax of selected spoilage bacteria; however, as APC reflects the total microbial community, it cannot be compared directly.
H 2 S-producing bacteria was not detected in the raw material. Pretreatment with CO 2 resulted in no detectable H 2 S-producing bacteria, whereas low counts (<3.8 log CFU × g −1 ) were detected from day four and throughout the storage period for the control-vacuum group (Figure 3). H 2 S-producing bacteria like Shwewanella putrefaciens is known to be CO 2 -sensitive (Boskou & Debevere, 1998;Dalgaard, 1995). The CO 2 -sensitivity of S. putrefaciens has been confirmed in MA-packages of for example, cod (Hansen et al., 2007) and saith (Lerfall et al., 2018a). The result also agrees with the previous results of Hansen et al. (2009b), which detected low counts of H 2 S-producing producing bacteria in MA-packed salmon. Boskou and Debevere (1998) reported that a pH-drop of 0.2 units, from 6.4 to 6.2, increased the lag-phase of S. putrefaciens in MA-packed cod. Thus, the pH difference between the groups in the presented study might also influence the growth of H 2 S-producing bacteria. Fuentes-Amaya et al. (2016) demonstrated the ability of H 2 S-producing bacteria to proliferate in vacuum-packed Atlantic salmon at 4 • C, thus including SGS-technology as an extra barrier can prevent spoilage due to the growth of this bacteria group.
A significantly higher LAB concentration (GLM, p = 0.016) was observed for the control-vacuum samples than the SGS-vacuum samples during storage, reaching a final concentration at day 15 of 6.2 ± 1.1 log CFU × g −1 and 2.0 ± 2.0 log CFU × g −1 , respectively. Lactic acid bacteria were not detected in the first 6 days of storage in neither group, but overall, a significant correlation between the concentration of APC and LAB was found for the control-vacuum group (p < 0.01, r = 0.896). Application of MA-and vacuum packaging selects for growth of LAB (Gram & Dalgaard, 2002). However, in the present study, we aimed to inhibit the development of LAB with the CO 2 pretreatment. The same phenomenon was reported by (Hansen et al., 2009a) by combining MA packaging and CO 2 emitters for pre-rigor Atlantic salmon fillets.
B.thermosphacta, a common spoilage bacteria in MAand vacuum-packed fish and meat products Mamlouk et al., 2012;Stanborough et al., 2017), was only detected in the control-vacuum samples at day 15 with a count of 2.3 ± 1.2 log CFU × g −1 indicating that CO 2 pretreatment of vacuum-packed salmon loins can be beneficial to prevent the growth of this spoilage bacteria. B. thermosphacta are also previously described to be inhibited by dissolved CO 2 in the food matrix, although it is a Gram-positive bacteria (Devlieghere & Debevere, 2000). The use of packaging atmosphere with less O 2 and more CO 2 has previously been proposed to reduce the spoilage potential of B. thermosphacta in MA-packed shrimps.

ATP degradation products
Fish deterioration and loss of seafood quality can be monitored by analyzing ATP degradation products (Hong et al., 2017;Howgate, 2006). Endogenous and bacterial enzymes catalyze post-mortem degradation of ATP in the fish muscle through the intermediate products ADP, AMP, IMP, Ino, and Hx. Inosine monophosphate is associated with fish freshness and the pleasant umami taste (Hong et al., 2017), and effort should be made to maintain the freshness of the IMP level in seafood. The further degradation from Ino to Hx result in the development of unpleasant flavors in stored fish (Hong et al., 2017;Howgate, 2006). In the present study, the samples showed a significant drop in the concentration of IMP and a rapid increase in Ino level between storage days 0 and 4 regardless of packing ( Figure 4). The highest concentration of Ino was found at day 11 for both groups, at the time point where IMP was depleted in the control-vacuum samples. The IMP concentration was significantly higher in the SGS-vacuum samples than the control-vacuum samples at storage days 11 and 15 (One-way ANOVA, p = 0.004 and p = 0.047, respectively). Findings from the present study indicate that SGStechnology prolongs the freshness of pre-rigor vacuumpacked salmon loins by slowing down the degradation of IMP to Ino.
A significant effect of the packaging conditions was found on the muscle concentration of Hx (GLM, p = 0.05). Between storage days 0 and 4, the value was low in both groups (Figure 4), however from day 11, the concentra-tion was higher in the control-vacuum samples than in the SGS-vacuum samples. On day 15, a maximum concentration of 1.48 and 0.94 mmol/g of Hx was reached in the control-vacuum and SGS-vacuum samples, respectively. The results are consistent with the significantly higher APC level in control-vacuum than SGS-vacuum samples, as Hx formation is catalyzed by both endogenous fish enzymes and bacterial enzymes (Hong et al., 2017). A significant correlation was found between the Hx concentration and the level of APC, H 2 S-producing bacteria, IMP and Ino (r = 0.842, r = 0.564, r = −0.848, r = 0.673, respectively, p < 0.01) The K value is used to evaluate the quality of raw fish freshness before significant microbiological spoilage is initiated (Hong et al., 2017). The raw material of day 0 had a low K value of 3.6 ± 0.4%, indicating fish of superior quality as the final degradation products of ATP (In and Hx) are barely present ( Figure 5). On day 11, the K value in control-vacuum samples reached 94.7 ± 1.4%, significantly higher than the SGS-vacuum samples at 81.7 ± 1.7% (Oneway ANOVA, p = 0.003).
Japanese researchers have suggested a K value of 20% as a limit for raw fish of sashimi quality (Hamada-Sato et al., 2005). However, Erikson et al. (1997) suggested salmon with K values lower than 50% to be of excellent quality. K value as a freshness indicator is disputed as it is highly dependent on species, and external factors such as season, handling conditions, and method of killing (Hong et al., 2017) H value is a quality parameter used to evaluate seafood spoilage as it reflects the amount of Hx accumulated. TA B L E 2 Color parameter (L*, a*, and b*) values of pre-rigor vacuum-packed salmon loins during 15 days of refrigerated storage at 4 • C (n = 3 for packed samples, n = 9 for raw material at day 0). The loins were pretreated with air (control-vacuum)  Hx are formed from the autolytical breakdown of ATP and produced by spoilage microorganisms (Hong et al., 2017). In the present study, the SGS-technology resulted in significantly lower H values at day 11 and 13 (One-way ANOVA, p = 0.011 and p = 0.046 respectively), reaching 11.4 ± 1.1% and 16.4 ± 0.84% for SGSvacuum and control-vacuum samples respectively at day 11. In addition, a significant correlation was found between H value and APC for both groups (r = 0.804 and r = 0.931 for SGS-vacuum and control-vacuum, respectively), indicating a microbial contribution to the Hx formation.

Color
Pre-rigor filleted RTE salmon is regarded as a high-quality sashimi product. The appearance is therefore of high importance to the consumers. Color is generally perceived as one of salmon fillets' most important quality parameters (Anderson, 2001). The price per kg of salmon is positively correlated to fillet redness (Haegermark, 2012). Thus, color deviations of products in new packaging compared to the conventional product can be unfavorable as they affect the consumers' perception of the product and willingness to pay (Alfnes et al., 2006). For the present experiment, the coulometric variables L*, a*, and b* with respect to storage day and packaging conditions are given in Table 2.
The flesh appearance changed after packing. The samples were lighter (higher L*-value) and more reddish (higher a*-value) at the first sampling point than the raw material at day 0. However, no differences in the flesh appearance were found between the SGS-vacuum and the controlvacuum samples. Furthermore, the storage time did not affect the visual appearance of the samples between days 4 and 15 of storage. Hence, the increased amount of CO 2 due to the CO 2 pretreatment did not pose any visual effect. Contradictory, Hansen et al. (2009a) found MA-packed Atlantic pre-rigor salmon with and without CO 2 emitters to have more redness and yellowness after 15 days of cold storage than vacuum-packed samples. Chan et al. (2021) reported that MA-packed (60% CO 2 , 40% N 2 ) salmon fillets were darker, more reddish, and yellowish than vacuum skin-packed fillets. Choubert and Baccaunaud (2006) demonstrated that a gas mixture composed of CO 2 :N 2 (40:60) preserved the color of the trout fillets to a greater extent than air during 3-week refrigerated storage. On the other side, bleaching of fillets may occur during storage in high headspace CO 2 concentrations (90%) (Barnett et al., 1982). Most of these studies do, however, report color changes in MA-packaged salmon portions allowing the surface to dry during storage. In the present study, this was not the case since the packaging film was in contact with the product maintaining the surface humidity and thereby giving better protection of the color of the products during storage.

CONCLUSION
The present study demonstrated significantly higher amounts of dissolved CO 2 in pre-rigor than in postrigor Atlantic salmon loins. However, the practical effect of the increased CO 2 solubility in a real-world application is probably of minor significance. In addition, favorable effects of SGS-treatment in combination with vacuum packing of pre-rigor loins of salmon stored at 4 • C were demonstrated. The SGS-technology resulted in lower growth rates of APC, total inhibition of H 2 S producing bacteria, and slower formation of Hx compared to the control group. Furthermore, no significant effect on the color appearance of the salmon filets due to the SGS-treatment was found. The results demonstrate that SGS technology combined with vacuum packaging is suitable for preserving high-quality pre-rigor filleted salmon products. The novel packaging solution represents a high industrial potential for value-added product development for a broader market.

A C K N O W L E D G M E N T
The Research Council of Norway financially supports this work through the research project "Concept development of full-scale soluble gas stabilization (SGS) technology for seafood" (project number 294641).

C O N F L I C T S O F I N T E R E S T
The authors declare no conflicts of interest.