Development of Lipid Changes Related to Quality Loss During the Frozen Storage of Farmed Coho Salmon (<Emphasis Type="Italic">Oncorhynchus kisutch</Emphasis>)

Lipid changes related to quality loss were evaluated during frozen storage of coho salmon for up to 15 months. Biochemical indices concerning lipid hydrolysis (free fatty acids, FFA) and oxidation (peroxide value, PV; thiobarbituric acid index, TBA-i; fluorescent compounds, FR; polyene index, PI) were determined and compared to sensory (odor and taste) and endogenous antioxidant (tocopherol isomers and astaxanthin) assessments. As a result of the frozen storage, lipid hydrolysis was shown to develop according to the increase in FFA content (p < 0.05). However, most biochemical lipid oxidation indices (PV, TBA-i and FR) led to a low degree of rancidity development (p < 0.05) when compared to other fatty fish species under similar frozen storage conditions. The PI value decreased (p < 0.05) at month 10 but then remained unchanged until the end of the experiment. Rancid odor and taste development were shown to be low throughout the experiment, according to the biochemical indices mentioned above. However, a progressive decrease (p < 0.05) in the original fresh odor and taste of salmon fish flesh occurred with increasing frozen storage time, such that fish samples had the poorest scores by month 15. Endogenous antioxidants were remarkably stable throughout the experiment and which might contribute to the oxidative stability of frozen farmed coho salmon lipids.     

Effect of advanced chilling methods on lipid damage during sardine (Sardina pilchardus) storage

Slurry ice is a biphasic system consisting of small spherical ice crystals surrounded by seawater at subzero temperature. Its effect on lipid damage (hydrolysis and oxidation) was evaluated during the chilled storage of a fatty fish species, sardine (Sardina pilchardus). Slurry ice treatment was checked alone and in combination with ozone and compared to traditional flake icing during a 22‐day storage. Different lipid damage indices (free fatty acids, FFA; peroxide value, PV; thiobarbituric acid index, TBA‐i; fluorescent compounds, FR) were checked and compared to sensory assessment and nucleotide degradation (K value). According to lipid hydrolysis (FFA) and oxidation (PV and FR) developments, slurry ice showed an inhibitory effect (p <0.05) on lipid damage during storage, as well as an inhibition of nucleotide autolytic degradation. Ozonised slurry ice did not provide differences (p >0.05) from slurry ice alone when considering lipid hydrolysis, nucleotide degradation and some lipid oxidation indices (PV and FR), although a higher (p <0.05) TBA‐i was observed at day 22 of storage when compared to flake ice and slurry ice treatments. However, a lower (p <0.05) fluorescence development was observed for fish treated under ozonised slurry ice when compared to traditionally iced fish. Sensory assessment showed a higher shelf life for fish samples treated under ozonised slurry ice than for their counterparts under slurry ice (15 d versus 12 d), while flake icing led to a far shorter shelf life (5 d). According to sensory and biochemical (lipid matter and nucleotide) analysis, slurry ice has proved to be a promising technology for damage inhibition and quality retention in a fatty fish species such as sardine. Ozonised slurry ice was also shown to be useful, since a longer shelf life was obtained in the present experiment and a pro‐oxidant effect of ozone on sardine lipids was not proved.     

Effects of thermal treatment on fish lipids‐amylose interaction

The paper describes a study on effects of thermal treatments (microwave heating and freezing) on fish lipids‐amylose starch interactions. Particular attention was paid to lipid availability (extractability from the system) and contribution of fatty acids to various groups of lipids after interactions produced by mixing, microwave heating and storage. Analyses were made on model systems containing fish lipids at different oxidation levels and gelatinised with 10% aqueous solution of amylose starch. The lipid:starch ratio was 1:1. The systems were assayed before and after mixing, microwave heating (4 min, 90 W), freezing (30 d, –18 °C), and heating followed by freezing. Lipid extractability (selective extraction), peroxide value (PV), anisidine value (AsV), fluorescence, and fatty acid profiles (GC/MS) were determined. Mixing of fish lipids with amylose was shown to result in lipid‐amylose interaction. The thermal treatments applied (microwave heating or freezing at –19 °C for 30 d) were observed to exert different, significant effects on fish lipids‐amylose starch interactions. The effects depended also on the extent of lipid oxidation. Fatty acids of the lipids were bound selectively, PUFA, and particularly DHA, were subjected to stronger binding and higher amounts of those acids were bound.     

Stability and Oxidizing Effect of Membraneless Electrolyzed Water

Membraneless electrolyzed water (MLEW) is generated by electrolysis of dilute salt water solution without a membrane separating cathodic and anodic water. The stability of MLEW under different storage temperatures and pHs, as well as its lipid oxidizing effects have not been extensively studied. In this study, three storage temperatures and two pH conditions were used to study the stability of the water. MLEW's lipid degradation potential was investigated using two methods by measuring the degree of oxidation of lipids containing polyunsaturated fatty acids. Lipid oxidation in an emulsion model free of other interferences was conducted with MLEW containing 30, 50 and 100 mg L^?1 free chlorine (FC), and lipid oxidation in a salmon fish fillet was conducted by washing the fillet pieces with MLEW containing 100 mg L^?1 FC. We found that the MLEW was relatively stable in capped containers under both acidic and neutral pH during the 21‐day period. Although there was significant lipid oxidation in the emulsion system, and the extent of oxidation was FC concentration dependent, we did not observe significantly more lipid oxidation in fish treated with MLEW compared to the control (stored under 4 °C for 10 days). Therefore, we conclude that MLEW may be used as a disinfectant in commercial operations without causing significant lipid oxidation and rancidity generation in foods.     

Effects of lipid oxidation on fish lipids — amylopectin interactions

The results of studies on fish lipid oxidation effects on lipid‐amylopectin starch interactions are presented. Particular attention is paid to fish lipid availability (extractability from the system) and fatty acid contributions to individual lipid groups after mixing‐provoked interaction and after a 30‐day storage at —18 °C. The study involved model systems containing fish lipids at different levels of oxidation, lipids containing an antioxidant (BHA), and gelatinised amylopectin starch in a 10% aqueous solution. The lipid to starch ratio was 1:1. The test systems were subjected to selective extraction. The extracts were assayed for lipid content, peroxide value, anisidine value, fluorescence, and fatty acid composition. Compared to fresh fish lipids, those lipids, which were oxidised to a higher extent were shown to be more amenable to complexing with amylopectin, but they were also more readily released during frozen storage. On the other hand, addition of BHA stabilised the lipid‐amylopectin starch interaction during storage at —18 °C. In the entire systems tested, polyunsaturated fatty acids, eicosapentaenoic and docosahexaenoic acids in particular, proved to be most susceptible to binding, up to 90% of the latter being complexed.     

Studies of chemical methods for assessing oxidative quality and storage stability of feeding oils

Peroxide value (PV), thiobarbituric acid number (TBA), anisidine value (AV), percent free fatty acids (FFA), and carbonyl value (CV) methods were compared in different experiments for their sensitivity and practicality in assessing the oxidative quality of four feeding oils: salmon oil, soybean oil, canola oil (oil from low erucic acid, low glucosinolate rapeseed), and canola soap stocks. In the first experiment, among the four methods (PV, AV, TBA and FFA) studied, PV appeared to be the most practical for its sensitivity, simplicity, and economy when the four oils were oxidized by bubbling air through at room temperature for 792 hr. In the second experiment, using herring oil, all four methods tested (PV, AV, TBA and FFA) were sufficiently sensitive once the oxidation of herring oil had passed the induction period and the oil was highly oxidized. In the third experiment, of the four methods (PV, AV, TBA and CV) compared, AV was the most sensitive for measuring the oxidative quality of canola oil aerated at 100 C for 240 hr. Results of further studies suggested that herring and canola oils stored under commercial conditions were stable for at least one year.     

Reaction of autoxidizing linoleate with coho salmon myosin

Reactions that contribute to denaturation, destruction and quality changes of fish muscle lipids and protein were studied. Compounds with characteristic fluorescence spectra were isolated by extraction and thin layer chromatography from an autoxidizing system consisting of sodium linoleate, Coho salmon myosin and buffer. Similar compounds were also present in extracts from freeze-dried salmon steaks and salmon kept frozen at −20 C for 1 year. TBA values and oxygen uptake of the autoxidizing system showed initial rapid increases with time followed by a significant decrease in TBA values and gradual leveling off of oxygen uptake upon prolonged oxidation. IR spectra before and after borohydride reduction and UV, visible and fluorescence spectra indicated the presence of C=N functional groups in extracts from the various samples. These compounds were not extractable from the control myosin solutions allowed to oxidize without addition of linoleate. Amino acid analyses of the myosin from the autoxidizing system, when compared with nonoxidized myosin-linoleate systems, indicated significant decreases in the amounts of histidine, lysine and methionine following oxidation. These reactions apparently contribute to the denaturation and destruction of the lipids and protein in the model system, as well as in a frozen or freeze-dried product. Michigan Agricultural Experiment Station Journal Article 6304.     

Fatty acid composition and oxidation of lipids in Korean catfish

This study determined the lipid content and FA composition of muscle and a mixture of muscle and viscera from Korean catfish as well as lipid oxidation and hydrolysis. Lipid content and FA compositions in Korean catfish, which were purchased every month or two during September 1999–July 2000, were analyzed. Lipid oxidation and hydrolysis were determined as PV, thiobarbituric acid value, and FFA in the muscle and the mixture during storage at 2°C for 12 d and −14°C for 9 wk. Lipid contents of the muscle and the mixture were 3.2 (w/w) and 5.4%, respectively. Oleic acid was the most abundant FA in the catfish lipids, constituting 28.0% (w/w) of total FA in the muscle and 25.6% in the mixture, followed by palmitic acid and linoleic acid, amounting to 20.2 and 12.2%, respectively, in the muscle and 20.2 and 12.5% in the mixture. EPA and DHA were 3.2 and 6.8%, respectively, in the muscle and 3.5 and 8.1% in the mixture. Seasonal variation in lipid contents and FA composition was minimal in catfish. Lipids in minced catfish oxidized and hydrolyzed readily at 2°C. Inclusion of viscera into the muscle increased lipid oxidation and hydrolysis. Frozen storage at −14°C and addition of ascorbic acid both reduced lipid oxidation. Frozen storage retarded lipid hydrolysis in catfish.     

MCT Oil Coating Improves the Oxidative Stability of Surface Lipids in Corn Extrudates

Lipid‐containing extrudates are highly susceptible to lipid oxidation because of their porous structure and high specific surface area. The objective of this study is to investigate the effect of a low‐oxidizable medium‐chain triglycerides (MCT) coating on the oxidative stability. Therefore, cornmeal is extrusion cooked with commercial vegetable oil and various water contents (10–18% on dry matter) and partially coated with 5% w/w MCT oil. Lipid oxidation in the extrudates is examined during storage at +40 °C by measuring hydroperoxide concentrations in different lipid fractions (surface, inner‐surface, and matrix‐incorporated lipids). Hexanal is analyzed using headspace‐gas chromatography. Coating allocation is studied by fatty acid profiles of the lipid fractions, by fluorescence microscopy, and computerized microtomography. Application of an MCT coating leads to lower hydroperoxide concentrations and significantly reduces hexanal formation during storage. In particular, the surface lipids are stabilized. MCT oil diffuse only to small extent inside the extrudates and its effect on lipid oxidation can be attributed to differenct effects: filling surface microcracks, reducing the release of volatiles, and dilution of oxidizable lipids. This effect is strongest in porous extrudates with low feed‐water content. Practical applications: Inhibition of lipid oxidation of fatty extrudates is of great importance for producers as well as consumers. In this study, we examine whether it is possible to improve the oxidative stability of extrudates by application of a lipid‐based coating. The results of this study suggest that it will be possible in the future to produce extrudates that are protected from oxidation by a thin natural oil film. We expect that especially products such as kibbles could benefit from treatment with a low‐oxidizable lipid‐based coating. In these products, fats are already used in combination with other functional ingredients. The combination of an MCT coating with a palatant or antioxidant is, therefore, an obvious alternative and of high relevance for manufacturers.     

Mackerel skin lipids as an unsaturated fat model system for the determination of antioxidative potency of TBHQ and other antioxidant compounds

A comparative study on the activity of antioxidation of butylhydroxy anisole, (BHA), butylhydroxy toluene (BHT),tert-butylhydroquinone (TBHQ), α-tocopherol, and tempeh oil has been investigated by using the readily oxidizable mackerel skin lipids as the tested model system. The oxidation rate of the tested lipids was mainly followed by measuring the weight gain, but some peroxide value, thiobarbituric acid (TBA), and free fatty acid (FFA), and the changes of composition of fatty acids and carbonyls were also determined and used for some practical evaluation of the various influences of these antioxidatns. Overall, the order of effectiveness for inhibiting the oxidation in mackerel skin lipids has found to be TBHQ > α-tocopherol > tempeh oil > BHA > BHT at the concentrations of 0.02% for all synthetic compounds, and 0.1% and 5% for α-tocopherol and tempeh oil, respectively. In addition, TBHQ has demonstrated not only to be the most powerful antioxidant for the unsaturated marine lipids but also to retard the formation of FFA and carbonyls from lipid hydrolysis and secondary oxidation reactions. The superior effectiveness of TBHQ in the highly unsaturated mackerel skin lipids is especially noteworthy in order to improve oxidative stability of marine oils and some fatty fishes.     

Influence of blanching treatment and drying methods on the drying characteristics and quality changes of dried sardine (Sardinella gibbosa) during storage

The aim of this study is to examine the drying characteristics of blanched and unblanched sardine during indoor and open sun drying processes. Changes in temperature and relative humidity of the air during drying were recorded. The color, peroxide value (PV), thiobarbituric acid reactive substances (TBARS), free fatty acid (FFA) content, fatty acid composition, and sensory attributes of dried samples were also evaluated once a month for 5 months of storage. High drying rates were obtained in all samples at the start of drying and then decreased with increasing drying time. The highest drying rate and effective water diffusivity (D_eff) were observed in blanched sardine during open sun drying. Blanching treatment slowed down the FFA progression during product storage but adversely affected the color, PV, and TBARS content as well as sensory properties. Although sardine dried for a longer time under indoor drying conditions, it attained a stable moisture ratio that was lower than in open sun-dried samples. Indoor drying produced a quality stable product with less lipid oxidation and the desired moisture content, higher polyunsaturated fatty acids and sensory properties. Blanching treatment negatively affected the fish quality and is therefore not recommended for commercial sardine drying.     

Effect of Temperature and Time on the Stability of Salmon Skin Oil During Storage

The effect of 45 days of storage at 25, 4, −18 and −80 °C on the quality indices; free fatty acid (FFA) content, peroxide value (PV), thiobarbituric acid reactive substances (TBARS), and changes in the fatty acid (FA) profile of crude oil recovered from salmon fish skins were evaluated at 5 day intervals using spectrophotometric and titrimetric methods. Higher temperatures and longer storage time resulted in higher quantities of oxidative products in the salmon skin oil (SSO). By day 45, SSO stored at 25 and 4 °C had 8.50 and 8.29% FFA, 32.43 and 26.33 μg malondialdehyde (MDA) eq g⁻¹ oil, and 88.19 and 64.53 mequiv peroxide kg⁻¹ oil, respectively. No significant (p > 0.05) changes in fatty acid profile were observed at all the storage temperature and time studied.     

Effects of rainbow trout freshness on n‐3 polyunsaturated fatty acids in fish offal

The effects of rainbow trout cold storage on the quality of offal left after fish processing to fillets with skin were determined. The intact farmed rainbow trout were kept at 2 °C in ice for 0, 4, 7, and 14 days of storage. The offal was, immediately after processing, frozen at ?20 °C and analysed after a month‐long frozen storage; fillets (non‐frozen) were analysed as well. Non‐protein nitrogen, volatile bases, trimethylamine, lipid oxidation (peroxide value, anisidine value, UV‐VIS spectra, and fluorescence) and fatty acid composition were determined. The offal consists in 15% of protein and in about 20% of chloroform/methanol‐extractable lipids, with n‐3 polyunsaturated fatty acids (n‐3 PUFA) accounting for 20.37 ± 1.25% of the fatty acids. The fish storage duration was found to exert a significant (p = 0.05) effect on the changes in lipids and nitrogen compounds. No losses of long‐chain n‐3 PUFA in the offal were detected during the 2 wk of storage in ice plus 1 month at ?20 °C. The rainbow trout offal is a valuable – rich and stable – source of n‐3 PUFA.     

Effects of Added Phosphates on Lipid Stability During Salt Curing and Rehydration of Cod (Gadus morhua)

Effects of added phosphates on retardation of lipid oxidation of salted cod during processing, storage and after rehydration were investigated. Lipid hydrolysis progress and development of color, primary and secondary lipid oxidation products and fluorescence intensities were determined. Added phosphates significantly retarded lipid hydrolysis and lipid oxidation progress, resulting in lower free fatty acid , lipid hydroperoxides (PV), thiobarbituric acid-reactive substances (TBARS) as well as fluorescence intensities (δF _or and δF _aq). Significant correlation between the lipid oxidation products (PV, TBARS, δF _or and δF _aq) and yellow/brownish discoloration (b* value) of salted cod was observed. Principal component analysis showed that TBARS, b* value and δF _or were the strongest indicators of lipid oxidation during salting and storage.     

Effect of high‐pressure treatment on microbial activity and lipid oxidation in chilled coho salmon

This work studies the effect of a previous hydrostatic high‐pressure (HHP) treatment on chilled farmed coho salmon (Oncorhynchus kisutch). Three different HHP conditions were applied (135 MPa‐30 s, 170 MPa‐30 s, and 200 MPa‐30 s for treatments T‐1, T‐2, and T‐3, respectively) and compared to untreated (control) fish throughout a 20‐day chilled storage. Microbial activity and lipid oxidation development were analyzed. Assessment of aerobe, psychrotroph, Shewanella spp. and Pseudomonas spp. counts and trimethylamine formation showed a marked inhibitory effect (p <0.05) of HHP treatment on microbial activity, with this effect increasing with the pressure value employed. Related to lipid oxidation development, higher peroxide mean values (day 10–20 period) were found in control samples and fish treated under T‐1 condition when compared to their counterparts corresponding to T‐2 and T‐3 treatments. On the contrary, quantification of thiobarbituric acid‐reactive substances and fluorescent interaction compounds showed higher levels (p <0.05) in fish samples corresponding to T‐2 and T‐3 treatments. In spite of the lipid oxidation development found, polyene index and tocopherol isomer (α and γ) content did not provide differences (p >0.05) as a result of previous HHP treatment.     

Preservative effect of an organic acid‐icing system on chilled fish lipids

This study provides a first approach concerning a novel chilling strategy, which employs a mixture of different preservative organic acids (ascorbic, citric and lactic) in the icing medium. Thus, ice prepared from water including two different concentrations of a commercial acid mixture‐formula (800 and 400 ppm; C‐800 and C‐400 conditions, respectively) were applied as icing system to three important commercial fish lean species (hake, Merluccius merluccius; megrim, Lepidorhombus whiffiagonis; angler, Lophius piscatorius). Lipid oxidation (peroxide value; thiobarbituric acid index; fluorescent compound formation) and hydrolysis (free fatty acid formation) were evaluated throughout the chilling time (up to 12–15 days) and compared to results obtained in fish kept under traditional ice prepared only from water (C‐0 condition); a complementary sensory evaluation was carried out. As a result of employing the C‐800 icing condition, a partial inhibition of lipid oxidation and hydrolysis development was obtained that was accompanied by a shelf life enhancement in all cases. Further research taking into account the complementary action of the present organic acids is envisaged. According to the lipid damage analysis, lipid hydrolysis was a more relevant event than lipid oxidation in all fish species tested.     

Lipid damage during frozen storage of Gadiform species captured in different seasons

Quality loss of two gadiform fish species (blue whiting, Micromesistius poutassou; hake, Merluccius merluccius) during frozen storage (–30 and –10 °C; up to 12 months) was studied. For this, hydrolytic (formation of free fatty acids, FFA) and oxidative (conjugated dienes, peroxide and interaction compound formation) lipid damage were analysed. For both species, individual fishes captured in two different trials (May and November) were considered. Increasing (p <0.05) lipid hydrolysis and oxidation (peroxide and interaction compound formation) were observed for all kinds of samples throughout the frozen storage. Interaction compound detection by fluorescence analysis showed the best correlation values with storage time. Some higher (p <0.05) hydrolysis development could be observed in hake captured in May than in its counterpart from the November trial, while frozen blue whiting did not provide definite differences for FFA formation between both trials. Concerning peroxide formation, higher (p <0.05) values were obtained for individual blue whiting and hake captured in November when compared to their corresponding May fish for both frozen storage conditions. Interaction compound formation was also found to be higher (p <0.05) for November hake fish than for its counterpart captured in May, while blue whiting did not provide definite differences between trials.     

Metal-catalyzed lipid oxidation and changes of proteins in fish

The addition of metals to pure fats and model systems has given us a picture of the part they play in lipid oxidation. But in complex substrates such as fish flesh the same metal ions catalyze reactions in other components of the muscle as well. A more complete picture of deterioration is obtained when we also examine the effect of metals on the other components of the muscle. Although lean fish muscle contains 0.5% to 1% unsaturated lipids, during frozen storage it rarely goes rancid, as indicated by thiobarbituric acid (TBA) values or rancid odors. The lipids do oxidize, but instead of forming carbonyls and other compounds associated with rancidity, they become bound up in lipid-protein complexes, which accounts for the toughened texture of overstored or poorly stored frozen fish. It has been generally accepted that the formation of these lipid-protein polymers is brought about by a reaction between the proteins and oxidizing fatty acids. For fish of the gadoid or cod family, this is further complicated by the enzymic production of formaldehyde, which forms in the muscle during cold storage. The direct addition of formaldehyde to give concentrations of 0.001% to 0.05% caused marked reductions in the extractable protein of cod muscle. Because formaldehyde and dimethylamine (DMA) are produced in the stored muscle by the same reaction, the accumulation of the DMA can be used as a measure of this process which leads to protein insolubility. Removal of the dark lateral muscle from the fillets before freezing reduced the production of DMA and formaldehyde during storage and resulted in less loss of extractable protein. One of 28 papers presented at the Symposium “Metal-Catalyzed Lipid Oxidation” presented at the ISF-AOCS, World Congress, Chicago, September 1970.     

Oxidative flavour deterioration of fish oil enriched milk

The oxidative deterioration of milk emulsions supplemented with 1.5 wt‐% fish oil was investigated by sensory evaluation and by determining the peroxide value and volatile oxidation products after cold storage. Two types of milk emulsions were produced, one with a highly unsaturated tuna oil (38 wt‐% of n‐3 fatty acids) and one with cod liver oil (26 wt‐% of n‐3 fatty acids). The effect of added calcium disodium ethylenediaminetetraacetate (EDTA) on oxidation was also investigated. Emulsions based on cod liver oil with a slightly elevated peroxide value (1.5 meq/kg) oxidised significantly faster than the tuna oil emulsions, having a lower initial peroxide value (0.1 meq/kg). In the tuna oil emulsions the fishy off‐flavour could not be detected throughout the storage period. Addition of 5—50 ppm EDTA significantly reduced the development of volatile oxidation products in the cod liver oil emulsions, indicating that metal chelation with EDTA could inhibit the decomposition of lipid hydroperoxides in these emulsions. This study showed that an oxidatively stable milk emulsion containing highly polyunsaturated tuna fish oil could be prepared without significant fishy off‐flavour development upon storage, provided that the initial peroxide value was sufficiently low.     

Fucosterol Causes Small Changes in Lipid Storage and Brassicasterol Affects some Markers of Lipid Metabolism in Atlantic Salmon Hepatocytes

Several feeding trials with Atlantic salmon fed naturally high phytosterol concentrations due to dietary rapeseed oil inclusion have shown changes in lipid metabolism and increased hepatic lipid storage in the fish. An in vitro trial with Atlantic salmon hepatocytes was, therefore, performed to study the possible direct effects of phytosterols on lipid storage and metabolism. The isolated hepatocytes were exposed to seven different sterol treatments and gene expression, as well as lipid accumulation by Oil Red O dyeing, was assessed. Fucosterol, a sterol found in many algae species, had an effect on the size of individual lipid droplets, leading to smaller lipid droplets than in the control without added sterols. A sterol extract from soybean/rapeseed led to an increase in the percentage of hepatocytes with visible lipid droplets at 20× magnification, while hepatocytes of both the sterol extract‐treated groups and fucosterol‐treated groups had a larger proportion of their area covered with lipids compared to control cells. Brassicasterol, a sterol characteristic of rapeseed oil, was the only sterol treatment leading to a change in gene expression, affecting the expression of the nuclear receptors, peroxisome proliferator‐activated receptor gamma (pparg) and retinoid X receptor (rxr). The current study thus shows that phytosterols can have direct, although subtle, effects on both hepatic lipid storage and gene expression of Atlantic salmon in vitro.