A Study in the Use of a High

Concentration of CO 2 in a Modified Atmosphere to Preserve Fresh Salmon

HAROLD J. BARNETT, FREDERICK E. STONE, GLENN C. ROBERTS, PATRICK J. HUNTER, RICHARD W. NELSON, and JOSEPHINE KWOK

Introduction recently, however, commercial use of Problems appeared in the early tests, controlled atmosphere or modified at­ e.g., undesirable changes in quality in The preservative effects of carbon mosphere (MA) to preserve fish was not the salmon were found in several ship­ dioxide (C02)on protein foods has been seriously considered for use by the in­ ments. Changes in quality were related known for many years and is used to dustry because of economic and techni­ to inadequate cooling of the fish after preserve perishables, such as meats and cal considerations. catching and during handling and pro­ poultry. Early research by Coyne (1932), Then in 1977, a salmon processor cessing as well as to problems in adapting Stansby and Griffiths (1935), and others demonstrated interest in the techniques packaging for shipment. The latter con­ showed that CO2 atmospheres have a by making a few exploratory shipments sisted of using Wet- Lok- type containers, beneficial preservative effect on certain of fresh salmon in special refrigerated which were perforated to allow for re­ species of bottomfish. vans from Alaska to the Pacific North­ moval of drip and free passage of the More recently, the National Marine west. The interest, of course, was created MA. Drying, caused by the vans' forced­ Fisheries Service (NMFS) in cooperation by the increased value of salmon and air refrigeration system, resulted in prod­ with Whirlpool Corporation' (Nelson and changing markets. The shipments, made uct shrinkage and loss of quality through Tretsven, 1975) made a series of labora­ in MA containing high concentrations oxidation and discoloration. The high tory experiments to determine the pre­ of CO2and 02' demonstrated the poten­ concentration of oxygen used in the MA servative effects of controlled atmos­ tial for improving the keeping quality of gas mixture was thought to be a contrib­ pheres (CA) on fresh Pacific salmon, fresh salmon shipped in large containers uting factor to the oxidation and dis­ Oncorhynchus spp. Results from these or vans. coloration problem. The problem of studies showed that the shelf life of drying was partly controlled by top icing refrigerated salmon could be significantly the boxed fish for shipment. extended by storage in a CA containing Harold 1. Barnett, Research Chemist; Frederick Because of the need for more infor­ E. Stone. Chemist; Glenn C. Roberts, Chemist; 11.5 percent CO2, 87 percent nitrogen Patrick J. Hunter, Engineering Technician; and mation, we initiated studies to investigate (N 2), and 1.5 percent oxygen (02)' Until Richard W. Nelson, Sur.ervisory Chemical En­ the variables of low temperature and gineer, are with the UtIlization Research Divi­ sion. Northwest and Alaska Fisheries Center. high concentration of CO2 gas as an National Marine Fisheries Service, NOAA. effective mechanism to control the qual­ 2725 Montlake Boulevard East, Seattle. WA ity of fresh Pacific salmon held in MA. 'Reference to trade names or commercial tirms 98112. Josephine Kwok. Food Technologist. does not imply endorsement by the National was with the New England Fish Company. Pier In recent studies by Williams et al. Marine Fisheries Service, NOAA. 89, Seattle. WA 98119. (1978), it was found that at SoC (41 OF),

ABSTRACT-Commercial use of modi­ combined with the use ofa high concenlration phere. Carbon dioxide absorbed in the llesh lied atmospheres 10 preserve fish during 0/ CO2 gas as an effective combination ./01' a/the salmon causes some swelling in the shipment has been limited because of eco­ controlling the qualily of/resh Pacific salmon canned product but the condition can be nomic and technical reasons. Recenl interest held in modified atmospheres. Salmon were controlled. The modi/iedatmosphere melhod in the use a/the technique to ship Pacific held in acceptable condition for 2f days in a described can be considered for commercial salmon. Oncorhynchus spp.. aUla/ Alaska modified atmosphere containing 90 percel1l use. Recommendations are'made tor han­ has demonstrated the need /01' additional CO2, The effect a/the CO2 was to slow the dling the salmon lor shipping and ./01' addi­ in/ormation 10 improve handling andshipping growth 0/ spoilage bacteria extending the lional research on the preservative effects 0/ procedures. Studies were conducted to de­ li'esh shelf lite of the salmon from aboul 12 CO2 and other atmospheres with respect to termine the effect oflow Slorage temperature davs on ice to 21 davs in the modified atmos- di/jerentlish species.

March IIJH2. 44t3; 7 ing the container airtight. One- quarter­ inch, stainless steel needle valves. located near the bottom of one side and on top of the lid, provided the necessary nlling and venting ports for gassing the storage container. A stainless steel "pop-off" valve located on the lid prevented pres­ surizing the container. A multipoint re­ corder, equipped with stainless steel, sabretype thermocouples, was used to measure and record the internal temper­ atures of the storage container and fish. The equipment described here was used only to simulate commercial con­ ditions. Less elaborate, light-weight, multipurpose, plastic fish totes that can be made airtight for use with gases and shipped in refrigerated vans or by air are now being designed for commercial use. Besides eliminating some of the pre­ viously described drawbacks of using ,- ;.....A>•. cardboard containers, airtight bulk con­ tainers could expedite handling and Figures 1.-Modified atmosphere container showing refrigeration flex lines. stain­ less steel "pop-off" valve on lid, and thermocouples connected to a multipoint packing ofsalmon as weU as reduce costs. recorder (shown to the left rear of container). Such containers could also increase the potential for using the MA method in remote areas, such as Alaska's Bristol Bay. The containers could be multipur­ pose and made in a variety of sizes for shipping on vessels, by air, and in refrig­ growth of bacteria, predominantly asso­ ning, a limited study was made of the erated vans or used to bulk store nsh in ciated with the spoilage of fish. was potential use of MA-stored salmon for processing plants. This could be espe­ almost totally inhibited by a 100 percent canning. cially useful during peak operating times. CO atmosphere. In tests made by the 2 The Modilied Atmosphere System orthwest and Alaska Fisheries Center's Utilization Research Division, Seattle, Laboratory equipment for this exper­ Materials and Methods Wash., in cooperation with a local fish­ iment consisteu of a portable freon- 12 Sample Preparation processing company, walleye pollock refrigeration unit modi ned with a hot­ and Procedures and Pacific cod stored at OOC (32°F) in a gas, bypass system to maintain a partial MA containing 90 percent CO2 and 10 load. at temperature, to prevent periodic The 310 pounds of salmon used in percent air were judged superior in qual­ cycling of the compressor. The system this experiment consisted of mature ity to walleye pollock and Pacific cod was uesigned to maintain uniform prod­ coho, Oncorhynchus kisutch, and chum, stored at OOC (32°F) in atmospheres uct temperatures within ± 1°F (0.6°C). 0. keta, salmon that had been gillnetted containing 20 percent and 55 percent The contai ner (Fig. 1) used to store in Puget Sound about 24 hours prior to CO2 , Based on this knowledge. it was the nsh was of a special design. Portable beginning the experiment. They were decided to investigate the use of the and constructed of nberglass, it had a headed and gutted and well iced shortly higher concentrations of CO2 than now 1.700- pound nsh- holding capacity and after being caught. The internal tem­ commercially used for shipping salmon was insulated with 4 inches of urethane perature of the fish when loaded in the inMA. foam. Stainless steel chiller plates (coils) container was about 32°F (OoC). Many In this experiment, fresh, fall-caught mounted on the inner walls were con­ nsh showed signs of water marking and butchered Pacinc salmon were held at nected via special bulkhead nttings to loss of brightness of color but were OOC (32°F) in the MA for 21 days. This the refrigeration unit by standard refrig­ otherwise of very good quality. In the report describes the rationale and results eration flex lines. The false floor of the laboratory, the salmon were removed of the experiment. In view of the con­ tank was perforated to allow excess from the ice and placed randomly in the tinuing interest of the salmon canning melted ice, water, and drip to drain from storage container. Thermocouples were industry in methods of extending the the nsh during storage. The top of the inserted into two of the nsh, one located keeping quality of salmon prior to can- container was gasketed to facilitate mak- on the bottom of the load and one on the

Marine h~heries Rel'ie\\' very top. Other probes were placed inside flesh. A Corning combination electrode Table 1. - Microbiological and chemical changes occur­ ring in the flesh from butchered salmon held in a refrig­ the holding tank to measure the internal was used to make the measurement. erafed, controlled atmosphere. ambient temperature which was main­ Analysis of CO2 content in the flesh Chemical Data Total tained at 32°F (O°C) during the experi­ of the salmon was made by the following Storage bacterial time cO2 Conc TBA counts ment. The storage container was then method: A sample of flesh was blended (days)' pH (ppm) ( J1 M/100g) (no./gl 6.22 247 2.8 x 10' sealed and purged with CO2 gas to obtain in an alkaline (pH lOA) Tris buffer (0.05 o 025 6 6.23 760 023 4.2 < 10' an atmosphere mixture containing 90 M). and then treated with an acid buffer 10 623 687 0.67 1.0' 10' percent CO and 10 percent air. The to liberate CO , Concentrations were 15 6.11 817 044 1.3 ~ 10' 2 2 20 620 1.118 34 x 10' exact concentration of the gas was de­ read from a calibration curve prepared 21 622 913 0.64 2.0' 10' termined with a Burrell gas analyzer. The from millivolt (Mv) readings obtained 'Storage time given is the number of days held at the laboratory. CO2 gassing procedure was repeated with an Orion carbon dioxide electrode each time the container was opened to connected to an Orion model 801 pH examine and sample the fish. meter. The salmon were periodically sampled The salmon were chemically tested and evaluated for quality during the for oxidative rancidity (TBA number)

21-day experiment. Criteria included using the method described by Lemon phere containing 80 percent C02120 sensory attributes, chemical and physical (1975). Essentially, the procedure calls percent O2 for 14 days. Similar observa­ changes, and total aerobic bacterial plate for blending tissue in an extracting solu­ tions were made by Barnett et al. (1971 ) counts. At each examination period, two tion (trichloroacetic acid, propyl gallate, on whole yellowtail rocknsh and chum llsh were randomly sampled from stor­ and EDTA) and llltering. The filtrate is salmon, Oncorhynchus keta, held in re­ age, split along the long axis with one- half reacted with TBA reagent and boiled. frigerated seawater modined with CO2, of each fish retained for bacteriological The cooled sample is read spectropho­ King and Nagel (1975) in studies on examination. Each of the remaining sides tometrically. Pseudomonas aeruginosa suggested in­ was evaluated chemically and organo­ hibition of microbial growth by CO is Sensory Tests 2 leptically. Chemical analyses were made "in part due to a mass action effect on in duplicate of each fish. At each sampling, the raw salmon certain decarboxylating enzymes" which Two additional fIsh were selected at were judged for appearance, texture (as interferes with metabolic pathways es­ each examination period, washed, fro­ determined by nnger pressure), odor, sential for normal growth. Pseudomonas zen, glazed, vacuum sealed in a poly­ and color of the flesh. The salmon were spp. bacteria are generally associated ethylene bag, and stored at -18°C prepared for sensory evaluation by bak­ with spoilage of fresh fish and are con­ (-20 0 P) for future chemical and sensory ing in covered aluminum containers for trolled by the presence of as little as 5 examination. In addition. two fish were 15 minutes at 350°F. The samples were percent CO2 in the CA (Coyne, 1933). randomly selected at intervals from MA evaluated by a five-member taste panel storage, washed, trimmed, sectioned, and for flavor, texture, and rancidity using a Chemical Measurements packed in I/r pound cans and thermally 5- point numerical scale. A score of 2 or pH processed. Cans were air-cooled and below for the sensory criteria indicates a stored at ambient room temperature. product of unacceptable quality. Canned The pH values of the salmon (Table 1) salmon was evaluated for fill weight, did not change significantly from their Analytical Methods vacuum, color, and general appearance normal (initial) physiological pH during Total aerobic plate counts were made and acceptability. the test. Small differences are attributed by the method described by Pelroy and more to biological variation among the Results and Discussions Eklund (1966). Briefly, the method is as fish than to the effects of the CO2 , How­ follows: 45 g of excised fish flesh is Bacteriological Measurements ever, Makashev (1959) observed a two­ homogenized aseptically with 180 ml of fold increase in the acidity (pH) of fish sterile 0.1 percent peptone solution at Results of the bacteriological exami­ muscle stored in a CO2 environment. 3.0oC (38°F). Serial dilutions in 0.1 nations are shown in Table 1. Total The pH of rockfish fillets (Parkin, 1979) percent peptone water were prepared bacterial counts made on the fish held in decreased from about 6.7 to 6.3 after 1 for pour plates from the homogenate. the MA remained low for the full 21 week in a MA containing 80 percent Plate counts were made using TPY solu­ days of the test. The results show that CO2 , It is assumed that the pH of the tion (1.5 percent trypticase, 0.5 percent the counts never exceeded 104 orga­ salmon in this experiment did not be­ peptone, 0.5 percent yeast extract, 0.2 nisms/g, which is normal for very fresh come more acid because of the strong percent glucose, 0.5 percent NaCl, and fish. A total bacterial plate count of 106 buffering capacity of the flesh or other 1.5 percent agar). Plates were incubated organisms/g is generally considered in­ physiological conditions existing during at 22°C (72°F) for 5 days. dicative of incipient spoilage. In a similar the experiment. The pH of the flesh of The pH of the flesh of the salmon was study, Parkin (1979) reported essentially the nsh would be expected to increase measured by homogenizing at 4:1 mix­ no growth of aerobic bacteria on rocknsh (become alkaline) if normal bacterial ture by weight of distilled water and fIsh nllets stored refrigerated in an atmos- spoilage had occurred.

March 19H2. 4413) CO2 Concentration (1971,1978) in experiments with use of values (Table I), oxidation of lipids in CO2 in refrigerated brine for preservation the flesh of the salmon, because of expo­ As shown in Table 1, the concentration of fish and shellfish. The increase is partly sure to a high concentration of CO2 , was of CO2 found in the muscle of the MA­ due to a complexing of the CO2 with not a significant factor in this experiment. stored fish increased by a factor of about muscle protein and/or combining with This is attributed to the relatively low 5 during the experiment. The largest physiological electrolytes. concentration of 02' in the MA envi­ increase occurred between the first and ronment and also to the low-storage TBA Measurements llfth days of storage. A similar phenom­ temperature. CO2 is known to retard enon was observed by Barnett et al. As indicated by the relatively low TBA oxidation of fats (Makashev, 1959) at low concentrations (20-30 percent) and is apparently more effective at higher concentrations (Tarr. 1948).

Table 2.-Sensory evaluations of butchered salmon held in a retrigerated CO 2-modified atmosphere. Sensory Evaluations Average sensory scores! Storage on baked salmon samples time The salmon from the MA storage were (days) Flavor Texture Rancidity General comments on raw salmon examined periodically for sensory 0 4.2 '" 0.8 4.6" 0.5 50'0 Fish are mature but of excellent quality changes. Samples were evaluated raw, cooked. and canned. 6 40 + 0.7 4.2 + OA 4.6 "0.5 Good color. odor. and texture

10 4.3 ' 0.8 4A "09 4.5 • OA Very good quality Raw Salmon

15 4.2 + 16 44 '·0.8 46 ·07 Normal color. odor. firm texture No serious deterioration of overall 19 30' 1.1 4A -'- 0.5· 46" 0.6 Slight ally odor. good color quality was observed in the fish during fIrm texture the experiment (Table 2). In studies using 21 3.7 .• 1.6 47 - 0.5 4A" 1.6 Slight oily odor. good color. and CO2 to preserve fresh sole and cod. texture; fish are of acceptable quailty Coyne (1933) found that the texture of 'Scores are based on a 5-point numerical scale' Score of 2 denotes a product of borderline the fish was slightly softer after prolonged quailty. storage in a high concentration (100 percent) of CO2 than fish exposed to CO2 concentrations of 40-60 percent. We did not observe subjectively any sig­ Table 3. - Physical and sensory evaluations of canned salmon prepared from salmon held in a refrigerated. nificant softening of texture in this ex­ CO -modified atmosphere. 2 periment. Days NetWeightloz.) Vacuum lin.) of Code' storage Range Avg. Range Avg. Color Remarks Cooked Salmon 0 (MOS) 76-85 7.9 1.0- 7.0 52 Avg Very slight oily ftavor, slight curd. and normal texture Flavor scores generally indicate that the MA- held fish, irrespective of species, 6 (BC) 7.7-8A 8.0 0 - 50 35 Avg( Normal flavor. odor. and texture (OS) 8.0-8A 83 10-100 6.5 Avg.( Normal flavor. muddy odor, slight remained of acceptable quality for the soft texture and water marked duration of the test (Table 2). No signifi­ (acceptable) cant differences (P 0.05) were found in 10 (BC) 7.5-8A 7.9 0-140 8.0 Avg. Normal flavor. odor. and texture. moderate curd the flavor and texture between the initial, (OS) 8.0 80 8.0 8.0 Avg.( TYPical flavor and odor lor water- untreated fish, and those held in the MA marked fish; slightly soft with moderate curd (acceplable) for 21 days. There was no significant difference in the organoleptic rancidity 14 (OCI 76-82 79 8.0-10.0 9.0 Avg Normal odor, flavor. and texture for water-marked fish: moderate curd scores between the untreated fish and (acceptable) the fish held in MA. The sensory scores (OS) 7.5-7.7 7.6 8.0-100 9.0 Avg. Normal odor. flavor; slight soft texture; moderate curd also indicate that the texture of the salmon was not adversely affected by 19 IBS) 7.3-8.1 78 6.0- 8.0 70 Avg. Normal odor; moderate. oily flavor and normal texture; moderate curd exposure to the MA. Sensory examina­ (OCI 7.2-8.1 7.7 60-120 8.5 Avg Normal odor and flavor for water- marked fish; slight. soft texture; tions for rancidity correlated well with moderate curd (acceptable) chemical tests for rancidity.

21 (OS) 76-80 7.7 80- 9.0 85 Avg.1 Slight oily odor and flavor; normal texture. moderate curd Canned Salmon (BCI 7.3-8.2 78 7.0- 8.0 7.7 Avg Normal flavor and odor; normal texture: moderate curd Canned salmon were evaluated for fill 'MOS = medium dark silver weight, vacuum, color, and general ap· BC bright chum. OS dark silver pearance and acceptability (Table 3). BS bright silver Net weights varied as a result of difficulty

10 Marine Fisheries Rel'iell' in filling each can by hand using estimated Summary and Recommendations conducted to further evaluate the pre­ in-fIll weights. The occasional loss of servative effects of CO2 and other at­ vacuum in the cans was caused byover­ The results of the experiment de­ mospheres with respect to different fish fIlling with product. None of the canned scribed here show that fresh, butchered species. The public health significance product examined was considered salmon can be bulk-held in acceptable is important in using processes in which spoiled or was rejected on the basis of condition for up to 21 days at OOC (32°Fl the normal microbiological spoilage decomposition. Low marks were mainly in a MA containing 90 percent CO2, pattern is modified as in the MA envi­ given on the basis of the maturity of the Butchered salmon usually spoil within ronment. The possible presence of fIsh, i.e., water marked, muddy odor, 12 days on ice. Clostridium botulinum and its ability to texture and curd defects, and oily flavor. The effect of the CO2 was to slow the grow and develop toxin in the CO MA growth of spoilage bacteria apparently 2 Some swelling of the air-cooled cans at higher (abuse) temperatures require by an inhibition of the CO on specific was noted: but after several days in 2 that future research include studies and metabolic pathways involved in the nor­ storage, the swelling subsided. The recommendations for insuring product mal growth of the bacteria. It also ap­ swelling of the cans was caused by excess safety during distribution and marketing. (free) CO liberated from the flesh during peared that the combination of low O2 2 and high CO concentration in the pre­ Literature Cited the cook, which resulted in occasional 2 servative gas protected the salmon from buckling of cans. Increasing the vacuum Barnett. H. 1.. R. W. Nelson. P. J. Hunter. S. (24 inches) cycle of the can seamer by the effects of oxidative rancidity. Bauer. and H. Groninger. 1971. Studies on Retention of excess CO in the flesh the use of carbon dioxide dissolved in re­ about 15 seconds prior to closing and a 2 of the salmon caused some swelling in frigerated brine for the preservation of whole slow release of pressure from the retort fish. Fish. Bull.. U.S. 69:433-442. the canned product. This condition was Barnett. H. 1.. R. W. Nelson. P. J. Hunter. and after the cook helped to control the the result of "free" CO liberated during H. Groninger. 1978. Use of carbon dioxide problem. Icing and holding the fIsh over­ 2 dissolved in refrigerated brine for the pre­ the cook. After cooling, swelling of the night prior to canning had no effect on servation of pink shrimp (Panda/us spp.l. Mar. FIsh. Rev. 40(9).24-28. resolving the excess CO and the buck­ cans was not evident. Sensory evaluation 2 of the canned product showed it to be Coyne. F. P. 1932. The effect of carbon dioxide ling problem. on bacterial growth with special reference to normal in appearance, odor, color, fla­ the preservation of fish. Part I. 1. Soc. Chem. vor, and texture for canned mature, Ind. (Lond.) 51: 119-121. Refrigerated Shelf-Life Test ___ . 1933. The effect of carbon dioxide net-caught salmon. For commercial ap­ on bacterial growth with special reference to plication in canning salmon of different the preservation of fish. Part II. Gas storage After 2 weeks in the MA environment. quality or species, additional tests should of fresh fish. J. Soc. Chem. Ind. (Lond.) 52: 19-24. 6 butchered salmon were individually be conducted. King. A. D.. Jr.. and C. W. Nagel. 1967. Growth packaged in polyethylene bags and Salmon removed after 2 weeks in MA and inhibition of a Pseudomonas by carbon stored in a 2°C (3SoF) refrigerated room dioxide. J. Food Sci. 32:575-579. storage and held on ice in a refrigerated __-,-' 1975. Influence of carbon dioxide until organoleptically spoiled. The fish room remained of acceptable quality for upon the metabolism of Pseudomonas aerug­ were periodically examined and evalu­ almost 1 week. inosa. J. Food Sci. 40:362-366. Lemon. D. W. 1975. An improved TBA test for ated for odor, color, and general appear­ The MA preservation method de­ rancidity. Environment Canada. Fish. Mar. ance. They retained most of their good scribed here can be considered for com­ Servo New Ser. Circ. 51.4 p. quality up to the fourth day of the test mercial use. It is recommended, how­ Makashev. A. P. 1959. Carbon dioxide as a means of prolonging the storage life of chilled when a slight off-odor not associated ever. that each user initiate shipments fish products. Trudy Vsesoyuz. Nauch-lssle­ with spoilage was detected. The odor on a small scale to gain familiarity and dovatel. Inst. Morsk. Rybn. Khozi i Abeanogr. was not detected after the fish were experience with the technique and its 37. 138 p. Nelson. R. w.. and W. I. Tretsven. 1975. Stor­ washed. No other signitlcant changes special handling requirements. Above all. age of Pacific salmon in controlled atmos­ were noted in the fish samples until the salmon to be shipped and marketed as phere. Proc. XIV Int. Congr. Refrig .. Mos­ seventh day of storage. At this time, the cow. USSR. 10 p. fresh via this method should be of the Parkin. K. L. 1979. The use of modified atmos­ remaining samples had a distinct sour highest quality. Furthennore, they should pheres for the preservation of fresh seafood odor on the skin, in the flesh, and in the be thoroughly washed and prechilled products. M.S. Thesis. Univ. California. Davis. Pelroy. G. A.. and M. W. Eklund. 1966. Changes poke (belly cavity). SignifIcant discolor­ prior to containerization. The tempera­ in the microflora of vacuum-packaged. irra­ ation of the peritoneal lining (not the ture of the tlsh should be maintained at diated petrale sole (Eopsella jordam) fillets underlying flesh) of the belly cavity was 0° ± O.SoC (32° ± 1°F) at all times ,stored at 0.5°C. Appl. Microbiol. 14:921-927. Stansby. M. E.. and F. P. Gnffiths. 1935. Car­ also observed. The samples were con­ when in transit. bon dioxide in handling fresh fish - haddock. sidered organoleptically unacceptable. Because of the preliminary nature of Ind. Engr. Chem. 27: 1452-1458. At this time, the fIsh had been out of Tarr. H. L. A. 1948. Control of rancidity in fish the work described here, it is recom­ flesh. Physical and chemical methods. J. Fish. water about 23 days. mended that additional research be Res. Board Can. 7:237- 247.

March /')82. 4413; /I