8. ASSESSMENT OF FISH QUALITY

8.1. Sensory methods
8.2. Biochemical and chemical methods
8.3. Physical methods
8.4. Microbiological methods

Most often "quality" refers to the aesthetic appearance and freshness or degree of spoilage which the fish has undergone. It may also involve safety aspects such as being free from harmful bacteria, parasites or chemicals. It is important to remember that "quality'' implies different things to different people and is a term which must be defined in association with an individual product type. For example, it is often thought that the best quality is found in fish which are consumed within the first few hours post mortem. However, very fresh fish which are in rigor mortis are difficult to fillet and skin and are often unsuitable for smoking. Thus, for the processor, slightly older fish which have passed through the rigor process are more desirable.

The methods for evaluation of fresh fish quality may be conveniently divided into two categories: sensory and instrumental. Since the consumer is the ultimate judge of quality, most chemical or instrumental methods must be correlated with sensory evaluation before being used in the laboratory. However, sensory methods must be performed scientifically under carefully controlled conditions so that the effects of test environment, personal bias, etc., may be reduced.

8.1 Sensory methods

Sensory evaluation is defined as the scientific discipline used to evoke, measure, analyze and interpret reactions to characteristics of food as perceived through the senses of sight, smell, taste, touch and hearing.

Most sensory characteristics can only be measured meaningfully by humans. However, advances are being made in the development of instruments that can measure individual quality changes.

Instruments capable of measuring parameters included in the sensory profile are the Instron, Bohlin Rheometer for measuring texture and other rheologic properties. Microscopic methods combined with image analysis are used to assess structural changes and "the artificial nose" to evaluate odour profile (Nanto et al., 1993).

Sensory process

In sensory analysis appearance, odour, flavour and texture are evaluated using the human senses. Scientifically, the process can be divided into three steps. Detection of a stimulus by the human sense organs; evaluation and interpretation by a mental process; and then the response of the assessor to the stimuli. Variations among individuals in the response of the same level of stimuli can vary and can contribute to a non-conclusive answer of the test. People can, for instance, differ widely in their response to colour (colour blindness) and also in their sensitivity to chemical stimuli. Some people cannot taste rancid flavour and some have a very low response to cold-storage flavour. It is very important to be aware of these differences when selecting and training judges for sensory analysis. Interpretation of the stimulus and response must be trained very carefully in order to receive objective responses which describe features of the fish being evaluated. It is very easy to give an objective answer to the question: is the fish in rigor (completely stiff), but more training is needed if the assessor has to decide whether the fish is post or pre-rigor. Subjective assessment, where the response is based on the assessor's preference for a product, can be applied in the fields like market research and product development where the reaction of the consumer is needed. Assessment in quality control must be objective.

Sensory methods

The analytical objective test used in quality control can be divided into two groups: discriminative tests and descriptive tests. Discriminative testing is used to determine if a difference exists between samples (triangle test, ranking test). Descriptive tests are used to determine the nature and intensity of the differences (profiling and quality tests). The subjective test is an affective test based on a measure of preference or acceptance.

Figure 8.1 Methods of sensory analysis

In the following, examples of discriminative and descriptive testing will be given. For further information concerning market testing, see Meilgaard et al. (1991).

Quality assessment of fresh fish

Quality Index Method

During the last 50 years many schemes have been developed for sensory analysis of raw fish. The first modern and detailed method was developed by Torry Research Station (Shewan et al., 1953). The fundamental idea was that each quality parameter is independent of other parameters. Later, the assessment was modified by collecting a group of characteristic features to be expressed in a score. This gives a single numerical value to a broad range of characteristics. In Europe today, the most commonly used method for quality assessment in the inspection service and in the fishing industry is the EU scheme, introduced in the council decision No. 103/76 January 1976 (Table 5.2). There are three quality levels in the EU scheme, E (Extra), A, B where E is the highest quality and below B is the level where fish is discarded for human consumption. The EU scheme is commonly accepted in the EU countries for sensory assessment. There is, however, still some discrepancy as the scheme does not take account of differences between species into account as it only uses general parameters. A suggestion for renewal of the EU scheme can be seen in Multilingual Guide to EU Freshness Grades for Fishery Products (Howgate et al., 1992),where special schemes for whitefish, dogfish, herring and mackerel are developed (Appendix E).

A new method, the Quality Index Method (QIM) originally developed by the Tasmanian Food Research unit (Bremner et al., 1985), is now used by the Lyngby Laboratory (Jonsdottir, 1992) for fresh and frozen cod, herring and saithe. In the Nordic countries and Europe it has also been developed for redfish, sardines and flounder.

Table 8.1 Quality assessment scheme used to identify the quality index demerit score (Larsen et al. 1992)  

QIM is based on the significant sensory parameters for raw fish when using many parameters and a score system from 0 to 4 demerit points (Jonsdottir, 1992). QIM is using a practical rating system, in which the fish is inspected and the fitting demerit point is recorded. The scores for all the characteristics are then summed to give an overall sensory score, the so-called quality index. QIM gives scores of zero for very fresh fish while increasingly larger totals result as fish deteriorate. The description of evaluation of each parameter is written in a guideline. For example, 0 demerit point for the appearance of the skin on herring means very bright skin only experienced in freshly caught herring. The appearance of the skin in a later state of decay turns less bright and dull and gives 2 demerit points. Most of the parameters chosen are equal to many other schemes. After the literal description, the scores are ranked for each description for all the parameters, giving scores 0-1, 0-2, 0-3 or 0-4. Parameters with less importance are given lower scores. The individual scores never exceed 4, so no parameter can excessively unbalance the score. A scheme for herring is shown in table 8.1; it is emphazised that it is neccessary to develop a new scheme for every species (the scheme for cod is seen in Appendix D).

There is a linear correlation between the sensory quality expressed as a demerit score and storage life on ice, which makes it possible to predict remaining storage life on ice. The theoretical demerit curve has a fixed point at (0,0) and its maximum has to be fixed as the point where the fish has been rejected by sensory evaluation of, e.g., the cooked product (see under structured scaling) or otherwise determined as the maximum keeping time. Using cooked evaluation the two parallel sensory tests demand an experienced sensory panel even though this is only required while developing the scheme, and later on it will not be necessary to assess cooked fish in order to predict the remaining shelf life. QIM does not follow the traditionally accepted S-curve pattern for deterioration of chilled fish during storage (Figure 5.1). The aim is a straight line which makes it possible to distiguish between fish at the start of the plateau phase and fish near the end of the plateau phase (Figure 8.2).

Figure 8.2 Combination of sensory curves for raw S(T) and cooked fish

When a batch of fish in Figure 8.2 reaches a sum of demerit points of 10, the remaining keeping time in ice will be 5 days. To predict remaining shelf life, the theoretical curve can be converted as shown in Figure 8.3.

Figure 8.3 A curve to predict the storage time remaining for herring stored in ice or sea water at 0°C

A fish merchant may want to know how long his purchase will remain saleable if the fish are stored on ice immediately. A buyer at a fish market might be interested in the equivalent number of days on ice where the fish have been stored since they were caught, and thus how much marketable time on ice is left. These condition indicators can be extracted for a fish sample with a known rate of change in demerit points using the quality index method.

Structured Scaling

Descriptive testing can also be used for quality determination and shelf life studies applying a structured scaling method. Structured scaling gives the panelist an actual scale showing several degrees of intensity. A few detailed attributes are chosen often based on work from a fully trained descriptive panel. Descriptive words must be carefully selected, and panelists trained so that they agree with the terms. Objective terms should be preferred rather than subjective terms. If possible, standards are included at various points of the scale. This can easily be done with different concentrations of salt but might be more difficult with conditions such as degree of spoilage. The most simple method (Table 8.2) can be 1. No off-odour/flavour, 2. Slight off-odour/flavour and 3. Severe off-odour/flavour, where the limit of acceptability is between 2 and 3. This has been further developed to an integrated assessment of cooked fish fillet of lean and fatty fish (see example in Appendix E).

A 10-point scale is used as described under 5.1 Sensory changes, and an overall impression of odour, flavour and texture is evaluated in an integrated way. For statistics, t-test and analysis of variance can be used (see example in Appendix F).

Table 8.2 Evaluation of cooked fish

Quality assessment of fish products

Assessment of fishery products can both be performed as a discriminative test and as a descriptive test.

Triangle test

The most used discriminative test in sensory analysis of fish is the triangle test (ISO standard 4120 1983), which indicates whether or not a detectable difference exists between two samples. The assessors receive three coded samples, are told that two of the samples are identical and one is different, and are asked to identify the odd sample.

Analysis of results from the triangle test is done by comparing the number of correct identifications with the number you would expect to obtain by chance alone. In order to test this the statistical chart in Appendix A must be consulted. The number of correct identifications is compared to the number expected by use of a statistical table, e.g., if the number of responses is 12, there must be 9 correct responses to achieve a significant answer (1% level).

Triangle tests are useful in determining, e.g., if ingredient substitution gives a detectable difference in a product. Triangle tests are often used when selecting assessors to a taste panel.

The samples marked A and B can be presented in six different ways:

 ABB    BBA    AAB
 BAB    ABA    BAA

Equal numbers of the six possible combinations are prepared and served to the panel members. They must be served randomly, preferably as duplicates. The number of panel members should be no less than 12 (an example of a triangle test from the ISO standard can be seen in Appendix B).

Table 8.3 Example of score sheet: triangle test

Ranking

In a ranking exercise, a number of samples are presented to the taste panel. Their task is to arrange them in order according to the degree to which they exhibit some specified characteristics, e.g, downward concentration of salt. Usually ranking can be done more quickly and with less training than evaluation by other methods. Thus ranking is often used for preliminary screening. The method gives no individual differences among samples and it is not suited for sessions where many criteria have to be judged simultaneously.

Profiling

Descriptive testing can be very simple and used for assessment of a single attribute of texture, flavour and appearance. Methods of descriptive analysis which can be used to generate a complete description of the fish product have also been developed. An excellent way of describing a product can be done by using flavour profiling (Meilgaard et al., 1991). Quantitative Descriptive Analysis provides a detailed description of all flavour characteristics in a qualitative and quantitative way. The method can also be used for texture. The panel members are handed a broad selection of reference samples and use the samples for creating a terminology that describes the product.

In Lyngby a descriptive sensory analysis for fish oil using QDA has been developed. A trained panel of 16 judges is used. Descriptive terms such as paint, nutty, grassy, metallic are used for describing the oil on an intensity scale. A moderately oxidized fish oil is given fixed scores and used as a reference.

Table 8.4 Profile of fish oil  

Advanced multivariate analysis is used for statistics and makes it possible to correlate single attributes to oxidative deterioration in the fish oil. The results can be reported in a "spider's web" (se Figure 8.5). The panel uses an intensity scale normally ranging from 0 to 9.

Profiling can be used for all kinds of fishery products, even for fresh fish when special attention is placed on a single attribute.

The results of QDA can be analyzed statistically using analysis of variance or multivariate analysis (O'Mahony, 1986).

Statistics

In any experiment including sensory analysis the experimental design (e.g., number of panel members, number of samples, time aspects, hypotheses to test) and statistical principles should be planned beforehand. Failure to do so may often lead to insufficient data and non-conclusive experiments. A guide to the most used statistical methods can be seen in Meilgaard et al. (1991). A panel used for descriptive testing shall preferably consist of no less than 8-10 persons, and it should be remembered that the test becomes statistically much stronger if it is done in duplicate. This can often be difficult using sensory analysis on small fish. Thus the experiment must include a sufficient number ofsamples to remove the sources of variability, and the testing must be properly randomized. For further information see O'Mahony (1986) and Smith (1989).

Figure 8.4 Flavour profiles of a fish oil after 2 weeks of storage at various temperatures (Rorbaek et al., 1993)

Training of assessors

Training of assessors for sensory evaluation is necessary in almost all sensory methods. The degree of training depends on the difficulty and complexity of the assessment. For example, for profiling a thorough training with presentation of a large range of samples is necessary in order to obtain proper definitions of the descriptors an equal use of the scoring system. The triangle test normally requires a minor degree of training.

Sensory quality control is often done by a few persons either at the fish market when buying fish or at quality inspection. The experience of these persons allows them to grade the fish. Starting as a fish inspector it is not necessary to know all the different methods of sensory assessment described in textbooks (Meilgaard et al., 1991), but some of the basic principles must be known. The assessor must be trained in basic tastes, the most common fish taste and must learn the difference between off- flavour and taints. This knowledge can be provided in a 2- day basic training course.

In bigger companies and for experimental work a further training of a sensory panel is necessary in order to have an objective panel. A laboratory panel must have 8-10 members, and the training and testing of panel members must be repeated regularly.

Facilities

The facilities required for sensory evaluation is described in textbooks on sensory evaluation.

The minimum requirement for evaluation is a preparation room and a room where the samples are served. The rooms should be well ventilated and provided with a good light (Howgate, 1994). There must be enough space on the tables for inspection of raw samples of fish.

Cooking and serving

The samples of fishery products should not be less than 50-100 g per person. Fillets can be served in loins and should be cooked to an internal temperature of 65°C. The samples should be kept warm when served, i.e., in insulated containers or on a hot plate. The fish can be heat treated by steaming in a water bath, packed as boiled-in-the-bag in a plastic pouche or in alufoil. An oven (microwave or steam-oven) can also be used for heat treatment. The fish can be packed in plastic or put on a small porcelain plate covered with alufoil. For cod loins (2,5x1,5x6cm) on a porcelain plate covered with alufoil the heating time in a steam-oven (convectomate) at 100°C must be 10 minutes. The samples should be coded before serving.

8.2 Biochemical and chemical methods

The appeal of biochemical and chemical methods for the evaluation of seafood quality is related to the ability to set quantitative standards. The establishment of tolerance levels of chemical spoilage indicators would eliminate the need to base decisions regarding product quality on personal opinions. Of course, in most cases sensory methods are useful for identifying products of very good or poor quality. Thus, biochemical/ chemical methods may best be used in resolving issues regarding products of marginal quality. In addition, biochemical/chemical indicators have been used to replace more time consuming microbiological methods. Such objective methods should however correlate with sensory quality evaluations and the chemical compound to be measured should increase or decrease with the level of microbial spoilage or autolysis. It is also important that the compounds to be measured must not be affected by processing (e.g., breakdown of amines or nucleotides in the canning process as a result of high temperatures).

The following is an overview of some of the most useful procedures for the objective measurement of seafood quality. Woyewoda et al. (1986) have produced a comprehensive manual of procedures (including proximate composition of seafood).

Amines - Total Volatile Basic Amines

Total volatile basic amines (TVB) is one of the most widely used measurements of seafood quality. It is a general term which includes the measurement of trimethylamine (produced by spoilage bacteria), dimethylamine (produced by autolytic enzymes during frozen storage), ammonia (produced by the deamination of amino-acids and nucleotide catabolites) and other volatile basic nitrogenous compounds associated with seafood spoilage. Although TVB analyses are relatively simple to perform, they generally reflect only later stages of advanced spoilage and are generally considered unreliable for the measurement of spoilage during the first ten days of chilled storage of cod as well as several other species (Rehbein and Oehlenschlager, 1982). They are particularly useful for the measurement of quality in cephalopods such as squid (LeBlanc and Gill, 1984), industrial fish for meal and silage (Haaland and Njaa, 1988), and crustaceans (Vyncke, 1970). However, it should be kept in mind that TVB values do not reflect the mode of spoilage (bacterial or autolytic), and results depend to a great extent on the method of analysis. Botta et al. (1984) found poor agreement among six published TVB procedures. Most depend upon either steam distillation of volatile amines or microdiffusion of an extract (Conway, 1962); the latter method is the most popular in Japan. For a comparative examination of the most common procedures for TVB analysis, see Botta et al. (1984).

Ammonia

Ammonia is formed by the bacterial degradation/deamination of proteins, peptides and amino- acids. It is also produced in the autolytic breakdown of adenosine monophosphate (AMP, Figure 5.4) in chilled seafood products. Although ammonia has been identified as a volatile component in a variety of spoiling fish, few studies have actually reported the quantification of this compound since it was impossible to determine its relative contribution to the overall increase in total volatile bases.

Recently, two convenient methods specifically for identifying ammonia have been made available. The first involves the use of the enzyme glutamate dehydrogenase, NADH and alpha-ketoglutarate. The molar reduction of NH₃ in a fish extract yields one mole of glutamic acid and NAD which can be monitored conveniently by absorbance measurements at 340 nm. Test kits for ammonia based on glutamate dehydrogenase are now available from Sigma (St. Louis, Missouri, USA) and Boehringer Mannheim (Mannheim, Germany). A third type of ammonia test kit is available in the form of a test strip (Merck, Darmstadt, Germany) which changes colour when placed in contact with aqueous extracts containing ammonia (ammonium ion). LeBlanc and Gill (1984) used a modification of the glutamate dehydrogenase procedure to determine the ammonia levels semi-quantitatively without the use of a spectrophotometer, but with a formazan dye, which changed colour according to the following reaction:

where INT is iodontrotetrazolium and MTT is 3 - [4,5-dimethylthiazol-2-yl] 2,5 diphenyl tetrazolium bromide 

Ammonia has been found to be an excellent indicator of squid quality (LeBlanc and Gill, 1984) and comprised a major proportion of the TVB value for chilled short-finned squid (Figure 8.7). However, ammonia would appear to be a much better predictor of the latter changes in quality insofar as finfish are concerned. LeBlanc (1987) found that for iced cod, the ammonia levels did not increase substantially until the sixteenth day of storage. It would appear that at least for herring, the ammonia levels increase far more quickly than trimethylamine (TMA) levels which have traditionally been used to reflect the growth of spoilage bacteria on lean demersal fish species. Thus ammonia has potential as an objective quality indicator for fish which degrades autolytically rather than primarily through bacterial spoilage.

Figure 8.7 Effect of storage time on production of ammonia. TVB and TMA in short finned squid (Illex illecebrosus), adapted from Gill (1990) 

Trimethylainine (TMA)

Trimethylamine is a pungent volatile amine often associated with the typical "fishy" odour of spoiling seafood. Its presence in spoiling fish is due to the bacterial reduction of trimethylamine oxide (TMAO) which is naturally present in the living tissue of many marine fish species. Although TMA is believed to be generated by the action of spoilage bacteria, the correlation with bacterial numbers is often not very good. This phenomenon is now thought to be due to the presence of small numbers of "specific spoilage" bacteria which do not always represent a large proportion of the total bacterial flora, but which are capable of producing large amounts of spoilage -related compounds such as TMA. One of these specific spoilage organisms, Photobactetium phosphoreum, generates approximately 10 - 100 fold the amount of TMA than that produced from the more commonly-known specific spoiler, Shewanella putrefaciens (Dalgaard, 1995) (in press).

As mentioned above, TMA is not a particularly good indicator of edibility of herring quality but is useful as a rapid means of objectively measuring the eating quality of many marine demersal fish. The correlations between TMA level or more preferably, TMA index (where TMA index = log (1 + TMA value)) and eating quality have been excellent in some cases (Hoogland, 1958; Wong and Gill, 1987). Figure 8.8 illustrates the relationship between odour score and TMA level for iced cod. The linear coefficient of determination was statistically significant at the P £ 0.05 level.

Figure 8.8 Relationship between odour score and TMA levels for iced cod. The straight line was fitted by linear regression analysis (P£ 0.05) and all data points were averages of data obtained for three individual cod, adapted from Wong and Gill (1987) 

The chief advantages of TMA analysis over the enumeration of bacterial numbers are that TMA determinations can be performed far more quickly and often reflect more accurately the degree of spoilage (as judged organoleptically) than do bacterial counts. For example, even high quality fillets cut with a contaminated filleting knife may have high bacterial counts. However, in such a case the bacteria have not had the opportunity to cause spoilage, thus TMA levels are bound to be low. The chief disadvantages of TMA analyses are that they do not reflect the earlier stages of spoilage and are only reliable for certain fish species. A word of caution should be given concerning the preparation of fish samples for amine analysis. TMA and many other amines become volatile at elevated pH. Most analytical methods proposed to date begin with a deproteinization step involving homogenization in perchloric or trichloroacetic acids. Volatilization of amines from stored samples may result in serious analytical errors. Therefore, samples should be neutralized to pH 7 immediately before analysis and should be left in their acidified form in sealed containers if being stored for extended time periods prior to analysis. It is also important to note that appropriate protection for hands and eyes be worn when handling perchloric and/or trichloroacetic acids. In addition, perchloric acid is a fire hazard when brought into contact with organic matter. Spills should be washed with copious quantities of water. Some of the methods of analysis reported to date include colorimetric (Dyer, 1945; Tozawa, 1971), chromatographic (Lundstrom and Racicot, 1983; Gill and Thompson, 1984) and enzymatic analysis (Wong and Gill, 1987; Wong et al., 1988), to name but a few. For a more comprehensive review of the analytical techniques for TMA see the recent review articles: (Gill 1990, 1992).

Dimethylarnine (DMA)

As outlined in section 5.2, certain types of fish contain an enzyme, TMAO dimethylase (TMAO-ase), which converts TMAO into equimolar quantities of DMA and formaldehyde (FA). Thus for fish in the cod (gadoid) family, DMA is produced along with FA in frozen storage with the accompanying FA-induced toughening of the proteins. The amount of protein denaturation is roughly proportional to the amount of FA/DMA produced, but it is most common to monitor the quality of frozen-stored gadoid fish by measuring DMA rather than FA. Much of the FA becomes bound to the tissue and is thus not extractable and cannot be measured quantitatively. The most common method for DMA analysis is a colorimetric determination of the DMA in deproteinized fish extracts. The Dyer and Mounsey (1945) procedure is still in use today although perhaps more useful is the colorimetric assay proposed by Castell et al. (1974) for the simultaneous determination of DMA and TMA, since both are often present in poor quality frozen fish. Unfortunately, many of the colorimetric methods proposed to date lack the specificity where mixtures of different amines are present in samples. The chromatographic methods including gas-liquid chromatography (Lundstrom and Racicot, 1983) and high performance liquid chromatography (Gill and Thompson, 1984) are somewhat more specific, and are not as prone to interferences as the spectrophotometric methods. Also, most of the methods proposed to date for the analysis of amines are destructive and not well suited for analyzing large numbers of samples. Gas chromatographic analysis of headspace volatiles has been proposed as a non- destructive alternative for amine determinations; however, none of the methods proposed to date are without serious practical limitations.

Dimethylamine is produced autolytically during frozen storage. For gadoid fish such as hake, it has been found to be a reliable indicator of FA-induced toughening (Gill et al., 1979). Because it is associated with membranes in the muscle, its production is enhanced with rough handling and with temperature fluctuations in the cold storage facility. Dimethylamine has little or no effect on the flavour or texture of the fish per se, but is an indirect indicator of protein denaturation which is often traceable to improper handling before and/or during frozen storage.

BiogenicAmines

Fish muscle has the ability to support the bacterial formation of a wide variety of amine compounds which result from the direct decarboxylation of amino-acids. Most spoilage bacteria possessing decarboxylase activity do so in response to acidic pH, presumably so that the organisms may raise the pH of the growth medium through the production of amines.

Histamine, putrescine, cadaverine and tyramine are produced from the decarboxylation of histidine, ornithine, lysine and tyrosine, respectively. Histamine has received most of the attention since it has been associated with incidents of scombroid poisoning in conjunction with the ingestion of tuna, mackerel, mahi-mahi (dolphinfish from Hawaii). However, the absence of histamine in scombroid fish (tuna, mackerel, etc.) does not ensure the wholesomeness of the product since spoilage at chill storage temperatures does not always result in the production of histamine. Mietz and Karmas (1977) proposed a chemical quality index based on biogenic amines which reflected the quality loss in canned tuna where:

They found that as the quality index ratio increased, the sensory scores on the canned' product decreased. Later, Farn and Sims (1987) followed the production of histamine, cadaverine and putrescine in skipjack and yellowfin tuna at 20°C and found that cadaverine and histamine increased exponentially after an initial lag period of about 36 hours. However, putrescine increased slowly after an initial lag period of 48 hours. Levels of cadaverine and histamine increased to maximum levels of 5-6µg/g tuna but the authors reported that the absence of such amines in raw or cooked product did not necessarily mean that the products were not spoiled.

It is interesting to note that most of the biogenic amines are stable to thermal processing, so their presence in finished canned products is a good indication that the raw material was spoiled prior to heating.

Some of the methods for biogenic amine analysis include high pressure liquid chromatography (Mietz and Karmas, 1977), gas chromatography (Staruszkiewicz andBond, 1981), spectrofluorometric (Vidal-Carou et al., 1990) and a newly-developed rapid enzymatic method for histamine using a microplate reader (Etienne and Bregeon, 1992).

Nucleotide Catabolites

A discussion of the analysis of nucleotide catabolites has been presented in section 5.2 -Autolytic Changes, although all of the catabolic changes are not due to autolysis alone. Most of the enzymes involved in the breakdown of adenosine triphosphate (ATP) to inosine monophosphate (IMP) are believed in most cases to be autolytic whereas the conversion of IMP to inosine (Ino) and then hypoxanthine (Rx) are believed mainly to be due to spoilage bacteria although Hx has been shown to accumulate slowly in sterile fish tissue. Since the levels of each of the catabolic intermediates rise and fall within the tissue as spoilage progresses, quality assessment must never be based upon levels of a single catabolite since the analyst has no way of knowing whether a single compound is increasing or decreasing. For example, if the IMP content of a fish sample were determined to be 5 µmoles/g tissue, the sample might well have been taken from a very fresh fish or a fish on the verge of spoilage, depending on whether or not AMP were present.

Thus, the analysis of the complete nucleotide catabolite profile is nearly always recommended. A complete analysis of nucleotide catabolites may be completed on a fish extract in 12-25 minutes using a high pressure liquid chromatographic (HPLC) system equipped with a single pump and spectrophotometric detector (wavelength 254 nm). Perhaps the simplest HPLC technique published to date is that proposed by Ryder (1985).

Several other approaches have been proposed for the analysis of individual or combination of nucleotide catabolites but none are more reliable than the HPLC approach. A word of caution is perhaps in order with regard to the quantitative analysis of nucleotide catabolites. Most methods proposed to date involve deproteinization of the fish samples by extraction with perchloric or trichloracetic acids. It is important that the acid extracts are neutralized with alkali (most often potassium hydroxide) as soon as possible after extraction to prevent nucleotide degradation in the extracts. Neutralized extracts appear to be quite stable even if held frozen for several weeks. One advantage of using perchloric acid is that the perchlorate ion is insoluble in the presence of potassium. Thus, neutralizing with KOH is a convenient method of sample "clean-up" before HPLC analysis and this procedure helps to extend the life of the HPLC column. Also, it should be noted that nucleotide determination on canned fish does not necessarily reflect the levels in the raw material. Gill et al. (1987) found recoveries of 50%, 75%, 64% and 92% for AMP, IMP, Ino and Hx standards which were spiked into canned tuna prior to thermal processing.

Several unusual but innovative approaches utilizing enzymatic assays have been proposed over the years and are presented in Table 8.3. All of the approaches to date rely on destructive sampling (tissue homogenization). It should be noted that regardless of the approach, enzymes denature with time and thus test kits, enzyme-coated strips, electrodes or sensors have a limited shelf life whereas the HPLC techniques do not.

Table 8.3 Fish Freshness Testing Using Enzyme Technology

The factors which have been shown to affect the nucleotide breakdown pattern include species, temperature of storage and physical disruption of the tissue. In addition, since nucleotide breakdown reflects the combined action of autolytic enzymes and bacterial action, the types of spoilage bacteria would no doubt affect the nucleotide patterns. The selection of which nucleotide or combination of nucleotide catabolites to be measured should be made carefully. For example, in certain cases one or two of the catabolites change rapidly with time of chilled storage, whereas the remaining components may change very little. The technical literature should be consulted for guidance on this matter. An excellent overview on the biological and technological factors affecting the nucleotide catabolites as quality indicators was presented by Frazer Hiltz et al. (1972).

Ethanol

Ethanol has been used for many years as an objective indicator for seafood quality although it is not nearly as common as the analysis of TMA. Since ethanol can be derived from carbohydrates via anaerobic fermentation (glycolysis) and/or deamination and decarboxylation of amino-acids such as alanine, it is a common metabolite of a variety of bacteria. It has been used to objectively measure the quality of a variety of fish including canned tuna (Iida et al., 1981 a, 1981b; Lerke and Huck, 1977), canned salmon (Crosgrove, 1978; Hollingworth and Throm, 1982), raw tuna (Human and Khayat, 1981), redfish, pollock, flounder and cod (Kelleher and Zall, 1983).

To date, the simplest and perhaps most reliable means of measuring ethanol in fish tissue is the use of the commercial enzyme test kits available from Boehringer Mannheim (German) or Diagnostic Chemicals (Charlottetown, P.E.I., Canada). One advantage of using ethanol as a spoilage indicator is that it is heat-stable (although volatile) and may be used to assess the quality of canned fish products.

Measurements of oxidative rancidity

The highly unsaturated fatty acids found in fish lipids (section 4.2) are very susceptible to oxidation (section5.4).The primary oxidation products are the lipid hydroperoxides. These compounds can be detected by chemical methods, generally by making use of their oxidation potential to oxidize iodide to iodine or to oxidize iron(II) to iron(III). The concentration of the hydroperoxides may be determined by titrimetricor by spectrophotometric methods, giving the peroxide value (PV) as milliequivalents (mEq) peroxide per 1 kg of fat extracted from the fish. A method for PV- determination by iodometry has been described by Lea (1952), and for determination by spectrophotometry of iron (III)thiocyanate by Stine et al. (1954).The methods for PV-determination are empirically based, and comparisons between PVs are only possible for results obtained using identical methods. For instance, the thiocyanate-method may give values 1.5 - 2 times higher than the iodine titration method (Barthel and Grosch, 1974).

For several reasons, interpretation of the PV as an index of quality is not straightforward. First, the hydroperoxides are odour- and flavour-less, thus the PV is not related to the actual sensory quality of the product analyzed. However, the peroxide value may indicate a potential for a later formation of sensorial-objectionable compounds. Second, lipid hydroperoxides break down with time, and a low PV at a certain point during the storage of a product can indicate both an early phase of autoxidation and a late stage of a severely oxidized product, where most hydroperoxides have been broken down (Kanner and Rosenthal, 1992),e.g., in dried, salted fish (Smith et al., 1990).

In later stages of oxidation secondary oxidation products will usually be present and thus be indicative of a history of autoxidation. These products (section 5.4) comprise aldehydes, ketones, short chain fatty acid and others, many of which have very unpleasant odours and flavours, and which in combination yield the fishy and rancid character associated with oxidized fish lipid. Some of the aldehydic secondary oxidation products react with thiobarbituric acid, forming a reddish coloured product that can be determined spectrophotometrically. Using this principle, a measure of thiobarbituric acid-reactive substances (TBA-RS) can be obtained. Several method variations exist; one method for fish lipids is described by Ke and Woyewoda (1979), and for fish by Vyncke (1975). The results are expressed in terms of the standard (di-)aldehyde used, malonaldehyde, and reported as micromoles malonaldehyde present in 1 g of fat. (A note of caution: Sometimes the TBA-results may be expressed as mg malonaldehyde in 1 g of fat, or as amount of malonaldehyde (µmol or µg) in relation to amount of tissue analyzed.) Several reports (e.g., by Hoyland and Taylor (1991) and by Raharjo et al. (1993)) speak of some correlation between TBA-RS and sensory assessments, but other authors fail to find a correlation (e.g., Boyd et al., 1993). Thus, caution is necessary in interpretation of TBA-RS values into measures of sensory quality.

Provided that the PV has not been lowered through extended storage or high temperature exposure, the PV (by iodometric titration) should not be above 10-20 meq/kg fish fat (Connell, 1975).

Examples of guidelines for TBA-RS-values: foods with TBA-RS above 1-2 µmol MDA-equiv per g fat (Connell, 1975) or above 10,µmol MDA-equiv per 1 kg fish (Ke et al., 1976) will probably have rancid flavours.

Modern instrumental methods allow analysis of better defined oxidation products (specific hydroperoxides, actual content of malonaldehyde), but for general quality estimation, methods that determine a broader range of oxidation products (such as PV and TBA-RS) are to be preferred, although these methods have their limitations as discussed above. Headspace analysis of the volatile oxidation products gives results correlating well with sensory evaluation (e.g., in catfish (Freeman and Hearnsberger, 1993)), but the method requires access to gas chromatographic equipment.

8.3 Physical methods

Electrical Properties

It has long been known that the electrical properties of skin and tissue change after death, and this has been expected to provide a means of measuring post mortem changes or degree of spoilage. However, many difficulties have been encountered in developing an instrument: for example, species variation; variation within a batch of fish; different instrument readings when fish are damaged, frozen, filleted, bled or not bled; and a poor correlation between instrument reading and sensory analysis. Most of these problems, it is claimed, are overcome by the GR Torrymeter (Jason and Richards, 1975). However, the instrument is not able to measure quality or freshness of a single fish, although it may find application in grading batches of fish, as shown in Figure 8.9.

Figure 8.9 Relationship between GR Torrymeter readings of various species of fish and freshness, adapted from Cheyne (1975)

Until recently, no instruments have been capable of on-line determination of quality although this type of mechanized quality evaluation would be highly desirable on the processing floor. The RT Freshness Grader development began in 1982 and, by 1990, a production model capable of sorting 70 fish per minute over 4 channels was made available. The developer was Rafagnataekni Electronics (Reykjavik, Iceland) based the sensing unit on the GR Torrymeter.

pH and Eh

Knowledge about the pH of fish meat may give valuable information about its condition. Measurements are carried out with a pH-meter by placing the electrodes (glass-calomel) either directly into the flesh or into a suspension of fish flesh in distilled water. Measurements of Eh are not carried out routinely, but it is likely that a freshness test can be based on this principle.

Measuring Texture

Texture is an extremely important property of fish muscle, whether raw or cooked. Fish muscle may become tough as a result of frozen storage or soft and mushy as a result of autolytic degradation. Texture may be monitored organoleptically but there has for many years been a need for the development of a reliable objective rheological test which would accurately reflect the subjective evaluation of a well-trained panel of judges. Gill et al. (1979) developed a method for evaluating the formaldehyde-induced toughening of frozen fish muscle. The method utilized an Instron Model TM equipped with a Kramer shear cell with 4 cutting blades. This method correlated well with data obtained from a trained texture panel. A method for measuring hardness/softness of fish flesh, designated as compressive deformability, has been reported by Johnson et al. (1980). An accuratelycut fish sample is compressed by a plunger, and the stress-strain curve recorded. A modulus of deformability is calculated from the recorded graph. The results from such measurements may, however, be difficult to interpret.

Another method, measuring the shear force of fish flesh, has been investigated by Dunajski (1980). From this work, it has been concluded that a thin-bladed shear force cell of the Kramer type can be applied.

These are but a few of the examples cited in the literature and until recently all involved expensive equipment and destructive sampling. Therefore, Botta (1991) developed a rapid non-destructive method for the measurement of cod fillet texture. It is a small, portable penetrometer which measures both firmness and resilience. Each test takes only 2-3 seconds to complete and results appear to correlate well with subjective texture grades.

8.4 Microbiological methods

The aim of microbiological examinations of fish products is to evaluate the possible presence of bacteria or organisms of public health significance and to give an impression of the hygienic quality of the fish including temperature abuse and hygiene during handling and processing. Microbiological data will in general not give any information about eating quality and freshness. However, as outlined in sections 5 and 6, the number of specific spoilage bacteria will be related to the remaining shelf life and this can be predicted from such numbers (see Figure 5.8).

Traditional bacteriological examinations are laborious, time-consuming, costly and require skill in execution and interpretation of the results. It is recommended that such analyses be limited in number and extent. Various rapid microbiological methods have been developed during the last decade and some of these automated procedures may be of use when large numbers of samples are to be analyzed.

Total counts

This parameter is synonymous with Total Aerobic Count (TAC) and Standard Plate Count (SPC). The total count represents, if carried out by traditional methods, the total number of bacteria that are capable of forming visible colonies on a culture media at a given temperature. This figure is seldom a good indicator of the sensoric quality or expected shelf life of the product (Huss et al., 1974). In ice-stored Nile perch, the total count was 109 cfu/g for days before the fish was rejected (Gram et al., 1989) and in lightly preserved fish products high counts prevail for long time before rejection. If a count is made after systematic sampling and a thorough knowledge of the handling of the fish before sampling, temperature conditions, packaging etc., it may give a comparative measure of the overall degree of bacterial contamination and the hygiene applied. However, it should also be noted that there is no correlation between the total count and presence of any bacteria of public health significance. A summary of different methods used for fish and fish products is given in Table 8.4.

Common plate count agars (PCA) are still the substrates most widely used for determination of total counts. However, when examining several types of seafood a more nutrient rich agar (Iron Agar, Lyngby, Oxoid) gives significantly higher counts than PCA (Gram, 1990). Furthermore, the iron agar yields also the number of hydrogen sulphide producing bacteria, which in some fish products are the specific spoilage bacteria. Incubation temperature at and above 30°C are inappropriate when examining seafood products held at chill temperatures. Pour plating and a 3-4 day incubation at 25°C is relevant when examining products where psychrotrophs are the most important organisms, whereas products where the psychrophilic Photobacterium phosphoreum occurs should be examined by surface plating and incubation at maximum 15°C.

Several attempts have been made to ease the procedures for determination of the content of bacteria (Fung et al., 1987). Both Redigel (RCR Scientific) and Petrifilm^TM SM (3M Company) are dried agars with a gelling agent to which the sample is added directly. The main advantage of Redigel and Petrifilm compared to conventional plate counts in addition to the costs, is the ease of handling. However, all agar-based methods share a common drawback in the lengthy incubation required.

Microscopic examination of foods is a rapid way of estimating bacterial levels. By phase contrast microscopy the level of bacteria in a sample can be determined within one log-unit. One cell per field of vision equals approximately 5 -10⁵cfu/ml at 1000 X magnification. The staining of cells with acridine orange and detection by fluorescence microscopy has earned widespread acceptance as the direct epifluorescence filter technique (DEFT). Whilst microscopic methods are very rapid, the low sensitivity must be considered their major drawback.

Bacterial numbers have been estimated in foods by measuring the amount of bacterial adenosine triphosphate (ATP) (Sharpe et al., 1970) or by measuring the amount of endotoxin (Gram-negative bacteria) by the Limulus amoebocytes lysate (LAL) test (Gram, 1992). The former is very rapid but difficulties exist in separating bacterial and somatic ATP.

Table 8.4 Methods for determination of the content of bacteria in seafood

Several methods (microcalorimetry, dye reduction, conductance and capacitance) used for rapid estimation of bacterial numbers are based on the withdrawal of a sample, incubation at high temperature (20-25°C) and detection of a given signal. In microcalorimetry the heat generated by the sample is compared to a sterile control, whereas in conductance and capacitance measurements of the change in electrical properties of the sample, as compared to a sterile control, is registered. The time taken before a significant change occurs in the measured parameter (heat, conductance, etc.) is called the Detection Time (DT). The DT is inversely related to the initial number of bacteria, i.e., early reaction indicates a high bacterial count in the sample. However, although the signal obtained is reversely proportional to the bacterial count done by agar methods, it is only a small fraction of the microflora that give rise to the signal and care must be taken in selection of incubation temperature and substrate.

Spoilage bacteria

The total number of bacteria on fish rarely indicates sensoric quality or expected storage characteristics (Huss et al., 1974). However, it is recognized that certain bacteria are the main cause of spoilage (see section 5.3). Different peptone-rich substrates containing ferric citrate have been used for detection of H₂S-producing bacteria such as Shewanella putrefaciens, which can be seen as black colonies due to precipitation of FeS (Levin, 1968; Gram et al., 1987). Ambient spoilage is often caused by members of Vibrionaceae that also will form black colonies on an iron agar to which an organic sulphur source is added (e.g., Iron Agar, Lyngby). No selective or indicative medium exists for the Pseudomonas spp. that spoil some tropical and freshwater fish or for Photobacterium phosphoreum that spoil packed fresh fish. At the Technological Laboratory, Lyngby, a conductance- based method for specific detection of P. phosphoreum is currently being developed (Dalgaard, personal communication). The presence or absence of pathogenic bacteria is often evaluated by methods based on immuno- or molecular biology techniques. Such techniques may also be developed for specific spoilage bacteria, and the Technological Laboratory has been currently investigating the use of antibodies specific for S. putrefaciens (Fonnesbech et al., 1993).Also, a gene-probe which is specific for S. putrefaciens has been developed but has not been tried on fish products (DiChristina and DeLong, 1993).

Spoilage reactions

Several spoilage reactions can be used for evaluation of the bacteriological status of fish products. As described above, agars on which H₂S producing organisms are counted have been developed. During spoilage of white lean fish, one of the major spoilage reactions is the bacteriological reduction of trimethylamine oxide to trimethylamine (Liston, 1980; Hobbs and Hodgkiss, 1982).When TMAO is reduced to TMA several physical changes occur: the redox-potential decreases, the pH increases and the electrical conductance increases. The measurement of such changes in a TMAO containing substrate inoculated with the sample can be used to evaluate the level of organisms with spoilage potential; thus the more rapid the change occurs the higher the level of spoilage organisms.

Several authors have inoculated a known amount of sample in a TMAO-containing substrate and recorded the time taken until a significant change in conductivity occurs (Gibson et al., 1984;Gram, 1985; Jorgensen et al., 1988). This time, the detection time, has been found to be inversely proportional to the number of hydrogen sulphide producing bacteria in fresh aerobically-stored fish, and rapid estimation of their numbers can be given within 8-36 hours.

The changes in redox-potential in a TMAO-containing substrate can be recorded either by electrodes or by observing the colour of a redox-indicator (Huss et al., 1987).As with the conductimetric measurements, the time taken until a significant change is recorded is inversely proportional to the initial amount of bacteria.

Pathogenic bacteria

Several pathogenic bacteria may either be present in the environment or contaminate the fish during handling. A detailed description of these organisms, their importance, and detection methods is given by Huss (1994).