Developing a Quality Index grading tool for hybrid striped bass (Morone saxatilis × Morone chrysops) based on the Quality Index Method

Abstract

A Quality Index Method (QIM) scheme was developed for hybrid striped bass (Morone saxatilis × M. chrysops). The new scheme is a convenient way to estimate storage time and remaining shelf life of fresh hybrid striped bass stored in ice. It consists of eight descriptors resulting in a total allowable score or Quality Index (QI) of fourteen points. Partial least squares regression (PLSR) was used to separate the effects of storage time and origin of fish. The calculated QI related linearly with storage days (QI = 0.483 × days in ice + 1.568, r² = 0.934). Storage time of unknown samples can be predicted from this relationship with a prediction error of ±1.2 days. Using Quantitative Descriptive Analysis (QDA), the maximum shelf life of fresh hybrid striped bass was determined to be 14 days in ice.

Introduction

Hybrid striped bass, a cross between striped bass (Morone saxatilis) and white bass (M. chrysops), is rapidly becoming a major product of U.S. aquaculture. Farm-raised hybrid striped bass is popular because of its favourable sensory characteristics and long-shelf life. The flesh is white, with a mild odour and flavour and a firm and juicy texture (Chambers & Robel, 1993). The hybrid striped bass industry in North Carolina began in 1987 (Hodson, 1995) and in 2005 nineteen farms produced 2.1 million lbs of fish, with a farm gate value of $6 million (Losordo et al., 2005). Current restraints on industry growth are inadequate supply of fingerlings, domestication of brood stocks and inconsistencies in the sensory quality and fresh shelf life of whole fish.

Fish quality can be evaluated in numerous ways, but the most reliable and easiest methods are sensory techniques (Ólafsdóttir et al., 1997). Raw or cooked fish are presented to a group of trained panellists, who evaluate the fish quantitatively using objective terms and numerical scales. The Quantitative Descriptive Method (QDA) has been widely used for the evaluation of seafood (Johnsen & Kelly, 1990; Schubring, 2000; Sveinsdóttir et al., 2002; Gines et al., 2004). With this method, cooked samples are evaluated for appearance, odour, flavour and texture on line scales (Stone & Sidel, 1993). The advantage is a thorough description of the quality, but the method requires a well-trained panel and can be time consuming. The Quality Index Method (QIM) was developed in Tasmania (Bremner, 1985) and is widely used in research and industry in Europe (Hyldig & Nielsen, 1997). All fish species have their own characteristic spoilage patterns and indicators, and consequently QIM schemes must be species specific. QIM schemes include parameters such as appearance of eyes and gills as well as odour and texture. The main advantage with the QIM system is the linear correlation between the Quality Index (QI) and the storage time in ice, making it possible to estimate the remaining shelf life. Other advantages include minor need for sample preparation, as whole raw fish are evaluated and not as much training is required as for some other sensory methods, such as QDA. The system can, therefore, be used on the farm or at the fish market, making it easier to perform objective quality evaluations. QIM schemes have been developed for a wide range of fish species, both wild and farmed (see e.g. Martinsdóttir et al., 2001). This work has mainly been done in Europe, but more work is needed to formulate schemes for fish species from other parts of the world. This study is one of the first of its kind in North America.

The overall goal of this project was to develop a QIM scheme for hybrid striped bass. This was achieved by (1) evaluating the shelf life and sensory quality of hybrid striped bass with a preliminary QIM scheme; (2) evaluating the cooked shelf life and sensory quality of hybrid striped bass with QDA; and (3) comparing the sensory qualities of fresh and cooked hybrid striped bass over storage time in ice.

Materials and methods

Initial development of the QIM scheme

Fish

Hybrid striped bass (c. 0.70 kg) were obtained in July 2002 from two farms in eastern North Carolina – nine fish from Farm I in 1 week and twelve fish from Farm II the following week. Fish were dipped from ponds by nets into chill tanks to render them senseless before packing in plastic boxes in ice (one box for each farm) for transport to the NCSU Seafood Laboratory. At the laboratory, half the fish were kept whole and the other half was gutted before they were stored in ice in a walk-in cooler (0–2°C) for up to 21 days.

QIM-descriptor determination

Nine to twelve fish were used from each of the two sample dates in this initial study. Five panellists from the laboratory participated in this study. They had not been trained prior to the evaluation, but were instructed in the QIM procedure as described by, e.g. Martinsdóttir et al. (2001) and Martinsdóttir (2002). All fish from Farms I and II were tagged with a three-digit code and were kept on ice in metal trays during the evaluations. The evaluations were performed three times a week during the storage period in a well-aerated room and under white fluorescent light. Panellists described all fish with their own words considering, the appearance of the whole fish, eyes, gills, cut surface, abdomen and texture.

All descriptors given by the panellists were sorted by the panel leader and grouped according to similarities. A preliminary scheme was constructed when fish from Farm I had been stored throughout the entire period. This scheme was then tested on the remaining samples from Farm II. During this test period, some additional changes were made to the scheme, mainly in the descriptions of the quality parameters to make each description more precise or easier to evaluate. As a result, a QIM scheme was proposed.

Shelf-life determination

Fish

Hybrid striped bass were obtained from two farms (A and B) in eastern North Carolina over a 4-week period in March to April 2005. The fish had been fed similar commercial feed and weighed when they were harvested 0.40–0.85 kg (mean = 0.59 kg, SD = 0.08 kg, n = 122). The fish for this study were harvested according to the procedures normally used on the farms. On Farm A, the fish were gathered from the pond with a large net, anaesthetised, transferred into slush ice, stunned (electro shock) and packed in ice in laminated cardboard boxes. At Farm B, fish were gathered from the pond 4–6 days before harvesting and placed in tanks with circulating clean fresh water in order to purge residual feed from the fish stomachs. They were then slaughtered as described for fish from Farm A.

Whole fish from both farms were transported iced to the NCSU Seafood Laboratory, re-iced and stored in ice in a walk-in cooler (0–2°C) for up to 16 days. New ice was added as the ice melted.

QIM evaluation

An in-house panel (n = 6) performed the evaluations, and three to six panellists participated in each of the five evaluation sessions. Panellists were specifically trained in the principles of QIM evaluation of hybrid striped bass, i.e. how to identify and quantify the quality parameters used in the QIM-scheme in hybrid striped bass from different storage times. Training was conducted during three sessions in the days preceding the evaluation sessions. In each session fish of different freshness was evaluated.

Each hybrid striped bass was marked with a three-digit code (with no reference to farm or storage time), and placed on white wax paper 15 min prior to the evaluations. The evaluations were conducted under standardised conditions, with as little interruption as possible at room temperature, in a well-aerated room, and under white fluorescent light. All panellists evaluated all fish by moving from one fish to the other and using the evaluation form developed during the initial study in 2002 (Table 1). The panellists were instructed to check that all descriptors were evaluated for all fish, before they handed in the evaluation forms. Eighteen to twenty-four fish from different storage times were evaluated in each session, so that three or six fish from four different storage times were evaluated in each session. Fish from Farm A were evaluated on day 2, 5, 7, 9, 12, 14 and 16 and fish from Farm B on day 1, 4, 6, 8, 11, 13 and 15. Afterwards data was entered manually into a spreadsheet.

QDA evaluation

Quantitative Descriptive Analysis was used to evaluate the quality of cooked samples of hybrid striped bass. In this method, a number of descriptors for appearance, odour, flavour and texture are evaluated independently on a line scale (Stone & Sidel, 1993).

The panel consisted of six panellists, with five to six panellists participating in each session. These panellists did not participate in the QIM evaluation, and were selected, tested and specifically trained in descriptive analysis (ISO, 1993, 1994). The panel was familiar with sensory evaluation of hybrid striped bass and therefore three training sessions that each lasted c. 3 h were sufficient. Vocabulary was developed during three sessions using hybrid striped bass from Farm A and B from different storage times (day 1–16) and stored as described above. The first session was qualitative, i.e. the aim was to develop a list of descriptors for appearance, odour, flavour, and texture of cooked hybrid striped bass. The following two sessions were quantitative, i.e. the panellists were trained to evaluate the descriptors on a line scale. Each descriptor was evaluated in an unstructured 15 cm line scale anchored 1.5 cm from both ends with 0 = none and 15 = strong presence of the descriptors. The descriptors used in our study are listed in Table 2.

The samples for QDA were prepared immediately after the QIM evaluations. Samples (c. 2 × 2 × 4 cm) were cut from the loin of the fish. Four samples were obtained from each fish, and the skin was left intact. Samples were placed skin down in aluminium containers (15 × 9 × 5 cm) and were covered with aluminium foil marked with individual three-digit codes. The codes were random numbers with no information about farm or storage time. The samples were cooked at 175°C for 15 min in a preheated oven and then served to the panel. Each panellist evaluated a total of eight samples in each session, i.e. duplicate samples from four storage times. There was a 2 min break between each sample and a 15 min break after the first four samples. Each session lasted c. 2 h. The evaluations were conducted under standardised conditions, with as a little interruption as possible at room temperature, in a well-aerated room, and under white fluorescent light. The panellists were instructed to check that all descriptors were evaluated for all fish, before they handed in the evaluation forms. All evaluations on the line scales were measured manually by the panel leader and entered into a spreadsheet.

Data analysis

Anova was performed on QIM and QDA results using Statistica^® version 5.5. (Statsoft Inc., Tulsa, OK, USA). The QIM-descriptors and the QI were related to storage time and farms using partial least squares regression (PLSR). Differences between individual fish from the same storage conditions and the difference between evaluations given by each panellist on the same fish were also explored with PLSR (Esbensen et al., 2000). The Jack-knife method (Martens & Martens, 1999) was used to determine significant variables with a significance level of 0.05. All models were calculated with full-leave-one-out-cross-validation and standardised data (weighted with 1/SD). The optimal number of components was determined as the number of components corresponding to the highest explained and validated Y-variation. All calculations were performed in Unscrambler^® version 9.1.2 (Camo Process AS, Oslo, Norway). The equation of best fit and correlation coefficients (r²) of QI vs. storage time in ice were calculated using Microsoft^® Excel 2002 (Microsoft Co., Redmond, WA, USA).

The QDA data was level corrected according to Nielsen et al. (2005), i.e. differences occurring because the panellists used different parts of the line scale were reduced. The level corrected data was explored by principal component analysis (PCA, Esbensen et al., 2000) and PLSR to study the effect of harvesting method and storage time on the quality parameters, and to estimate the cooked shelf life.

Results and discussion

The QIM scheme

The initial objective was to develop a QIM scheme for both round and gutted hybrid striped bass. The first version of the scheme included descriptions for the cut surface and the abdomen. However, the shelf-life changes in these parameters were not consistent or gave us any additional information, and the parameters were therefore omitted. The colour and discoloration of the skin were the quality parameters the panellists found most difficult to evaluate. The colour, i.e. the darkness of the stripes, not only faded when the fish were stored in ice, but were different initially, with fish from Farm I having darker stripes than fish from Farm II. The discoloration, i.e. red bruising on the belly, also seemed to depend on the farm in addition to storage time, although fish from the same farm could vary greatly with regard to bruising. It is not clear what causes the variation in discoloration. One possible explanation is that it arises from stressful harvesting as indicated by Harms et al. (1996). The farmers suggest that it is related to spawning, because spawning fish seem to be more easily bruised, according to the growers’ observations. It is also likely to suggest that the effect is temperature dependent, as the bruising occurs more frequently during summer time.

The final QIM scheme (Table 1) resulted in descriptions for nine quality parameters scored 0–2 demerit points, with 0 point given to fresh fish and two points to spoiled fish. The QI is the total sum of demerit points given to each fish. This scheme is comparable with the schemes previously developed in Europe on several cold water fish species and shrimp (Martinsdóttir et al., 2001).

The initial study showed that ice stored hybrid striped bass had minor changes until after 10 days, and after that changes were observed in odour and texture. Major quality defects observed were in discoloration (red spots and bruising) along the ventral and anal fins, gill cover and belly areas. Eye (pupil) colour began to loose its bright appearance on day 5, while gill colour (bright red) decreased about the same time and odour development was noted at day 10. Texture of raw fish muscle remained firm up to 14 days in ice.

The results from this study were used to evaluate hybrid striped bass in a shelf life study.

Development of individual descriptors in the QIM scheme

In Fig. 1 is shown the development in the individual quality parameters during the storage period. Skin colour, discoloration and odour of the whole fish differed between fish from the two farms. Especially, the dark distinct stripes on the fish as evaluated under skin colour faded during ice storage, but fish from Farm B had significantly darker stripes than fish from Farm A during the entire storage period (P < 0.001).

Discoloration did not seem to be affected by the time in ice. During the first half of the storage period, fish from Farm B were more bruised than fish from Farm A. This was opposite during the second half of the storage period. Of the forty-eight fish evaluated from each farm, 58% had no bruising, while 31% of the fish from Farm A had minor bruising, and 29% of the fish from Farm B had severe bruising. The QIM-system is ideally constructed with quality parameters that receive increasing demerit score with increased storage time (Hyldig & Nielsen, 1997; Martinsdóttir et al., 2001). This was not the case for discoloration in this study, and it should therefore be excluded from the QI-calculation.

The odour of the whole fish changed from nearly ‘neutral’ to ‘green’ notes within the first 10 days, and then ‘fishy’ and ‘putrid’ notes were observed. Fish from Farm B smelled significantly more ‘fishy’ (P < 0.05) than fish from Farm A after day 9. The texture changed from being stiff due to rigour to firm within the first 5 days and remained relatively firm during the entire storage period. Approximately half of the sampled fish were soft at the end of the storage period. The eyes changed after c. 8–10 days from being black and convex, to grey and flat, and ultimately to (after c. 12 days) hazy and concave. No differences were found between fish from the two farms with regards to changes in texture or eyes. Neither was there any difference regarding the parameters describing the gills. The mucus changed gradually from transparent to milky, while the colour changed from bright red to pale red. Only a few fish (c. 2/20) were observed with brown gills instead, the pale red colour tended to dominate as the fish spoiled. The odour in the gills followed the same pattern as the odour of the whole fish. It changed from ‘fresh’ to ‘green’ within the first 7 days and then to ‘fishy’ at the end of the storage period.

Partial least squares regression modelling was used to better separate the effects of storage time and farms in order to verify that discoloration should be taken out of the QIM scheme. The correlation loadings from this analysis are in Fig. 2. The first factor (horizontal) spans the variation of data due to increased storage time, while the second factor (vertical) spans the variation due to differences between farms. Samples aligned to the left have received high scores for the parameters aligned on the first factor, and the first axis divides samples from Farm A (under the line) and Farm B (over the line). Discoloration differs between farms and does not change with increased storage time. Skin colour is also dependent on the origin of the fish, i.e. the farm, but changes also with time. All the other parameters, including odour of whole body that was suggested above to differ between farms, are independent of where the fish came from and increase with increasing storage time.

Quality Index vs. storage time

The QI was calculated as the total sum of demerit points given to each fish. The QI represents an average for all evaluations made on fish from the same farm and time in ice. The QI formed a linear relationship with storage time with a slope of 0.468 and a y-intercept of 2.418 (r² = 0.904, figure not shown). No significant difference was found in data collected from the two farms separately. The slope is comparable with the QI-storage time relationship for farmed salmon (Sveinsdóttir et al., 2002), but lower than for, e.g. wild cod (Martinsdóttir et al., 2001) and herring (Nielsen & Hyldig, 2004). The y-intercept is higher than in other species (Martinsdóttir et al., 2001). Because the aim is to develop a scheme with a y-intercept as close to 0.0 as possible (Hyldig & Nielsen, 1997) the data was analysed to see how the individual quality parameters changed with increased storage time.

There were variations within groups, i.e. fish from the same farm and storage day did not always get the same QI (P < 0.05). This variation was significant on days 1, 8, 11 and 13, where the range spanned three to four QI-units. Differences also occurred between individual panellists (P < 0.05), with panellist no. 6 and no. 2 usually scoring the fish higher or lower, respectively, than the other panellists. This was also seen by Sveinsdóttir et al. (2002), who concluded that including more fish per code (five instead of three) decreased the variation and resulted in better prediction of remaining storage time.

The results above indicated that discoloration should be omitted from the calculation of the QI, because it was unrelated to storage time. When this was done, the linear relation between QI and storage time changed to a slope of 0.483 and y-intercept of 1.568 (r² = 0.934, Fig. 3). Hence, the slope remained almost unchanged, the y-intercept decreased one unit, and the correlation increased slightly. No difference was found between the linear regression lines obtained from fish from the two farms.

Discoloration is a very important quality defect in hybrid striped bass. Although it was taken out of the QIM scheme it is highly recommended that it is evaluated and the score is given separately. More research is needed to describe why discoloration occurs and how it can be prevented.

Predicting the storage time for hybrid striped bass from the QI

This study resulted in a QIM scheme where QI has a high correlation to the storage time. This relationship can be used to estimate the storage time from a QI obtained from fish with an unknown storage history. For instance, this is the case, when a buyer receives an order and wants to estimate how long he/she can store the fish before it has to be processed. A prediction model between QI and storage time was calculated using PLSR to estimate how accurately the QIM scheme could estimate the storage age of the fish. The model was calculated using three-fourth of the results obtained in the QIM evaluations. The samples were randomly chosen and the storage time of the last one-fourth was then estimated from the QI.

The results from the prediction model showed that the standard error of performance (SEP) for the QI was 1.2. Since the QI was the sum of eight parameters from the QIM scheme (discoloration omitted) it can be assumed to be normally distributed and 2 × SEP can be regarded as a 95% confidence interval (Esbensen et al., 2000). Therefore, it can be assumed that when individual fish are evaluated from a batch, the storage time can be predicted within a margin of ±1.2 days. Including more than one fish and averaging the obtained results will result in an increased accuracy. With results from three fish the margin of error can be decreased to ±0.8 days.

Estimating the shelf life of hybrid striped bass

The QIM evaluations gave an indication of the shelf life of hybrid striped bass. However, it is not possible to estimate the shelf life from the QIM evaluations alone, because they are performed on whole raw fish, and the sensory parameters in raw fish are not necessarily directly related to the sensory parameters in cooked fish. For example, some odours or flavours may disappear when the fish is cooked while others are enhanced. Looking at Fig. 2 that shows the development in descriptors in raw hybrid striped bass over time, a grouping of storage days is seen along the first component (from right to left). Evaluations from the first 5 days (D01-D05) are grouped, then evaluations from the next 4 days (D06–D09), followed by D11–D14 and finally are days 15 and 16 positioned on the left end of the factor. The gap between days 11–14 and days 15–16 could indicate that one major shift in the quality parameters occurs at this point. In order to be able to relate the quality of raw and cooked hybrid striped bass, QDA of cooked samples was also used in this study. A little variation was found in the sensory properties of the cooked samples of hybrid striped bass during most of the storage period from the beginning of storage, and this made it difficult to determine the shelf life. Some parameters remained unchanged during the entire storage period (Fig. 4). Existing work has estimated the shelf life at 1–4°C to be c. 10 days (Karahadian et al., 1995) to 14 days (Eifert et al., 1992), while others have reported that a trained panel did not detect any significant fishy odour and flavour in samples stored for 21 days at −2°C or 2°C and only slight differences in texture (Boyd et al., 1992).

In addition to evaluating the parameters in Table 2, the panellists were also asked to add additional descriptors if they were encountered during the study. At the end of the study, around days 13–14, the comments for odour and flavour became more negative although the fish was not rated particularly ‘fishy’. This would have been expected, based on the study by Karahadian et al. (1995), where an increased production of hydroperoxides and nucleotides during refrigerated storage of striped bass was observed resulting in ‘fishy’ odour. Scores for ‘fishy’ odour and flavour in the present study were between 15 and 40 (on a scale from 0 to 150) during the entire storage period. ‘Fishy’ flavour increased significantly (P < 0.05) in fish from Farm B until day 12, but no significant change was found in fish from Farm A.

According to both PLSR and Anova, only a few parameters changed over time and only marginally (data not shown). The appearance and odour of the cooked samples did not change significantly over time (P =0.42–0.96). The intensity of the ‘sweet’ flavour decreased over time (P = 0.05). This was more evident in fish from Farm B, which also had a higher intensity of sweetness than fish from Farm A over the entire storage period. The parameters that changed most were firmness and juiciness. Both decreased significantly (P = 0.007 and 0.05, respectively), i.e. the fish softened and became dryer, after c. 9 days of storage. Fish from Farm B were firmer than fish from Farm A during the entire storage period. While the texture of fish from Farm A remained relatively firm, it decreased significantly (P < 0.05) from day 8 in fish from Farm B (Fig. 4). These fish were also described by the panellists as mushy by days 15–16. Based on especially the texture scores, we estimate the shelf life of hybrid striped bass to be 14 days.

Other parameters that were not on the evaluation sheet, but used frequently by the panellists in their comments, were ‘earthy’ and ‘muddy’ odour and flavour. These descriptors were not on the evaluation form, because they were not present in the fish that was used for the word development and training sessions. These descriptors were only used for fish from Farm A, but not from all storage days or harvests. This off odour is frequently found in farm-raised catfish and its occurrence depends on numerous parameters (Maga, 1987; Howgate, 2004). It has not been reported previously in pond raised hybrid striped bass. Earthy and muddy odours were not as intense in the whole round fish as in the cooked fillet samples.

Calculating the remaining shelf life

With an estimate of the shelf life and the regression line from the QI-storage time relationship, it is possible to predict the remaining shelf life of the hybrid striped bass stored in ice. This is the most important feature of the QIM system. It provides the buyer with a tool not only to determine the quality of the fish, but also to calculate how many days he or she can store the fish before it loses its quality. The remaining shelf life is found by subtracting the predicted storage time (calculated from the QI regression line), from the estimated shelf life (estimated in this study to be 14 days in ice for hybrid striped bass). Example: a QI score of 6 corresponds to a storage time of (6–1.568)/0.483 = 9.2 days in ice which gives a remaining shelf life of 14–9.2 = 4.8 days in ice.

Validating the results

The study was repeated in the fall of 2005 with forty-two fish from Farm A using the same methods described earlier to validate the results and evaluate the robustness of the scheme against seasonal variations. The QI and storage time in ice were related linearly (r² = 0.989) with a slope of 0.573 and y-intercept of 1.725. This is not significantly different from the relation obtained in the spring, even though three of four trained QIM panellists in this part of the study had not participated in earlier QIM sessions. The maximum shelf life in ice was again estimated to be 14 days and again the only descriptor that changed significantly over time was texture, and the discoloration seen in the original study was again apparent and the earthy and muddy odours and flavours were present.

Conclusion

This study has resulted in a QIM evaluation scheme for hybrid striped bass. The scheme results in a linear relationship between QI and storage time in ice: QI =0.483 × days in ice + 1.568 (r² = 0.934), and can be used on individual fish to estimate the remaining shelf life in ice with an error of maximum ±1.2 days. The results show that it can be used on fish from different farms using different harvesting methods, and it can grade hybrid striped bass freshness. The study also shows that the QIM principles can be applied to develop evaluation schemes that can grade the fish on characteristics that do not depend solely on storage time, e.g. colour of skin and discoloration. It also emphasises that there is need for more information on the quality of hybrid striped bass. Especially, the occurrence of earthy/muddy odours and flavours needs to be investigated so the problem can be reduced. There is also need for more information about the causes behind the red bruising that occurs frequently.

Acknowledgments

The authors want to acknowledge Kathy Robinson, Ted Davis and Steve Locke from The Fish Connection Cooperative, Washington, NC for providing the hybrid striped bass, the sensory panels for their skilled work, Mike Frinsko from the North Carolina Cooperative Extension Service, Trenton, NC for valuable information about hybrid striped bass aquaculture, and Greg Bolton, NCSU Seafood Laboratory, Morehead City, NC for assisting with the sample preparation. This study was funded in part through a grant (R/MG-0416) from the North Carolina Sea Grant Program, the North Carolina Cooperative Extension Service, and the Center for Marine Sciences and Technology. Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the authors and do not necessarily reflect the views of North Carolina State University.

References