- Email: [email protected]
Evaluation of sea bream (Sparus aurata) shelf life using an optoelectronic nose
Food Chemistry 138 (2013) 1374–1380
Contents lists available at SciVerse ScienceDirect
Food Chemistry journal homepage: www.elsevier.com/locate/foodchem
Evaluation of sea bream (Sparus aurata) shelf life using an optoelectronic nose Patricia Zaragozá a, Ana Fuentes b, Isabel Fernández-Segovia b, José-Luis Vivancos a,c,⇑, Arantxa Rizo b, José V. Ros-Lis a, José M. Barat b, Ramón Martínez-Máñez a,c a
Centro de Reconocimiento Molecular y Desarrollo Tecnológico, Unidad Mixta Universitat Politècnica de València – Universitat de València, Camino de Vera s/n, 46022 Valencia, Spain Departamento de Tecnología de Alimentos, Universitat Politècnica de València, Camino de Vera s/n, 46022 Valencia, Spain c CIBER de Bioingeniería, Biomateriales y Nanomedicina (CIBER-BBN), Spain b
a r t i c l e
i n f o
Article history: Received 9 July 2012 Received in revised form 20 September 2012 Accepted 22 October 2012 Available online 12 November 2012 Keywords: Sea bream Colorimetric array Shelf life Chromogenic sensors
a b s t r a c t A new optoelectronic nose for the shelf-life assessment of fresh sea bream in cold storage has been developed. The chromogenic array used eight sensing materials (based on aluminium oxide and silica gel) containing pH indicators, Lewis acids and an oxidation–reduction indicator. The colour changes of the sensor array were characteristic of sea bream spoilage. Colour modulations were measured on day 0 and for the samples held in cold storage for 2, 4, 7, 9 and 11 days. Determination of moisture content, pH, total volatile basic nitrogen (TVB-N), drip loss, ATP-related compounds and K1-value and microbial (mesophilic bacteria and Enterobacteriaceae) analyses were carried out on the same days. The changes in the chromogenic arrays data were processed by statistical analysis (PCA). Moreover, PLS statistical studies allowed the creation of models to correlate the chromogenic data with concentrations of mesophilic and Enterobacteriaceae. The results suggest the feasibility of this system to help develop optoelectronic noses for fish freshness monitoring. Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction Fish is a highly perishable food product whose freshness and quality rapidly decline post-mortem. Loss of freshness and spoilage of fish are complicated processes, and various factors such as species and different storage conditions influence the spoilage pattern. In fact, the development of reliable methods to assess fish freshness and to evaluate quality criteria has been the goal of fish research for many years (Di Natale et al., 2001; Gil et al., 2008). The current methods used in the food industry to determine fish shelf life are based on physical, chemical, microbiological and sensory measurements (Ólafsdóttir et al., 2004). However, some of these methods are tedious, expensive, time-consuming and require skilled personnel. As an alternative, the development of rapid nondestructive quality control techniques, which can be applied at any stage of the supply chain, is of much importance (Pons-SánchezCascado, Vidal-Carou, Nunes, & Veciana-Nogués, 2006). A possible solution to this issue is to design smart or intelligent packaging capable of providing information about fish quality. Intelligent packaging is an emerging technology that uses the communication function of the package to facilitate decision making in order to
⇑ Corresponding author at: Centro de Reconocimiento Molecular y Desarrollo Tecnológico, Unidad Mixta Universitat Politècnica de València – Universitat de València, Camino de Vera s/n, 46022 Valencia, Spain. Tel.: +34 96 387 7000; fax: +34 96 387 9869. E-mail address: [email protected] (J.-L. Vivancos). 0308-8146/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.foodchem.2012.10.114
provide the benefits of enhanced food quality and safety. Smart packaging is defined as a system that monitors the condition of packaged food to provide information about the quality during storage, transportation and distribution (Yam, Takhistov, & Miltz, 2005). The potential of smart packaging technology is far-reaching, ranging from food safety and drug use monitoring to postal delivery tracking and embedded security tags (Butler, 2004). In the context of this work, smart packaging for fish monitoring would involve a fish freshness ‘‘indicator’’, which uses metabolites as ‘‘information’’ to monitor the status of fish, which can be directly related to fish spoilage (Kuswandi et al., 2012). However, studies on the application of this type of technology in fish are very limited. From another viewpoint, the use of chromogenic chemosensors is a promising technique for the simple design of freshness and quality indicators. Colorimetric probes are usually cheap, versatile, can be printed on the package and colour changes can be easily measured using simple image capturing systems (e.g. cameras) or seen by the naked eye through transparent films (Huang, Xin, & Zhao, 2011). In fact, there are very few technologies which are as inexpensive as visual imaging. Although some chromogenic indicators have been described for food spoilage detection, they are generally based on a single compound and have their limitations such as lack of specificity (offering false positives or false negatives). In particular, the possibility of false negatives is likely to dissuade producers from adopting indicators unless specific indications of actual spoilage can be guaranteed. Thus, although
P. Zaragozá et al. / Food Chemistry 138 (2013) 1374–1380
1375
some attempts have been made in single analyte indicators, one of the most promising approaches is the use of optoelectronic noses, built by an array of dyes able to offer spoilage information through suitable colour changes (Loughran & Diamond, 2000; Pacquit et al., 2006). Nonetheless, despite these encouraging examples, chromogenic arrays for food quality analyses are still scarce (Alimelli et al., 2007; Huang et al., 2011; Kuswandi et al., 2012). Based on the above detailed information, we aimed to develop a rapid, easy-to-use optoelectronic system for shelf-life assessments of fish samples. As fish, we selected fresh sea bream, which was kept in cold storage. Fish spoilage was monitored simultaneously by using an array of chromogenic dyes, and through physicochemical and microbial analyses. 2. Materials and methods 2.1. Chromogenic array preparation The five dyes used in the study were bromocresol purple, Resorufin, bromophenol blue, phenol red (acquired from Sigma Aldrich (St. Louis, MO, USA)), and a dinuclear complex of rhodium (synthesised according to known procedures (Esteban et al., 2010; Moragues et al., 2011)). Selection of these dyes was based on previously reported optoelectronic noses and on our own experience in designing colorimetric probes (Salinas et al., 2012). These arrays include pH indicators, (phenol red, bromocresol purple, bromophenol blue), Lewis acids (dinuclear rhodium complex), and oxidation–reduction indicators (Resorufin). The chemical structures of the five dyes are shown in Fig. 1. The supports (aluminium oxide and silica gel) and the analytical-grade solvents (dichloromethane and ethanol) were purchased from Scharlau Chemie, S.A. (Barcelona, Spain). The final sensing materials were readily obtained by preparing suspensions of the inorganic support (aluminium oxide or silica gel) with a certain amount of the corresponding dye in an appropriate solvent (dichloromethane or ethanol). These suspensions were stirred for 8 h to confer maximum dye absorption in the inorganic support. Afterwards, the solvent was evaporated in a rotary evaporator to obtain the corresponding sensing material. A total of eight chromogenic materials were prepared using this simple approach. Table 1 shows in detail the amounts of silica, dye and solvent used in their preparation. The array was prepared by placing approximately 40 mg of each material into a microplate to obtain a 2 4 array (8 sensing materials). 2.2. Sample preparation Sea bream from a Spanish farm (Frescamar Alimentación, S.L., Burriana, Castellón, Spain) of commercial size 400–600 g were used as the raw material. Fish were transported to the laboratory in polyspan boxes with ice. Eighteen fish were weighed and fifteen were individually placed in aluminium trays of 2.7 L. The eight different sensing materials described in Table 1 were placed in a microplate inside the tray close to the fish samples (Fig. 2). In addition, three trays containing the colorimetric array were also packaged in the absence of fish and used as controls. Petri dishes filled with water were placed in the control trays to reproduce moisture conditions. The studies with the samples were repeated three times. All the trays were wrapped with plastic film and stored at 4 °C for 11 days. The physico-chemical and microbiological analyses of the samples were performed on days 0, 2, 4, 7, 9 and 11 of storage. Before the analyses, samples were headed, gutted and filleted, and two fillets were obtained from each fish. Photographs of the colorimetric arrays were obtained on each analysis day in all the trays stored.
Fig. 1. Chemical structures of the dyes used.
2.3. Physico-chemical analyses Moisture content, pH, total volatile basic nitrogen (TVB-N), drip loss, ATP-related compounds and K1-value were determined in all the samples in triplicate (n = 3). 2.3.1. Moisture, pH, TVB-N and drip loss The moisture content was determined according to the AOAC method 950.46 (1997). pH measurements were taken out using a digital pH-meter micropH 2001 (Crison Instruments, S.A., Barcelona, Spain) with a puncture electrode (Crison 5231) at six different sample locations. The TVB-N contents were determined by steam distillation according to the method described by Malle and Tao (1987). To evaluate the drip loss, samples were weighed before being placed in the tray and stored, as mentioned above. After storage, samples were blotted with filter paper and weighed again. The drip loss was expressed as the percentage of exudate. 2.3.2. ATP related compounds and K1-value The ATP-related compounds, consisting of inosine-50 -monophosphate (IMP), inosine (Ino), and hypoxanthine (Hx), were assayed by HPLC by following the method described by Barat et al. (2008), with some minor modifications. The extraction method used was similar to that described by Burns and Kee (1985). The analysis was conducted on a Hitachi LaChrom Elite liquid chromatography (Hitachi Ltd., Tokyo, Japan) with a pump (model L-2130), an auto-sampler (model L-2200) and a UV detector (model L-2400). Data acquisition was performed with the EZChrom Elite software (Agillent Technologies, Palo Alto, CA, USA). Separations were made on a reverse-phase Ultrabase C18 250 4.6 mm, with an internal particle diameter of 5 lm (Análisis Vínicos, S.L., Tomelloso, Spain). A guard column containing the same C18 packing as above was positioned in front of the analytical column to protect
1376
P. Zaragozá et al. / Food Chemistry 138 (2013) 1374–1380
Table 1 Details for the preparation of the sensing materials. Code
Dye
Support
mg dye (g solid)1
mmol dye (g solid)1
Solvent
Solvent amount (ml)
1 2 3 4 5 6 7 8
Resorufin Bromocresol purple Bromocresol purple Bromophenol blue Phenol red Dinuclear rhodium complex Dinuclear rhodium complex Dinuclear rhodium complex
Aluminium oxide Silica gel Aluminium oxide Silica gel Silica gel Silica gel Silica gel Silica gel
40 40 4 40 40 40 80 140
0.170 0.074 0.007 0.059 0.113 0.042 0.083 0.145
Ethanol Ethanol Ethanol Ethanol Dichloromethane Dichloromethane Dichloromethane Dichloromethane
8 8 8 8 5 6 6 6
2.6. Statistical analysis
Fig. 2. Scheme of a tray with a sea bream sample and the chromogenic array.
it from contamination. Compounds were identified by using a retention time comparison of the unknown with those of the standards, and by a standard addition or ‘‘spiking’’ (Johnson & Stevenson, 1978). IMP, Ino, and Hx were quantified according to the external standard method using calibration curves of the peak area of compound versus the compound concentration under identical chromatographic conditions. K1-values were calculated according to Eq. (1):
K 1 ð%Þ ¼
½Ino þ ½Hx 100 ½IMP þ ½Ino þ ½Hx
ð1Þ
where IMP is inosine 50 -monophosphate; Ino, inosine; Hx, hypoxanthine. All the chemical reagents were provided by Sigma–Aldrich.
2.4. Microbial analyses Mesophilic counts were performed according to the method provided in the ISO 4833:2003 standard (ISO Standard, 2003). Enterobacteriaceae were enumerated according to the method described by Pascual and Calderón (2000). All the culture media were provided by Scharlau. All the analyses were performed in triplicate and the results were expressed as log cfu g1.
2.5. Photographs and colour measurements The colour measurement was performed on each component by taking photographs with a Nikon camera D3000. The colour RGB and CIELAB coordinates of the array were collected on each analysis day by taking a photograph and by using the Photoshop Pro 5 software.
Data are reported as mean ± standard deviation. Data were statistically processed using Statgraphics Centurion (Statpoint Technologies, Inc., Warrenton, VA, USA). An analysis of variance (One-Way ANOVA), in which storage time was the factor, was conducted for each parameter evaluated to test whether there were significant differences throughout cold storage. The LSD procedure (least significant difference) was used to test for differences between averages at the 5% significance level. In order to assess the feasibility of the colorimetric array to determine the fish shelf life, a Principal Components Analysis (PCA) and a Partial Least Square (PLS) were conducted. All the analyses were performed using the MATLABÒ PLS Tool-box (Eigenvector Research Inc.). A PCA was also performed with the data from the physico-chemical and microbiological analyses.
3. Results and discussion 3.1. Physico-chemical analyses 3.1.1. Moisture, pH, TVB-N and drip loss The moisture content, pH, TVB-N and drip loss values of fresh sea breams (day 0) and of the samples in cold storage for 2, 4, 7, 9 and 11 days are shown in Table 2. The moisture values ranged from 71.03 to 75.57 g/100 g, which is in accordance with different studies carried out on cultured sea bream (Alasalvar et al., 2001; Goulas & Kontominas, 2007; Grigorakis, 2007; Grigorakis, Taylor, & Alexis, 2003). The pH values remained stable up to day 4 and slightly increase thereafter. The higher values observed at the end of storage could be related to microbial development since the metabolism of microorganisms produces basic compounds. The pH changes pat-
Table 2 Moisture, pH, TVB-N and drip loss of fresh sea bream (day 0) and samples in cold storage for 2, 4, 7, 9, and 11 days (means and standard deviations, n = 3). Storage time (days)
Moisture (g H2O/100 g)
pH
TVB-N (mg N/100 g)
Drip loss (% exudates)
0 2 4 7 9 11
71.61 ± 2.30a 71.03 ± 1.45a 72.96 ± 0.93ab 71.96 ± 0.77ab 75.57 ± 0.53c 74.28 ± 1.45bc ⁄
6.29 ± 0.02a 6.29 ± 0.04a 6.30 ± 0.05a 6.34 ± 0.08a 6.47 ± 0.03b 6.37 ± 0.04a ⁄⁄
18.88 ± 0.74a 18.23 ± 1.24a 18.84 ± 0.72a 17.13 ± 0.11a 18.20 ± 0.37a 21.33 ± 2.54b ⁄
– 0.34 ± 0.18ab 0.45 ± 0.35ab 0.99 ± 0.41c 0.66 ± 0.32bc 1.73 ± 0.11d ⁄⁄⁄
Same letters in the same column indicate homogeneous group membership. Significance level (a). * p < 0.05. ** p < 0.01. *** p < 0.001.
P. Zaragozá et al. / Food Chemistry 138 (2013) 1374–1380
tern was similar to that observed by Grigorakis et al. (2003) and by Barat et al. (2008). The TVB-N values slightly increased on day 11 of storage. The slight increase observed in this study was similar to the evolution of TVB-N in sea bream during ice and cold storage reported in several studies (Grigorakis et al., 2003; Méndez, 2011). Civera, Turi, Parisi, and Fazio (1995, cited by Grigorakis et al., 2003) also observed a minor increase of TVB-N in three Sparidae species during refrigerated storage. According to several authors, the rise in this parameter is related to the activity of spoilage bacteria and endogenous enzymes, storage conditions, hygienic practices, etc. (Fernández-Segovia, Escriche, Fuentes, & Serra, 2007; Kykkidou, Giatrakou, Papavergou, Kontominas, & Savvaidis, 2009; Özogul, Özyurt, Kuley, & Özkutuk, 2009). The drip loss values exhibited a slight increase during storage due to the protein denaturation that took place in the fish muscle, causing softening of tissues and loss of water-holding capacity. 3.1.2. ATP-related compounds and K1-value Fig. 3 depicts the changes of inosine 50 -monophosphate (IMP), inosine (Ino) and hypoxanthine (Hx) during a cold storage of 11 days. The evolution of major adenine nucleotides and their related compounds can provide information about the biochemical changes occurring during storage, and may well be correlated with sea bream freshness. In this study, ATP, ADP and AMP were not analysed because the main changes during storage occurred in IMP, Ino and Hx (Alasalvar et al., 2001). Several authors have reported that the conversion of ATP into IMP is usually completed within 1 day and is presumed to be totally autolytic (Hiltz, Dyer, Nowlan & Dingle, 1972; Jones, 1965; cited by Alasalvar et al., 2001). The IMP levels at the beginning of the study (days 0 and 2) were high (over 10 lmol/g). The concentration of this metabolite progressively decreased throughout storage, with levels over 5 lmol/g at the end of the study. The initial Ino contents were low, but increased during the storage period to reach values of 3.6 lmol/g. However, the Hx values remained at low levels during the 11 days of this study, implying that the breakdown of Ino into
a
1377
Hx was very slow with Ino being the major IMP metabolite at the end of storage. Similar slow Hx formation rates have been observed for several fish species (Ryder, Buisson, Scott, & Fletcher, 1984; Scott, Fletcher, Charles, & Wong, 1992). The ATP catabolism to IMP has been reported to be caused essentially by endogenous enzymes, as mentioned above. Nevertheless, the hydrolysis of inosine and Hx formation may also result from bacterial enzymes (Dalgaard, 2000; Surette, Gill, & LeBlanc, 1988). In addition, several authors have found that loss of sensory quality correlates with increases in the Hx level during storage (Giménez, Roncalés, & Beltrán, 2000; Greene & Bernatt-Byrne, 1990). Although Hx has been reported to be a good indicator of freshness during the storage period, we found it did not correlate with cultured sea bream freshness, as other studies have done with this fish species (Alasalvar et al., 2001; Grigorakis et al., 2003; Lougovois, Kyranas, & Kyrana, 2003). This could be due to the fact that Hx formation has been reported to vary considerably within a given species (Huss, 1988) and an individual fish as its formation may be greater in red muscle than in white muscle (Murata & Sakaguchi, 1986; Watabe, Kamal & Hashimoto, 1991; cited by Alasalvar et al., 2001). This general pattern of ATP degradation with low Hx formation, IMP preservation and low Ino levels, as well as similar concentrations of these compounds, has also been found in previous studies into the same fish species (Alasalvar et al., 2001; Grigorakis et al., 2003; Lougovois et al., 2003). Due to variations in the original amount of ATP present in the muscles of different fish, it is possible that the determination of the individual components (IMP, Ino, or Hx) may not provide accurate fish freshness information. The K1-value, which measures the extension of IMP degradation, reduces this variability. This parameter has been demonstrated to be a good freshness index for a large number of fish species (Ehira & Uchiyama, 1987), and has been widely used in studies into fish quality and freshness (Barat et al., 2008; Fernández-Segovia, Escriche, & Serra, 2008; Mørkøre et al., 2010; Shahidi, Chong, & Dunajski, 1994). The K1-value increased during storage from values of 8% (day 0) up to 43% (day 11) (Fig. 3). According to Lougovois et al. (2003), very fresh sea
b
c
Fig. 3. (a) Inosine 50 -monophosphate (IMP), inosine (Ino) and hypoxanthine (Hx), (b) K1-value and (c) mesophilic bacteria and Enterobacteriaceae counts of sea bream (day 0) and samples in cold storage for 2, 4, 7, 9, and 11 days (means and standard deviations, n = 3). Upper areas of the horizontal lines in (c) are unacceptable (solid line for mesophilic and dashed line for Enterobacteriaceae).
1378
P. Zaragozá et al. / Food Chemistry 138 (2013) 1374–1380
bream will have a K1-value lower than 10%, while fish at the end of its shelf life will have a K1-value of 33–35%. These findings agree with our results, although in this study, the K1-value dropped slightly when fish reached the microbial spoilage limits. 3.2. Microbial analyses The evolution of mesophilic bacteria and Enterobacteriaceae counts of the different samples are offered in Fig. 3. The low levels of both microorganisms at the beginning of storage indicate the raw material’s excellent quality. Moreover, the mesophilic counts gradually increased throughout storage. In this work, 106 cfu g1 of mesophilic bacteria were used as a limit to evaluate microbial spoilage, which is based on other studies (Arashisar, Hisar, Kaya, & Yanik, 2004; Pascual & Calderón, 2000; Özogul, Polat, & Özogul, 2004). Samples reached this acceptability limit at day 7 of the study. No Enterobacteriaceae were found on days 0, 2 and 4 of storage. However, by taking into account the limit of 3 log cfu g1 of Enterobacteriaceae established in other studies (Fuentes, Fernández-Segovia, Barat, & Serra, 2011), high levels of this microorganism (2.8 log cfu g1) were found on day 7. From these data, it can be concluded that the sea bream used in this study had a shelf life of 7 days. 3.3. Chromogenic array Difference maps were obtained by the difference between the red, green and blue (RGB) of each sensing material from day n and the initial values measured on day 0. To facilitate visualisation only, the colour palette of the difference map was enhanced by expanding the colour range from 0–100 to 0–255; any RGB change of <4 was treated as background noise and was ignored, while changes of >100 were mapped to 255. The difference vector is conveniently visualised in Fig. 4 as a map of the absolute values of colour changes. For clarity purposes, the materials distribution in the array has been incorporated into the day 0 array. At a glance, Fig. 4 clearly shows the presence of characteristic colour fingerprints for each day, thus confirming the possibility of using this array to monitor fish spoilage. We emphasise two tendencies in colour differences: almost
1
2
3
all the dyes changed colour and enhanced colour change strength during storage. The former suggests that all the dyes were affected by atmospheric changes during the storage time, whereas the latter indicates that the enhanced colour modulation is a suitable approach to differentiate among the different sampling days. The colour data were analysed using a Principal Component Analysis (PCA). The main feature of a PCA is the coordination of the data in the new base (scores plot) and the contribution to each component of the sensors (loads plot). RGB and CIELAB colour data were used to perform the PCA analysis. Principal component (PC) scores were then used in the discrimination analysis to assign each sample to a particular group. The original variables, in this case, were the responses of the eight sensing materials used in the optoelectronic array. A PCA was used here as a simple method to project data to a two-dimensional plane. The mean centering pre-processing technique was applied to a dataset of 36 measurements (i.e., 18 initial samples and control samples plus six sampling days 3 replicates) and 48 colour features (i.e., eight R, G and B coordinates and eight L⁄, a⁄ and b⁄ coordinates). A PCA study of the full set of patterns revealed a high degree of dispersion among the independent dimensions created by the linear combinations of the colour features of the eight dyes used in these arrays. The first five PCs explained 77.74% of variance, whereas twelve PCs were needed to account for 95% of variance. This large number of independent dimensions is in agreement with the use of different responsive compounds that employ several forms of intermolecular interactions between dyes and the volatile compounds generated during fish decay. This increasing dimensionality also helps discriminate among highly related samples (e.g., different days) of complex matrixes (fish spoilage). The PCA results are shown in Fig. 5a, which captured 58.94% of the variance observed in the experiment in the first two PCs, representing the largest fraction of overall variability in the samples. Different groups can be observed. The data of days 0 and 2, as well as the controls (colorimetric array without fish), were mixed in a unique group, whereas the other days (i.e., 4, 7, 9 and 11) clustered well in four separate groups, with the distance from the first days being longer as storage time increased.
4
a
b
5
6
7
8
c
d
e
f
Fig. 4. Colour change profiles after exposure to the atmosphere generated throughout storage. Numbers indicate the code of the dye used (codes given in Table 1). Letters a, b, c, d, e, and f correspond to 0, 2, 4, 7, 9, and 11 days of storage, respectively.
P. Zaragozá et al. / Food Chemistry 138 (2013) 1374–1380
a
1379
b
Fig. 5. Principal component analysis (PCA) score plot of sea bream in cold storage for different times (0, 2, 4, 7, 9, and 11 days). (a) The PCA performed using RGB and the L⁄, a⁄ and b⁄ coordinates of the colorimetric array (b) PCA performed with the data from the physico-chemical and microbial analyses.
a
b
Fig. 6. Experimental values versus the values predicted by the PLS statistical model (dashed lines). The solid line represents ideal behaviour. (a) Mesophilic bacteria, (b) Enterobacteriaceae.
A PCA analysis was also performed with the data from the physico-chemical and microbial analyses (Fig. 5b). The mean centering pre-processing technique was applied to a dataset of 18 measurements (i.e., six sampling days 3 replicates) and 10 parameters (moisture content, pH, TVB-N, drip loss values, mesophilic bacteria Enterobacteriaceae counts, ATP-related compounds and K1-value). The first two factors explained 81.21% of total variance. In general, the groups identified were the same as those from the colorimetric arrays data. Since the PCA analyses of the colorimetric measurements and physico-chemical and microbial values classified the samples into similar groups, the PLS statistical tool was used to predict physicochemical and microbial values from the optoelectronic array data. Statistical models were tested for all the physico-chemical and microbial parameters. From these studies, a good remarkable correlation was observed for mesophilic bacteria and Enterobacteriaceae. The experimental values versus the values predicted by the PLS statistical models for both microorganisms are shown in Fig. 6. A preliminary evaluation of the accuracy of the created prediction model can be made by visually inspecting the difference between the measured and predicted values. However, a more rigorous analysis was achieved by linearly fitting the experimental points. Thus by using a simple linear model (i.e., y = p1⁄x + p2), a fitting line and also fitting parameters (p1, p2 and RMSEC) were obtained. Parameters p1 (slope of the fitting line) and p2 (intercept
with the y axis) relates to accuracy in the prediction, whereas the root mean square error of calibration (RMSEC) relates to the PLS model’s precision. Ideally, the predicted values should lie along the diagonal line, indicating that the predicted and actual values are the same. Thus, the PLS model is much better as the p1 value approaches 1 and the RMSEC value approaches 0. As stated above, the most robust models were obtained for mesophilic bacteria and Enterobacteriaceae. In both cases, values of p1 close to 1 were found, indicating a high degree of accuracy in the prediction (p1 values of 0.9177 and 0.9510 for mesophilic bacteria and Enterobacteriaceae, respectively). Moreover, values of 0.3952, 0.0799 and 0.6692, 0.3694 were found for p2 and RMSEC for mesophilic bacteria and Enterobacteriaceae, respectively. Bearing in mind that the microbial count is one of the most important parameters to establish fish spoilage, the PLS studies results obtained confirm the potential usefulness of the colorimetric array in assessing the shelf life of sea bream in cold storage.
4. Conclusions An optoelectronic nose for monitoring sea bream spoilage has been developed. This was based on eight chromogenic sensing materials and was developed by the incorporation of five dyes into two inorganic supports (alumina and silica gel). Changes in colour
1380
P. Zaragozá et al. / Food Chemistry 138 (2013) 1374–1380
of the array and changes in the target physico-chemical parameters (i.e., pH, total volatile basic nitrogen (TVB-N), drip loss, ATP-related compounds and the K1 value and microbial (mesophilic bacteria and Enterobacteriaceae) analyses) were correlated with fish ageing. These parameters and colour coordinates (RGB and L⁄, a⁄ and b⁄ coordinates) in the optoelectronic array were measured on days 0, 2, 4, 7, 9 and 11, and were used for the PCA analysis. The data of days 0, 2, and of the controls were mixed in a unique group, whereas the other days (i.e., 4, 7, 9 and 11) clustered well in four separate sets. Interestingly, PLS statistical analyses allowed the creation of models to correlate the chromogenic data with mesophilic and Enterobacteriaceae concentrations. By considering the relation between fish decay and the counts of such bacteria, the results of this study indicate the possibility of developing simple-to-use optoelectronic noses for fish spoilage detection. Acknowledgements The financial support from the Spanish Government (Project MAT2009-14564-C04), the Generalitat Valenciana (Project PROMETEO/2009/016) and Universitat Politècnica de València (Project Ref. 2711) is gratefully acknowledged. References Alasalvar, C., Taylor, K. D. A., Öksüz, A., Garthwaite, T., Alexis, M. N., & Grigorakis, K. (2001). Freshness assessment of cultured sea bream (Sparus aurata) by chemical, physical and sensory methods. Food Chemistry, 72, 33–40. Alimelli, A., Pennazza, G., Santonico, M., Paolesse, R., Filippini, D., D’Amico, A., et al. (2007). Fish freshness detection by a computer screen photoassisted based gas sensor array. Analytica Chimica Acta, 582(2), 320–328. AOAC (1997). Official methods of analysis (16th ed.). Washington: Association of Official Analytical Chemists. Arashisar, S., Hisar, O., Kaya, M., & Yanik, T. (2004). Effects of modified atmosphere and vacuum packaging on microbiological and chemical properties of rainbow trout (Oncorynchus mykiss) fillets. International Journal of Food Microbiology, 97, 209–214. Barat, J. M., Gil, L., Breijo, E., Aristoy, M. C., Toldrá, F., Martínez-Máñez, R., et al. (2008). Freshness monitoring of sea bream (Sparus aurata) with a potentiometric sensor. Food Chemistry, 108, 681–688. Burns, B. G., & Kee, P. J. (1985). Liquid chromatographic determination of hypoxanthine content in fish tissue. Journal of the Association of Official Analytical Chemists, 68, 444–448. Butler, P. (2004). Smart packaging. Cambridge, UK: IDTechEx Ltd. Dalgaard, P. (2000). Freshness, quality and safety in seafoods. Flair-flow Europe technical manual F-FE 380A/00. 31 January, 2012. Di Natale, C. D., Olafsdottir, G., Einarsson, S., Martinelli, E., Paolesse, R., & D’Amico, A. (2001). Comparison and integration of different electronic noses for freshness evaluation of cod-fish fillets. Sensors and Actuators B: Chemical, 77(1–2), 572–578. Ehira, S., & Uchiyama, H. (1987). Determination of fish freshness using the K-value and comments on some other biochemical changes in relation to freshness. In D. E. Kramer & J. Liston (Eds.), Seafood quality determination (pp. 185–207). Amsterdam: Elsevier Science Publishers. Esteban, J., Ros-Lis, J. V., Martínez-Máñez, R., Marcos, M. D., Moragues, M., Soto, J., et al. (2010). Sensitive and selective chromogenic sensing of carbon monoxide by using binuclear rhodium complexes. Angewandte Chemie International Edition, 49(29), 4934–4937. Fernández-Segovia, I., Escriche, I., Fuentes, A., & Serra, J. A. (2007). Microbial and sensory changes during refrigerated storage of desalted cod (Gadus morhua) preserved by combined methods. International Journal of Food Microbiology, 116(1), 64–72. Fernández-Segovia, I., Escriche, I., & Serra, J. A. (2008). Evolution of volatile fraction and ATP related compounds during storage of desalted cod (Gadus morhua). Food Science and Technology International, 14(5), 37–47. Fuentes, A., Fernández-Segovia, I., Barat, J. M., & Serra, J. A. (2011). Influence of sodium replacement and packaging on quality and shelf life of smoked sea bass (Dicentrarchus labrax L.). LWT – Food Science and Technology, 44(4), 917–923. Gil, L., Barat, J. M., Escriche, I., García-Breijo, E., Martínez-Máñez, R., & Soto, J. (2008). An electronic tongue for fish freshness analysis using a thick-film array of electrodes. Microchimica Acta, 163, 121–129. Giménez, B., Roncalés, P., & Beltrán, J. A. (2000). Different packaging conditions to improve shelf-life of filleted rainbow trout. In A. Gudjónsson & O. Niclasen (Eds.), Proceedings of 30th WEFTA plenary meeting Tórshavn (pp. 43–50). Faeroe Islands: Froya Fródskaparfelag.
Goulas, A. E., & Kontominas, M. G. (2007). Combined effect of light salting, modified atmosphere packaging and oregano essential oil on the shelf-life of sea bream (Sparus aurata): Biochemical and sensory attributes. Food Chemistry, 100(1), 287–296. Greene, D. H., & Bernatt-Byrne, E. I. (1990). Adenosine triphosphate catabolites as flavor compounds and freshness indicators in Pacific Cod (Gadus macrocephalus) and Pollock (Theragra chalcogramma). Journal of Food Science, 55(1), 257–258. Grigorakis, K. (2007). Compositional and organoleptic quality of farmed and wild gilthead sea bream (Sparus aurata) and sea bass (Dicentrarchus labrax) and factors affecting it: A review. Aquaculture, 272(1–4), 55–75. Grigorakis, K., Taylor, K. D. A., & Alexis, M. N. (2003). Seasonal patterns of spoilage of ice-stored cultured gilthead sea bream (Sparus aurata). Food Chemistry, 81(2), 263–268. Huang, X., Xin, J., & Zhao, J. (2011). A novel technique for rapid evaluation of fish freshness using colorimetric sensor array. Journal of Food Engineering, 105(4), 632–637. Huss, H. H. (1988). Fresh fish: Quality and quality changes. Rome: Food and Agriculture Organization of the United Nations. ISO Standard (2003). 4833: Microbiology of food and animal feeding stuffs – Horizontal method for the enumeration of microorganisms – Colony-count technique at 30 °C. Geneva, Switzerland: International Organization for Standardization. Johnson, E., & Stevenson, R. (1978). Quantitative analysis in basic liquid chromatography. Palo Alto, CA: Varian Associates, pp. 223. Kuswandi, B., Jayus Restyana, A., Abdullah, A., Heng, L. Y., & Ahmad, M. (2012). A novel colorimetric food package label for fish spoilage based on polyaniline film. Food Control, 25(1), 184–189. Kykkidou, S., Giatrakou, V., Papavergou, A., Kontominas, M. G., & Savvaidis, I. N. (2009). Effect of thyme essential oil and packaging treatments on fresh Mediterranean swordfish fillets during storage at 4 °C. Food Chemistry, 115(1), 169–175. Loughran, M., & Diamond, D. (2000). Monitoring of volatile bases in fish sample headspace using an acidochromic dye. Food Chemistry, 69(1), 97–103. Lougovois, V. P., Kyranas, E. R., & Kyrana, V. R. (2003). Comparison of selected methods of assessing freshness quality and remaining storage life of iced gilthead sea bream (Sparus aurata). Food Research International, 36(6), 551–560. Malle, P., & Tao, S. H. (1987). Rapid quantitative determination of trimethylamine using steam distillation. Journal of Food Protection, 50(9), 756–760. Méndez, E. L. (2011). Aplicación de la espectroscopía de impedancia en la evaluación de la calidad de dorada. Master Thesis. Universitat Politècnica de València. Moragues, M. E., Esteban, J., Ros-Lis, J., Martínez-Máñez, R., Marcos, M. D., Martínez, M., et al. (2011). Sensitive and selective chromogenic sensing of carbon monoxide via reversible axial CO coordination in binuclear rhodium complexes. Journal of the American Chemical Society, 133(39), 15762–15772. Mørkøre, T., Rødbotten, M., Vogt, G., Fjæra, S. O., Kristiansen, I. Ø., & Manseth, E. (2010). Relevance of season and nucleotide catabolism on changes in fillet quality during chilled storage of raw Atlantic salmon (Salmo salar L.). Food Chemistry, 119(4), 1417–1425. Ólafsdóttir, G., Nesvadba, P., Di Natale, C., Careche, M., Oehlenschläger, J., Tryggvadóttir, S. V., et al. (2004). Multisensor for fish quality determination. Trends in Food Science and Technology, 15(2), 86–93. Özogul, K., Özyurt, G., Kuley, E., & Özkutuk, S. (2009). Sensory, microbiological and chemical assessment of the freshness of red mullet (Mullus barbatus) and goldband goatfish (Upeneus moluccensis) during storage in ice. Food Chemistry, 114, 505–510. Özogul, F., Polat, A., & Özogul, Y. (2004). The effects of modified atmosphere packaging and vacuum packaging on chemical, sensory and microbiological changes of sardines (Sardina pilchardus). Food Chemistry, 85(1), 49–57. Pacquit, A., Lau, K. T., McLaughlin, H., Frisby, J., Quilty, B., & Diamond, D. (2006). Development of a volatile amine sensor for the monitoring of fish spoilage. Talanta, 69(2), 515–520. Pascual, M., & Calderón, V. (2000). Recuento de microorganismos aerobios mesófilos revivificables. In Microbiología Alimentaria (pp. 13–15). Madrid: Díaz de Santos. Pons-Sánchez-Cascado, S., Vidal-Carou, M. C., Nunes, M. L., & Veciana-Nogués, M. T. (2006). Sensory analysis to assess the freshness of Mediterranean anchovies (Engraulis encrasicholus) stored in ice. Food Control, 17(7), 564–569. Ryder, J. M., Buisson, D. H., Scott, D. N., & Fletcher, G. C. (1984). Storage of New Zealand jack mackerel (Trachurus novaezelandiae) in ice: Chemical, microbiological and sensory assessment. Journal of Food Science, 49(6), 1453–1456. Salinas, Y., Ros-Lis, J. V., Vivancos, J.-L., Martínez-Mañez, R., Marcos, M. D., Aucejo, S., et al. (2012). Monitoring of chicken meat freshness by means of a colorimetric sensor array. Analyst, 137, 3635–3643. Scott, D. N., Fletcher, G. C., Charles, J. C., & Wong, R. J. (1992). Spoilage changes in deep water fish smooth oreo dory during storage in ice. International Journal of Food Science and Technology, 27, 577–587. Shahidi, F., Chong, X., & Dunajski, E. (1994). Freshness quality of harp seal (Phoca groenlandica) meat. Journal of Agriculture and Food Chemistry, 42, 868–872. Surette, M. E., Gill, T. A., & LeBlanc, P. J. (1988). Biochemical basis of postmortem nucleotide catabolism in cod (Gadus morhua) and its relationship to spoilage. Journal of Agricultural and Food Chemistry, 36(1), 19–22. Yam, K. L., Takhistov, P. T., & Miltz, J. (2005). Intelligent packaging: Concepts and applications. Journal of Food Science, 70(1), R1–R10.
Contents lists available at SciVerse ScienceDirect
Food Chemistry journal homepage: www.elsevier.com/locate/foodchem
Evaluation of sea bream (Sparus aurata) shelf life using an optoelectronic nose Patricia Zaragozá a, Ana Fuentes b, Isabel Fernández-Segovia b, José-Luis Vivancos a,c,⇑, Arantxa Rizo b, José V. Ros-Lis a, José M. Barat b, Ramón Martínez-Máñez a,c a
Centro de Reconocimiento Molecular y Desarrollo Tecnológico, Unidad Mixta Universitat Politècnica de València – Universitat de València, Camino de Vera s/n, 46022 Valencia, Spain Departamento de Tecnología de Alimentos, Universitat Politècnica de València, Camino de Vera s/n, 46022 Valencia, Spain c CIBER de Bioingeniería, Biomateriales y Nanomedicina (CIBER-BBN), Spain b
a r t i c l e
i n f o
Article history: Received 9 July 2012 Received in revised form 20 September 2012 Accepted 22 October 2012 Available online 12 November 2012 Keywords: Sea bream Colorimetric array Shelf life Chromogenic sensors
a b s t r a c t A new optoelectronic nose for the shelf-life assessment of fresh sea bream in cold storage has been developed. The chromogenic array used eight sensing materials (based on aluminium oxide and silica gel) containing pH indicators, Lewis acids and an oxidation–reduction indicator. The colour changes of the sensor array were characteristic of sea bream spoilage. Colour modulations were measured on day 0 and for the samples held in cold storage for 2, 4, 7, 9 and 11 days. Determination of moisture content, pH, total volatile basic nitrogen (TVB-N), drip loss, ATP-related compounds and K1-value and microbial (mesophilic bacteria and Enterobacteriaceae) analyses were carried out on the same days. The changes in the chromogenic arrays data were processed by statistical analysis (PCA). Moreover, PLS statistical studies allowed the creation of models to correlate the chromogenic data with concentrations of mesophilic and Enterobacteriaceae. The results suggest the feasibility of this system to help develop optoelectronic noses for fish freshness monitoring. Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction Fish is a highly perishable food product whose freshness and quality rapidly decline post-mortem. Loss of freshness and spoilage of fish are complicated processes, and various factors such as species and different storage conditions influence the spoilage pattern. In fact, the development of reliable methods to assess fish freshness and to evaluate quality criteria has been the goal of fish research for many years (Di Natale et al., 2001; Gil et al., 2008). The current methods used in the food industry to determine fish shelf life are based on physical, chemical, microbiological and sensory measurements (Ólafsdóttir et al., 2004). However, some of these methods are tedious, expensive, time-consuming and require skilled personnel. As an alternative, the development of rapid nondestructive quality control techniques, which can be applied at any stage of the supply chain, is of much importance (Pons-SánchezCascado, Vidal-Carou, Nunes, & Veciana-Nogués, 2006). A possible solution to this issue is to design smart or intelligent packaging capable of providing information about fish quality. Intelligent packaging is an emerging technology that uses the communication function of the package to facilitate decision making in order to
⇑ Corresponding author at: Centro de Reconocimiento Molecular y Desarrollo Tecnológico, Unidad Mixta Universitat Politècnica de València – Universitat de València, Camino de Vera s/n, 46022 Valencia, Spain. Tel.: +34 96 387 7000; fax: +34 96 387 9869. E-mail address: [email protected] (J.-L. Vivancos). 0308-8146/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.foodchem.2012.10.114
provide the benefits of enhanced food quality and safety. Smart packaging is defined as a system that monitors the condition of packaged food to provide information about the quality during storage, transportation and distribution (Yam, Takhistov, & Miltz, 2005). The potential of smart packaging technology is far-reaching, ranging from food safety and drug use monitoring to postal delivery tracking and embedded security tags (Butler, 2004). In the context of this work, smart packaging for fish monitoring would involve a fish freshness ‘‘indicator’’, which uses metabolites as ‘‘information’’ to monitor the status of fish, which can be directly related to fish spoilage (Kuswandi et al., 2012). However, studies on the application of this type of technology in fish are very limited. From another viewpoint, the use of chromogenic chemosensors is a promising technique for the simple design of freshness and quality indicators. Colorimetric probes are usually cheap, versatile, can be printed on the package and colour changes can be easily measured using simple image capturing systems (e.g. cameras) or seen by the naked eye through transparent films (Huang, Xin, & Zhao, 2011). In fact, there are very few technologies which are as inexpensive as visual imaging. Although some chromogenic indicators have been described for food spoilage detection, they are generally based on a single compound and have their limitations such as lack of specificity (offering false positives or false negatives). In particular, the possibility of false negatives is likely to dissuade producers from adopting indicators unless specific indications of actual spoilage can be guaranteed. Thus, although
P. Zaragozá et al. / Food Chemistry 138 (2013) 1374–1380
1375
some attempts have been made in single analyte indicators, one of the most promising approaches is the use of optoelectronic noses, built by an array of dyes able to offer spoilage information through suitable colour changes (Loughran & Diamond, 2000; Pacquit et al., 2006). Nonetheless, despite these encouraging examples, chromogenic arrays for food quality analyses are still scarce (Alimelli et al., 2007; Huang et al., 2011; Kuswandi et al., 2012). Based on the above detailed information, we aimed to develop a rapid, easy-to-use optoelectronic system for shelf-life assessments of fish samples. As fish, we selected fresh sea bream, which was kept in cold storage. Fish spoilage was monitored simultaneously by using an array of chromogenic dyes, and through physicochemical and microbial analyses. 2. Materials and methods 2.1. Chromogenic array preparation The five dyes used in the study were bromocresol purple, Resorufin, bromophenol blue, phenol red (acquired from Sigma Aldrich (St. Louis, MO, USA)), and a dinuclear complex of rhodium (synthesised according to known procedures (Esteban et al., 2010; Moragues et al., 2011)). Selection of these dyes was based on previously reported optoelectronic noses and on our own experience in designing colorimetric probes (Salinas et al., 2012). These arrays include pH indicators, (phenol red, bromocresol purple, bromophenol blue), Lewis acids (dinuclear rhodium complex), and oxidation–reduction indicators (Resorufin). The chemical structures of the five dyes are shown in Fig. 1. The supports (aluminium oxide and silica gel) and the analytical-grade solvents (dichloromethane and ethanol) were purchased from Scharlau Chemie, S.A. (Barcelona, Spain). The final sensing materials were readily obtained by preparing suspensions of the inorganic support (aluminium oxide or silica gel) with a certain amount of the corresponding dye in an appropriate solvent (dichloromethane or ethanol). These suspensions were stirred for 8 h to confer maximum dye absorption in the inorganic support. Afterwards, the solvent was evaporated in a rotary evaporator to obtain the corresponding sensing material. A total of eight chromogenic materials were prepared using this simple approach. Table 1 shows in detail the amounts of silica, dye and solvent used in their preparation. The array was prepared by placing approximately 40 mg of each material into a microplate to obtain a 2 4 array (8 sensing materials). 2.2. Sample preparation Sea bream from a Spanish farm (Frescamar Alimentación, S.L., Burriana, Castellón, Spain) of commercial size 400–600 g were used as the raw material. Fish were transported to the laboratory in polyspan boxes with ice. Eighteen fish were weighed and fifteen were individually placed in aluminium trays of 2.7 L. The eight different sensing materials described in Table 1 were placed in a microplate inside the tray close to the fish samples (Fig. 2). In addition, three trays containing the colorimetric array were also packaged in the absence of fish and used as controls. Petri dishes filled with water were placed in the control trays to reproduce moisture conditions. The studies with the samples were repeated three times. All the trays were wrapped with plastic film and stored at 4 °C for 11 days. The physico-chemical and microbiological analyses of the samples were performed on days 0, 2, 4, 7, 9 and 11 of storage. Before the analyses, samples were headed, gutted and filleted, and two fillets were obtained from each fish. Photographs of the colorimetric arrays were obtained on each analysis day in all the trays stored.
Fig. 1. Chemical structures of the dyes used.
2.3. Physico-chemical analyses Moisture content, pH, total volatile basic nitrogen (TVB-N), drip loss, ATP-related compounds and K1-value were determined in all the samples in triplicate (n = 3). 2.3.1. Moisture, pH, TVB-N and drip loss The moisture content was determined according to the AOAC method 950.46 (1997). pH measurements were taken out using a digital pH-meter micropH 2001 (Crison Instruments, S.A., Barcelona, Spain) with a puncture electrode (Crison 5231) at six different sample locations. The TVB-N contents were determined by steam distillation according to the method described by Malle and Tao (1987). To evaluate the drip loss, samples were weighed before being placed in the tray and stored, as mentioned above. After storage, samples were blotted with filter paper and weighed again. The drip loss was expressed as the percentage of exudate. 2.3.2. ATP related compounds and K1-value The ATP-related compounds, consisting of inosine-50 -monophosphate (IMP), inosine (Ino), and hypoxanthine (Hx), were assayed by HPLC by following the method described by Barat et al. (2008), with some minor modifications. The extraction method used was similar to that described by Burns and Kee (1985). The analysis was conducted on a Hitachi LaChrom Elite liquid chromatography (Hitachi Ltd., Tokyo, Japan) with a pump (model L-2130), an auto-sampler (model L-2200) and a UV detector (model L-2400). Data acquisition was performed with the EZChrom Elite software (Agillent Technologies, Palo Alto, CA, USA). Separations were made on a reverse-phase Ultrabase C18 250 4.6 mm, with an internal particle diameter of 5 lm (Análisis Vínicos, S.L., Tomelloso, Spain). A guard column containing the same C18 packing as above was positioned in front of the analytical column to protect
1376
P. Zaragozá et al. / Food Chemistry 138 (2013) 1374–1380
Table 1 Details for the preparation of the sensing materials. Code
Dye
Support
mg dye (g solid)1
mmol dye (g solid)1
Solvent
Solvent amount (ml)
1 2 3 4 5 6 7 8
Resorufin Bromocresol purple Bromocresol purple Bromophenol blue Phenol red Dinuclear rhodium complex Dinuclear rhodium complex Dinuclear rhodium complex
Aluminium oxide Silica gel Aluminium oxide Silica gel Silica gel Silica gel Silica gel Silica gel
40 40 4 40 40 40 80 140
0.170 0.074 0.007 0.059 0.113 0.042 0.083 0.145
Ethanol Ethanol Ethanol Ethanol Dichloromethane Dichloromethane Dichloromethane Dichloromethane
8 8 8 8 5 6 6 6
2.6. Statistical analysis
Fig. 2. Scheme of a tray with a sea bream sample and the chromogenic array.
it from contamination. Compounds were identified by using a retention time comparison of the unknown with those of the standards, and by a standard addition or ‘‘spiking’’ (Johnson & Stevenson, 1978). IMP, Ino, and Hx were quantified according to the external standard method using calibration curves of the peak area of compound versus the compound concentration under identical chromatographic conditions. K1-values were calculated according to Eq. (1):
K 1 ð%Þ ¼
½Ino þ ½Hx 100 ½IMP þ ½Ino þ ½Hx
ð1Þ
where IMP is inosine 50 -monophosphate; Ino, inosine; Hx, hypoxanthine. All the chemical reagents were provided by Sigma–Aldrich.
2.4. Microbial analyses Mesophilic counts were performed according to the method provided in the ISO 4833:2003 standard (ISO Standard, 2003). Enterobacteriaceae were enumerated according to the method described by Pascual and Calderón (2000). All the culture media were provided by Scharlau. All the analyses were performed in triplicate and the results were expressed as log cfu g1.
2.5. Photographs and colour measurements The colour measurement was performed on each component by taking photographs with a Nikon camera D3000. The colour RGB and CIELAB coordinates of the array were collected on each analysis day by taking a photograph and by using the Photoshop Pro 5 software.
Data are reported as mean ± standard deviation. Data were statistically processed using Statgraphics Centurion (Statpoint Technologies, Inc., Warrenton, VA, USA). An analysis of variance (One-Way ANOVA), in which storage time was the factor, was conducted for each parameter evaluated to test whether there were significant differences throughout cold storage. The LSD procedure (least significant difference) was used to test for differences between averages at the 5% significance level. In order to assess the feasibility of the colorimetric array to determine the fish shelf life, a Principal Components Analysis (PCA) and a Partial Least Square (PLS) were conducted. All the analyses were performed using the MATLABÒ PLS Tool-box (Eigenvector Research Inc.). A PCA was also performed with the data from the physico-chemical and microbiological analyses.
3. Results and discussion 3.1. Physico-chemical analyses 3.1.1. Moisture, pH, TVB-N and drip loss The moisture content, pH, TVB-N and drip loss values of fresh sea breams (day 0) and of the samples in cold storage for 2, 4, 7, 9 and 11 days are shown in Table 2. The moisture values ranged from 71.03 to 75.57 g/100 g, which is in accordance with different studies carried out on cultured sea bream (Alasalvar et al., 2001; Goulas & Kontominas, 2007; Grigorakis, 2007; Grigorakis, Taylor, & Alexis, 2003). The pH values remained stable up to day 4 and slightly increase thereafter. The higher values observed at the end of storage could be related to microbial development since the metabolism of microorganisms produces basic compounds. The pH changes pat-
Table 2 Moisture, pH, TVB-N and drip loss of fresh sea bream (day 0) and samples in cold storage for 2, 4, 7, 9, and 11 days (means and standard deviations, n = 3). Storage time (days)
Moisture (g H2O/100 g)
pH
TVB-N (mg N/100 g)
Drip loss (% exudates)
0 2 4 7 9 11
71.61 ± 2.30a 71.03 ± 1.45a 72.96 ± 0.93ab 71.96 ± 0.77ab 75.57 ± 0.53c 74.28 ± 1.45bc ⁄
6.29 ± 0.02a 6.29 ± 0.04a 6.30 ± 0.05a 6.34 ± 0.08a 6.47 ± 0.03b 6.37 ± 0.04a ⁄⁄
18.88 ± 0.74a 18.23 ± 1.24a 18.84 ± 0.72a 17.13 ± 0.11a 18.20 ± 0.37a 21.33 ± 2.54b ⁄
– 0.34 ± 0.18ab 0.45 ± 0.35ab 0.99 ± 0.41c 0.66 ± 0.32bc 1.73 ± 0.11d ⁄⁄⁄
Same letters in the same column indicate homogeneous group membership. Significance level (a). * p < 0.05. ** p < 0.01. *** p < 0.001.
P. Zaragozá et al. / Food Chemistry 138 (2013) 1374–1380
tern was similar to that observed by Grigorakis et al. (2003) and by Barat et al. (2008). The TVB-N values slightly increased on day 11 of storage. The slight increase observed in this study was similar to the evolution of TVB-N in sea bream during ice and cold storage reported in several studies (Grigorakis et al., 2003; Méndez, 2011). Civera, Turi, Parisi, and Fazio (1995, cited by Grigorakis et al., 2003) also observed a minor increase of TVB-N in three Sparidae species during refrigerated storage. According to several authors, the rise in this parameter is related to the activity of spoilage bacteria and endogenous enzymes, storage conditions, hygienic practices, etc. (Fernández-Segovia, Escriche, Fuentes, & Serra, 2007; Kykkidou, Giatrakou, Papavergou, Kontominas, & Savvaidis, 2009; Özogul, Özyurt, Kuley, & Özkutuk, 2009). The drip loss values exhibited a slight increase during storage due to the protein denaturation that took place in the fish muscle, causing softening of tissues and loss of water-holding capacity. 3.1.2. ATP-related compounds and K1-value Fig. 3 depicts the changes of inosine 50 -monophosphate (IMP), inosine (Ino) and hypoxanthine (Hx) during a cold storage of 11 days. The evolution of major adenine nucleotides and their related compounds can provide information about the biochemical changes occurring during storage, and may well be correlated with sea bream freshness. In this study, ATP, ADP and AMP were not analysed because the main changes during storage occurred in IMP, Ino and Hx (Alasalvar et al., 2001). Several authors have reported that the conversion of ATP into IMP is usually completed within 1 day and is presumed to be totally autolytic (Hiltz, Dyer, Nowlan & Dingle, 1972; Jones, 1965; cited by Alasalvar et al., 2001). The IMP levels at the beginning of the study (days 0 and 2) were high (over 10 lmol/g). The concentration of this metabolite progressively decreased throughout storage, with levels over 5 lmol/g at the end of the study. The initial Ino contents were low, but increased during the storage period to reach values of 3.6 lmol/g. However, the Hx values remained at low levels during the 11 days of this study, implying that the breakdown of Ino into
a
1377
Hx was very slow with Ino being the major IMP metabolite at the end of storage. Similar slow Hx formation rates have been observed for several fish species (Ryder, Buisson, Scott, & Fletcher, 1984; Scott, Fletcher, Charles, & Wong, 1992). The ATP catabolism to IMP has been reported to be caused essentially by endogenous enzymes, as mentioned above. Nevertheless, the hydrolysis of inosine and Hx formation may also result from bacterial enzymes (Dalgaard, 2000; Surette, Gill, & LeBlanc, 1988). In addition, several authors have found that loss of sensory quality correlates with increases in the Hx level during storage (Giménez, Roncalés, & Beltrán, 2000; Greene & Bernatt-Byrne, 1990). Although Hx has been reported to be a good indicator of freshness during the storage period, we found it did not correlate with cultured sea bream freshness, as other studies have done with this fish species (Alasalvar et al., 2001; Grigorakis et al., 2003; Lougovois, Kyranas, & Kyrana, 2003). This could be due to the fact that Hx formation has been reported to vary considerably within a given species (Huss, 1988) and an individual fish as its formation may be greater in red muscle than in white muscle (Murata & Sakaguchi, 1986; Watabe, Kamal & Hashimoto, 1991; cited by Alasalvar et al., 2001). This general pattern of ATP degradation with low Hx formation, IMP preservation and low Ino levels, as well as similar concentrations of these compounds, has also been found in previous studies into the same fish species (Alasalvar et al., 2001; Grigorakis et al., 2003; Lougovois et al., 2003). Due to variations in the original amount of ATP present in the muscles of different fish, it is possible that the determination of the individual components (IMP, Ino, or Hx) may not provide accurate fish freshness information. The K1-value, which measures the extension of IMP degradation, reduces this variability. This parameter has been demonstrated to be a good freshness index for a large number of fish species (Ehira & Uchiyama, 1987), and has been widely used in studies into fish quality and freshness (Barat et al., 2008; Fernández-Segovia, Escriche, & Serra, 2008; Mørkøre et al., 2010; Shahidi, Chong, & Dunajski, 1994). The K1-value increased during storage from values of 8% (day 0) up to 43% (day 11) (Fig. 3). According to Lougovois et al. (2003), very fresh sea
b
c
Fig. 3. (a) Inosine 50 -monophosphate (IMP), inosine (Ino) and hypoxanthine (Hx), (b) K1-value and (c) mesophilic bacteria and Enterobacteriaceae counts of sea bream (day 0) and samples in cold storage for 2, 4, 7, 9, and 11 days (means and standard deviations, n = 3). Upper areas of the horizontal lines in (c) are unacceptable (solid line for mesophilic and dashed line for Enterobacteriaceae).
1378
P. Zaragozá et al. / Food Chemistry 138 (2013) 1374–1380
bream will have a K1-value lower than 10%, while fish at the end of its shelf life will have a K1-value of 33–35%. These findings agree with our results, although in this study, the K1-value dropped slightly when fish reached the microbial spoilage limits. 3.2. Microbial analyses The evolution of mesophilic bacteria and Enterobacteriaceae counts of the different samples are offered in Fig. 3. The low levels of both microorganisms at the beginning of storage indicate the raw material’s excellent quality. Moreover, the mesophilic counts gradually increased throughout storage. In this work, 106 cfu g1 of mesophilic bacteria were used as a limit to evaluate microbial spoilage, which is based on other studies (Arashisar, Hisar, Kaya, & Yanik, 2004; Pascual & Calderón, 2000; Özogul, Polat, & Özogul, 2004). Samples reached this acceptability limit at day 7 of the study. No Enterobacteriaceae were found on days 0, 2 and 4 of storage. However, by taking into account the limit of 3 log cfu g1 of Enterobacteriaceae established in other studies (Fuentes, Fernández-Segovia, Barat, & Serra, 2011), high levels of this microorganism (2.8 log cfu g1) were found on day 7. From these data, it can be concluded that the sea bream used in this study had a shelf life of 7 days. 3.3. Chromogenic array Difference maps were obtained by the difference between the red, green and blue (RGB) of each sensing material from day n and the initial values measured on day 0. To facilitate visualisation only, the colour palette of the difference map was enhanced by expanding the colour range from 0–100 to 0–255; any RGB change of <4 was treated as background noise and was ignored, while changes of >100 were mapped to 255. The difference vector is conveniently visualised in Fig. 4 as a map of the absolute values of colour changes. For clarity purposes, the materials distribution in the array has been incorporated into the day 0 array. At a glance, Fig. 4 clearly shows the presence of characteristic colour fingerprints for each day, thus confirming the possibility of using this array to monitor fish spoilage. We emphasise two tendencies in colour differences: almost
1
2
3
all the dyes changed colour and enhanced colour change strength during storage. The former suggests that all the dyes were affected by atmospheric changes during the storage time, whereas the latter indicates that the enhanced colour modulation is a suitable approach to differentiate among the different sampling days. The colour data were analysed using a Principal Component Analysis (PCA). The main feature of a PCA is the coordination of the data in the new base (scores plot) and the contribution to each component of the sensors (loads plot). RGB and CIELAB colour data were used to perform the PCA analysis. Principal component (PC) scores were then used in the discrimination analysis to assign each sample to a particular group. The original variables, in this case, were the responses of the eight sensing materials used in the optoelectronic array. A PCA was used here as a simple method to project data to a two-dimensional plane. The mean centering pre-processing technique was applied to a dataset of 36 measurements (i.e., 18 initial samples and control samples plus six sampling days 3 replicates) and 48 colour features (i.e., eight R, G and B coordinates and eight L⁄, a⁄ and b⁄ coordinates). A PCA study of the full set of patterns revealed a high degree of dispersion among the independent dimensions created by the linear combinations of the colour features of the eight dyes used in these arrays. The first five PCs explained 77.74% of variance, whereas twelve PCs were needed to account for 95% of variance. This large number of independent dimensions is in agreement with the use of different responsive compounds that employ several forms of intermolecular interactions between dyes and the volatile compounds generated during fish decay. This increasing dimensionality also helps discriminate among highly related samples (e.g., different days) of complex matrixes (fish spoilage). The PCA results are shown in Fig. 5a, which captured 58.94% of the variance observed in the experiment in the first two PCs, representing the largest fraction of overall variability in the samples. Different groups can be observed. The data of days 0 and 2, as well as the controls (colorimetric array without fish), were mixed in a unique group, whereas the other days (i.e., 4, 7, 9 and 11) clustered well in four separate groups, with the distance from the first days being longer as storage time increased.
4
a
b
5
6
7
8
c
d
e
f
Fig. 4. Colour change profiles after exposure to the atmosphere generated throughout storage. Numbers indicate the code of the dye used (codes given in Table 1). Letters a, b, c, d, e, and f correspond to 0, 2, 4, 7, 9, and 11 days of storage, respectively.
P. Zaragozá et al. / Food Chemistry 138 (2013) 1374–1380
a
1379
b
Fig. 5. Principal component analysis (PCA) score plot of sea bream in cold storage for different times (0, 2, 4, 7, 9, and 11 days). (a) The PCA performed using RGB and the L⁄, a⁄ and b⁄ coordinates of the colorimetric array (b) PCA performed with the data from the physico-chemical and microbial analyses.
a
b
Fig. 6. Experimental values versus the values predicted by the PLS statistical model (dashed lines). The solid line represents ideal behaviour. (a) Mesophilic bacteria, (b) Enterobacteriaceae.
A PCA analysis was also performed with the data from the physico-chemical and microbial analyses (Fig. 5b). The mean centering pre-processing technique was applied to a dataset of 18 measurements (i.e., six sampling days 3 replicates) and 10 parameters (moisture content, pH, TVB-N, drip loss values, mesophilic bacteria Enterobacteriaceae counts, ATP-related compounds and K1-value). The first two factors explained 81.21% of total variance. In general, the groups identified were the same as those from the colorimetric arrays data. Since the PCA analyses of the colorimetric measurements and physico-chemical and microbial values classified the samples into similar groups, the PLS statistical tool was used to predict physicochemical and microbial values from the optoelectronic array data. Statistical models were tested for all the physico-chemical and microbial parameters. From these studies, a good remarkable correlation was observed for mesophilic bacteria and Enterobacteriaceae. The experimental values versus the values predicted by the PLS statistical models for both microorganisms are shown in Fig. 6. A preliminary evaluation of the accuracy of the created prediction model can be made by visually inspecting the difference between the measured and predicted values. However, a more rigorous analysis was achieved by linearly fitting the experimental points. Thus by using a simple linear model (i.e., y = p1⁄x + p2), a fitting line and also fitting parameters (p1, p2 and RMSEC) were obtained. Parameters p1 (slope of the fitting line) and p2 (intercept
with the y axis) relates to accuracy in the prediction, whereas the root mean square error of calibration (RMSEC) relates to the PLS model’s precision. Ideally, the predicted values should lie along the diagonal line, indicating that the predicted and actual values are the same. Thus, the PLS model is much better as the p1 value approaches 1 and the RMSEC value approaches 0. As stated above, the most robust models were obtained for mesophilic bacteria and Enterobacteriaceae. In both cases, values of p1 close to 1 were found, indicating a high degree of accuracy in the prediction (p1 values of 0.9177 and 0.9510 for mesophilic bacteria and Enterobacteriaceae, respectively). Moreover, values of 0.3952, 0.0799 and 0.6692, 0.3694 were found for p2 and RMSEC for mesophilic bacteria and Enterobacteriaceae, respectively. Bearing in mind that the microbial count is one of the most important parameters to establish fish spoilage, the PLS studies results obtained confirm the potential usefulness of the colorimetric array in assessing the shelf life of sea bream in cold storage.
4. Conclusions An optoelectronic nose for monitoring sea bream spoilage has been developed. This was based on eight chromogenic sensing materials and was developed by the incorporation of five dyes into two inorganic supports (alumina and silica gel). Changes in colour
1380
P. Zaragozá et al. / Food Chemistry 138 (2013) 1374–1380
of the array and changes in the target physico-chemical parameters (i.e., pH, total volatile basic nitrogen (TVB-N), drip loss, ATP-related compounds and the K1 value and microbial (mesophilic bacteria and Enterobacteriaceae) analyses) were correlated with fish ageing. These parameters and colour coordinates (RGB and L⁄, a⁄ and b⁄ coordinates) in the optoelectronic array were measured on days 0, 2, 4, 7, 9 and 11, and were used for the PCA analysis. The data of days 0, 2, and of the controls were mixed in a unique group, whereas the other days (i.e., 4, 7, 9 and 11) clustered well in four separate sets. Interestingly, PLS statistical analyses allowed the creation of models to correlate the chromogenic data with mesophilic and Enterobacteriaceae concentrations. By considering the relation between fish decay and the counts of such bacteria, the results of this study indicate the possibility of developing simple-to-use optoelectronic noses for fish spoilage detection. Acknowledgements The financial support from the Spanish Government (Project MAT2009-14564-C04), the Generalitat Valenciana (Project PROMETEO/2009/016) and Universitat Politècnica de València (Project Ref. 2711) is gratefully acknowledged. References Alasalvar, C., Taylor, K. D. A., Öksüz, A., Garthwaite, T., Alexis, M. N., & Grigorakis, K. (2001). Freshness assessment of cultured sea bream (Sparus aurata) by chemical, physical and sensory methods. Food Chemistry, 72, 33–40. Alimelli, A., Pennazza, G., Santonico, M., Paolesse, R., Filippini, D., D’Amico, A., et al. (2007). Fish freshness detection by a computer screen photoassisted based gas sensor array. Analytica Chimica Acta, 582(2), 320–328. AOAC (1997). Official methods of analysis (16th ed.). Washington: Association of Official Analytical Chemists. Arashisar, S., Hisar, O., Kaya, M., & Yanik, T. (2004). Effects of modified atmosphere and vacuum packaging on microbiological and chemical properties of rainbow trout (Oncorynchus mykiss) fillets. International Journal of Food Microbiology, 97, 209–214. Barat, J. M., Gil, L., Breijo, E., Aristoy, M. C., Toldrá, F., Martínez-Máñez, R., et al. (2008). Freshness monitoring of sea bream (Sparus aurata) with a potentiometric sensor. Food Chemistry, 108, 681–688. Burns, B. G., & Kee, P. J. (1985). Liquid chromatographic determination of hypoxanthine content in fish tissue. Journal of the Association of Official Analytical Chemists, 68, 444–448. Butler, P. (2004). Smart packaging. Cambridge, UK: IDTechEx Ltd. Dalgaard, P. (2000). Freshness, quality and safety in seafoods. Flair-flow Europe technical manual F-FE 380A/00.
Goulas, A. E., & Kontominas, M. G. (2007). Combined effect of light salting, modified atmosphere packaging and oregano essential oil on the shelf-life of sea bream (Sparus aurata): Biochemical and sensory attributes. Food Chemistry, 100(1), 287–296. Greene, D. H., & Bernatt-Byrne, E. I. (1990). Adenosine triphosphate catabolites as flavor compounds and freshness indicators in Pacific Cod (Gadus macrocephalus) and Pollock (Theragra chalcogramma). Journal of Food Science, 55(1), 257–258. Grigorakis, K. (2007). Compositional and organoleptic quality of farmed and wild gilthead sea bream (Sparus aurata) and sea bass (Dicentrarchus labrax) and factors affecting it: A review. Aquaculture, 272(1–4), 55–75. Grigorakis, K., Taylor, K. D. A., & Alexis, M. N. (2003). Seasonal patterns of spoilage of ice-stored cultured gilthead sea bream (Sparus aurata). Food Chemistry, 81(2), 263–268. Huang, X., Xin, J., & Zhao, J. (2011). A novel technique for rapid evaluation of fish freshness using colorimetric sensor array. Journal of Food Engineering, 105(4), 632–637. Huss, H. H. (1988). Fresh fish: Quality and quality changes. Rome: Food and Agriculture Organization of the United Nations. ISO Standard (2003). 4833: Microbiology of food and animal feeding stuffs – Horizontal method for the enumeration of microorganisms – Colony-count technique at 30 °C. Geneva, Switzerland: International Organization for Standardization. Johnson, E., & Stevenson, R. (1978). Quantitative analysis in basic liquid chromatography. Palo Alto, CA: Varian Associates, pp. 223. Kuswandi, B., Jayus Restyana, A., Abdullah, A., Heng, L. Y., & Ahmad, M. (2012). A novel colorimetric food package label for fish spoilage based on polyaniline film. Food Control, 25(1), 184–189. Kykkidou, S., Giatrakou, V., Papavergou, A., Kontominas, M. G., & Savvaidis, I. N. (2009). Effect of thyme essential oil and packaging treatments on fresh Mediterranean swordfish fillets during storage at 4 °C. Food Chemistry, 115(1), 169–175. Loughran, M., & Diamond, D. (2000). Monitoring of volatile bases in fish sample headspace using an acidochromic dye. Food Chemistry, 69(1), 97–103. Lougovois, V. P., Kyranas, E. R., & Kyrana, V. R. (2003). Comparison of selected methods of assessing freshness quality and remaining storage life of iced gilthead sea bream (Sparus aurata). Food Research International, 36(6), 551–560. Malle, P., & Tao, S. H. (1987). Rapid quantitative determination of trimethylamine using steam distillation. Journal of Food Protection, 50(9), 756–760. Méndez, E. L. (2011). Aplicación de la espectroscopía de impedancia en la evaluación de la calidad de dorada. Master Thesis. Universitat Politècnica de València. Moragues, M. E., Esteban, J., Ros-Lis, J., Martínez-Máñez, R., Marcos, M. D., Martínez, M., et al. (2011). Sensitive and selective chromogenic sensing of carbon monoxide via reversible axial CO coordination in binuclear rhodium complexes. Journal of the American Chemical Society, 133(39), 15762–15772. Mørkøre, T., Rødbotten, M., Vogt, G., Fjæra, S. O., Kristiansen, I. Ø., & Manseth, E. (2010). Relevance of season and nucleotide catabolism on changes in fillet quality during chilled storage of raw Atlantic salmon (Salmo salar L.). Food Chemistry, 119(4), 1417–1425. Ólafsdóttir, G., Nesvadba, P., Di Natale, C., Careche, M., Oehlenschläger, J., Tryggvadóttir, S. V., et al. (2004). Multisensor for fish quality determination. Trends in Food Science and Technology, 15(2), 86–93. Özogul, K., Özyurt, G., Kuley, E., & Özkutuk, S. (2009). Sensory, microbiological and chemical assessment of the freshness of red mullet (Mullus barbatus) and goldband goatfish (Upeneus moluccensis) during storage in ice. Food Chemistry, 114, 505–510. Özogul, F., Polat, A., & Özogul, Y. (2004). The effects of modified atmosphere packaging and vacuum packaging on chemical, sensory and microbiological changes of sardines (Sardina pilchardus). Food Chemistry, 85(1), 49–57. Pacquit, A., Lau, K. T., McLaughlin, H., Frisby, J., Quilty, B., & Diamond, D. (2006). Development of a volatile amine sensor for the monitoring of fish spoilage. Talanta, 69(2), 515–520. Pascual, M., & Calderón, V. (2000). Recuento de microorganismos aerobios mesófilos revivificables. In Microbiología Alimentaria (pp. 13–15). Madrid: Díaz de Santos. Pons-Sánchez-Cascado, S., Vidal-Carou, M. C., Nunes, M. L., & Veciana-Nogués, M. T. (2006). Sensory analysis to assess the freshness of Mediterranean anchovies (Engraulis encrasicholus) stored in ice. Food Control, 17(7), 564–569. Ryder, J. M., Buisson, D. H., Scott, D. N., & Fletcher, G. C. (1984). Storage of New Zealand jack mackerel (Trachurus novaezelandiae) in ice: Chemical, microbiological and sensory assessment. Journal of Food Science, 49(6), 1453–1456. Salinas, Y., Ros-Lis, J. V., Vivancos, J.-L., Martínez-Mañez, R., Marcos, M. D., Aucejo, S., et al. (2012). Monitoring of chicken meat freshness by means of a colorimetric sensor array. Analyst, 137, 3635–3643. Scott, D. N., Fletcher, G. C., Charles, J. C., & Wong, R. J. (1992). Spoilage changes in deep water fish smooth oreo dory during storage in ice. International Journal of Food Science and Technology, 27, 577–587. Shahidi, F., Chong, X., & Dunajski, E. (1994). Freshness quality of harp seal (Phoca groenlandica) meat. Journal of Agriculture and Food Chemistry, 42, 868–872. Surette, M. E., Gill, T. A., & LeBlanc, P. J. (1988). Biochemical basis of postmortem nucleotide catabolism in cod (Gadus morhua) and its relationship to spoilage. Journal of Agricultural and Food Chemistry, 36(1), 19–22. Yam, K. L., Takhistov, P. T., & Miltz, J. (2005). Intelligent packaging: Concepts and applications. Journal of Food Science, 70(1), R1–R10.