Experimental animals and design

In May 2013, November 2013 and May 2014, a total of 160 farmed Atlantic salmon (Salmo salar) with mean value±standard deviation (S.D.) mass of (4.4±1.2) kg were sampled during commercial slaughtering at two different plants in Norway. A full factorial design (Fig. 1) with four factors was designed; season (spring/autumn), location (north/south), rigor status (pre/post) and filleting (HOG/fillet) resulting in 16 different groups where the effect on water-holding capacity (WHC) and water content was investigated. In addition, 160 fillets with mean±S.D. mass of (1.6±0.4) kg from the entire design apart from the filleting were stored on ice for 14 days to quantify the drip loss during storage and to determine the composition of the drip loss (Fig. 1).

Fig. 1

Schematic illustration of the experimental design, where spring and autumn refer to May and November, respectively, south and north to N58-59° and N63-64°, respectively, pre- and post-rigor to point of filleting, and head on gutted (HOG) and fillet to when the analysis was conducted (before and after filleting, respectively). The time scale shows when the specific analyses were conducted for each group. WHC=water-holding capacity, DM=dry matter, DL=drip loss

Fish from the north site was farmed around N63-64°, while the fish from the south site was farmed around N58-59° along the west coast of Norway. Spring part of the trial was performed in May, while the autumn part was performed in November. The sea temperature was 9.5 and 5.0 °C at the south location, while the north location had 9.5 and 6.0 °C in November and May, respectively. Before sampling, the fish were slaughtered according to the commercial practice at the plants; the fish was crowded, pumped in a double-chamber vacuum pump, live chilled up to 45 min, stunned with percussive captive bolt and gills were cut before bleeding in a bleeding tank for up to 60 min. The fishes were then gutted to end up as head on gutted (HOG), and all the fish were measured (mass, length and pH), tagged and randomly subjected to pre- or post-rigor analysis and WHC or storage trial, respectively. The fish subjected to pre-rigor analysis were analyzed consecutively, while fish subjected to post-rigor analysis were stored in expanded polystyrene boxes on ice in a chilled room (0-4 °C) for four days to go through rigor mortis.

pH, temperature, condition factor and drip loss

pH and temperature were measured right after gutting in the anterior dorsal muscle close to the gills, using a Mettler Toledo SevenGo Pro^TM pH-meter (Mettler Toledo Inc., Columbus, OH, USA) connected to an Inlab puncture electrode (ibid). In addition, post-rigor fish were measured after 4 days of storage. During storage of fillets, muscle pH and temperature were measured on the cranial part of the fillet.

Before analysis, the fish were dried off and weighed, and the fork length was measured. Fulton’s condition factor (CF) (13) was calculated for gutted fish according to the following formula:CF=m/l³·0.1 /1/where m is gutted mass (kg) and l is fork length (m).

In the storage trial, HOG subjected to post-rigor filleting were weighed after being dried off on day 0 and day 4 to quantify the drip loss. Fillets for the storage trial were machine filleted, dried off, weighed and packaged in a closed bulk polyethylene bag inside expanded polystyrene boxes with top icing. Thereafter, all samples were transported within 24 h to the laboratory where the fillets were stored in chilled rooms (0-4 °C) for 14 days. During storage, melting water was removed through weep holes in the bottom of each box, while drip loss was gathered. Drip loss (DL/%) from the fillets was measured as the difference in fillet mass between day 0 and day x of both the right and left fillet:
where m₀ is fillet mass (g) on day 0, and m_x is fillet mass (g) on day x.

Drip loss was assessed on days 7 and 14 after filleting (i.e. day 7 and 14 for pre-rigor, and day 4 (HOG), day 11 and 18 for post-rigor). In addition, drip loss of post-rigor fillets is reported as cumulated drip loss including drip loss as HOG. An average of both fillets was used for statistical analysis.

Water-holding capacity and water content

Water-holding capacity was measured in the dorsal muscle. Samples for determination of WHC and water content of whole fish were first sampled from the left side before the fish was machine filleted, and samples from the right fillet were taken approx. 45 min later. Samples were cut in triplicates (diameter 31 mm, hight 20 mm) from the cranial part on the dorsal back and backwards, above the lateral line on identical locations on all fishes using the white muscle tissue (Fig. 2). Three portions from each sample, approx. 5 g were made, and the first two (Fig. 2c, I and II) were weighed and dried at 105 °C for 16 to 24 h (14) for estimating the dry content of the muscle and hence the water content. The last sample (Fig. 2c, III) was weighed and placed into metal carriers (Hettich Lab Technology, Tuttlingen, Germany) and centrifuged (Rotina 420 R; Hettich Lab Technology) for 15 min at 4 °C, using a free swing rotor at 530×g. WHC was calculated from the following formula (15):
Fig. 2

Schematic illustration of the: a) head on gutted and b) fillet areas used for determination of water-holding capacity (WHC) and water content (WC). In addition, insert c) shows the splitting and use of the sample, where two parts were used for WC (I and II) and one part for WHC (III) determination

Composition of drip loss

During storage, the fillets were stored in closed bags to gather the drip loss and to prevent contamination from the melting ice. The amount of water, protein, fat and astaxanthin were quantified in triplicate according to the design.

Nitrogen content was measured on a Tecator Kjeltec system (model 2020 digestor and 1026 distilling unit; Tecator, Höganäs, Sweden) (16). Protein content was calculated from nitrogen measurements using the formula:

Fat and astaxanthin were extracted by a method after Bligh and Dyer (17). Total contents of fat were calculated and total contents of astaxanthin were quantified by HPLC using a H₃PO₄ -modified silica gel column (17) and all-E-astaxanthin (A-9335; Sigma-Aldrich, Merck, St. Louis, MO, USA) as an external standard. The eluent was acetone/n-hexane (86:14) and the flow was 1.2 mL/min. Detection wavelength was set at 470 nm. The employed molar absorption coefficients, ε_{1 cm} (1%), at 470 nm in hexane containing 4% (by volume) CHCl₃ were 2100 for all-E-astaxanthin (18), and 1350 and 1750 for 13Z- and 9Z-astaxanthin, respectively. The ε_{1 cm} (1%) for 13Z- and 9Z-astaxanthin were estimated from the HPLC response factors of Schüep and Schierle (19). Water content was measured as described previously.

Statistics

All data were analyzed by general linear model (GLM), regression (R) and Pearson’s correlation coefficient (r), using Minitab v. 17 statistical software (20). All results are given as mean value± S.D., unless stated otherwise.

Biometrics

Comparing different populations of fish during commercial conditions certainly poses many challenges, as the fish used in this trial was sampled during a period of one year at two different locations. Even if they differed significantly (p<0.002) in gutted mass between the two locations (north (4.1±1.1) kg, south (4.7±1.2) kg), none of the design factors varied significantly in initial pH (p>0.084) or condition factor (p>0.065). Average initial pH was 6.9±0.3, while condition factor was 1.1±0.1.

Water-holding capacity, water content and pH

Water-holding capacity is known to be related to pH, but including pH as a covariate in the general linear model did not give any significant result (p=0.645). The WHC was measured at three different positions on the fillet (Fig. 2), but no effect of position on the fillet (p=0.295) was detected, so the average of the three measurements was used in the following analysis of the results.

The water content in the muscle was not affected by season, but there was a significant difference between the two localizations (p<0.001, F=14) (Table 1), and a significant (p<0.001, F=147) interaction between localization and season was detected (Fig. 3). In addition, pH was significantly affected by localization (p<0.001, F=78) and season (p=0.001, F=12) (Table 1) together with an interaction effect between localization and rigor (p<0.001, F=29, data not shown). No other significant (p>0.190) effects on pH were detected.

Fig. 3

Interaction plot between localization (north: N63-64°, and south: N58-59°) and season (spring and autumn) for: a) water-holding capacity (WHC; p<0.001, F=24) and b) water content in the muscle (p<0.001, F=147) (N=80)

Furthermore, a significant difference (p<0.001, F=148) between HOG and fillet was detected in WHC (Table 1), where fillet had a decrease of 2.6% compared to HOG. The water content in the muscle was significantly affected by filleting (p<0.001, F=78), where HOG had approx. 2% more water than the fillet, measured on the same fish. Filleting opens up the muscle structure of the fish and removes amongst other the peritoneum. In addition, it is likely that the filleting machine induces micro-ruptures in the muscle structure. These expose the mucle and physically decrease its water-holding property and the water content. Jørpeland et al. (21) found the same effect in farmed cod, where whole fish had 0.5% more water than hand-filleted cod. Filleting by machine as done in the present study might give the muscle increased pressure compared to hand filleting. This can have squeezed more water out of the fillet, hence the increased difference than that observed in farmed cod.

Rigor status had a significant (p<0.001) impact on WHC, with a decrease of 5.5% on average from pre- to post-rigor, and explained a major part of the observed variation (F=656) (Table 1). No effect of rigor was detected (p>0.234) on the water content of the muscle. Assessing WHC of pre-rigor Atlantic salmon has as far as we know never been done before. Several papers report WHC of pre-rigor filleted salmon, but the time of measurement was at least 1 day after filleting (22, 23). However, a study of ground rainbow trout (Oncorhynchus mykiss gairdneri) did not reveal any differences in WHC measured immediately or 24 h post mortem (24). The observed decrease in WHC may be caused by several mechanisms related to the development of rigor mortis. WHC is known to be affected by pH, as pH close to the isoelectric point is known to lower the WHC (4), while some other researchers have stated that there is no strong connection between the pH and WHC in salmon muscle due to small variations during storage (11). Even though pH was not detected as a covariate to WHC, a positive correlation between pH and WHC was found (r=0.66, p>0.001). This connection was most obvious when comparing pre- vs. post-rigor fish, as both pH (p<0.001, F=1589) and WHC were significantly decreased in post-rigor fish compared to pre-rigor. pH had an average drop of approx. 0.6 pH units from pre- to post-rigor, ending up at pH=6.29. This drop is caused by the conversion of glycogen to lactate post mortem (25). In addition, the muscles form actomyosin during rigor, leaving less space for water (26), and together with changes in the salt ion balance (27), these factors might explain the observed drop in WHC between pre- and post-rigor samples.

Higher water content in the muscle implies more loosely bound water that is easier to remove, hence lower WHC and increased drip loss. Both WHC and water content have a significant interaction between localization and season in this trial, where increased amount of water in the muscle was accompanied by decrease in WHC. Water content and fat content are proved to have a strong negative correlation in salmon (28), so the observed difference in water content might imply increased fat content in the fish and that the amount of fat in the fish affects both WHC and drip loss, which might explain the observed differences of WHC and pH during the season. Observed differences in pH between the two localities might be explained by stress before slaughter.

Drip loss during storage

Storage time is a continuous variable and pH is known to have an influence on water-holding capacity (4), so incorporating both storage time (days from slaughtering) and pH as covariates into general linear model revealed increased drip loss with both increasing storage time (r=0.846, p<0.001, F=877) and increasing pH (r=0.222, p<0.001, F=14).

Season affected the drip loss significantly (p<0.001, F=16) (Table 2). Significant interactions between localization and season (p<0.001, F=22, data not shown) and localization and rigor (p=0.027, F=5, data not shown) were detected. No other significant effects were detected (p>0.169). Season is known to affect both quality and composition of salmon, where fat, texture and drip loss increase after the strong growth in the summer (12, 29). The growth of farmed salmon is strongly linked to the temperature of the sea (30), where the fish increase growth up to a certain point, before it drops (31). Both photoperiod and water temperature are different along the long Norwegian coast, where the south part has higher water temperature than the north, but north has longer daylight in the summer. Salmon slaughtered in the spring had 0.27% increased drip loss compared to those slaughtered in the autumn, while localization had no significant effect (p>0.05) on either drip loss or WHC in our trial. In addition, WHC, drip loss and water content had a significant interaction with localization and season, where increased WHC was accompanied by decreased drip loss and decreased water content and vice versa. This is in contradiction with the findings of Mørkøre et al. (12), who found an increased drip loss in October compared to April in salmon, while Roth et al. (32) found no effect of season on turbot.

The drip loss was moreover significantly affected by the rigor status (p<0.001, F=49) (Table 2). By splitting the data set in post-rigor and pre-rigor and performing a linear regression with time (Fig. 4), the drip loss of post-rigor fillets increased on average 0.14% each day (R²=0.945), whereas the drip loss of pre-rigor fillets increased on average 0.18% each day (R²=0.985), reaching (2.50±0.46) % on day 18 and (2.62±0.50) % on day 14 respectively. The drip loss of HOG after 4 days of storage was (0.14±0.34) %. Previous studies on the effect of rigor status during filleting on drip loss are inconclusive. Skjervold et al. (33) and Rosnes et al. (34) found higher drip loss in pre-rigor fillets after 14 and 16 days of storage on ice than in post-rigor fillets, while Einen et al. (25) and Skjervold et al. (22) found no effect of filleting time on drip loss. Fillets in the present study had a drip loss of 2.5–2.6% at the end of storage, while in the cited studies it ranged from 0.9–3%. All these reports related the drip loss to the time from filleting, not the time from slaughtering. In the present study, however, the HOG had almost ignorable drip loss, where some of the HOG even gained mass. One could expect less drip loss in pre-rigor filleted fillets since pre-rigor fillets have higher WHC. The drip loss was however higher in pre-rigor filleted samples than in those filleted post rigor. During filleting the muscle is exposed, which increases the flesh surface and subsequently the potential for drip loss. In addition, WHC in pre-rigor filleted salmon measured post rigor showed the same WHC as post-rigor filleted salmon (22, 23). It is important to remember that the major variation of drip loss in the present study was explained by storage time. If the drip loss in the present study was related to the time of filleting, post-rigor would have 0.18% drip loss on average each day, almost identical to pre-rigor (data not shown).

Fig. 4

Accumulated drip loss of pre-rigor (circle) and post-rigor (square) fillets stored on ice, where the post-rigor ones were stored as head on gutted for the first 4 days (N=80)

Composition of drip loss

Storage time was a significant (p<0.001) covariate for all the four analyzed constituents, where the amount of water (r=-0.94, p>0.001, F=271), fat (r=-0.85, p>0.001, F=92) and astaxanthin (r=-0.68, p<0.001, F=27) decreased significantly with storage time (Table 2). Protein, on the other hand, increased with storage time (r=0.91, p<0.001, F=205).

The water content was significantly higher in drip loss from post-rigor fillets (p=0.005, F=8.5), and in fillets from the north location (p=0.001, F=4.6), while the protein content was significantly lower in post-rigor fillets (p=0.002, F=12) and in the fillets from the autumn group (p=0.014, F=5.6). In addition, significant interaction effects between location and rigor status were found on both water content (p<0.001, F=22) and protein content (p=0.003, F=11) (Fig. 5). Degradation of protein is an effect of autolysis, where the degradation increases with time. Proteins lost in the drip loss are the water-soluble proteins. The composition of proteins in salmon muscle varies during storage, and water-soluble proteins decrease significantly from 5 to 14 days post mortem (11). Post-rigor filleted salmon was 4 days older at the time of assessment than the pre-rigor salmon, and this time difference can explain lower amount of protein in the drip loss. In farmed cod, pre-rigor fillets lost more protein than post-rigor fillets (35), most likely due to the same effect. In addition, the amount of protein increased during storage, confirming the degradation of the fillet.

Fig. 5

Interaction plot between localization (north: N63-64°, and south: N58-59°) and rigor (pre and post) status for: a) water content in the drip loss (p<0.001, F=22) and b) protein content in the drip loss (p=0.003, F=11) (N=80)