The five fish species of interest were captured with purse seine between September 2020 and April 2021 by local fishermen at the Norwegian west coast. The catch occurred in area 2.a.2, according to the FAO Major Fishing Areas (14). The fish was gutted immediately after capture and kept on ice until the end of rigor mortis (41–57 h post-mortem). Fish were either filleted directly or frozen as whole (−80°C) and subsequently thawed before filleting. In order to have a thorough understanding of the chemical composition and identify any potential changes in the nutritional content within different body regions, muscle samples were taken from two fillets per flatfish. The two fillets, being the lower loin (LL) and upper belly (UB), of the flatfish samples were considered for analyses (n_Flounder = 7 × 2, n_{Lemon sole} = 5 × 2, n_Plaice = 10 × 2, n_Megrim = 5 × 2), as shown in Figure 1A. In contrast, only one fillet was kept for thornback ray (n = 5), as it consists of two central fillets (Figure 1B). All fillets were frozen directly and stored at –80°C until further use.

Dry matter and ash content was determined following the AOAC 925.10 method (15). Samples were homogenized and between 1 and 2 g were weighed in duplicates in porcelain crucibles. The samples were placed in a dehydrator at 105°C for 24 h (TS8056; Termaks, Norway). After 24 h, the samples were placed in a desiccator to cool down to room temperature, weighed and water content was calculated according to equation 1. The dried samples were transferred to an ash oven and burned at 550°C for 20 h (B410; Nabertherm,Germany). The samples were placed in a desiccator to cool down, weighed and the inorganic matter was then determined following the principle of equation 1.

Total crude protein content (%) was determined using the Kjeldahl method (15). A Kjeldahl apparatus (K-449 and K-375, Büchi Labortechnik, Schwitzerland) was used for measurements. The sample digestion and titration were carried out following the application manual No: 114/2013 of Büchi Switzerland. Briefly, sulfuric acid (H₂SO₄, 95–97%) and two Kjeldahl Tablets Eco (3.5 g K₂SO₄/0.105 g CuSO₄ × 5H₂O/0.105 g TiO₂) were added to samples (1.5 g) before digestion. Digested samples were first neutralized with NaOH (32%, 15–90 ml) and H₂SO₄ (0.25 mol/l) was used as the titration solution subsequently. To determine the total protein concentration, a conversion factor of 6.25 × nitrogen (%) was applied (16).

Total lipids (%) were determined following the method of Bligh et al. (17). Samples (2 g ww) were weighed into chloroform-resistant tubes and chloroform was added. The solvent-sample mixture was subjected to extensive homogenization and centrifugation to achieve phase separation. The aqueous phase and chloroform-lipid phase were separated. Chloroform was evaporated from samples by applying liquid nitrogen. To accelerate the evaporation, the samples were placed on a heating block turned to 40°C (Stuart™ block heater type: SBH130D/3, Cole-Parmer, United States). The total lipids (%) were calculated according to Bligh et al. (17). The remaining chloroform phase containing lipids was frozen and stored at −80°C for fatty acid analysis.

Fatty acids were prepared as methyl esters for analysis by gas chromatography. For fatty acid methyl ester (FAME) preparation, the method of Metcalfe et al. (18) was used. Chloroform phases containing lipids from individual fish were systematically merged to obtain five samples for thornback ray (n = 5), four samples for megrim (n = 4), and three samples each for lemon sole (n = 3) and flounder (n = 3). As these fish are very lean and individual fish were limited in size, it was necessary to merge samples to obtain at least 0.02 g of lipids per sample. Samples from European plaice were not merged due to bigger fish sizes, and 6 individuals (n = 6) were chosen for analysis. Nitrogen evaporation was conducted at 30°C until all chloroform was removed from the samples, and 3 ml KOH in methanol (0.5 M) was added to the samples and vortexed to saponify the lipids. Samples were incubated in a water bath at 70°C for 20 min, vortexed, and cooled on ice. Afterwards, 5 ml of boron trifluoride-methanol (14%, BF₃) was added to allow acid-catalyzed esterification of the fatty acids. The samples were re-incubated in the water bath at 70°C for 5 min and cooled on ice. N-butyl acetate (2 ml) was added, and the samples were shaken. Subsequently, saturated NaCl (around 1.5 ml), and two spatulas of powdered sodium sulfate (Na₂SO₄) were added to the samples, and the samples were rested at room temperature (21°C) to allow phase separation. Around 0.5 ml of hexane was added, and the lipid phase was then pipetted out and filtered using a 0.2 μm PTFE membrane (VWR International, United States) into GC vials.

The FAMEs were analyzed by gas chromatography (GC) using a GC apparatus (Agilent 6850, Agilent Technologies, United States). The samples (2–3 μl) were introduced by an evaporation injector (inlet: 260°C, pressure: 18.1 psi). Hydrogen was used as a carrier gas to pass the samples onto a polyethylene glycol column (HP-INNOWAX, i.D.: 0.25 mm; film: 0.25 μm, Merck Life Sciences, Norway), where FAMEs were separated at different times along the stationary phase. A flame ionization detector (FID) adjusted to 310°C was used to detect the samples. The oven program was set to a constant temperature of 160°C for 3 min, with an increase of 3°C/min to 240°C and held for 3 min.

Fatty acids were identified by comparing relative retention times (RRTs) of the external FAME standard mix containing 37 fatty acid methyl esters (Supelco 37 Component FAME Mix, Merck Life Sciences, Norway) with sample peaks. Chromatogram peaks, showing similar RRTs to the external standard were considered for determination. The intensity of each peak was calculated against the total intensity of FAMEs, to determine the percentage distribution of the individual fatty acids in each sample.

For analyzing potentially elevated levels of contaminants in the samples, a variety of trace elements and polychlorinated biphenyls (PCBs) were chosen, and samples were pooled together. Each sample contained two individuals (three for European plaice) of same size, equally distributed and homogenized. Per species, two pooled samples (n = 2) were examined.

The samples were analyzed for 20 elements, including both toxic and essential trace elements such as Ag, Al, As, Ca, Cd, Co, Cr, Cu, Fe, Hg, K, Mg, Mn, Mo, Na, Ni, Pb, Se, V, and Zn. An inductive coupled plasma mass spectroscopy (ICP-MS, 8800 Triple Quadrupole; Agilent Technologies, United States) system was used. The system was linked to an autosampler (prepFAST M5, ESI, United States). To test the accuracy of the analysis, certified reference materials (CRM) were used, namely MODAS-5 (cod tissue, Nr. 0496) and MODAS-3 (herring tissue, Nr. 0958). The procedure of sample preparation, including microwave digestion and subsequent steps were previously described in detail by Kendler et al. (21) following the method of Sørmo et al. (22).

The analysis for PCBs included PCB-3, 8, 28, 52, 101, 118, 138, 153, 180, 195, 206, and 209, including the ICES-6 PCBs (PCB: 28, 52, 101, 138, 153, 180) and the dioxin-like PCB 118. A GC–MS system (7890A, Agilent Technologies, United States) was employed to detect PCBs. The system included split liner injection, an inert mass selective detector (5,975, Agilent Technologies, United States) and a Thermo TG 5MS column (length: 30 m; i.D.: 250 μm; film: 0.5 μm). A detailed description of the procedure can be found in Kendler et al. (21). The sample extraction followed the method described by Teunen et al. (23).

All statistical analyses were performed using Minitab 19¹ (Minitab Inc., United States). A Grubbs Outlier test with a significance level of α < 0.05 was conducted to find outliers in the data set. Data were analyzed using univariate analysis of variance (ANOVA) combined with Tukey HSD post hoc test when significance was detected to investigate the differences between groups. Statistical differences were reported at the level of α < 0.05. For flatfish representatives, analyses were carried out in 2 × 2 parallels (2 parallels for each UB and LL fillet; 4 in total per sample) and are presented as means ± standard deviation (SD) if not other stated. For thornback ray, the same analyses were performed in duplicates.

The proximate composition of flounder, lemon sole, megrim, plaice and thornback ray are shown in Table 1. Significant differences were observed for the species’ ash, water, protein and lipid content. Megrim was found to have the lowest average water content of 79.2%, being significantly lower than flounder (p = 0.003). The measured water content for megrim equaled the values found by Afonso et al. (31), and Barbosa et al. (30), who showed values from 75–79%. Thornback ray showed a similar water content (80.1%) compared to lemon sole (81.4%) and plaice (80.5%), but differed from previous investigations by Colakoglu et al. (32) and Turan et al. (33) with water contents of 77%. The water contents of all investigated species are similar to those found by Karl et al. (34), investigating different flatfish species with average values ranging from 78.1 to 82.1%.

The ash content of the four flatfish representatives was between 1.0–1.25%, while fillets from thornback ray (0.9%) showed lower values of inorganic material. This was in line with previous investigations on flatfish, although previous studies on thornback ray found slightly higher ash values (1.1–1.4%) for this species (32, 33). Ash content in plaice is higher than previously investigated by Karl et al. (34) of 0.9% but similar to results from three different seasons by Kendler et al. (21) of values ranging from 1.07 to 1.28%. Plaice has a significantly higher ash content compared to flounder (p = 0.005) and thornback ray (p < 0.001).

For protein content, the differences were more significant between the investigated species. Megrim had a significantly higher protein content than lemon sole (p = 0.036), flounder (p = 0.003) and plaice (p = 0.024) with an average of 19.6%, being marginally higher than the studies by Barbosa et al. (30) with 16.6 to 18.6%. Lemon sole and flounder had a protein content of 16–18%, in line with previous investigations of plaice (16.6%) and yellowfin sole (16.0%) of Karl et al. (34). Significantly higher protein values were observed for thornback ray with 24.0%, being considerably higher than all investigated flatfish species (p < 0.001). The measured protein content of thornback ray was not in correspondence with previous studies and was 4–5% higher than observed in studies by Colakoglu et al. (32) and Turan et al. (33) with 18.6 and 20%, respectively. It must be mentioned that the proximate composition of thornback ray exceeds a total of 100%, which indicates an overestimation of the protein content measured by the Kjeldahl method using a conversion factor of 6.25. The Kjeldahl method measures the total nitrogen content, assuming approximately 16% nitrogen in proteins. However, other non-proteins in the cells also contain nitrogen, which can lead to an overestimation of the protein nitrogen in the food (35). Ray tissue contains around 350–400 mM urea, a nitrogen containing non-proteinaceous component, which might have been an interfering substance in the Kjeldahl analysis (36). Moreover, as the relative nitrogen content of amino acids fluctuates and the amino acid composition depends on the protein source, the assumption of 16% nitrogen content is rather general. Studies have shown that using a conversion factor of 6.25 can overestimate the total protein content of some foods (35, 37, 38). For this reason, attention has been given to creating conversion factors that are species/ food-specific (37, 38). Nevertheless, the established conversion factor of 6.25 is still widely used being officially recognized by the AOAC as a standard analytical method for protein determination, which makes results better comparable (39). The results suggest that protein contents of the four flatfish species are not overestimated, as the proximate composition (100%) is not notably exceeded and standard deviations are acceptable (1%).

Regarding the lipid content, all species can be considered as lean species, having values below 2%. Plaice was found to have significantly higher lipid values of in average 1.25% compared to lemon sole (p = 0.025) with 0.74% and thornback ray (p = 0.036) with 0.76% total fat content. The findings in this study show lower lipid contents for megrim (0.98%) compared to the study of Pastoriza et al. (40) finding a lipid content of up to 1.9%. Karl et al. (34) found lower lipid values for plaice of around 0.8%, but similar values of around 1% of other investigated flatfish. In a previous study of Kendler et al. (21), significant differences in the lipid content of plaice depending on fishing season were discovered ranging from 0.75 to 1.55%. For thornback ray, the measured lipid content of 0.76% was marginally higher than the finding of Turan et al. (33) of 0.5%, while much lower than the finding of Colakoglu et al. (32) of 3.4%. Two previous studies on deep-sea fish found that general deep-sea elasmobranchs like thornback ray had a lipid content of around 0.7 to 1.0%, which support the findings of the present study (41, 42).

Total amino acids (TAA) were investigated to determine the nutritional value of the proteins in the fish. The TAA results for the species are given in Table 2. The most abundant TAA for all species were leucine and lysine, as well as glutamic and aspartic acid, under physiological conditions in the form of glutamate and aspartate. The same abundant amino acids were found in an investigation of three flatfish species by Kim et al. (43), although showing higher amounts of glycine than in the present study. The contents of glutamate and aspartate can be regarded as overestimated as glutamine and asparagine were converted to these two amino acids during acid analysis (44). Consequently, asparagine and glutamine were detected in the lowest amounts in all investigated samples. Significant differences (p < 0.05) were found for most of the amino acids between the five species. All species have a preferable distribution of indispensable amino acids (IAA), accounting for more than 50% of the TAA distribution. Megrim was found to have significantly higher amounts of IAA (7.24 g/ 100 g; p = 0.002) compared to flounder (6.14 g/100 g), lemon sole (6.07 g/100 g), plaice (5.62 g/100 g), and thornback ray (5.88 g/100 g). Furthermore, megrim has comparably more total amino acids (13.80 g/100 g; p = 0.002) than the other species. This was also found in three of the four species for the total protein content, with the exception of thornback ray, where an overestimated protein content is suspected. The amino acid determination by Blackburn (19) directly calculates the protein amount by considering only amino acid residues in the analysis and does not take into account possible interfering non-proteinaceous components. Amino acid hydrolysis can therefore be regarded as a good approach to determine the total amino acid content and gives an indication of the protein content of foods. However, the concentration of some of the amino acids might be lowered significantly due to the hydrolysis step prior to HPLC-analysis, which leads to an underestimation of the total protein content (35). Furthermore, it has to be mentioned that due to their instability during acid hydrolysis, cysteine and tryptophan were not determined in the applied method. Consequently, these two amino acids should be analyzed separately if the TAA measurement is used to indicate the total protein content.

The DIAA ratios were calculated and the DIAAS was calculated based on the DIAA ratios. The DIAA reference ratios are displayed in Table 3, and indicate whether all DIAA were present in the protein in adequate amounts to meet the requirements for adults set by FAO et al. (28). To calculate the ratios, the measured TAA levels were converted to mg/g protein for the different species as described in section 2.6.1. Scores above 1 indicate sufficient IAA levels, while ratios below 1 indicate insufficient levels. In addition, the DIAAS (%), was calculated, being the lowest amino acid ratio per species multiplied by 100 to convert the ratio to a percentage score. As indicated in Table 3, all amino acids show ratios above 1, despite methionine and cysteine (as combined values) in lemon sole (0.7), implying an excellent overall protein quality of the species. Cysteine and tryptophan were not analyzed in this study, which explains the low methionine + cysteine ratio, as it only consists of methionine. Ratios up to 2.5 for threonine were observed for thornback ray, megrim and lemon sole, pointing out the relevance of these species for a sufficient intake of indispensable amino acids. Moreover, DIAAS of over 100% were discovered for flounder (120%), megrim (120%), plaice (110%), and thornback ray (120%), stressing the high protein quality of the studied fish. Lemon sole shows a poorer DIAAS (70%), which is possibly due to the lack of data from cysteine.

In addition to their nutritional importance, amino acids are associated with taste when occurring unbound in the form of free amino acids (FAA) in biological systems. Figure 2 shows the FAA distribution (mg/100 g sample ww) of the five investigated species. FAA are grouped according to their distinctive flavor as described by Fuke et al. (45), Kirimura et al. (46) and Sarower et al. (47). FAAs have been identified as essential taste contributors in seafood (45–47). Glutamic acid, in the form of glutamate, glycine, and alanine are commonly identified among the most important taste contributors. Glycine and alanine are linked to sweetness, while FAAs such as valine, arginine, and methionine are linked to bitter taste in seafood (47). Aspartate and glutamate both provide a sour taste. However, especially relevant in seafoods, these amino acids also give an umami taste in the presence of sodium salts, such as the familiar monosodium glutamate (MSG). Phenylalanine and tyrosine also have a bitter taste, but can enhance the umami flavor (47). Related to the present study, significant differences between species were found for all FAAs, but arginine/glycine (p = 0.114; Arg/Gly as combined values), threonine (p = 0.866) and methionine (p = 0.872). Lysine showed significant differences amongst the species (p < 0.001), with flounder and megrim having notably higher values. Flounder, lemon sole and plaice were identified to have significantly higher levels of histidine (p < 0.001), while megrim and thornback ray show comparatively low values of 0.51 and 0.84 mg/100 g sample, respectively. The most prevalent FAA are allocated in the group of “Sweet” amino acids with arginine/glycine having the highest content, followed by lysine and alanine.

Referring to the TAA and FAA distribution of the five investigated species, high variances within the species were observed, which is expressed by relatively large standard deviations as shown in Table 2 and Figure 2. External factors such as sex, maturity, and feeding behavior can influence the chemical composition of fish. Moreover, seasonality can play an important factor, as pointed out in a previous study on chemical composition of European plaice caught during three seasons by Kendler et al. (21).

The following sections (section 3.3 and 3.4) highlight the potential of the studied species as sources for eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) in human nutrition. Flounder, plaice and thornback ray were found to have significantly higher total PUFA compositions than megrim with 50.7, 47.2, and 49.3%, respectively, as can be seen in Table 4. Significant differences were also found in the amount of present DHA. Thornback ray shows considerable higher values (36.2%) comparing to the other species and profoundly lower values were observed in lemon sole (14.7%).

Although the studied fish are categorized as lean species, the fatty acid composition is of high importance. Health promoting effects due to prevention of cardiovascular diseases (CVD), tumor cell proliferation and inflammation processes as well as beneficial effects on brain, retina and neurodevelopment in children are primarily attributed to eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) (9).

The total EPA and DHA intake when consuming the five investigated species was calculated to highlight how the species contribute to providing these indispensable FAs to human diet. The contribution to the daily recommendations of 250 mg EPA and DHA by the EFSA Panel on Dietetic Products et al. (48) are shown in Figure 3. Moreover, the average contribution of all IAA given as DIAAS by FAO (24) including the daily requirements of IAA for an 80 g adult as proposed by FAO et al. (28) are shown in Figure 3. The values refer to a portion size of 200 g as recommended in a report on dinner serving sizes of foods by the Norwegian Food Authority (Mattilsynet) and were used for the calculations (49).

Figure 3 emphasizes on the nutritional value of the studied fish related to the content of DIAA as well as EPA and DHA in a 200 g fillet portion. The green bars show that the total required daily amount of DIAA for an 80 kg adult are met by all species. A fillet of 200 g of all species covers more than 120% of the total required DIAA, with thornback ray fulfilling the requirements to 172%. Nevertheless, to fully cover the daily intake of every individual indispensable amino acid (100%), a larger portion size than 200 g is needed for all investigated fish species. The daily recommended intake of, e.g., valine was restrictively covered in megrim (69%), hence a portion size of 371 g would be necessary to cover the daily demand of this particular amino acid (100%). Furthermore, valine is also the restrictive amino acid for thornback ray and flounder, where 54 and 61% of the daily demand is covered by a 200 g portion, respectively. In plaice, phenylalanine (53%) and in lemon sole, methionine (33%) are not fully covered with regards to a portion size of 200 g.

With regards to EPA and DHA, all investigated fish species contribute significantly to the daily suggested intake of 250 mg by EFSA (48). The relative contribution of DHA and EPA in 200 g fillets is shown in dark/light blue shading and was converted to mg/100 g edible portion using the conversion factors proposed by Weihrauch et al. (29) as described in section 2.6.2. Even tough being lean species, a 200 g portion of all of the five species contributes in average to more than 100% of the recommended average daily intake (AI) of EPA and DHA set by EFSA (48). The highest contribution was found in a 200 g portion of flounder (217%), followed by megrim (194%), thornback ray (182%), plaice (136%) and lemon sole (129%). With regards to the weekly recommended intake of 1.75 g EPA + DHA (250 mg × 7) this would mean a consumption of 3.2 portions of 200 g flounder, 3.6 of megrim, 3.8 of thornback ray, 5.2 of plaice and 5.4 portions of lemon sole, respectively. These results are highly relevant, as lean fish is usually not associated with providing sufficient levels of n3 fatty acids and the focus for covering n3 fatty acids was previously put on fatty fish such as salmon or trout in the past (50). All investigated species contain higher relative amounts of DHA, despite lemon sole, which has a 50:50 share of EPA and DHA.

Marine fish are good sources for both macro and trace elements, including minerals like calcium, magnesium or selenium, being vital for human health. All fish contain sufficient amounts of potassium (K) and magnesium as shown in Table 5. Significant differences between species were observed for the elements manganese (p < 0.001), magnesium (p = 0.002) and iron (p = 0.047). High values in selenium, ranging from 0.25 mg kg⁻¹ in thornback ray to 0.49 mg kg⁻¹ in lemon sole were found in this study. Compared to the study of Karl et al. (34) on different flatfish, selenium values ranging from 0.13 to 0.31 mg kg⁻¹ were reported. When setting dietary recommendations, the dietary reference value (DRV) is used. The DRV in this study refers to either the average requirement (AR), the population reference intake (PRI) or the adequate intake (AI), depending on the available data from the expert panel on Dietetic Products, Nutrition and Allergies from EFSA (51–60). Even though no significant difference (p = 0.213) in selenium content between species was detected, an effect on the contribution to meet the DRV is visible, given as % of DRV. Hence, a 200 g fillet of lemon sole covers the selenium intake to 140%, whereas a 200 g fillet of thornback ray reaches only 74.3% of the daily selenium coverage. Moreover, all five species are a good source of potassium, covering around 20% of the DRV. Differences in species are visible for the magnesium contribution, where a 200 g portion of plaice covers 10% of the DRV and megrim contributes up to 16% of the daily recommended intake.

Being demersal fish species, Pleuronoectiformes and Rajiformes are more likely to accumulate PCBs and hazardous trace elements than other fish species (13). Therefore, when frequently eating flatfish or ray, the matter of food safety must be considered. However, both intrinsic and environmental variables have a role in the bioaccumulation of hazardous as well as beneficial compounds (31). Individuals different in sizes and sexes, exhibit different concentrations in trace elements and contaminants, due to a variety in habitat and migration behaviors. In this study several toxic trace elements were determined (Table 6). The analysis on PCBs, including both non-dioxin and dioxin-like PCB congeners, show values lower than the detection limit (LOD) for flounder, lemon sole, megrim, and thornback ray. Hence no significant accumulation of any of the investigated PCB congeners were detected in those four species. Concerning plaice, traces of PCB 3, 52, 101, 118, 138, 153, and 180 were detected, as previously reported in the study of Kendler et al. (21) that looked into seasonal differences. A reason for the lower PCB values compared to plaice could be the lower fat content as well as overall lower fish size in the three other flatfish and the thornback ray. With regards to toxic trace elements, significantly different values between the species were detected for the elements chromium (p = 0.002), nickel (p < 0.001), cadmium (p < 0.001) and lead (p < 0.001). The highest accumulation of chromium and cadmium were found in thornback ray with 56.36 ± 29.12 μg kg⁻¹ and 0.58 ± 0.05 μg kg⁻¹ respectively. Large varieties within species can be seen for all elements, with particular great SD in the mercury content of plaice (112.9 ± 54.1 μg kg⁻¹) and thornback ray (174.37 ± 169 μg kg⁻¹). Despite differences between species as well as individuals, the maximum levels of cadmium (0.1 mg kg⁻¹), lead (0.3 mg kg⁻¹) and mercury (0.5 mg kg⁻¹) set by the EC (61) are not exceeded. When considering arsenic as potential hazardous component, organic and inorganic arsenic must be differentiated. Inorganic arsenic is the toxic form and according to Sloth et al. (62) a maximum of 1% of total arsenic in marine species is found in the form of hazardous arsenite and arsenate. The calculations in the previous study of Kendler et al. (21) on arsenic content in European plaice were followed for the other four species in this study. Considering the suggestion of 1% inorganic arsenic (58), it is safe to consume the recommended portion size by the Norwegian Food Authority (49) of 200 g of each of the five investigated species.