Abstract Fatty acids (FAs) were analysed in Baltic herring (Clupea harengus membras) stored in the Swedish Environmental Specimen Bank for up to 40 years. The purpose was to evaluate the retrospective use of FA signatures to detect temporal and spatial changes in the Baltic ecosystem. Fish from northern and central Baltic captured in the 1970s, the 1980s, in 1990, 2000, and 2009 and stored at − 25 °C were analysed. From the 1980s and onward herring from the south Baltic were included. A total of 55 FA and 4 alkenyl chains (detected as dimethyl acetals) were identified, and 28 of these (present at > 0.5% by weight) were used in evaluation of the data. The amount of some 20–22 carbon polyunsaturated fatty acids (PUFA) was related to time with lower amounts in older samples while other PUFAs were not related to time. Principal component analysis with saturated FAs and monounsaturated FAs showed similar sample groupings as the one obtained by including the PUFAs. The differences found in herring FA in this longitudinal study could be attributed to location of sampling, year of collection and storage time. However, the clearly distinguishable pattern in the FA composition in herrings from different locations in the Baltic Sea seen at all decades indicate that this technique can be used retrospectively. Introduction Tissue fatty acid signatures (FASs) have been widely used to investigate diets of predators and structure of food webs (Grahl-Nielsen and Mjaavatten, 1991; Käkelä and Hyvärinen, 1996; Budge et al., 2002; Daalsgaard et al., 2003; Budge et al., 2006; Hebert et al., 2006; Iverson, 2009). If this technique could be applied to material that has been saved in environmental specimen banks one could detect food web changes retrospectively. The Swedish Environmental Specimen Bank (SESB) at the Swedish Museum of Natural History (SMNH) has collected and stored tissue samples of fish, seabirds and seals since the 1960s. These samples could be valuable in ecological studies of the Baltic Sea where long-term temporal or spatial changes of food webs are being investigated, if either the whole FAS or at least certain indicative fatty acids (FAs) in the stored material have remained fairly intact. Herring (Clupea harengus) is a widely distributed species found in the North Atlantic and North Pacific temperate waters, as well as in several locations of southern oceans. The species is commercially important and extensively used in monitoring programmes all over the world. Baltic herring (Clupea harengus membras) is a subspecies of herring, harvested in the region for animal and human food. Specimens of this fish have been collected in various parts of the Baltic Sea and stored in SESB since the early 1970s. These samples could be a valuable tool in studying changes in the Baltic ecosystem over time. The Baltic herring tolerates broad salinity range. Together with sprat (Sprattus sprattus) and cod (Gadhus morhua), it is the dominating fish species in the south and middle parts of the Baltic Sea. However, in the northern parts, the salinity is normally too low for cod and sprat, leaving Baltic herring as the dominating fish species. The herring feeds on zooplankton and mysids and is the main prey species for many large fish, seabirds, and seals (Möllman and Köster, 1999; Peltonen et al., 2004). Thus, the herring plays a key role in the Baltic food web and is an ideal target for ecosystem monitoring. Currently the ecosystem function in the Baltic Sea is changing due to both global environmental and local anthropogenic factors affecting the Baltic community structure of organisms at low trophic levels, which changes are then reflected in the prey available for the fish (Casini et al., 2006; Österblom et al., 2007). Even the initial natural conditions in the Baltic Sea have been variable. The water is referred to brackish and the sea is shallow (average depth only 53 m) with very large catchment including densely populated areas. The sea is divided into basins that are separated by shallow sills that slow down the turnover rate and promote water stagnation. Each year there is a considerable influx of fresh water from a number of rivers in the northern parts, while in the south the input of saltwater through the narrow straits of Denmark is occasional. The outcome is that different parts of the Baltic Sea differ considerably in water salinity, temperature, and stratification and thus also in the conditions for organisms living there. During the last decades, the Baltic Sea has been subject to increasing anthropogenic inputs of nutrients and pollutants. These along with the slowly decreasing salinity and increasing water temperature have caused an ecosystem change so profound that during the latter part of the last century the Baltic Sea went from being an oligotrophic to a eutrophic system (Vuorinen et al., 1998). The changed conditions have affected the structures of phytoplankton and zooplankton communities, and since 1980s the total phytoplankton biomass has increased with gains in dinoflagellates, and concomitantly there have been losses in diatoms and Pseudocalanus copepods (Alheit et al., 2005). At the same time, shifts in dominating top predators were observed, first from seals to cod and then to herring and other clupeids (Österblom et al., 2007). The purpose of this study was to find out if retrospective analysis of the FAS in the Baltic herring stored in SESB for up to 37 years could be used to identify temporal and spatial differences in the Baltic ecosystem. In this study, we provide data on large temporal and spatial differences in the FAS of Baltic herring obtained from an environmental specimen bank and discuss which of the compositional changes may be due to degradation of tissue lipids and FAs during the long-term storage and which changes in the FAS likely represent genuine food web changes. We also test the possibility to use normalized subsets of specific FA known to be resistant to oxidation, and compare the revealed temporal trends obtained by using the full FAS and the resistant partial FAS. Material and methods Fish samples Since the 1970s, samples of Baltic herring have been caught once or twice a year from a number of locations by either trawl or gillnet. The herrings analysed in this study were caught from three separate areas in the Baltic Sea situated in ICES subdivision 31(North), 27 (Central), and 25 (South) (Figure 1). All herrings used were caught during the autumn. Figure 1. View largeDownload slide Sampling areas for fish in this study. In 1973, the sampling place in the North and Central Baltic was Kalix and Mysingen, respectively. The following years, herrings from North and Central Baltic were collected at Harufjärden and Landsort. Figure 1. View largeDownload slide Sampling areas for fish in this study. In 1973, the sampling place in the North and Central Baltic was Kalix and Mysingen, respectively. The following years, herrings from North and Central Baltic were collected at Harufjärden and Landsort. The fish samples were frozen as soon as possible after capture. In SESB, each frozen individual was wrapped in aluminium foil and put into a small plastic bag and then all the fish from the same sampling occasion were placed together in a larger, sealed plastic bag. Thus the fish specimens used for this study had been stored in those bags untouched at −25°C since the days of capture (Table 1). Table 1. Sampling places and years for fish (n = 5 for each catch) used in this study. Place year year year year year North Baltica (ICES 31) 1973 1981 1990 2000 2009 Central Baltica (ICES 27) 1973 1983 1990 2000 2009 South Baltic (ICES 25) – 1980 1990 2000 2009 Place year year year year year North Baltica (ICES 31) 1973 1981 1990 2000 2009 Central Baltica (ICES 27) 1973 1983 1990 2000 2009 South Baltic (ICES 25) – 1980 1990 2000 2009 a See Figure 1 for the sampling places 1973 compared with sampling places 1981–2009 in North Baltic and 1983–2009 in Central Baltic. To eliminate the influence of sex, only females were analysed in this study. In addition, all fish were approximately in the same state of the reproductive cycle identified by gonadal maturity. The age of the herrings was determined from otoliths. Fish condition as Fulton’s condition factor (CF) was calculated according to the formula: CF=W/L3*100, W, weight (g) and L, length (cm). Complete data of length, weight, condition and age of the fish, and the year and month of capture are shown in Supplementary Table S1. Head, backbone and tailfin was removed at SMNH before the fish were sent for FAS analysis to the Helsinki University Lipidomics Unit, where the remaining fish carcasses were homogenized in two steps. First, the carcasses were ground by using a mechanical mincer and then to produce fine course uniform mass the homogenization was finalized manually with a mortar and pestle. Three subsamples of 0.4–0.5 g were taken for the analysis from each homogenized fish. The careful homogenization procedure and taking several large subsamples ensured the representativeness of the sample, regarded important in the in situ direct methods (Thiemann et al., 2004). Fatty acid analysis Fatty acid methyl esters (FAMEs) were prepared by direct transmethylation of tissue lipids (Carrapiso and Carcía, 2000), i.e. heating the fish tissue homogenates in 1% methanolic sulphuric acid with hexane as co-solvent in closed vials under nitrogen atmosphere at a temperature of 95 °C. The resulting FAMEs were extracted into hexane and the solutions were dried with anhydrous sodium sulphate and concentrated under nitrogen atmosphere. The samples were analysed using a Shimadzu GC-2010 Plus gas chromatograph with an AOC-20i auto injector and an AOC-20s auto sampler. The gas chromatograph had a Zebron ZB-wax capillary column (0.25 mm internal diameter and 30 m length) with inner wall coated with 0.25 µm polyethylene glycol (Phenomenex, Torrence CA, USA). Before each sample batch, an equimolar standard mixture (Larodan AB, Malmö, Sweden) was run and the peak areas of several saturated, monounsaturated, and polyunsaturated FA (SFA, MUFA, and PUFA, respectively) standards were monitored. To avoid errors commonly seen in automated integration results, the same experienced operator manually integrated each gas chromatogram by using GC Solutions software (Shimadzu) and the peak areas of the 51 quantitatively most important FAs were converted to their respective weight percentages per total FAs. In the data, we included the found four dimethyl acetals (DMAs) formed from the alkenyl chains of tissue plasmalogen phospholipids (PLs) during the derivatization of the sample. Since the alkenyl chains are known to be easily oxidized compared with the FAs (Mangold and Weber, 1987) we kept them in the FAS to serve as sensitive indicators of oxidative break down during the long-term storage of the herring samples. Data analysis The means of the three subsamples were used in further analyses of the data. The variation between the three subsamples of each fish was small, typically around ±3% of the replicate means for each FA and never exceeded 5%. The main purpose was to identify changes in the herring FAS with time, both changes that could be attributed to storage time and condition and those that could be attributed to changes in the Baltic food webs, and to retrieve this information both multivariate and univariate statistical methods were employed. As the multivariate approach, principal component analysis (PCA) was performed on the whole data, first by using as loadings the full FA selection and then, to diminish the possible effects of sample deterioration on the FAS (Frankel, 1998), a subset of FA containing only the oxidation resistant SFAs and MUFAs (composition data were renormalized to 100% to fully remove the influence of the during storage reduced PUFA levels on the remaining proportional data). For each year of sampling, the full FAS of the herrings from different locations were subjected for soft independent modelling of class analogy (SIMCA). The compositional differences (first described by a PCA biplot) between the samples from different experimental groups were quantified by creating space-ﬁlling models for two of the groups at a time and subsequently computing the distances of the samples to these two models at p < 0.05 level (Wold and Sjöström, 1977). As the amount of each FA was expressed in % by weight per total FAs loaded to the PCA, to improve normality of the data distribution arc sine transformation of the square root of the % values was performed prior to the analyses. To ensure that the PCA analyses, even when using the transformed data, were not biased by the proportional nature of the composition data, the analyses were repeated by using non-metric multidimensional scaling (nMDS). In addition, as the sample size in each year/location group was small and the normal distribution of the data could not be assured, also the univariate comparisons in order to find out temporal and spatial differences were made with non-parametric tests. Kruskal-Wallis analysis of ranks was used when comparing more than two groups and Mann-Whitney U-test was used when comparing two groups. Software used for the statistical analyses were Dell Statistica version 13 (1884-2015 Dell Inc.), Pattern Recognition Systems Sirius 8.5 (Pattern Recognition Systems, Bergen, Norway), and Primer 7 (PRIMER-E, Auckland, New Zealand). Results CF and age A decreasing trend was observed in CF of the fish from the northern (1973–2009) and southern Baltic (1980–2009) but not in the fish from the central Baltic. In the central Baltic on the other hand a significand (p = 0.008) decrease in CF was found between fish from 2000 and 2009. In 2000, there was also a significant difference in CF between fish from different sampling locations (Figure 2). Figure 2. View largeDownload slide CF in herring from south, central and north Baltic analysed in this study (n = 5 for all groups). Dots represent median values, box represent the interquartile range (25–75%), and whisker shows min and max values. Figure 2. View largeDownload slide CF in herring from south, central and north Baltic analysed in this study (n = 5 for all groups). Dots represent median values, box represent the interquartile range (25–75%), and whisker shows min and max values. The median age of fish was 5 years at all sampling locations. The variation in age was 3–9 in the southern herrings, 2–9 in the central herrings and 2–7 in the northern herrings. The 2-year-old herrings were all captured in the 1973. Fatty acids A total of 51 FAs and 4 DMAs (full data in Supplementary Table S2) were identified. A PCA on all of the 55 variables revealed that despite their very low concentrations of <0.5% (by weight) a few trace FAs and the DMAs had a strong influence on the PCA. The DMAs were significantly more abundant in the most recent samples (Supplementary Figure S1). None of the single DMAs was present in concentrations >0.5% in any sample and ∑DMAs was 0.10–1.22% by weight. To eliminate the strong influence of the trace components it was decided that only the FAs that were present with >0.5% in at least one sample should be used in multivariate statistical analyses. That left 6 SFAs, 11 MUFAs, and 11 PUFAs (in all 28 FAs) that were used for further statistical analyses (Table 2). Table 2. The 28 FAs in Baltic herring (>0.5% by weight) used in the statistical evaluation of data. 14:0 16:1n-9 18:2n-6 15:0 iso 16:1n-7 18:3n-3 15:0 16:1n-5 18:4n-3 16:0 17:1-branch 20:2n-6 17:0 17:1n-8 20:4n-6 18:0 18:1n-9 + 11 20:3n-3 18:1n-7 20:4n-3 18:1n-5 20:5n-3 20:1n-9 22:5n-6 20:1n-7 22:5n-3 22:1n-9 22:6n-3 14:0 16:1n-9 18:2n-6 15:0 iso 16:1n-7 18:3n-3 15:0 16:1n-5 18:4n-3 16:0 17:1-branch 20:2n-6 17:0 17:1n-8 20:4n-6 18:0 18:1n-9 + 11 20:3n-3 18:1n-7 20:4n-3 18:1n-5 20:5n-3 20:1n-9 22:5n-6 20:1n-7 22:5n-3 22:1n-9 22:6n-3 When using the 28 FAs as loadings, the PCA displayed a V-shaped biplot where the herring samples collected in 1973–2000 from the northernmost Baltic Sea were grouped on one side of the plot away from the concomitantly collected samples from the central and southern Baltic Sea (Figure 3 upper panel). In contrast, the recent samples of 2009 from different locations did not differ in their FAS and thus formed the tip of the pattern V. Similar sample separation and V-shaped pattern were obtained by using nMDS, which showed that the PCA analyses using arc sine transformed data were valid (thus only the PCA results are shown). The most important factor behind the separation of the samples i.e. principal component 1 (PC1) accounted for 29% of the total compositional variation in the dataset and incorporated most of the spatial variation. PC2 accounted for an additional 26% of the variation (not yet addressed in PC1) and incorporated most of the temporal variation. Figure 3. View largeDownload slide PCA on 28 FA (>0.5% by weight) in Baltic herring caught in the North (blue), Central (red), and South Baltic (black) in 1973–2009 (upper panel). The same dataset without the PUFAs (lower panel). Figure 3. View largeDownload slide PCA on 28 FA (>0.5% by weight) in Baltic herring caught in the North (blue), Central (red), and South Baltic (black) in 1973–2009 (upper panel). The same dataset without the PUFAs (lower panel). According to the univariate statistics, the samples of the year 2009 from all locations contained significantly higher (Kruskal-Wallis ANOVA p < 0.01) levels of total PUFA than the 1973–1990 samples, whereas the SFA and MUFA totals in the recent samples were significantly lower (p < 0.01) than in the other samples. In addition, at all locations the samples collected in 2000–2009 had higher ratios of total n-3 PUFA to total n-6 PUFA (n-3/n-6 ratio) than the older samples. A significant positive correlation with collection year was found for the sum of n-3 PUFAs but not for the sum of n-6 PUFAs (Figure 4). Figure 4. View largeDownload slide Correlation between collection year and Σn-3PUFA (left) and Σn-6PUFA (right). Σn-3PUFA are significantly correlated with collection year (r = 0.833) while Σn-6 PUFA are not (r = 0.033). Figure 4. View largeDownload slide Correlation between collection year and Σn-3PUFA (left) and Σn-6PUFA (right). Σn-3PUFA are significantly correlated with collection year (r = 0.833) while Σn-6 PUFA are not (r = 0.033). The member of n-3 PUFAs, 22:6n-3 was the most abundant individual PUFA in all samples. The relative levels of 22:6n-3 showed a significant increasing trend from the 1970–1980s until 2009 (shown as example in Figure 5). The same was also true for 20:5n-3 and the n-6 PUFAs, 22:5n-6, and 20:4n-6 (not shown). The levels of the 18–carbon n-3 PUFAs, 18:3n-3, and 18:4n-3 showed no temporal change and the most abundant individual n-6 PUFA, 18:2n-6 remained either at the same level or had higher relative levels in the samples of 1970–1980s than in those from the years 1990 to 2009 (also shown in Figure 5). PUFAs are known to be more susceptible to oxidation (Frankel, 1998). To rule out the effect of possible PUFA oxidation during storage on the temporal FAS patterns, the PCA was repeated by using as loadings a subset of SFAs and MUFAs. The resulting biplot (Figure 3, lower panel) closely resembled the one obtained by using the full FA set (Figure 3, upper panel) (the same result was seen by using nMDS). There is a slightly larger (34%) separation on the PC1 axis and a slightly smaller (24%) separation on the PC2 axis compared with when the PUFAs were included (Figure 3, lower panel). Figure 5. View largeDownload slide The most abundant n-3 PUFA, 22:6n-3 (above) and n-6 PUFA, 18:2n-6 (below) in autumn collected Baltic herrings from South Baltic (1980–2009), Central Baltic (1973–2009) and North Baltic (1973–2009) (n = 5 for all areas and years). Dots represent median values, box represent interquartile range (25–75%), whiskers shows min and max values. Figure 5. View largeDownload slide The most abundant n-3 PUFA, 22:6n-3 (above) and n-6 PUFA, 18:2n-6 (below) in autumn collected Baltic herrings from South Baltic (1980–2009), Central Baltic (1973–2009) and North Baltic (1973–2009) (n = 5 for all areas and years). Dots represent median values, box represent interquartile range (25–75%), whiskers shows min and max values. Further analysis of the 28 FAS of each decade showed that the herrings from the northern Baltic were clearly separated from those from the central and southern Baltic in the 1980s, in 1990, in 2000, and in 2009. Furthermore, herrings from the central and southern Baltic were separated in the 1980s, in 1990, and in 2000 (p < 0.05 SIMCA) (Figures 6–9). In general, 16:1n-7 and its elongation products (18:1n-7 and 20:1n-7) and PUFAs with 20–22 carbons were present with higher relative concentrations in herring from the northern Baltic, while 18:1n-9 and PUFAs with 18 carbons were abundant in herring from the middle and south Baltic. In 2009, the herrings from northern Baltic were still significantly separated from those from the southern and central Baltic although the south and central herrings were not significantly separated at this time (Figure 9). The same pattern was seen in all decades when the PUFAs were excluded from the analyses (Figures 6–9, lower panels). The trend of the spatial differences in herring FAS becoming smaller with time was seen in the power of the PC1 axis in explaining FAS variation: for the samples of 1980s, the PC1 captured 68% of the variation, for 1990 48%, for 2000 40%, and for 2009, the PC1 captured 35% of the variation. If the analysis was repeated with only the SFAs and MUFAs, the overall pattern did not change and the PC1 axis capturing the spatial variation explained 70% (1980s), 45% (1990), 41% (2000), and 33% (2009) of the total data variation. Figure 6. View largeDownload slide Upper panel: 28 FA from herring captured in Northern (N), Central (C), and South (S) Baltic in early 1980s. Lower panel: the same dataset without the PUFAs. The N, C, and S areas are significantly separated in both cases. Figure 6. View largeDownload slide Upper panel: 28 FA from herring captured in Northern (N), Central (C), and South (S) Baltic in early 1980s. Lower panel: the same dataset without the PUFAs. The N, C, and S areas are significantly separated in both cases. Figure 7. View largeDownload slide Upper panel: 28 FA from herring captured in Northern (N), Central (C), and South (S) Baltic in 1990. Lower panel: the same dataset without the PUFAs. The N, C, and S areas are significantly separated in both cases. Figure 7. View largeDownload slide Upper panel: 28 FA from herring captured in Northern (N), Central (C), and South (S) Baltic in 1990. Lower panel: the same dataset without the PUFAs. The N, C, and S areas are significantly separated in both cases. Figure 8. View largeDownload slide Upper panel: 28 FA from herring captured in Northern (N), Central (C), and South (S) Baltic in 2000. Lower panel: the same dataset without the PUFAs. The N, C, and S areas are significantly separated in both cases. Figure 8. View largeDownload slide Upper panel: 28 FA from herring captured in Northern (N), Central (C), and South (S) Baltic in 2000. Lower panel: the same dataset without the PUFAs. The N, C, and S areas are significantly separated in both cases. Figure 9. View largeDownload slide Upper panel: 28 FA from herring captured in Northern (N), Central (C), and South (S) Baltic in 2009. Lower panel: the same dataset without the PUFAs. N is significantly separated from C and S in both cases. C and S are not significantly separated. Figure 9. View largeDownload slide Upper panel: 28 FA from herring captured in Northern (N), Central (C), and South (S) Baltic in 2009. Lower panel: the same dataset without the PUFAs. N is significantly separated from C and S in both cases. C and S are not significantly separated. Discussion This work showed that it was possible to distinguish herring from different geographical areas in the Baltic Sea by analysing FA composition although the herrings had been stored in SESB for up to 40 years. Likely, there was deterioration with time in certain PUFAs, mainly the 20–22 carbon PUFAs with 4–6 double bonds. However, we cannot rule out the possibility that the PUFA levels would partly reflect alterations in the PUFA supply of the fish due to food web changes. In this study, we could detect a V-shape in the PCA with the more recent samples at the tip of the V, indicating that samples had become more similar in recent time. The pattern was very similar even if the PUFAs were omitted. The PC1-axis incorporated much of the geographical variation while the PC2-axis captured much of the temporal variation. The samples from the northern Baltic were vertically orientated indicating that the differences between these samples are mostly due to time. However, the samples from central and southern Baltic were distributed along a diagonal axis indicating that some other factor besides time is responsible for this distribution. From the 1970s and onward, the PC1 axis had a decreasing importance while the PC2 axis had an increasing importance in explaining the variation in the data indicating that towards present the geographical component became less central. It was obvious that the 20–22 carbon PUFAs had undergone partial deterioration due to storage indicating that including these FAs in the data analysis could hamper detecting spatial or temporal differences. However, when the PUFAs were excluded, the main PCA pattern did not change. Among different FA structures the PUFAs are regarded as the most vulnerable for oxidative damage (Frankel, 1998). In line with this, the herring PUFA totals were the highest in the recently collected samples at all three sampling locations, indicating deterioration of PUFAs in the older samples. The presence of DMAs indicates the presence of the phospholipid class plasmalogens. Plasmalogens are regarded as integral antioxidants of biological membranes and are consumed in the oxidative chain reactions (Brosche and Platt, 1998; Nagan and Zoeller, 2001; Wallner and Schmitz, 2011). The lower levels of DMAs in the older samples also indicate oxidation of plasmalogens due to storage. Correspondingly, there is most likely deterioration involved in the apparently lower levels of 22:6n-3 and 20:5n-3 in older samples. However, the trend of decreasing PUFA level along with increasing storage time was not consistent for every individual PUFA. For example, the clear decrease in the 22:6n-3 level found when comparing the oldest samples to the most recent ones largely contributed to the adverse correlations between the total n-3 PUFA and collection year. In contrast, no obvious losses due to storage time were found for total n-6 PUFAs, which is surprising since the n-6 PUFAs are also believed to be prone to oxidative damage. However, the same pattern of decreasing levels for the 20–22 carbon PUFAs (mainly with 4–6 double bonds) but not for the 18 carbon PUFAs (mainly with 2–3 double bonds) was seen for both the n-3 PUFAs and n-6 PUFAs. The explanation to the different relative losses of different n-3 and n-6 PUFAs may at least partly be due to their different patterns of incorporation into different classes of structural and storage lipids. The long chain and highly unsaturated PUFAs are more commonly present in PLs, which constitute cellular membranes, than in triacylglycerols (TAGs), residing in storage lipid droplets (Linko, 1967). The FA composition of TAGs is regarded to give more information on the diet while the FA composition of structural lipids (PLs) is deemed more conservative (Iverson, 2009). To maintain proper environment for membrane proteins there are strict preferences of incorporating different FAs into the structural PLs, and dietary PUFAs are incorporated after further elongation and desaturation (Abbott et al., 2012). Thus, it can be argued that in order to use FAS analysis as a tool for retrospective analyses of food web changes, the dietary influence on the FA composition of isolated storage lipids may be easier to recognize. On the other hand, when herring is preyed upon by piscivorous birds and mammals, the whole fish is often consumed and thus from the point of view of many food web studies the FA composition of the whole fish is more relevant. In a previous study on the effects of long-term storage on seal blubber FAS, the most surprising finding was that although the blubber both looked yellowish and had a strong smell, the blubber FAS had changed little (Lind et al., 2012). The herring samples of this study that were collected in the 1970s and 1980s were slightly discoloured and had a stronger odour when compared with the more recent samples. The levels of a number of 20–22 carbon PUFAs had significantly decreased levels in these whole fish stored for long periods of time but not in the seal blubber samples stored under the same conditions. The reason for the different degree of structural damage of PUFAs may be that the lipids in seal blubber predominantly are TAGs, and lipids in the whole fish homogenates are mixtures of different types of PLs (been in contact with ice in the frozen fish) and TAGs. A study on human adipose tissue that also mainly consisted of TAGs, reported that n-3 PUFAs were stable for up to 6 years when kept either at room temperature of 20 °C or frozen at −80 °C (Katan et al., 2003). This could indicate that PUFAs bound to TAGs in large droplets of adipocytes in seal blubber or other adipose tissues are protected from water and oxygen and thus more resistant to deterioration than the PUFAs bound to PLs. The polar head groups of PLs locate them in water-lipid interphases and make them more accessible to hydrolysis. A study on frozen rainbow trout muscle stored in sealed aluminium bags at −15 °C for up to 34 weeks concluded that hydrolysis was the major cause of the lipid deterioration (Ingemansson et al., 1995). The herrings used for this study had been wrapped in aluminium foil and stored in sealed plastic bags, which according to Ingemansson and co-workers provide good protection against oxidation. Besides the work cited above, there is a lack of studies on the fate of lipids and FAs in fish during long-term storage., A large number of studies have been conducted on the changes of lipid and FA profiles in fish stored for up to approximately 1 year (Ackman et al., 1968; Christophersen et al., 1992; Ingemansson et al., 1995; Aubourg et al., 2005, 2007; Passi et al., 2005; Özogul et al., 2006). However, most of these studies have been conducted in order to validate optimal storage of fish for human consumption, and thus the focus has been on detecting compounds that affects taste, smell and texture of fish muscle as a food item, not on changes in fish FAS. There has been a decline in herring weight-at-age since the 1980s. One of the hypotheses are that decreasing salinity and increasing temperature has changed the zooplankton community (Flinkman et al., 1998; Möllmann et al., 2005). In this study, we found a decreasing trend in CF during the time-period covered in herring from northern and southern Baltic but not in herring from central Baltic. In central Baltic, on the other hand there was a significant drop in CF between 2000 and 2009. Furthermore, the only significant difference in CF between fish from different sampling locations was found in 2000. Thus, feeding conditions likely do not explain the different FA patterns between the different sampling locations. However, it is tempting to speculate that the fattier herring of the past, likely had higher body TAG/PL ratio than the specimens caught in 2009, which may have allowed more diet-derived variation in the body FAs in the early years than was possible in the lean fish of 2009, in which the strictly regulated FA composition from PL presumably had larger influence on the body FA composition. The eutrophication of the Baltic has proceeded gradually but the process has accelerated during the last decades (Finni et al., 2001; Weckström et al., 2004; HELCOM, 2009). Eutrophication reduces silicate concentration of water, which affects the structure of phytoplankton communities by decreasing the diatom biomass (Conley et al., 1993). At the same time, the increased runoff from rivers in the northern Baltic has decreased the salinity, which also decreased the silicate load to the Baltic Sea and has further reduced the diatom biomass. In addition, the climate change with increased winter temperatures has been found to have negative effect on the Baltic diatom biomass (Humborg et al., 2008; Jaanus et al., 2011; Wasmund et al., 2013). Furthermore, convincing time-series data of the increasing biomass of cyanobacteria have been detected in the Baltic Sea from 1970s until present (Kahru and Elmgren, 2014). The phytoplankton FA profiles have been shown to be affected especially by the phytoplankton community composition, various diatoms, cryptophytes and dinoflagellates increasing the basal supply of highly unsaturated essential PUFAs, and the chlorophytes and cyanobacteria decreasing it (Galloway and Winder, 2015). Thus, the basal FA supply of the aquatic food web is largely determined by phytoplankton community composition. In food web studies, the commonly used tracer for diatoms is the level of the 16 carbon MUFA, 16:1n-7 or the ratio of 16:1n-7 to 16:0 (Daalsgaard et al., 2003). Thus the higher level of 16:1n-7 as well as the higher ratio of 16:1n-7 to16:0 found in this study for the northern Baltic herring indicated that in this part of the Baltic, diatoms have been more important in the basis of the food web but have become less significant during the time covered by this study (Olli et al., 2011; Suikkanen et al., 2013). Dietary 22:6n-3 and 20:5n-3 are essential for fish development, growth and health (Sargent et al., 1999), and their supply via food web is affected by environmental conditions such as eutrophication and brownification (Taipale et al., 2016). The uncertainty of detecting the original levels of these PUFAs in the fish samples stored for decades is a limitation and a challenge for future long-term ecological monitoring programmes. Lowering of the storage temperature considerably below the −25° C used for the herrings of this study and/or utilizing vacuum or protection gas techniques might store these valuable marker FAs. Although there has been an obvious deterioration of highly unsaturated, long-chain PUFAs due to the very long time storage, the current study showed that fish FAS can be used for retrospective studies to indicate environmental and food web changes. Only two such PUFAs, 22:6n-3 and 22:5n-3 was present with significantly reduced levels in the oldest samples. Despite this, the herrings caught several decades ago from different locations of the Baltic Sea had clearly distinguishable patterns of FA composition. The 20 carbon PUFAs and the 16 carbon MUFAs were present with larger relative amounts in the herring from the northern Baltic while the C18 MUFAs and PUFAs were more abundant in the herring from middle and southern Baltic. Concurrent analyses can be made by using such a set of FA variables that are not prone to oxidation i.e. SFAs and MUFAs. In addition to showing that the FAS method is a useful tool even for long-term monitoring of food web changes, we revealed a clear trend of the lipids of food webs of the Baltic herring losing their spatial heterogeneity in the 40-year period, 1970−2010, the reasons for which warrant further studies. Supplementary data Supplementary material is available at the ICESJMS online version of the manuscript. Acknowledgements We thank two anonymous reviewers for constructive comments and suggestions. References Abbott S. K., Else P. L., Atkins T. A., Hulbert A. J. 2012. Fatty acid composition of membrane bilayers: importance of diet polyunsaturated fat balance. Biochimica et Biophysica Acta , 1818: 1309– 1317. Google Scholar CrossRef Search ADS PubMed Ackman R. G., Ke P. J., MacCallum W. A., Adams D. R. 1968. Newfoundland capelin lipids: fatty acid composition and alterations during frozen storage. Journal of the Fisheries Research Board of Canada , 26: 2037– 2060. Google Scholar CrossRef Search ADS Alheit J., Möllmann C., Dutz J., Kornilovs G., Loewe P., Mohrholz V., Wasmund N. 2005. Synchronous regime shifts in the central Baltic and the North Sea in the late 1980s. ICES Journal of Marine Science , 62: 1205– 1215. Google Scholar CrossRef Search ADS Aubourg S. P., Lago H., Sayar N., González R. 2007. Lipid damage during frozen storage of Gadiform species captured in different seasons. European Journal of Lipid Science and Technology , 109: 608– 616. Google Scholar CrossRef Search ADS Aubourg S. P., Rodríguez A., Gallardo J. M. 2005. Rancidity development during frozen storage of mackerel (Scomber scombrus): effect of catching season and commercial presentation. European Journal of Lipid Science and Technology , 107: 316– 323. Google Scholar CrossRef Search ADS Brosche T., Platt D. 1998. The biological significance of plasmalogens in defense against oxidative damage. Experimental Gerontology , 33: 363– 369. Google Scholar CrossRef Search ADS PubMed Budge S., Iverson S. J., Bowen D. W., Ackman R. G. 2002. Among- and within species variability in fatty acid signatures of marine fish and invertebrates on the Scotian Shelf, Georges Bank, and southern Gulf of St. Lawrence. Canadian Journal of Fisheries and Aquatic Sciences , 59: 886– 898. Google Scholar CrossRef Search ADS Budge S. M., Iverson S. J., Koopman H. N. 2006. Studying trophic ecology in marine ecosystems using fatty acids: a primer on analysis and interpretation. Marine Mammal Science , 22: 759– 801. Google Scholar CrossRef Search ADS Carrapiso A. I., Carcía C. 2000. Development in lipid analysis: some new extraction techniques and in situ transesterification. Lipids , 35: 1167– 1177. Google Scholar CrossRef Search ADS PubMed Casini M., Cardinale M., Hjelm J. 2006. Inter-annual variation in herring, Clupea harengus, and sprat, Sprattus sprattus, condition in the central Baltic Sea: what gives the tune? Oikos , 112: 638– 650. Google Scholar CrossRef Search ADS Christophersen A. G., Bertelsen G., Andersen H. J., Knuthsen P., Skibsted L. H. 1992. Storage life of frozen salmonids. Zeitschrift fur Lebensmittel-Undersuchung und Forschung , 194: 115– 119. Google Scholar CrossRef Search ADS Conley D. J., Schelske C. L., Stoermer E. F. 1993. Modification of the biogeochemical cycle of silica with euthrophication. Marine Ecology Progress Series , 101: 179– 192. Google Scholar CrossRef Search ADS Daalsgaard J., St. John M. A., Kattner G., Muller-Navarra D., Hagen W. 2003. Fatty acid trophic markers in the pelagic marine environment. Advances in Marine Biology , 46: 227– 339. Finni T., Laurila S., Laakkonen S. 2001. The History of Eutrophication in the Sea Area of Helsinki in the 20th Century. AMBIO: A Journal of the Human Environment , 30: 264– 271. Google Scholar CrossRef Search ADS Flinkman J., Aro E., Vuorinen I., Viitasalo M. 1998. Changes in northern Baltic zooplankton and herring nutrition from 1980s to 1990s: top-down and bottom-up processes at work. Marine Ecology Progress Series , 165: 127– 136. Google Scholar CrossRef Search ADS Frankel E. N. 1998. Lipid Oxidation , The Olily Press, Dundee, Scotland. 303 pp. Galloway A. W. E., Winder M. 2015. Partitioning the relative importance of phylogeny and environmental conditions on phytoplankton fatty acids. PLoS One , 10: e0130053. Google Scholar CrossRef Search ADS PubMed Grahl-Nielsen O., Mjaavatten O. 1991. Dietary influence on fatty-acid composition of blubber fat of seals as determined by biopsy - a multivariate approach. Marine Biology , 110: 59– 64. Google Scholar CrossRef Search ADS Hebert C. E., Arts M. T., Weseloh D. V. C. 2006. Ecological tracers can quantify food web structure and change. Environmental Science and Technology , 40: 5618– 5623. Google Scholar CrossRef Search ADS PubMed HELCOM. 2009. Eutrophication in the Baltic Sea – an integrated thematic assessment of the effects of nutrient enrichment and eutrophication in the Baltic Sea region. Baltic Sea Environment Proceedings , 115B. 152 pp. Humborg C., Smedberg E., Medina M. R., Mörth C.-M. 2008. Changes in dissolved silicate loads to the Baltic Sea — the effects of lakes and reservoirs. Journal of Marine Systems , 73: 223– 235. Google Scholar CrossRef Search ADS Ingemansson T., Kaufmann P., Ekstrand B. 1995. Multivariate evaluation of lipid hydrolysis and oxidation data from light and dark muscle of frozen stored rainbow trout (Oncorhynchus mykiss). Journal of Agriculture and Food Chemistry , 43: 2046– 2052. Google Scholar CrossRef Search ADS Iverson S. J. 2009. Tracing aquatic food webs using fatty acids: from qualitative indicators to quantitative determination. In Lipids in Aquatic Ecosystems , pp. 281– 308. Ed. by Kainz M., Brett M. T., Arts M. T.. Springer New York, New York, NY. Google Scholar CrossRef Search ADS Jaanus A., Andersson A., Olenina I., Toming K., Kaljurand K. 2011. Changes in phytoplankton communities along a north-south gradient in the Baltic Sea between 1990 and 2008. Boreal Environment Research , 16(suppl A): 191– 208. Kahru M., Elmgren R. 2014. Multidecadal time series of satellite-detected accumulations of cyanobacteria in the Baltic Sea. Biogeosciences , 11: 3619– 3633. Google Scholar CrossRef Search ADS Katan M., Harryvan J. L., van de Bovenkamp P. 2003. n-3 fatty acids in human fat tissue aspirates are stable for up to 6 y. European Journal of Clinical Nutrition , 57: 816– 818. Google Scholar CrossRef Search ADS PubMed Käkelä R., Hyvärinen H. 1996. Site-specific fatty acid composition in adipose tissues of several northern aquatic and terrestrial mammals. Comparative Biochemistry and Physiology , 115B: 501– 514. Google Scholar CrossRef Search ADS Lind Y., Bäcklin B.-M., Lundström K., Budge S. M., Walton M., Karlsson O. 2012. Stability of fatty acid composition in seal blubber during long-term storage. Marine Ecology Progress Series , 461: 283– 291. Google Scholar CrossRef Search ADS Linko R. 1967. Fatty acids and other components of Baltic herring flesh lipids, Turun yliopiston julkaisuja. 101, Turku, 121 pp. Mangold H. K., Weber N. 1987. Review: Biosythesis and biotransformation of ether lipids. Lipids , 22: 789– 799. Google Scholar CrossRef Search ADS PubMed Möllman C., Köster F. W. 1999. Food consumption by clupeids in the Central Baltic: evidence for top-down control? ICES Journal of Marine Science , 56(Suppl.): 100– 113. Google Scholar CrossRef Search ADS Möllmann C., Kornilovs G., Fetter M., Köster F. W. 2005. Climate, zooplankton, and pelagic fish growth in the central Baltic Sea. ICES Journal of Marine Science , 62: 1270– 1280. Google Scholar CrossRef Search ADS Nagan N., Zoeller R. A. 2001. Plasmalogens: biosynthesis and functions. Progress in Lipid Research , 40: 199– 229. Google Scholar CrossRef Search ADS PubMed Olli K., Klais R., Tamminen T., Ptacnik R., Andersen T. 2011. Long term changes in the Baltic Sea phytoplankton community. Boreal Environment Research , 16 (Suppl. A): 3– 14. Österblom H., Hansson S., Larsson U., Hjerne O., Wulff F., Elmgren R., Folke C. 2007. Human-induced trophic cascades and ecological regime shifts in the Baltic Sea. Ecosystems , 10: 877– 889. Google Scholar CrossRef Search ADS Özogul Y., Özogul F., Özkütük S., Kuley E. 2006. Hydrolysis and oxidation of European eel oil during frozen storage for 48 weeks. European Food Research and Technology , 224: 33– 37. Google Scholar CrossRef Search ADS Passi S., Cataudella S., Tiano L., Littarru G. P. 2005. Dynamics of lipid oxidation and antioxidant depletion in Mediterranean fish stored at different temperatures. Biofactors , 25: 241– 254. Google Scholar CrossRef Search ADS PubMed Peltonen H., Vinni M., Lappalainen A., Pönni J. 2004. Spatial feeding patterns of herring (Clupea harengus L.), sprat (Sprattus sprattus L.), and the three-spined stickleback (Gasterosteus acueatus L.) in the Gulf of Finland, Baltic Sea. Ices Journal of Marine Science , 61: 966– 971. Google Scholar CrossRef Search ADS Sargent J., Bell G., McEvoy L., Tocher D., Estevez A. 1999. Recent developments in the essential fatty acid nutrition of fish. Aquaculture , 177: 191– 199. Google Scholar CrossRef Search ADS Suikkanen S., Pulina S., Engstrom-Ost J., Lehtiniemi M., Lehtinen S., Brutemark A. 2013. Climate Change and Eutrophication Induced Shifts in Northern Summer Plankton Communities. Plos One , 8: e66475. Google Scholar CrossRef Search ADS PubMed Taipale S. J., Vuorio K., Strandberg U., Kahilainen K. K., Järvinen M., Hiltunen M., Peltomaa E., Kankaala P. 2016. Lake eutrophication and brownification downgrade availability and transfer of essential fatty acids for human consumption. Environment International , 96: 156– 166. Google Scholar CrossRef Search ADS PubMed Thiemann G. W., Budge S. M., Iverson S. J. 2004. Determining blubber fatty acid composition: a comparison of in situ direct and traditional methods. Marine Mammal Science , 20: 284– 295. Google Scholar CrossRef Search ADS Wallner S., Schmitz G. 2011. Plasmalogens the neglected regulatory and scavenging lipid species. Chemistry and Physics of Lipids , 164: 573– 589. Google Scholar CrossRef Search ADS PubMed Wasmund N., Nausch G., Feistel R. 2013. Silicate consumption: an indicator for long-term trends in spring diatom development in the Baltic Sea. Journal of Plankton Research , 35: 393– 406. Google Scholar CrossRef Search ADS Weckström K., Juggins S., Korhola A. 2004. Quantifying background nutrient concentrations in coastal waters: a case study from an urban embayment of the Baltic Sea. AMBIO: A Journal of the Human Environment , 33: 324– 327. Google Scholar CrossRef Search ADS Wold S., Sjöström M. 1977. SIMCA: a method for analyzing chemical data in terms of similarity and analogy. In Chemometrics: Theory and Application vol. 52, pp. 243– 282. Ed. by Kowalski B. R., ACS Symposium Series, Washington DC. Google Scholar CrossRef Search ADS Vuorinen I., Hänninen J., Viitasalo M., Helminen U., Kuosa H. 1998. Proportion of copepod biomass declines with decreasing salinity in the Baltic Sea. Ices Journal of Marine Science , 55: 767– 774. Google Scholar CrossRef Search ADS © International Council for the Exploration of the Sea 2017. All rights reserved. For Permissions, please email: email@example.com
ICES Journal of Marine Science – Oxford University Press
Published: Jan 1, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 12 million articles from more than
10,000 peer-reviewed journals.
All for just $49/month
Read as many articles as you need. Full articles with original layout, charts and figures. Read online, from anywhere.
Keep up with your field with Personalized Recommendations and Follow Journals to get automatic updates.
It’s easy to organize your research with our built-in tools.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera