Use of fatty acid profiles to monitor the escape history of farmed Atlantic salmon

Farmed Atlantic salmon can escape from fish farms at various stages of their life, from juveniles to large mature fish. Escapees that enter rivers to spawn pose a threat to the genetic integrity of wild populations. Knowledge about the timing of these escapes can provide important information for wildlife management and the aquaculture industry, enabling them to prevent or mitigate the negative impacts of escapees. Farmed salmon food has a high content of terrestrial lipids; thus, we used fatty acid (FA) profiling to monitor the escape history of farmed salmon. Escaped salmon captured in rivers (n = 251) presented a wide range of FA profiles that we used to classify the fish as (1) early-escaped wild-like fish that were assumed to have escaped at smolt or early post-smolt stage (24%), (2) recently escaped fish with high levels of FAs typically found in commercial salmon food (61%) and (3) intermediate escapees whose FA profiles lay between those 2 groups (15%). To estimate the size at escape of the intermediate escapees, we performed a feeding experiment that monitored the development of FA profiles after a shift in diet from terrestrial to marine lipids. Most intermediate escapees appeared to have escaped when they were <3 kg, and ranged from 3 to 11 kg when recaptured in rivers. We conclude that FA profiling is a promising tool to monitor escape histories, and that the proportion of post-smolt escapees in this study was high compared to official escape statistics which include very few reports of young fish escaping.


INTRODUCTION
Atlantic salmon Salmo salar L. farming is a growing aquaculture industry, with a total global production of 1.426 million t in 2010 (FAO 2012).Escapes of farmed fish have raised concerns regarding their potential impacts on the environment.Escaped salmonids may spread diseases (Johansen et al. 2011) and parasites such as salmon lice (Heuch & Mo 2001, Skilbrei 2012).If they enter rivers to spawn (Saegrov et al. 1997), they may compromise the genetic integrity of local wild salmon populations (Crozier 1993, Clifford et al. 1998, Skaala et al. 2006, Glover et al. 2013).Young farmed salmon that escape from net pens in the sea as smolts or post-smolts during spring and summer migrate rapidly towards the feeding areas of wild salmon in the open sea, returning after 1 to 3 yr to spawn (Skilbrei 2010a,b).The spawning behaviour and success of these escapees are believed to be similar to those of wild salmon, in contrast to the less optimal behaviour of adult fish that escape from the pens just prior to their entry into the river (Fleming et al. 1996(Fleming et al. , 1997)).
Do escaped smolts represent a high proportion of the escaped salmon that enter in rivers to spawn, or have most of the escapees escaped recently from net pens as adults?According to Norwegian escape statistics, < 4% of escapees escaped as smolts between 2005and 2011(Skilbrei et al. 2015)).However, escaped salmon can leave the net pens unreported and possibly also undetected by the fish farmer (Skilbrei & Wennevik 2006, Glover 2010, Zhang et al. 2013), and therefore estimates of actual escapes of both smolts and adults are significantly higher than reported numbers (Skilbrei et al. 2015).Knowledge of the escape history of these salmon is important for risk assessments and for the development of strategies to prevent future escapes and reduce their ecological impacts.
A suitable method of monitoring the escape history of Atlantic salmon is needed.Reading scales is an effective means of distinguishing between escaped farmed and wild salmon due to the much faster growth rate of farmed juvenile salmon in freshwater (Lund & Hansen 1991), but the age and time of escape of farmed fish is often difficult to determine from scales (Erkinaro et al. 2010).Scale analysis depends on the assumption that farmed salmon exhibit a seasonal rhythm in growth similar to that of wild salmon.However, this interpretation may be obscured by aquaculture practices.Smolt development can be affected by photoperiod manipulations in the hatcheries, and smolts are transferred to net pens in the sea at various sizes from early spring to late autumn (Berge et al. 1995, Duston & Saunders 1995, Skilbrei 2013).Furthermore, outbreaks of disease on farms may influence growth patterns (Damsgård et al. 1998, McLoughlin & Graham 2007).Complementary methods to scale readings are therefore required.
Genetic methods have been used to trace escaped farmed salmon to the fish farm of origin (Glover et al. 2008, Glover 2010).This approach uses genetic samples from fish farms for comparisons with recaptured escaped salmon, but has primarily been used to trace recently escaped salmon and does not provide information about the time of escape.Several methods have been used to discriminate between wild and farmed salmon, such as fatty acid (FA) distributions (Axelson et al. 2009), chemical profiling with elemental analysis (Anderson et al. 2010), 13 C nuclear magnetic resonance (Aursand et al. 2009), and the use of stable isotopes (Dempson & Power 2004, Schröder & de Leaniz 2011).These methods all depend on differences in the composition of the food consumed by wild and farmed fish.Therefore, they have the potential to be used to study the escape history of farmed fish in more detail, but have not yet been fine-tuned enough to do so.
The use of FA profiles can be employed to distinguish between salmon that escaped early in their life versus as adults, because of the use of terrestrial lipids in salmon feed.These terrestrial lipids are low in typical marine long-chain polyunsaturated FAs (PUFAs) and high in medium-chain PUFAs 18:2n-6 and 18:3n-3 (Olsen et al. 2013).The depot lipids (triacylglycerols, TAG) mirror the composition of the feed and give the best resemblance to dietary history compared with analysis of total lipids, especially in mature salmon with low lipid content (Olsen et al. 2013).The content of the TAG FA 18:2n-6 in farmed salmon is typically above 10%.In contrast, the level of 18:2n-6 was low (< 2.5%) in sea-ranched salmon in Iceland (Jonsson et al. 1997) and in farmed salmon released as smolts and recaptured in Norway after 1 to 3 yr at sea (Olsen & Skilbrei 2010).
Quantifying tissue lipids may therefore provide a good estimate of an individual's feeding history, at least if the intention is to distinguish between escapes of young fish that have grown to adult size in the Norwegian Sea and recently escaped adult salmon.However, salmon of any size may escape during the production cycle from young to adult.If a fish escapes at an intermediate size and then switches to natural prey, it could be expected to present a more diffuse lipid profile.Therefore, to estimate the size at escape, we need to estimate the elimination rate of terrestrial FAs after the switch to natural prey.
We tested the use of lipid profiling to monitor the escape histories of farmed salmon captured in salmon rivers.We also report results of a laboratory experiment that examined the development of the FA profiles of farmed salmon after replacement of the terrestrial lipids in their feed with fish oils, in order to estimate the size of the escapee at the time of the switch from food pellets to natural prey.

Feed experiment
Feed preparation.A total of 3 experimental terrestrial diets based on linseed, rapeseed and soya oil (produced by Nofima) were fed to the fish during a priming phase (Table 1).Thereafter, all fish were fed the same fishmeal/fish oil-based diet.All diets were made using double-screwed extrusion with similar standards and additives as commercial diets.The initial pellets (4.5 mm) had a lipid content of 12% in order to ensure that the requirements of essential FAs were met.Thereafter, 22% of the various terrestrial oils were coated using a vacuum.The fishmeal/ fish oil-based diet (8 mm) was produced in a similar way, and were based on the FA composition in wild-caught salmon (Olsen & Skilbrei 2010).The chemical compositions of the 2 diets used are shown in Table 1.
Fish.In June 2011, 210 Atlantic salmon Salmo salar (NLA strain; Norwegian breeding programme) were transferred to six 1.5 × 1.5 × 1 m 3 standard fiberglass tanks equipped with feed collectors, supplied with aerated seawater (mean ± SD 9 ± 0.5°C) and kept under a 12 h light:12 h dark daylight regime at the Matre Re search Station, Norway.The fish were weighed (mean ± SD: 304 ± 50 g) 1 d before the start of the experiment (29 July 2011).The fish were fed the 3 experimental terrestrial diets in duplicate tanks every day using disc feeders from 08:00 to 12:00 h from 23 July to 24 October 2011 (92 d).Thereafter, all fish were fed the marine diet until the end of the experiment on 11 July 2012, at which point there were 128 fish remaining in the experiment.
All fish were measured for length and weight on 29 July 2011, 24 October 2011, 17 to 21 February 2012 and 11 July 2012.Tissue samples were taken from 4 to 6 individuals in each treatment group (2-4 tank −1 ) on 24 October 2011, 24 November 2011, 6 January 2012, 17 February 2012, 25 April 2012 and 11 July 2012.The fish were first anaesthetized with a strong solution of MS222 and then killed by a stroke to the head.On 20 February, all fish were tagged with T-bar tags of different colours (according to treatment group), and transferred to a 5 m diameter fiberglass tank for the remainder of the experiment.During sampling, they were anaesthetized in a 0.4% solution of MS222 and measured to the nearest 0.1 g and 1 mm.

Capture of escaped farmed salmon
Escaped farmed salmon were collected from 6 rivers that drain into the Hardangerfjord basin (Table 2).This fjord system houses a large aquaculture industry (Fig. 1) that produced 80 000 t of salmon in 2014 (Skaala et al. 2014).There is a high incidence of escaped farmed sal mon in the spawning population of these rivers.The River Etne supports by far the largest wild salmon stock in the region, and it also at tracts the highest numbers of esca pees (Vollset et al. 2014).
Escaped salmon were captured in the River Etne from 27 September to 10 November 2011, and in 2 bag-nets operating near the River Etne estuary from 6 August to 15 October 2011 by a fishery intended to target and remove escaped salmon.Wild salmon unintentionally taken in the bag-nets were used as controls for the FA analysis.Escaped salmon and some wild salmon were also captured in the Rivers Steinsdal, Opo and Kinso (see Fig. 1) from 5 September to 16 November 2011.The fish were categorized as wild or farmed salmon according to the scale-reading method described by Lund & Hansen (1991) and Fiske et al. (2005).The most important criteria for the identification of escaped farmed salmon were smolt size, smolt age and transition from fresh to salt water.In addition, a detailed analysis of sea (winter) bands of a sample of 61 escaped farmed salmon was performed.It was assumed that the presence of irregularities in sea growth (i.e. a period of reduced growth that did not look like a winter band), was indicative of the time of escape.The fish were grouped into the following 4 categories: (1) escaped early in life (i.e. of smolt-like size), ( 2) recently escaped, (3) escaped at an intermediate size (be tween smolt size and size at capture) and (4) recent/intermediate escape group.

Escape event with large broodstock salmon
In late September 2011, a fish farm reported to the authorities that 1937 broodstock salmon with a mean size of ~10 kg (largest individuals being almost 16 kg) had escaped from a fish farm in the fjord basin.The swimming distance from the broodstock farm to the River Etne is ~18 km.According to the fish farmer, the fish had been fed standard fish food until De cember 2010, and thereafter Skret ting Vitalis SA until 25 July 2011, when feeding was stopped due to the low appetite of the maturing fish.Skretting Vitalis SA is a specialized broodstock feed with a high content of marine fish oils and a relatively low proportion of terrestrial ingredients, which the producer recommend during the final 9 to 12 mo before spawning.

FA analysis
In accordance with the recommendations of Olsen et al. ( 2013), the adipose fin was used for TAG-derived FA profiling of the river-caught salmon.A total of 251 escapees and 42 wild salmon were analysed, and 14 samples of food pellets from different producers collected at various fish farms were analysed for comparison.Lipids were extracted from the adipose fin (field trial) or white muscle posterior to the pectoral fin (feeding experiment) of individual fish according to Folch et al. (1957), and stored under nitrogen at −80ºC until analysis.TAGs were separated from the total lipid by high-performance thin-layer chromatography (HPTLC) using the neutral solvent system of Olsen & Henderson (1989).All samples were methylated, and the respective fatty acid methyl esters (FAME) were analysed on a HP-7890A gas chromatograph (Agilent) with a flame ionization detector (GC-FID) as described previously in Olsen et al. ( 2004) for the feeding experiment, and in Meier et al. (2006) for the field trial.In total, 61 well-defined peaks in the chromatogram were selected, and identified by comparing retention times with a FAME standard (GLC-463 from Nu-Chek Prep) and retention index maps and mass spectral libraries (GC-MS) (www.chrombox.org/ index.html) performed under the same chromatographic conditions as the GC-FID (Wasta & Mjøs 2013).Chromatographic peak areas were corrected by empirical response factors calculated from the areas of the GLC-463 mixture.The chromatograms were integrated using the EZChrom Elite software (Agilent Technologies).The 36 FAs with a content higher than 0.2% of total FA in the samples were included in the analysis.

Relationships between fish size, diet and FA levels
The decline in terrestrial FAs after the diet change from the experimental terrestrial to the fish oil-based pellets was plotted and compared with the dilution model of Robin et al. (2003).According to these authors, the dilution of an initial quantity of a mixture of FAs by an increasing quantity of a new mixture can be logically described by the following equation for each FA: where Pi is the percentage of the FA i, with 0 for its original (initial) value and f for its final value.Q 0 represents the initial quantity of total FAs, and Q t the same at time t.By assuming that W t / W 0 (where W is the weight of the fish) can be used as an approximation for Q t / Q 0 , and presenting Pi t as relative values from Pi 0 = 100% to Pi f = 0%, the formula for the dilution model simplifies to: Based on the weight and percentage of the FA 18:2n-6 of escaped farmed salmon at the time of river capture (i.e.W t and P t ), a logarithmic function between fish size and relative content of 18:2n-6 derived from the diet experiment (see 'Results') was applied to estimate the size of escaped farmed sal mon at the time of switch from commercial pellets to natural food (i.e.W 0 ).In the model, 18:2n-6 = 12% of the total FA corresponds to P 0 (recalculated to 100% in the model), and 18:2n-6 = 2.5% corresponds to P f (0% in the model).18:2n-6 = 2.5% was chosen because this appeared to be close to the maximum level in wild fish, and is used as an estimate of the level in farmed escaped fish, where this FA has been 'washed out'.18:2n-6 = 12% represents a normal level of this FA in salmon food pellets, and therefore estimates the level of this FA at the time of escape.

Classification of time of escape of farmed salmon
By taking scale readings and the level of the FA 18:2n-6 into account, the fish were classified into 4 main categories: (1) wild salmon, (2) farmed sal mon believed to have escaped at an early age, with levels of 18:2n-6 < 2.5%, (3) recently escaped farmed salmon with 18:2n-6 > 7% and (4) fish with intermediate levels of 18:2n-6 (i.e. between 2.5 and 7%) that are believed to have escaped between the 2 former stages.

Statistics
The mean weights of the 3 terrestrial diet groups and differences between tanks were tested using general linear model (GLM) and Newman-Keuls multiple test (Statistica v.12, StatSoft).There were no significant effects of Tank during the first part of the experiment until all fish were reared in 1 common tank, so Tank was not included in the further comparisons.Normality was checked by normal pro bability plots.FA compositions of the tank trial were analysed using GLM (SPSS v.21) and Tamhane's post hoc test assuming unequal variances.Prin cipal Component Analysis (PCA) was carried out using Sirius v.8.1 (PRS AS).Before PCA, the relative values (i.e.percent of the sum) were scaled by dividing each value by the mean of the values of all samples for that particular FA, with the intention of leveling out the quantitative difference among the FAs, leaving them all to vary around 1. All FAs > 0.2% were included in the PCA (36 FAs).Score and loading plots from PCA analysis were generated in Sirius.

Feed experiment
The fish grew from an average (± SD) weight of 304 ± 50 g in July to 844 ± 131 g in October, when the diets were changed to a fully marine fish meal and oil-based diet.From October until the end of the experiment in July, the fish grew to an average weight of 3359 ± 591 g (Fig. 2).There were no significant differences be tween the mean sizes of the 3 terrestrial diet groups at the start of the experiment (GLM; F = 1.18, p = 0.31), in February (F = 0.08, p = 0.93) or at the end of the experiment in July (F = 1.21, p = 0.30) (Newman-Keuls tests; p > 0.1).
Feeding the marine-based oil to the fish reduced the level of terrestrial FAs as the size of the fish increased (Fig. 2).At the end of the trial, all the FA profiles were similar (Table 4), although there were small but significant differences between the levels of 18:1n-9, 18:3n-3 and 18:2n-6 (Table 4).The de crease in 18:2n-6 and 18:3n-3 did not completely match a theoretical dilution model (y = 100 /x, simplified from Robin et al. 2003; see 'Materials and methods'), which predicts how levels decline when fish increase in biomass by consuming feed with a lower content of these FAs (Fig. 3).The decline was more rapid, and was fitted by the logarithmic function y = 4.1583 − 1.5871× log 10 (x).

FA profiles of salmon caught in rivers
The levels of 18:2n-6 ranged from 1 to 12%.The distribution of 18:1n-9 produced a similar picture but displayed larger variability than 18:2n-6 among wild fish and early and intermediate escapees (Figs. 4 & 5).In addition to escaped farmed salmon believed to have escaped early, late or at an intermediate stage of life, an additional group of large sized salmon with 18:2n-6 between 6 and 8% were present (Table 5).They were assumed to have come from the escape of large broodfish that had been fed with a broodstock feed with a lower content of terrestrial oils (Table 6).
Both the differences and similarities between the categories of fish were visualized by including all measured FAs > 0.2% (n = 36) in a PCA.The 2 groups of recently escaped adult fish were clearly distinguishable from the wild fish and early escapees along principal component 1 (PC1), which explained 58% of the variation in the data (Comp. 1 in Fig. 6).
Most of the intermediate escapees were estimated to have been between 1 and 3 kg when they switched from commercial food pellets to natural feed (see 'Material and methods' for calculations), and between 3 and 11 kg when captured in rivers (Fig. 9).

FA profiles of food pellets
The content of FAs like 18:1n-9 and 18:2n-6 were high, but variable, in food pellets from different producers obtained from fish farms.The content of the typical terrestrial lipids were lower in food for small fish (smolts) and broodfish (Table 6).All detected FAs in the food are shown in Table S2 in the Supplement at www.int-res.com/articles/suppl/q007p001_supp.pdf.

DISCUSSION
This study demonstrates that analysis of FAs is a powerful tool to monitor the escape history of escaped farmed Atlantic salmon, enabling us to distinguish between fish that escaped at different stages during the production cycle, from the juvenile and post-smolt stages to large adult salmon.It was possible to observe the effects of the initial priming of the FAs up to 9 mo (end of the trial), or until the fish had grown to 4 times their initial weight, provided the FAs had high initial values.The pattern of disappearance of 18:1n-9 was different from that of 18:2n-6 and 18:3n-3.The high background level of 18:1n-9 is most likely due to adjustment of the synthesis or the metabolism since 18:1n-9 is an important product of the de novo synthesis being deposited in salmonid tissues (Olsen et al. 1991).
Normally, it is assumed that FA oxidation for energy is directly fuelled by extracellular FAs with only minimal contribution from the intracellular TAG pool (Gibbons et al. 2000).This means that the level of plant FAs should decrease in fish on a marine diet as a function of dilution.This was, however, not the case in the present study, were the FAs 18:2n-6 and 18:3n-3 disappeared at a rate ca.30% faster than expected from a pure dilation model (Robin et al. 2003).This indicates some TAG turnover despite being nutritional sufficient, and probably reflects the fact that hormone selective lipase is active in adipose tissues and is regulated by factors other than diet alone (Weil et al. 2013).Observations have also suggested that these particular FAs are reduced at a higher rate than other FAs following dilution, indicating selective catabolism from the tissue (Robin et al. 2003, Jobling 2003, 2004).This is also evident in some salmon trials that used finishing diets (Bell et al. 2003).The dilution model may overestimate their content in the tissues, as was the case in the present study.We therefore used a modification of the model, i.e. the logarithmic function fitted to our data when esti mating size at diet change of river-captured escaped farmed salmon.
Most of the escaped farmed salmon had either escaped early in life or relatively recently as large salmon weighing 2 to 14 kg.Only 15% appear to have escaped after the post-smolt stage, then switched to natural prey and continued growth until maturity.This is in accordance with the following tagging/ release experiments.Post-smolts that escape during their first summer in net pens in seawater may return to freshwater to spawn after 1 to 3 yr at sea (Skilbrei 2010a,b), with much higher long-term survival rates For these reasons, it has been suggested that fully grown salmon have a limited ability to switch to natural prey if released, that immature fish will suffer from high mortality, and that fish that have started the sexual maturation process at the time of escape will have a much higher chance of surviving until they can enter a river to spawn, normally in the same year that they escaped (Skilbrei et al. 2015).The present study supports these suggestions.
FA profiles of large salmon showing intermediate levels of lipids characteristic of terrestrial oils suggest that some fish that escaped after the post-smolt stage were able to switch to natural food.It is not known whether these fish switched rapidly to natural prey, but be havioural and other data suggest that the time of year of escape may influence their ability and opportunity to do so.Escaped brood fish 18:2n-6 between 6 and 8%, 7.14 ± 0.51 10.3 ± 1.9 57 weight > 8 kg Recently escaped adults 18:2n-6 > 7% 10.56 ± 0.93 4.6 ± 1.6 96 Intermediate escapees 18:2n-6 between 2.5 and 7% 4.35 ± 1.33 5.3 ± 1.9 37 Escaped as young 18:2n-6 < 2.5% 1.58 ± 0.31 4.9 ± 1.7 61 Wild salmon Based on scales 1.48 ± 0.31 4.4 ± 1.3 42 Table 5. Salmo salar.Mean (± SD) fatty acid (FA) composition and weight of wild and escaped farmed salmon.Classification of the stage at which salmon was assumed to have escaped was based on levels (%) of the FA 18:2n-6 and weight.See 'Materials and methods' for additional information about the escaped brood fish   2013).However, they may remain in fjords and coastal areas for some time; in one autumn release study, the released farmed salmon stayed in the vicinity of the fish farm and fed on excess food pellets for several months (Olsen & Skilbrei 2010).Under such circumstances, the switch to natural food will be delayed compared with the time of escape if these fish eventually learn to catch natural prey.
We determined that 24% of the escapees had escaped early in life, probably as smolts and postsmolts.This contrasts with official escape statistics, which report that very few smolts and post-smolts escape, and supports the suggestion of Skilbrei et al. (2015) that most smolt escape events go unnoticed and/or unreported.Their FA profiles were similar to those of wild salmon with respect to the low levels of the typical terrestrial FAs (18:2n-6, 18:1n-9 and 18:3n-3) and the high contribution of marine FAs such as long-chain MUFAs and long-chain n-3 PUFAs.
Fish from all groups were separated ac cording to the relative amount of the long-chain MUFAs (20:1and 22:1) or short-chain PUFAs (18:4n-1, 16:2n-4, 18:4n-3), and this indicated different food sources among in dividuals in the groups.Long-chain MUFAs such as 20:1n-9 and 22:1n-11 originate from zooplankton such as Calanus spp. in the North Atlantic Ocean (Albers et al. 1996, Falk-Petersen et al. 2000).These FAs are used as FA trophic markers (FATM) for fish foraging in the Calanus food web (Dalsgaard et al. 2003).The short-chain PUFAs produced by algae (16:2n-4 and 18:4n-1) are FATMs for diatoms, and 18:4n-3 is a FATM for dinoflagellates, and indicate a diet low in the food web (Dalsgaard et al. 2003).Thus, the FA profiles may be used to study diets and trophic interaction in early escaped and wild fish (Petursdottir et al. 2008), however that type of analysis is beyond the scope of the present study.The accuracy of FA profiling will depend on variability in the composition of different salmon feeds on the market.The feed samples we collected demonstrate that the content of lipids characteristic of terrestrial oils varies among producers, and in specialized feeds used for specific purposes.A higher content of marine oils in pellets given to juvenile fish and smolts during their adaptation to sea water will widen the difference between smolt and adult escapees, while the influence of terrestrial oil on large broodstock fish may be less than expected if the fish have been offered a special diet just prior to spawning.The availability of marine resources used for production of salmon food varies on the global market, so differences in the composition of the feed among producers and over time must be expected.Monitoring this variability in escapehistory studies is recommended, and might be helpful if FA profiling is used to identify the farm of origin of recent escape incidents (Martinez et al. 2009).
The mismatch between the results of scale readings and the use of FA levels to classify escaped farmed salmon according to stage at escape supports the view of Erkinaro et al. (2010) that the age and time of escape of farmed fish is often difficult to determine from scales alone.We conclude that FA profiling, when used in combination with scale analysis to determine the wild or farmed origin of the fish, is a promising method of studying the history of escaped salmon.This type of information is important for risk analysis and for implementing measures to prevent and mitigate the undesired impacts of farmed salmon escapes.For example, national regulations regarding operational procedures in aquaculture came into force in Norway in 2008 in order to minimize the risk of smolt escapes, since these fish are believed to have higher spawning success than recently escaped salmon (Fleming et al. 1996(Fleming et al. , 1997)).Although smolt or post-smolt escapes are not reflected to any significant degree in the escape statistics (Skilbrei et al. 2015, www.fiskeridir.no),our study indicates that the problem of early life-stage salmon escaping from fish farms has not been completely solved.
Atlantic salmon replace the original salmon stock in the River Vosso, western Norway.ICES J Mar Sci 54: 1166−1172 Schröder V, de Leaniz CG (2011)

Fig. 1 .
Fig. 1.Sites of salmon farming in Hardangerfjord basin (southwestern Norwegian coast) and surrounding areas (source: Norwegian Directorate of Fisheries), and location of rivers where samples of escaped farmed fish were collected

Fig. 2 .
Fig. 2. Salmo salar.(A) Weight of fish and (B-D) relative percentages of the 3 fatty acids (FAs) (B) 18:2n-6, (C) 18:1n-9 and (D) 18:3n-3 in salmon after the shift from the 3 initial feeds based on linseed, rapeseed and soybean oils to a common feed based on marine FAs (see 'Materials and methods' for full description).The content of the 3 FAs in the 3 original diets (large symbols) and in the marine diet (dotted lines) are included

Fig. 3 .
Fig. 3. Salmo salar.Relative change (%) from original level of the fatty acids (FAs) 18:2n-6 and 18:3n-3 with increased weight (y times initial weight, W 0 , on 24 October 2011) of the fish after the change of diet to marine feed; 0% relative FA is defined as the FA level of the marine diet.A logarithmic function is fitted to the data (solid line).A theoretic curve describing how the initial percentage of FAs would be lowered if its reduction had been caused by simple dilution is also shown (y = 100 /x; dashed line) (simplified from Robin et al. 2003, see 'Material and methods')

Fig. 6 .
Fig. 6.Salmo salar.Principal component analysis score plot of levels of fatty acids in triacylglycerols of wild salmon (black squares) and farmed salmon escaped recently as brood fish (green) and adults (blue), early in life (orange) or at an intermediate stage (red)

Acknowledgements.
Fig. 8.Comparison between the use of the fatty acid 18:2n-6 and scale readings to group 61 escaped farmed Atlantic salmon according to time of escape.Weight at capture was also used to classify the 'Broodfish' group.Numbers of individuals are given above bars Discrimination between farmed and free-living invasive salmonids in Chilean Patagonia using stable isotope analysis.Biol Invasions 13: 203−213 Skaala O, Wennevik V, Glover KA (2006) Evidence of temporal genetic change in wild Atlantic salmon, Salmo salar L., populations affected by farm escapees.ICES J Mar Sci 63: 1224−1233 Skaala O, Johnsen GH, Lo H, Borgstrøm R and others (2014) A conservation plan for Atlantic salmon (Salmo salar) and anadromous brown trout (Salmo trutta) in a region with intensive industrial use of aquatic habitats, the Hardangerfjord, western Norway.Mar Biol Res 10: 308−322 Skilbrei OT (2010a) Reduced migratory performance of farmed Atlantic salmon post-smolts from a simulated escape during autumn.Aquacult Environ Interact 1: 117−125 Skilbrei OT (2010b) Adult recaptures of farmed Atlantic salmon post-smolts allowed to escape during summer.Aquacult Environ Interact 1: 147−153 Skilbrei OT (2012) The importance of escaped farmed rainbow trout (Oncorhynchus mykiss) as a vector for the salmon louse (Lepeophtheirus salmonis) depends on the hydrological conditions in the fjord.Hydrobiologia 686: 287−297 Skilbrei OT (2013) Migratory behaviour and ocean survival of escaped out-of-season salmon smolts.Aquacult Environ Interact 3: 213−221 Skilbrei OT, Jørgensen T (2010) Recapture of cultured salmon following a large-scale escape experiment.Aquacult Environ Interact 1: 107−115 Skilbrei OT, Wennevik V (2006) The use of catch statistics to monitor the abundance of escaped farmed salmon and rainbow trout in the sea.ICES J Mar Sci 63: 1190−1200 Skilbrei OT, Holst JC, Asplin L, Mortensen S (2010) Horizontal movements of simulated escaped farmed Atlantic salmon (Salmo salar L.) in a western Norwegian fjord.ICES J Mar Sci 67: 1206−1215 Skilbrei OT, Skulstad OF, Hansen T (2014) The production regime influences the migratory behaviour of escaped farmed Atlantic salmon.Aquaculture 424−425: 146−150 Skilbrei OT, Heino M, Svåsand T (2015) Using simulated escape events to assess the annual numbers and destinies of escaped farmed Atlantic salmon of different life stages from farm sites in Norway.ICES J Mar Sci 72: 670−685 Solem Ø, Hedger RD, Urke HA, Kristensen T, Økland F, Ulvan EM, Uglem I (2013) Movements and dispersal of farmed Atlantic salmon following a simulated-escape event.Environ Biol Fish 96: 927−939 Soto D, Jara F, Moreno C (2001) Escaped salmon in the inner seas, southern Chile: facing ecological and social conflicts.Ecol Appl 11: 1750−1762 Vollset KW, Skoglund H, Barlaup BT, Pulg U and others (2014) Can the river location within a fjord explain the density of Atlantic salmon and sea trout?Mar Biol Res 10: 268−278 Wasta Z, Mjøs SA (2013) A database of chromatographic properties and mass spectra of fatty acid methyl esters from omega-3 products.J Chromatogr A 1299: 94−102 Weil C, Fefevre F, Bugeon J (2013) Characteristics and metabolism of different adipose tissues in fish.Rev Fish Biol Fish 23: 157−173 Zhang Z, Glover KA, Wennevik V, Svåsand T and others (2013) Genetic analysis of Atlantic salmon captured in a netting station reveals multiple escapement events from commercial fish farms.Fish Manag Ecol 20: 42

Table 3 .
Percentage of total fatty acids (%) of the terrestrial diets (Lin: linseed oil; Rap: rapeseed oil; Soy: soybean oil) fed to Atlantic salmon from July to October, and the marine diet fed from October to July.SFA: saturated fatty acid; MUFA: monounsaturated fatty acid; PUFA:

Table 6 .
Content of selected fatty acids (FAs) (> 4%) in salmon food pellets obtained from fish farms between 2011and 2013.All FAs are presented in TableS2in the Supplement at www.int-res.com/articles/suppl/q007p001_supp.pdfralprey to survive.Most experimental releases during autumn have demonstrated that the fish disperse relatively rapidly from the escape location(Skil brei et  al. 2010, Skilbrei & Jørgensen 2010, Solem et al.