Aquaculture has expanded rapidly over the past decades with an average growth rate of 8·8 % per year since 1970 compared with only 1·2 % for capture fisheries(1). This increase in fish production has led to an increase in aquafeed production concurrent with a greater demand for fish oil (FO) and fishmeal(Reference Naylor, Goldburg and Primavera2, Reference Miller, Nichols and Carter3). Consequently, the rise in FO demand from aquafeed industries has added further pressure on wild fisheries, which are generally considered to be finite, fully exploited and at times unpredictable due to El Niño events(Reference Naylor, Goldburg and Primavera2, Reference Naylor, Hardy and Bureau4). It is predicted that the future needs of the aquaculture industry for FO will outstrip the current supply within the next decade(Reference Pickova and Morkore5). In addition to the predicted shortfall in FO supply, there has been concern about the levels of dioxins and dioxin-like poly-chlorinated biphenyls in some FO depending on the source fishery(Reference Jacobs, Covaci and Schepens6, Reference Bell, McGhee and Dick7), which presents a potential health hazard. Therefore, the aquaculture industry is faced with a major challenge in finding suitable oil sources for the replacement of FO(Reference Miller, Nichols and Carter3, Reference Naylor, Hardy and Bureau4).
In an effort to sustain Atlantic salmon aquaculture, a wide variety of commercial vegetable oils (VO) have been investigated as FO replacements(Reference Miller, Nichols and Carter3, Reference Turchini, Torstensen and Ng8). The use of VO rarely affects fish growth performance(Reference Bell, Henderson and Tocher9–Reference Torstensen, Froyland and Lie11). However, the low levels of n-3 long-chain ( ≥ C20) PUFA (n-3 LC-PUFA), in particular EPA and DHA, in fish fed on VO remain a major shortcoming. Generally with increasing increments of VO in diets, there has been a corresponding decrease in n-3 LC-PUFA content in fish. In Atlantic salmon fed on 100 % VO, flesh EPA and DHA levels were reduced to 30 and 35 %, respectively(Reference Bell, Tocher and Henderson12), because oils derived from vegetable sources lack n-3 LC-PUFA, and the capacity for fish, especially of marine origin, to endogenously biosynthesise n-3 LC-PUFA from the VO substrates is limited(Reference Miller, Bridle and Nichols13). In addition, VO are usually characterised by high levels of n-6 PUFA and low n-3:n-6 ratios, hence feeding diets rich in VO has the potential to reduce the important health benefits derived from the consumption of n-3 LC-PUFA obtained by eating fish and other seafoods(Reference Simopoulos14, Reference Ruxton, Reed and Simpson15).
So far, there has been only one report of the endogenous biosynthesis of n-3 LC-PUFA from the metabolic precursors capable of matching levels of EPA and DHA present in fish fed a FO diet. The flesh of Atlantic salmon parr (in freshwater) fed Echium oil (EO) rich in stearidonic acid (SDA; 18 : 4n-3), a precursor of EPA, contained comparable levels of EPA and DHA with that obtained using a FO diet(Reference Miller, Nichols and Carter16). However, in a follow-up trial with smolt, high levels of n-3 LC-PUFA usually found in seawater Atlantic salmon fed diets rich in FO were not attained via biosynthesis from precursors in the EO diet(Reference Miller, Bridle and Nichols13), though gene expression of Δ5 desaturase and elongase enzymes in the liver was up-regulated(Reference Miller, Bridle and Nichols13). Nutritional history might be important, and the smolts had not been fed SDA-rich diets before the EO diet(Reference Miller, Bridle and Nichols13). In the present study, we have attempted to determine whether feeding EO from parr to smolt would result in increased n-3 LC-PUFA biosynthesis. A whole-body fatty acid (FA) mass balance (FAMB) approach(Reference Turchini, Francis and De Silva17–Reference Turchini and Francis19) was used to investigate the metabolism of individual FA along the n-3 pathway.
In the present study, three diets were formulated to compare rapeseed oil (RO), EO and FO (Table 1). Fishmeal (Skretting Australia, Cambridge, TAS, Australia) was defatted using a mixture of hexane and ethanol (400 ml/100 ml fishmeal). EO was provided as Crossential SA14 (Croda Chemicals, East Yorkshire, UK). FO was from Jack mackerel, Trachurus symmetricus L. (Skretting Australia), and a domestic RO was used (Steric Trading Pty Limited, Villawood, NSW, Australia). The diets were manufactured using a California Pellet Mill (CL-2), dried and stored at 5°C(Reference Carter, Bransden and Lewis20).
CMC, carboxymethyl cellulose; ALA, α-linolenic acid; SDA, stearidonic acid; LA, linoleic acid; GLA, γ-linolenic acid.
* Skretting Australia, Cambridge, TAS, Australia.
† MP Biomedicals Australasia Pty Limited, Seven Hills, NSW, Australia.
‡ Starch Australasia, Lane Cove, NSW, Australia.
§ Hamlet Protein A/S, Horstens, Denmark.
∥ Croda Chemicals, East Yorkshire, UK.
¶ Steric Trading Pty Limited, Villawood, NSW, Australia.
** Penford Limited, Lane Cove, NSW, Australia.
†† Vitamin mix (ASV4) supplied per kg of feed: 2·81 mg thiamin HCl, 1·0 mg riboflavine, 9·15 mg pyridoxine HCl, 25 mg nicotinic acid, 54·32 mg calcium d-pantothenate, 750 mg myo-inositol, 0·38 mg d-biotin, 2·5 mg folic acid, 0·03 mg cyanocobalamin, 2·8 mg retinol acetate, 0·1 mg cholecalciferol, 250 mg α-tocopherol acetate, 5 mg menadione sodium bisulphate (Sigma-Aldrich, Castle Hill, NSW, Australia) and 100 mg Roche Rovimix E50.
‡‡ Mineral mix (TMV4) supplied/kg of feed: 117 mg CuSO4.5H2O, 7.19 mg KI, 1815 mg FeSO4.7H2O, 307 mg MnSO4.H2O, 659 mg ZnSO4.7H2O, 329 mg Na2SeO3, 47.7 mg CoSO4.7H2O (Sigma-Aldrich).
§§ Roche Vitamins Australia, Frenchs Forest, NSW, Australia.
¶¶ Degussa, Frankfurt, Germany.
The experiment was conducted at the University of Tasmania (Launceston, TAS, Australia) in accordance with the University of Tasmania Animal Ethics guidelines (Investigation A0009731). Atlantic salmon (Salmo salar L.) parr (approximately 25 g) were obtained from Wayatinah Salmon Hatchery (SALTAS, TAS, Australia) and acclimatised for 14 d in a 3000 litre partial recirculation system. Fish were maintained on the FO diet (Table 1) before starting the experiment.
At the start of the experiment, fish were anaesthetised (benzocaine, 50 mg/l) and weighed, and fork length was measured. For the measurement of initial FA and chemical composition of the whole carcass, eight fish were euthanised (100 mg/l) and stored at − 20°C. The experiment used a partial recirculation system, equipped with a protein skimmer and physical, UV and biological filters(Reference Bransden, Carter and Nichols10). Water temperature was kept constant at 15°C with continuous daily replacement of approximately 15 % volume. Dissolved O2, pH, NH3, nitrate and nitrite were monitored daily to ensure that water quality remained within the parameters recommended for Atlantic salmon(Reference Wedemeyer21).
A total of forty-four fish were randomly allocated into each of twelve 300 litre tanks, and the three diets were hand-fed in quadruplicate at a fixed ration of 2·0 % body weight/d in two equal rations. Every 2 weeks, feed intake was monitored to adjust the feeding ration. Since experimental fish had missed the natural window for smolting, photoperiod was manipulated following normal commercial procedures to trigger smoltification. At 28 d intervals, fish were bulk weighed, and sixteen fish per treatment were weighed, and fork length was measured for the calculation of condition factor (K). Based on physical characteristics of fish undergoing parr–smolt transformation such as silvering of the body, loss of parr marks and darkening of fin margins(Reference McCormick, Moriyama and Björnsson22), fish were transferred to seawater at 112 d. Before seawater transfer, fish were bulk weighed, and two fish per tank were euthanised (benzocaine 100 mg/l) and stored at − 20°C for FA and chemical composition analyses of whole carcass. At seawater transfer, one FO replicate tank had less fish than expected (due to escapees). Data from this tank were omitted from analysis due to the different feeding history. After 7 d in seawater, fish were randomly culled to twenty-four fish per tank, and blood from five fish per treatment was taken from below the anal fin with a heparinised syringe for the measurement of plasma osmolality on a Vapro 5250 vapour pressure osmometer (Wescor® Inc., Logan, UT, USA) to confirm smolt status of fish. Blood plasma osmolality values for all groups (314–358 mOsmol/kg) were within the range considered to be normal for Atlantic salmon smolts(Reference Arnesen, Johnsen and Mortensen23).
At the end of the experiment (196 d), fish were bulk weighed, and three fish per tank were euthanised (benzocaine 100 mg/l) and stored at − 20°C for whole-carcass FA and chemical composition analyses.
Specific growth rate was calculated as specific growth rate (%/d) = 100 × (ln BWf/ln BWi)/d, where BWf and BWi are final and initial wet weights (g), respectively, and d is the number of days of the experiment. Feed consumption was calculated as the total average amount of feed (g) consumed/fish over the number of days of the experiment. Feed efficiency ratio was calculated as feed efficiency ratio (g/g) = total weight gain (g)/feed consumption (g). K was calculated as K (%) = 100 × (BW/FL3), where FL is the fork length (cm).
Diets included yttrium oxide (1 g/kg) as a digestibility marker(Reference Carter, Lewis and Nichols24). On days 108–111 (freshwater phase), faecal samples from all tanks were collected from faecal settlement collectors (Guelph system) attached to the tanks between 11.00–17.00 and 19.00–09.00 hours(Reference Carter, Lewis and Nichols24, Reference Ward, Carter and Townsend25). At the end of the experiment (seawater phase), fish were stripped for collection of faeces(Reference Percival, Lee and Carter26). Faecal samples were freeze-dried before chemical analysis. Apparent digestibility was calculated using the standard formula: apparent digestibility (%) = 100 − (100 (Y diet/Y faeces) × (FAfaeces/FAdiet)), where Y is the percentage of yttrium oxide and FA is the percentage of particular FA(Reference Maynard and Loosli27).
Standard methods were used to determine DM (freeze dry to constant weight followed by drying at 135°C for 2 h)(28); total lipid(Reference Bligh and Dyer29), nitrogen (Kjeldahl using Se catalyst; crude protein was calculated as N × 6·25), energy (bomb calorimeter; Gallenkamp Autobomb, Loughborough, Leics, UK, calibrated with benzoic acid) and ash were measured by combustion at 600°C for 2 h(28). Apart from DM, freeze-dried samples were used for chemical analyses and corrected for DM.
Lipid extraction and isolation
Whole carcass and faecal samples were freeze-dried and extracted overnight using a modified Bligh & Dyer protocol(Reference Bligh and Dyer29). This involved a single-phase extraction using CHCl3–MeOH–H2O (1:2:0·8, by vol.), followed by phase separation to yield a total lipid extract.
An aliquot of the total lipid extract was transmethylated in MeOH–CHCl3–HCl (10:1:1, by vol.) for 2 h at 100°C. After addition of Milli-Q water (1 ml), the mixture was extracted with hexane–chloroform (4:1, v/v) to obtain FA methyl esters. Samples with an internal injection standard (19 : 0 FA methyl esters) added were analysed by GC using an Agilent Technologies 7890B GC (Palo Alto, CA, USA) equipped with an Equity™-1 fused silica capillary column (15 m × 0·1 mm internal diameter and 0·1 μm film thickness), a flame ionisation detector, a split/splitless injector and an Agilent Technologies 7683 B Series autosampler. Helium was used as the carrier gas. Samples were injected in splitless mode at an oven temperature of 120°C. After injection, oven temperature was raised to 270°C at 10°C/min and finally to 310°C at 5°C/min. Peaks were quantified with Agilent Technologies ChemStation software. GC results are typically subject to an error of up to ± 5 % of individual component area.
Individual components were identified by mass spectral data and by comparing retention time data with authentic and laboratory standards. GC-MS analyses were performed on a Finnigan ThermoQuest GCQ GC-MS fitted with an on-column injector and using ThermoQuest Xcalibur software (Austin, TX, USA). The GC was equipped with an HP-5 cross-linked methyl silicone-fused silica capillary column (50 m × 0·32 mm internal diameter). Helium was used as the carrier gas, with operating conditions as described previously(Reference Miller, Nichols and Barnes30).
Fatty acid mass balance
A whole-body FAMB was performed on the n-3 biosynthetic pathway to compare individual FA appearance or disappearance and the accretion of individual n-3 PUFA as described previously(Reference Turchini, Francis and De Silva17).
Values are reported as means with their standard errors. Normality and homogeneity of variance were confirmed, and the percentage data were arcsine-transformed before analysis. Comparison between treatments for FA, growth performance, osmolality, K and mass balance means was done by one-way ANOVA, followed by multiple comparison using Tukey–Kramer's honestly significant difference wherever applicable. Significance was accepted at P ≤ 0·05. Statistical analysis was performed using SPSS for Windows version 16.0 (SPSS, Chicago, IL, USA).
There were no significant differences (P ≥ 0·05) in blood plasma osmolality of fish fed on all diets 7 d post-transfer to seawater. Mean K for all groups was similar during the short-day period, but shortly after switching to continuous light, RO and EO fish displayed higher K compared with FO fish before seawater transfer (P < 0·05; Fig. 1). After seawater transfer, elevated K was observed particularly for RO fish. The pattern was similar for all groups and was typical for smolting fish, with a steady drop in K after onset of continuous light, which was carried on for 28 d in seawater before increasing thereafter.
Growth (final weight, weight gain and specific growth rate) and efficiency were higher (P < 0·05) for RO fish compared with EO fish in freshwater, whereas FO fish did not differ significantly (P ≥ 0·05) in performance and efficiency from the other two treatments. Feed consumption and survival were not significantly different (P ≥ 0·05) between treatments (Table 2).
SGR, specific growth rate; FC, feed consumption; FER, feed efficiency ratio.
a,b Mean values within a column with unlike superscript letters belonging to either freshwater or seawater were significantly different (P < 0·05).
Growth (final weight, weight gain and specific growth rate) was higher for RO fish compared with fish fed on the EO and FO diets in seawater, with no difference (P ≥ 0·05) in growth performance for EO and FO fish (Table 2). Survival was higher in RO fish compared with FO and EO fish.
During the freshwater phase, feeds had no significant effect on carcass DM (pooled mean 326·1 (sem 2·5) g/kg), crude protein (159·5 (sem 1·1) g/kg, w/w), total lipid (124·4 (sem 3·4) g/kg, w/w) or ash (26·3 (sem 0·5) g/kg, w/w). During the seawater phase, feeds had no significant effect on carcass crude protein (pooled mean 146·6 (sem 1·0) g/kg, w/w), total lipid (120·2 (sem 2·7) g/kg, w/w) or ash (25·4 (sem 0·4) g/kg, w/w). Significant differences for carcass DM (P < 0·05) were obtained (EO 314·4 (sem 3·3); FO 320·8 (sem 4·6) and RO 328·0 (sem 3·6)), with EO fish having a lower DM than RO fish.
FO fish had higher whole-carcass EPA and DHA content than either EO or RO fish in freshwater. EO fish had significantly higher (P < 0·01) SDA, α-linolenic acid (ALA; 18 : 3n-3) and eicosatetraenoic acid (ETA; 20 : 4n-3) compared with fish fed on the other two diets (Table 3). Total n-3 for EO fish was comparable with FO fish, and total PUFA was higher for EO fish compared with FO fish, whereas for RO fish, both total n-3 and total PUFA were lowest (P < 0·01). The n-3:n-6 ratios for FO, EO and RO fish were in the order FO>EO>RO.
f, Mean sum of squares; ALA, α-linolenic acid; SDA, stearidonic acid; ETA, eicosatetraenoic acid; DPA, docosapentaenoic acid; LA, linoleic acid; GLA, γ-linolenic acid; ARA, arachidonic acid.
a,b,c Mean values across the row with unlike superscript letters were significantly different as determined by Tukey–Kramer's honestly significant difference (df = 3, P < 0·01).
† Includes 15 : 0, 17 : 0, 20 : 0, 22 : 0 and 24 : 0.
‡ Includes 16 : 1n-9, 16 : 1n-5, 18 : 1n-5, 20 : 1n-7, 22 : 1n-9, 22 : 1n-11 and 24 : 1n-9.
§ Includes 21 : 5n-3 and 24 : 6n-3.
∥ Includes 20 : 2n-6, 22 : 4n-6 and 24 : 5n-6.
¶ Includes 16 : 2n-4, 16 : 3n-4 and 18 : 2n-9.
FO fish had higher whole-carcass EPA and DHA content than either EO or RO fish in seawater. EO fish had significantly higher (P < 0·01) ALA, SDA and ETA compared with fish fed on the other two diets (Table 4). There was significantly higher (P < 0·05) EPA and docosapentaenoic acid (DPA; 22 : 5n-3) in EO fish compared with RO fish. Total n-3 and total PUFA were higher (P < 0·01) for EO fish compared with FO and RO fish. The n-3:n-6 ratios for FO, EO and RO fish were in the order FO>EO>RO. Total FA was significantly higher (P < 0·05) in RO fish compared with EO and FO fish.
f, Mean sum of squares; ALA, α-linolenic acid; SDA, stearidonic acid; ETA, eicosatetraenoic acid; DPA, docosapentaenoic acid; LA, linoleic acid; GLA, γ-linolenic acid; ARA, arachidonic acid;
a,b,c Mean values across the row with unlike superscript letters were significantly different as determined by Tukey–Kramer's honestly significant difference (df = 3, P < 0·01, *P < 0·05).
† See Table 1 for diet definitions.
‡ Includes 15 : 0, 17 : 0, 20 : 0, 22 : 0 and 24 : 0.
§ Includes 16 : 1n-9, 16 : 1n-5, 18 : 1n-5, 20 : 1n-7, 22 : 1n-9, 22 : 1n-11 and 24 : 1n-9.
∥ Includes 21 : 5n-3 and 24 : 6n-3.
¶ Includes 20 : 2n-6, 22 : 4n-6 and 24 : 5n-6.
** Includes 16 : 2n-4, 16 : 3n-4 and 18 : 2n-9.
Fatty acid mass balance
Biosynthesis of ETA and EPA was higher (P < 0·05) for EO fish compared with both RO and FO fish and with no difference in the biosynthesis of DPA among all groups in freshwater (Table 5). DHA showed a positive mass balance, which was not different between EO and RO fish. The main negative mass balance (NMB) in EO fish was for ALA, which represented 69 % of the net intake (Table 5) followed by SDA (42 %). Total elongated/desaturated n-3 LC-PUFA products (ETA+EPA+DPA+DHA = 1249 μmol/fish) represented 15 % of the combined NMB for ALA and SDA in EO fish (8414 μmol/fish). There was a NMB of 2327 μmol/fish for ALA in RO fish, which represented 63 % of net intake (Table 5). A NMB of 1096 μmol/fish (47 %) for ALA was obtained in RO fish as a result of elongation and desaturation, out of which SDA biosynthesis represented 42 % (464 μmol/fish). Total elongated/desaturated n-3 LC-PUFA products (ETA+EPA+DPA+DHA) in RO fish were 632 μmol/fish. In FO fish, the main NMB was for EPA, with a disappearance of about 66 % of net intake (4264 μmol/fish), mainly due to β-oxidation as only 228 μmol/fish were used for the biosynthesis of DPA.
ALA, α-linolenic acid; SDA, stearidonic acid; ETA, eicosatetraenoic acid; DPA, docosapentaenoic acid.
a,b,c Mean values within a column with unlike superscript letters belonging to either freshwater or seawater were significantly different (P < 0·05).
Higher biosynthesis of ETA (P < 0·05) occurred in EO fish compared with RO and FO fish in seawater. There were no significant differences (P>0·05) in EPA, ETA and DHA between EO and RO fish, but negative values of EPA, ETA and DHA were obtained for RO fish (Table 5). In contrast, there was a positive mass balance of all n-3 LC-PUFA along the n-3 pathway for EO fish (Table 5).
There was a NMB of 12 836 and 3472 μmol/fish for ALA and SDA, respectively, for EO fish in seawater accounting for 73 and 51 % of net intakes. Total elongated/desaturated products of n-3 LC-PUFA (ETA+EPA+DPA+DHA) were 696 μmol/fish for EO fish representing only 4 % of combined NMB for ALA and SDA. There was a NMB of 74 % (5468 μmol/fish) for ALA net intake from RO fish, with only 13 % (689 μmol/fish) elongated/desaturated along the pathway largely as SDA (633 μmol/fish). There was no positive mass balance of FA for FO fish in seawater, and the main NMB was for EPA (79 % net intake) followed by DHA (52 % net intake).
Higher accretion of Δ5 desaturated and elongated FA (P < 0·05) in EO fish compared with RO fish was observed in freshwater. The accretion of elongated and desaturated FA could not be computed for RO and FO fish in seawater as a result of high NMB along the n-3 pathway.
Growth and parr–smolt transformation
The transfer and growth of fish in seawater were key elements in the present study, and both the K (Fig. 1) and osmolality values for all groups were indicative of smolted fish. Feeding Atlantic salmon a diet of 100 % RO had a positive impact on growth over the duration of the experiment. Alterations in lipid metabolism are regarded as an integral part of the parr–smolt transformation, and a VO diet might be better suited for Atlantic salmon to adapt to seawater since its FA composition more closely resembles those of fish from the wild(Reference Bell, Ghioni and Sargent31). The importance of dietary VO during smoltification might be multiple: higher osmoregulatory capacity(Reference Bell, Tocher and Farndale32, Reference Tocher, Bell and Dick33), increased growth(Reference Bendiksen, Arnessen and Jobling34) and acting as a protection barrier against translocation of pathogens(Reference Jutfelt, Olsen and Björnsson35). In agreement with previous findings on the beneficial effects of VO on fish undergoing smoltification, RO fish showed higher growth and survival, indicating that RO fish were better prepared for transition in seawater than EO and FO fish. To the best of our knowledge, in all previous studies involving complete substitution of FO by VO in Atlantic salmon undergoing parr–smolt transformation, fishmeal contributed some n-3 LC-PUFA to the diet, with EPA and DHA composition ranging from 1 to 4 % of total FA(Reference Bell, Tocher and Farndale32–Reference Bendiksen, Arnessen and Jobling34). In the present study, defatted fishmeal was used, and only trace amounts of n-3 LC-PUFA were present in the VO diets, with EPA and DHA composition ranging from 0·2 to 0·5 % of total FA (0·3–0·9 g/kg; Table 1). An important finding of the present study in relation to parr–smolt transformation was that fish fed exclusively on the VO diets (EO or RO) successfully smolted. This finding could prove to be important for feed formulation in the context of FO substitution in Atlantic salmon in freshwater and also extends to fishmeal replacement in the freshwater life cycle of Atlantic salmon.
n-3 Fatty acid metabolism – freshwater phase
Our prime objective was to test whether feeding Atlantic salmon a diet rich in SDA from parr to smolt would result in higher biosynthesis of n-3 LC-PUFA. In freshwater, EO fish had higher ETA compared with fish fed the other diets (Table 3). Atlantic salmon parr towards the end of the freshwater period were at an important phase in their life cycle, preparing for the transfer to seawater. This critical period for the fish is accompanied by an increase in desaturation and elongation activities along both the n-3 and n-6 pathways for the production of LC-PUFA(Reference Bell, Tocher and Farndale32, Reference Tocher, Bell and Dick33). Therefore, if provided with enough substrate (ALA), fish can meet their EPA and DHA requirements as observed in RO fish, and the presence of SDA in the EO diet, which is a precursor for EPA and allows bypassing of the initial Δ6 desaturase enzyme(Reference Miller, Bridle and Nichols13, Reference Miller, Nichols and Carter16), has enabled higher biosynthetic activity along the n-3 LC-PUFA pathway resulting in an increase in ETA. In a previous study on Atlantic salmon parr fed an EO diet, comparable levels of EPA and DHA in muscle were obtained for FO fish(Reference Miller, Nichols and Carter16). In the present study, we did not obtain a similar result. However, in the present study, the K value was higher because the diet contained 200 g/kg total lipids compared with 129 g/kg in the previous study. Therefore, due to the lower dietary lipid level, there was a reduction in TAG proportion relative to polar lipids in the white muscle, which might explain the retention of DHA(Reference Miller, Nichols and Carter16). Furthermore, the previous study was conducted over a short duration (42 d), which coincided with the critical parr–smolt transformation period characterised by an increase in desaturation and elongation activities.
A point of focus was the biosynthetic activity along the n-3 pathway, and the FAMB approach was used to verify our hypothesis. For EO fish, ETA and EPA biosynthesis was higher compared with RO fish due to the presence of high SDA (9·2 % of total FA) in the EO diet. In contrast, for RO fish, ALA had to be desaturated to SDA, adding an extra step along the pathway at the cost of 42 % of total elongation and desaturation products. As a result, there was a twofold increase in total n-3 LC-PUFA biosynthesis (1249 v. 631 μmol/fish) in EO fish. There was no difference in DHA biosynthesis between EO and RO fish, hereby underlying the importance of DHA in fish undergoing smoltification and life in seawater. Previous research has demonstrated that in Atlantic salmon undergoing parr–smolt transformation, there was an increase in DHA in gill and liver polar lipids in fish fed on VO diets(Reference Bell, Tocher and Farndale32, Reference Tocher, Bell and Dick33). In other studies(Reference Bell, McEvoy and Tocher36, Reference Bell, Henderson and Tocher37), there were preferential deposition and retention of DHA in muscle lipids irrespective of the concentration in the diet, which was attributed to the specificity of fatty acyl transferase enzymes towards incorporation of DHA into flesh TAG and polar lipids.
FO fish in freshwater showed some degree of biosynthetic activity especially for DPA, and while we do not neglect the fact that EPA and DHA could have been produced, it might have been masked by the high dietary presence of these FA. This masking is regarded as a shortcoming associated with the FAMB approach(Reference Turchini, Francis and De Silva17, Reference Turchini and Francis19, Reference Francis, Peters and Turchini38). Consequently, when computing the last step of the method, it was not possible to detect any increment of FA as a result of Δ5 and Δ6 desaturases, and elongase enzymes except for the conversion of EPA to DPA (Fig. 2) with a positive mass balance obtained for DPA. Other limitations may occur; the computation of the FAMB proceeds only in the direction of its specific pathway, therefore it does not take into account possible FA chain shortening such as retroconversion of DHA to DPA or the β-oxidation of FA previously elongated and desaturated(Reference Turchini, Francis and De Silva17, Reference Turchini and Francis19). The production of eicosanoids, resolvins and protectins from their precursor's arachidonic acid, EPA and DHA is also not quantified. However, the production of these functional metabolites is minimal, probably having little impact on the total FAMB(Reference Turchini, Francis and De Silva17, Reference Turchini and Francis19). In rats, the production of eicosanoids, as measured by their urinary excretion, does not exceed 1 μg/d(Reference Hansen and Jensen39). Most of these metabolites are generally involved in inflammatory processes and are either potent pro- or anti-inflammatory at nanomolar concentrations, and their production is measured in minute amounts (ng/mg of protein) in tissues(Reference Marcheselli, Hong and Lukiw40, Reference Hudert, Weylandt and Lu41). Arguably, the FAMB might be more limited with respect to production of these functional metabolites in cases of infected or injured fish. Similar limitations also occur with other methods employing labelled FA to assess FA metabolism in fish(Reference Turchini and Francis19). As the use of the FAMB approach expands, including via incorporation of measurement of metabolites present at low abundance, the method may be further fine-tuned.
Similarly, for EO fish, the high presence of SDA in the diet might have masked its desaturation from ALA, therefore any accretion of Δ6 desaturated SDA could not be obtained (Fig. 2). Hence, when assessing the biosynthetic activity along the n-3 pathway, this method might be best suited when comparing between oils with ALA as the main precursor and very low amounts of other n-3 PUFA. In freshwater, ALA was the main FA that is β-oxidised in RO fish (33 % of net intake), while in EO fish, 69 and 42 % of ALA and SDA net intake showed a NMB. The NMB for 42 % of SDA equated to 1607 μmol/fish, which was greater than the total elongated/desaturated products (1249 μmol/fish). Therefore, theoretically 78 % of SDA NMB was biosynthesised along the n-3 pathway, and dietary ALA did not contribute to any n-3 LC-PUFA biosynthesis in EO fish. Since some ALA might have been elongated to SDA and then further metabolised, as mentioned earlier, the high dietary amount of SDA might have masked this step.
In FO fish, 66 % of EPA net intake showed a NMB mainly through β-oxidation. It has been well documented that excess dietary EPA is readily β-oxidised(Reference Turchini and Francis19, Reference Francis, Peters and Turchini38, Reference Stubhaug, Lie and Torstensen42). Moreover, it has been shown that at around seawater transfer, there is an increase in β-oxidation capacity in the liver and muscle of Atlantic salmon(Reference Stubhaug, Lie and Torstensen42, Reference Stubhaug, Lie and Torstensen43), which would explain the apparent β-oxidation of EPA. The FAMB approach was developed to determine enzymatic activity(Reference Turchini, Francis and De Silva17); however, since the enzyme activity is usually measured over a limited incubation time, it was proposed to report either any FA accretion as an indication of enzymatic activity(Reference Turchini and Francis19) or apparent enzyme activity(Reference Francis, Peters and Turchini38). In the present study, the accretion of certain FA could not be computed due to the masking effect of the FA present in high amounts in the diet. However, in the freshwater phase, we confirmed that in EO fish, there was higher n-3 LC-PUFA production as a result of higher accretion of desaturated and elongated FA (Fig. 2).
n-3 Fatty acid metabolism – seawater phase
The same scenario as in freshwater was observed in seawater for whole-carcass FA content, with fish fed on the FO diet having higher amounts of EPA and DHA (Table 4). However, between the VO diets, the presence of SDA in EO fish in seawater resulted in greater n-3 LC-PUFA biosynthesis since higher ETA, EPA and DPA were accumulated. This observation was confirmed through the FAMB (Table 5), where positive values were obtained at all levels of desaturation/elongation along the n-3 pathway leading to a net gain of 696 μmol/fish in total n-3 LC-PUFA for EO fish compared with a net loss ( − 439 μmol/fish) for RO fish. Yet again, the presence of SDA in the EO diet resulted in these differences because of the extra step involved in producing SDA in RO fish at the expense of 633 μmol/fish (Table 5). In a similar study(Reference Miller, Bridle and Nichols13), it has been found that an EO diet promoted an increase in elongase and Δ5 desaturase gene expression in Atlantic salmon smolt when compared with fish fed a FO diet, and that the increase in activity led to higher EPA in the liver compared with fish fed a RO diet.
It was evident that the biosynthetic activity along the n-3 pathway was negligible in seawater for all three diets (Table 5). This observation has been previously documented, whereby marine fish cannot convert dietary ALA from VO sources to EPA and DHA at a physiologically significant rate(Reference Tocher44, Reference Sargent, Tocher, Bell, Halver and Hardy45) due to the evolutionary consequence of a natural diet rich in n-3 LC-PUFA(Reference Tocher, Bell and Dick46). Ex vivo approaches were used to assess n-3 biosynthetic capacity of Atlantic salmon in isolated hepatocytes in previous studies(Reference Bell, Tocher and Farndale32, Reference Tocher, Bell and Dick33, Reference Tocher, Bell and Dick46) and showed lower hepatic desaturation of ALA to n-3 LC-PUFA in Atlantic salmon post-smolts compared with parr. Our in vivo approach has also shown different biosynthetic capacity for Atlantic salmon in fresh and seawater. Due to the low n-3 biosynthetic activity at the cellular level of key tissues of Atlantic salmon in seawater and the NMB of substantial amounts of FA through β-oxidation, any significant accretion of n-3 LC-PUFA could not be detected when examined at the whole organism level in the present study. We have analysed the whole carcass, therefore it will also be important to examine the individual tissues (muscle and liver) in future studies.
In seawater fish, the FAMB was characterised by high NMB of specific substrates mainly due to β-oxidation: ALA for RO fish (74 % net intake), ALA and SDA for EO fish (73 and 51 % net intake), EPA and DHA for FO fish (79 and 52 % net intake). While there is a preferential order of FA for β-oxidation, this is subservient to β-oxidation of excess FA(Reference Turchini and Francis19). ALA has been shown to be readily β-oxidised(Reference Torstensen and Stubhaug47, Reference Stubhaug, Froyland and Torstensen48) in Atlantic salmon, while SDA in EO fish and EPA in FO fish were probably supplied surplus to requirements from their respective diets. In general, immediately after seawater transfer, the feed intake of Atlantic salmon smolt is reduced due to stress, and this period is accompanied by a reduction in condition (K) as fish use their lipid stores as an energy source. Therefore, it is very likely that fish were initially using their lipid reserves in seawater, which also contributed to the large NMB in FA.
The use of FO can be regarded as a rather wasteful practice due to the β-oxidation of substantial amounts of EPA and DHA; however, high amounts were still accumulated and resulted in the observed three- to fivefold difference compared with fish fed on the VO diets. It should be stressed that from a human health perspective, the SDA-enriched oil might be a more suitable substitute due to the improved n-3:n-6 ratio, high levels of total n-3 (mostly as ALA and SDA) and total PUFA. Furthermore, high intake of SDA, from genetically modified soyabean oil, increased the n-3 index and lowered the risk of cardiac events in humans(Reference Harris, Lemke and Hansen49). EO in the present study was used as a model oil due to its high SDA naturally, but the use of EO is currently uneconomic as a substitute for FO(Reference Miller, Nichols and Carter3). Nevertheless, plant genomics research is underway to increase the synthesis of SDA in commercially viable oil seed plants and also to further improve the n-3:n-6 ratio(Reference Ursin50). In the near future, such plant oils, containing EPA and DHA(Reference Naylor, Goldburg and Primavera2, Reference Miller, Nichols and Carter3), might be commercially available and could be suitable for FO substitution in the diet of Atlantic salmon(Reference Graham, Cirpus and Rein51, Reference Venegas-Caleron, Sayanova and Napier52).
Several major findings were drawn from the present study. First, complete substitution of FO with both EO and RO in the diets led to successful parr–smolt transformation without any additional input of n-3 LC-PUFA. Second, in freshwater, both RO and EO fish were able to biosynthesise n-3 LC-PUFA to meet their requirements, and the presence of SDA in the EO diet resulted in higher n-3 LC-PUFA biosynthesis. Third, in seawater, n-3 LC-PUFA biosynthetic activity was non-existent at the whole-body level for RO and FO fish, whereas some n-3 LC-PUFA biosynthesis occurred in EO fish probably as a result of the long feeding history on SDA, but to a lesser extent than in freshwater fish. In addition, the FAMB approach has been a useful tool to assess FA metabolism at the whole-body level in the present study, although further research is required to fine-tune the method. Finally, although an EO diet increased the n-3 LC-PUFA biosynthesis, EPA and DHA contents in both fresh and seawater fish were still lower compared with in those fed the FO diet. However, due to higher ALA, SDA and total n-3 PUFA obtained in fish, oil enriched with SDA in aquafeeds would be more beneficial from a consumer perspective compared with conventional VO.
We thank H. King (SALTAS) for the provision of Atlantic salmon parr. We are also grateful to N. P. Sanga, R. S. Katersky and K. Latif (University of Tasmania) for their assistance during sampling and to D. Holdsworth for managing the CSIRO GC–MS facilities. The present study was supported by an Endeavour International Postgraduate Scholarship from the University of Tasmania and a CSIRO Food Futures Flagship postgraduate award. The authors disclose no conflicts of interest. The authors' contributions were as follows: M. B. C., A. R. B., P. D. N. and C. G. C. designed the study. M. B. C. and A. R. B. conducted the study. P. D. N. and C. G. C. provided the essential reagents/materials/instruments. M. B. C. performed the statistical analysis. M. B. C., P. D. N. and C. G. C. wrote the manuscript. M. B. C. had primary responsibility for the final content. All authors read and approved the final manuscript.