The effect of gradual addition of camelina seeds in the diet of rainbow trout ( Oncorhynchus mykiss ) on growth, feed efficiency and meat quality

Camelina ( Camelina sativa ) is a robust crop to cultivate and a potential source of protein, oil and n- 3 fatty acids for aquaculture. The aim of this study was to examine the effects of camelina seed in feed on rainbow trout ( Oncorhynchus mykiss ) performance and nutritional composition of the muscle. A mixture of faba bean, wheat gluten meal and rapeseed oil was gradually replaced by camelina seed (0%, 10% and 20% camelina in the diet). Fifty 0+ old fish were placed in each of the 12 recirculation aquaculture tanks to enable three treatments with four replicates. Camelina seed inclusion did not affect the growth performance, crude protein or fat content of fish body. The moisture content was the lowest ( p = 0.021) with 20% camelina. The vitamin D 3 contents in fish muscle were low, indicating that camelina may slightly impair the absorption of vitamin D 3 ( p = 0.055). Camelina diets decreased ( p < 0.001) the proportion of monounsaturated fatty acids and increased ( p < 0.001) polyunsaturated fatty acids and n- 3 fatty acids, particularly α - linolenic acid, in fish muscle. Camelina seed proved to be a potential plant- based ingredient for rainbow trout feed.

Inclusion of camelina in rainbow trout feed may be limited by antinutritional factors, such as glucosinolates (GSL), sinapine, phytic acid and condensed tannins, which are found in camelina seeds (Matthäus & Zubr, 2000;Russo & Reggiani, 2012;Schuster et al., 1998). Antinutrients are concentrated when seeds are processed further to products with higher protein content. Even though rainbow trout has been shown to be less sensitive for antinutritive effects of GSL than Atlantic salmon, it can still suffer from compromised growth when intake of GSL increases (Burel et al., 2000). However, due to the high crude lipid content, beneficial fatty acid composition and moderately high contents of protein, whole camelina seeds could fit the diets better than camelina meals and cakes after oil removal. The lack of various processing steps needed when using whole seed makes it cost effective raw material for feed. Important fish quality factors for humans are nutritionally beneficial fatty acids and vitamin D. Many of the health benefits of eating fish have been linked to n-3 fatty acids, especially to 20:5n-3 and 22:6n-3. In clinical trials, fish oils have been shown to prevent arteriosclerosis, type 2 diabetes and memory disorders in the elderly (Fard et al., 2018). The fat content and the fatty acid composition of fish depend on the species, age, nutrition, body part and maturity of the fish as well as on season and living environment in general (Välimaa et al., 2019).
Fish and fish products are regarded as the most important dietary source of vitamin D for humans. Vitamin D is a hormone-like vitamin, and its deficiency in humans is common globally. The bestknown function of vitamin D is its antirachitic property, but it has also numerous noncalcemic functions in the body (Autier et al., 2014). The predominant vitamin D compound in fish is cholecalciferol (vitamin D 3 ), which occurs in highly variable concentrations in different species. No correlation between fat or vitamin D 3 content of feed and vitamin D 3 content of the rainbow trout muscle has been found (Mattila et al., 1997(Mattila et al., , 1999. However, the transfer efficiency of vitamin D 3 from feed to rainbow trout muscle may be hampered due to the general composition of the feed (Ferreira et al., 2020).
The aim of this study was to examine the usability of camelina seeds as a component of rainbow trout feed and its effects on fish performance and nutritional composition of fish muscle including fatty acid composition and vitamin D 3 contents. To maintain the raw composition of the diet constant upon increasing camelina seed level, a mixture containing faba bean (Vicia faba), wheat (Triticum aestivum) gluten meal and rapeseed (Brassica rapa subsp. Oleifera) oil was replaced by camelina seed accordingly.

| Ethical statement
The authors confirm that the ethical policies of the journal, as noted on the journal's author guidelines page, have been adhered to. The use of animals in scientific experimentation was in line with Directive 2010/63/EU, and the study followed the protocols approved by the Regional State Administrative Agency, Helsinki, Finland.

| Experimental diets
A mixture of three plant ingredients, that is faba bean, wheat gluten meal and rapeseed oil, was replaced by increasing concentrations of camelina seeds (variety Calena, cultivated in Finland, Table 1).
The mixture had a similar crude protein and crude fat content than camelina seed.
Three isonitrogenous and isolipidous experimental diets were formulated to meet the nutrient requirements of rainbow trout by using ingredients typically used in commercial rainbow trout feed (National Research Council (NRC), 2011; Tables 2 and 3). In all diets, 80% of the added oil was from a plant source (rapeseed oil or camelina seed oil) and 20% from fish oil. In the test diets, the mixture of plant ingredients was gradually replaced by camelina seeds (camelina seed in diet; CS0 = 0%, CS10 = 10% and CS20 = 20%).
When rapeseed oil, faba bean and wheat gluten meal were partially replaced with camelina seed, the amino acid composition of the diets showed only minor variation (  (Table 1). Phytic acid concentration in the camelina seed was 20.5 g kg −1 . The diet CS0 did not contain GSL, and in CS10 and CS20, the total GSL content was 3.28 mmol kg −1 and 5.03 mmol kg −1 respectively (

| Fish and experimental conditions
The study was carried out between November 2018 and March 2019 in a recirculation aquaculture system (RAS) at the Parainen re- fish used were 0+ hatchery-reared all female rainbow trout (initial weight 155.1-156.5 g), originating from a private company (Hanka-Taimen). Acclimation to the experimental conditions commenced 4 weeks before the experiment started and during that time the fish were kept in four tanks and fed a commercial trout diet (Hercules 5.0 mm, Raisioaqua). Then, the fish were randomly divided into 12 round, 500 L fibreglass tanks, and 50 fish were placed in each tank to enable three treatments with four replicate tanks each.
The RAS consisted of 12 similar bottom-drained rearing tanks with one common water treatment system; a drum filter (mesh size of 60 µm) for solids removal, two moving bed and one fixed bed bioreactors, cascade aeration column to remove dissolved carbon dioxide, a forced-ventilated cascade to add pure oxygen and UVdisinfection unit. The brackish water (salinity 6-7 ppm) from coastal area of Baltic Sea was used as the clean replacement water at the relative water renewal rate of about 4 000-5000 L kg −1 feed fed.
During acclimatization and experiment, fish were fed twice a day (50% and 50% of the daily dose at 05.00-05.15 and 11.00-11.15) by computer-controlled feeding system (Arvo-Tec) using published restricted autumn-feeding table of Raisioaqua. The feeding system calculates feeding and biomass growth based on tank biomass, feeding level (% of biomass) and feed conversion ratio (FCR). The feeding activity of fish was observed daily during the second feeding session and if the feeding activity began to decline, the feeding level was restricted to ensure that all the feed offered was eaten. The fish were exposed to continuous illumination provided by artificial overhead lighting. The use of animals in scientific experimentation was in line with Directive 2010/63/EU.

| Sampling and chemical analyses
Fish were fasted for two days prior to each sampling. Biomasses were recorded at the beginning, three times during the experiment and at the end of the experiment. After each intermediate weigh- ing, the amount of feed fed was recorded and FCR was calculated and adjusted in the feeding system accordingly if needed. A random sample of 10 fish and five fish per tank was taken at the beginning and at the end of the experiment, respectively, for chemical analy- Samples of camelina seed, faba bean, wheat gluten meal and experimental diets were freeze-dried, ground to a fine powder and stored in freezer until analysed. All other analyses except fatty acid composition and contents of vitamin D 3, phytic acid and GSL were analysed in Eurofins laboratories.
Fatty acid methyl esters (FAME) of lipid in freeze-dried experimental diet samples (200 mg) were prepared in a one-step extractiontransesterification procedure using chloroform and 2% (v/v) sulfuric acid in methanol (Shingfield et al., 2003). The lipids in freeze-dried trout samples (80 mg) were extracted with methanol, MQ-water and methyl tert-butyl ether and transesterified to FAME using acetylchloride in methanol (1:9) (Ostermann et al., 2014). Fatty acid content was determined using tritridecanoin (T-135; Nu-Chek-Prep) as an internal standard and tripalmitin (T-5888; Sigma-Aldrich) as an external standard. The FAME were quantified using a gas chromatograph and flame ionization detector by using a temperature gradient programme (Shingfield et al., 2003) and hydrogen as the carrier gas operated at constant pressure (206.8 kPa) and nominal initial flow rate of 2.1 mL min −1 . The fatty acid composition as weight percentages was calculated using theoretical response factors (Wolff et al., 1995).
An internal standard method previously described by Mattila et al. (1992) and Mattila et al. (1999)  the conditions were the same as described by Mattila et al. (1992).
The recovery of spiked cholecalciferol was 110% as calculated using internal standard method. The coefficient of variation of triplicated performed analyses varied between 3.3% and 10.1%.
Determination of phytic acid content consisted of ferric precipitation, release of the phytate by addition sodium hydroxide and determination of the phosphorous content by the ICP-OES (Mattila et al., 2018;Plaami & Kumpulainen, 1991).

| Calculations and statistical analysis
Average initial and final weight of fish (wet weight, WW), feed intake (FI %), specific growth rate (SGR %), thermal growth coefficient (TGC × 1000), FCR and protein efficiency ratio (PER) were calculated as follows.
where t is the number of feeding days

| Fish performance
Fish in all treatment groups had similar initial body wet weight (155.1-156.5 g), and one fish died during the experiment (group CS20). By the end of the trial, fish in all dietary groups more than tripled their initial wet weights (504.9-517.5 g). At the end of the experiment, the dietary treatments did not have statistically significant effects on the feed intake and growth performance (

| Effect of camelina seeds on the vitamin D 3 content and fatty acid composition in rainbow trout muscle
Feeding with CS0, CS10 and CS20 resulted in 7.7, 6.1 and 6.0 µg kg −1 vitamin D 3 in rainbow trout muscle samples respectively. There was a tendency that camelina enriched diets led to lower vitamin D 3 concentrations in the muscles (p = 0.055; Table 7).
Contrary to MUFA, the proportion of PUFA increased (p < 0.001) in muscle from rainbow trout fed camelina diets compared with CS0, being the highest with CS20. Compared with CS0, rainbow trout fed camelina seed contained higher proportions of total n-3 fatty acids (p < 0.001) and the highest values were observed with CS20.

| Fish performance and antinutrient content of feeds
Camelina seed tested proved to be safe and potential feed ingredient.
The effect of using camelina seed as a component of rainbow trout feed on the large (150-500 g) fish seems to be rather neutral, not detrimental nor beneficial. Inclusion of camelina seed did not result in statistically significant decrease in fish performance measures even though the total GSL content of diets (CS10 3.28 mmol kg −1 and CS20 5.03 mmol kg −1 ) exceeded the safe upper limit of GSL intake for rainbow trout of 1.4 mmol kg −1 (Burel et al., 2000;NRC, 2011).
The main GSL in camelina seed is 10-methylsulfinyldecyl-gluco sinolate (50%-60% of the total GSLs), while the amounts of 9-me thylsulfinylnonyl-glucosinolate and 11-methylsulfinylundecyl-gluc osinolate are smaller. The total GSL content of camelina seed was 35.6 mmol kg −1 (i.e. 58.8 mmol kg −1 in oil-free material), which is higher than reported by Russo and Reggiani (2012) (Fenwick et al., 1986;Tripathi & Mishra, 2007). The GSL contents in the CS10 and CS20 diet were 15% and 35% lower, respectively, than the theoretically calculated amounts. This indicates that the GSLs in camelina seeds were not efficiently decomposed by intrinsic myrosinase during the preparation and processing of the feed. The period of time between grounding of diet mixture including camelina seeds and extruding it to pellets was 45-60 min, which probably resulted in partial decomposition of the GSL and minor effects of dietary GSL on the growth performances of fish.
Phytic acid concentration in the camelina seed was 20.5 g kg −1 , which was slightly less than reported by Russo & Reggiani, 2012 (25.4-32.3 g kg −1 ). The processing used in this study did not reduce the phytic acid content. The phytic acid contents of all experimental diets were rather similar (9.0-12.0 g kg −1 ), so the possible effects of phytic acid to the results cannot be perceived.
However, based on the current data, higher than 10% inclusion rate of camelina should be thoroughly considered as there appears to be a numerical tendency for decreasing growth indicators with increasing inclusion percentage. The results are in accordance with earlier reports of Bullerwell et al. (2016) and Anderson et al. (2018). They reported that 10%-15% dietary addition of ground camelina seed had only a small effect on the performances of fingerling (3-110 g) rainbow trout. At the commercial production scale and production cycles, these small differences might turn economically significant in practise.

| Fatty acid composition of rainbow trout muscle
As expected, the dietary fatty acid composition was reflected in the rainbow trout muscle. This has been shown in many earlier reports (Caballero et al., 2002;Hixson et al., 2014a;Pettersson et al., 2009;Thanuthong et al., 2011). However, due to the metabolism of fatty acids, such as chain elongation, desaturation, β-oxidation and de novo synthesis occurring in fish tissues, the fatty acid composition of feed is not directly proportional to the fatty acid composition of fish tissues (Tocher, 2003). In the present study, the proportion of 18:1n-9 was much higher in CS0 diet (47.7%) compared to fish muscle on diet CS0 (41.9%). This was TA B L E 6 Composition of the rainbow trouts fed the experimental diets (mean ± standard deviation, n = 4) probably a result of the relatively rapid oxidation of 18:1n-9 for energy in rainbow trout muscle (Kiessling & Kiessling, 1993) when fish were fed CS0 rich in 18:1n-9. Thus, the partial replacement of rapeseed oil, faba bean and wheat gluten meal with camelina seed reduced the proportion of 18:1n-9 by 15 and 31% in CS10 and CS20 diets, respectively, whereas the decrease in the proportion of 18:1n-9 in rainbow trout muscle was 12 and 23% for CS10 and CS20 respectively. Σ Polyunsaturated fatty acids 32.8 ± 0.2 c 36.2 ± 0.6 b 38.6 ± 0.8 a <0.001 Σ n-3 fatty acids 15.9 ± 0.3 c 19.9 ± 0.7 b 22.9 ± 1.0 a <0.001 Σ n-6 fatty acids 16.9 ± 0.1 a 16.3 ± 0.2 b 15.6 ± 0.2 c <0.001 n-6/n-3 ratio 1.1 ± 0.0 a 0.8 ± 0.1 b 0.7 ± 0.1 c <0.001 Note: a,b,c Values within rows that are denoted by different superscripts are significantly different from each other, p < 0.05. †Camelina seed in the diet: 0%, 10% or 20%.
The previous studies suggest that the bioconversion of 18:3n-3 to 22:6n-3 is active in rainbow trout and that the concentrations of n-3 long-chain PUFA increase with the increase in dietary supply of 18:3n-3 even though the bioconversion to 22:6n-3 is insufficient to compensate the reduced intake of 22:6n-3 (Thanuthong et al., 2011;Turchini and Francis, 2009), which is consistent with the present results. It has also been shown that the Δ6 desaturase has higher affinity towards 18:3n-3 than 18:2n-6 and Δ5 desaturase has greater affinity towards 20:4n-3 over 20:3n-6 in rainbow trout (Thanuthong et al., 2011). In addition, in Atlantic salmon the expression of Δ5 and Δ6 desaturases increased when vegetable oils replaced fish oil in the diet (Jordal et al., 2005;Zheng et al., 2005) and Δ6 desaturase had a preference for 18:3n-3 over 18:2n-6 (Zheng et al., 2005). In the present study, the proportions of 18:2n-6 and total n-6 fatty acids were slightly decreased in the diets with increasing amount of camelina seed, suggesting that the competition between n-6 and n-3 substrates in rainbow trout tissues was even lower in CS10 and CS20 compared with CS0. This could result in increasing the efficiency of chain elongation and desaturation of n-3 fatty acids in camelina seed diets.
The decrease in the proportion of total n-6 fatty acids, especially 18:2n-6, and the increase in total n-3 fatty acids in rainbow trout muscle when rapeseed oil, faba bean and wheat gluten meal were replaced with camelina seed resulted in lower n-6/n-3 ratio. In addition, although the total SFA increased and MUFA decreased in camelina fed rainbow trout, the total PUFA increased together with n-3, which is beneficial to human health. Fish are the primary source of the highly unsaturated n-3 PUFA for humans, and regarding the eating quality of cultured rainbow trout and cultured fish in general, the quantity of long-chain n-3 fatty acids in the fish muscle is an important attribute. The long-chain n-3 PUFA in seafood reduce the risk for congestive heart failure, coronary heart disease, ischaemic stroke and sudden cardiac death (Rimm et al., 2018). Despite being subtle, the increases in long-chain n-3 PUFA in rainbow trout fed camelina seed in the present study indicate increased nutritional quality inducing positive implications on human health. The results also show that by feeding camelina seeds, rainbow trout can be used to produce health promoting very long-chain n-3 PUFA to consumers. However, vegetable oils rich in 18:3n-3 cannot fully replace fish oil in the feed of cultured rainbow trout as the pathway of long-chain n-3 PUFA biosynthesis is not efficient enough to maintain the same level of 20:5n-3 and 22:6n-3 in vegetable oil-fed fish as in fish oil-fed fish (Tocher, 2015).

| Vitamin D
The vitamin D 3 levels in rainbow trout muscle samples (Table 7) were very low (6.0-7.7 µg kg −1 ) compared with earlier published values. According to Finnish National Food Composition Table of  D 3 in fillet, respectively, in rainbow trout fed commercial diets. In our earlier study (Mattila et al., 1999), the contents of vitamin D 3 in individual rainbow trouts varied from 57 to 156 µg kg −1 fillet. The low levels observed in the present study may be due to the differences in the composition of the experimental diets and the fish farming systems. For example, in Mattila et al. (1999), the feeds contained mostly ingredients of fish origin and the experiment was conducted outdoors in earthen ponds during summer. In the present study, most of the feed ingredients were of vegetable origin and the experiment was performed indoors in round, 500 L fibreglass tanks using RAS.
According to our previous studies, the origin of vitamin D 3 in fish is not easily explained. We have shown that vitamin D 3 contents vary greatly between different fish species and within the same species caught from different places (Mattila et al., 1995(Mattila et al., , 1997. Further, individual variation between fish of the same species may be great, even if they are caught from the same place. This individual variation was not explained by the season, fat content, weight, sex or age of the fish (Mattila et al., 1997(Mattila et al., , 1999. The study of Mattila et al. (1997) gave indication that the diet would be a likely factor which causes high variation. However, later we showed that the content of vitamin D 3 in rainbow trout diet did not correlate with the vitamin D 3 content of the fillet (Mattila et al., 1999). Accordingly, in the present study, the experimental diets contained over target levels of vitamin D 3 (Table 2) and still the vitamin contents in fish muscle were low.
However, one important factor may be the general composition of fish feeds. Rainbow trout feeds have changed significantly during the last 20 years. Fish-derived ingredients have been replaced with plant-derived ones (faba bean, rapeseed oil, etc.). These plant ingredients contain antinutrients which may hamper the transfer efficiency of vitamin D 3 from feed to muscle. There was indication in the present study that replacing part of faba bean, wheat gluten and rapeseed oil in the fish diet with camelina seeds may impair the transfer efficiency of vitamin D 3 to the fish muscle even more (p = 0.055).
Another factor affecting vitamin D 3 content in rainbow trout may be illumination. Quite recently Pierens and Fraser (2015) showed that rainbow trout can produce vitamin D 3 when exposed either to full spectrum simulated solar light (290-1200 nm) or to visible blue light in the wavelength range of 380-480 nm. Sunlight simulation produced more vitamin D 3 than blue light. According to this study, instead of obtaining vitamin D via diet, some fish, such as rainbow trout, may need visible light from the sun to induce vitamin D 3 production in skin to meet their vitamin D requirements. In the present study, the rainbow trout was kept indoors and was exposed to normal artificial industry lighting. In addition to feed composition, this can cause the generally low vitamin D 3 levels in the fish muscle samples.
Vitamin D deficiency is a global public health problem, especially in northern countries and amongst older people and ethnic minority groups. Generally, the daily recommendations vary between 5 and 20 µg in Europe depending on country, age and physiological condition (Mendes et al., 2020;Spiro & Buttriss, 2014). Only few foods contain naturally vitamin D, and fish is one of the most important (Spiro & Buttriss, 2014). It seems that the conditions used in this study led to very low vitamin D 3 content in rainbow trout muscle which is undesirable in terms of vitamin D intake of consumers.
Currently, RAS is not widely used for rainbow trout farming but there is a tendency to increase its use due to environmental concerns as it causes less eutrophication compared to conventional fish farming practices.

| CON CLUS IONS
Camelina seed proved to be a potential plant-based ingredient for rainbow trout feed. No negative effects on productivity indicators were observed upon intermediate to relatively high inclusion rates.
The present study confirms the earlier findings that plant oils rich in 18:3n-3 increase slightly the proportion of long-chain n-3 fatty acids in rainbow trout, which increases the quality of the fish in terms of human nutrition. This can play an important role in the future when marine resources are needed more, and they need to be utilized more sparingly. The vitamin D 3 contents in rainbow trout muscle samples were very low. The reason for low levels may be poor transfer efficiency of vitamin D 3 from feed to muscle due to great concentration of antinutrient containing plant components in feeds, non-optimal illumination conditions during fish growth or both. This requires further research not only in RAS but also in conventional cultivation system to ensure vitamin D-rich rainbow trout for consumers.
Camelina production is competitive as it is a robust plant, and when it is used as seed, no elaborate processing is required. Currently high demand of camelina oil, however, might limit the availability of whole camelina seeds for fish feeding.

ACK N OWLED G EM ENTS
This work received financial support from the Strategic Research Program of the Academy of Finland (Grant no 293045 and 314243; Novel Protein Sources of Food Security), which is gratefully acknowledged.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon request.