This experiment was carried out at the Fish Nutrition Laboratory of the Fish Farm Teaching, Research and Extension Unit (UEPE-Piscicultura) of the Animal Biology Department of the Federal University of Viçosa (UFV), Viçosa, Minas Gerais, Brazil. In addition, it was approved by the Ethics Committee on the Use of Production Animals (CEUAP/UFV) of the UFV (protocol nº 035/2020).

A completely randomized design experiment with four treatments and five replicates was performed. The treatments consisted of four isonitrogenous (422.6 g kg^-1) and isoenergetic (1891.0 MJ kg^-1) diets, formulated with different vegetable lipid sources, soybean, canola, linseed or olive oil (Mundo dos Óleos LTDA, Brasília, DF, Brazil) and containing 138.82 g kg^-1 of total lipids (Table 1). Experimental diets were formulated following the chemical composition of ingredients described by Rostagno et al. (2005). Dietary protein and lipid levels were based on studies carried out by Veras et al. (2010) and Faria et al. (2019).

The ingredients were finely ground, mixed, moistened with water (50°C) and pelleted. Then, the diets were dried in a forced air oven at 50°C for 24 h, crushed in a manual mill and manually passed through granulometric sieves (Tecnal, Piracicaba, SP, Brazil) to obtain pellets sizes proportional to the mouth size of the fish (2 mm).

Samples of the diets were collected for fatty acid composition analysis using a gas chromatograph (Shimadzu GC-17A, Kyoto, Japan) equipped with a fused silica SP-2560 (bis-cyanopropil polysiloxane) chromatography column (Supelco Inc., Bellefonte, PA, USA) 100 m long and with a 0.25 mm internal diameter (Table 2). The fatty acid profiles of the experimental diets were determined at the Laboratory of Food Analysis, Department of Nutrition and Health, UFV.

The fish used in this study were obtained from the spawning of breeders housed in culture tanks of the Fish Farm Teaching, Research and Extension Unit (UEPE-Piscicultura) of the Animal Biology Department of the Federal University of Viçosa (UFV) and previously trained to accept processed diets following the methodology proposed by Luz et al. (2002) and adapted by Kasai et al. (2011).

Giant trahira (H. lacerdae) juveniles approximately three months old, with an initial length of 4.76 ± 0.50 cm and weight of 1.97 ± 0.20 g (mean ± SD), were selected and distributed in 20 aquariums (35 × 30 × 14 cm) blue polyethylene containing 7 L of water, at the density of eight fish per aquarium. The aquariums were arranged in a semi-static system, with 80% water volume renewal every three days, with constant aeration and biological filter. All aquariums were covered with plastic screens to prevent fish escape. The laboratory was maintained in photoperiod of 12 h through fluorescent lamps (60 W) and analog timer. Fish were fed the experimental diets until apparent satiation, three times a day (8:00, 13:00 and 17:00 hours) for 50 days.

During the experimental period, the water temperature was maintained at 26 ± 1.0°C (mercury thermometer) and the dissolved oxygen at 7.5 mg L^-1 (multiparameter YSI-550a, Life Science, Greene, MS, USA), while pH and unionized ammonia remained at around 6.8 and 0.0 mg L^-1 respectively (Labcon® analysis kits, Florianópolis, SC, Brazil).

At the end of the experiment, all fish from each aquarium (N=40 per treatment) were counted and weighed on a precision scale (model MB45 Toledo® 0.01 g, São Bernardo do Campo, São Paulo, Brazil) and measured to evaluate growth performance parameters. The following indices were calculated:

Length gain (cm) = final length (g) - initial length (cm);

Weight gain (g) = final weight (g) - initial weight (g);

Specific growth rate (%) = [(ln final weight - ln initial weight)/ 50 days] × 100;

Feed intake (g fish^-1) = amount of food consumed (g)/ number of fish;

Feed conversion rate = amount of food consumed (g)/ weight gain (g);

Survival rate (%) = (final number of fish/initial number of fish) × 100.

The chemical composition of diets and whole-body of fish (dry matter, ash, crude protein, crude lipids and crude energy) were determined according to the AOAC (2000). The carcasses of three fish from each aquarium were previously grouped (N=5 per treatment), ground in a blender and homogenized. The carcass was considered fish without scales and viscera (stomach, intestine, gonads, heart, liver, gall bladder and swim bladder). Moisture was made by constant drying in an oven at 110°C until weight. The ash was obtained by incinerating the samples in a muffle furnace at 600°C for 3 h. The Kjeldahl method (N × 6.25) was used for crude protein analysis. Gross energy was measured by burning as a sample in a bomb calorimeter. Lipids analysis followed the Folch et al. (1957) method, and the fatty acid derivatization reaction followed the previously established method by IUPAC (1987). The fatty acid profile was obtained using a gas chromatograph (Shimadzu GC-17A, Kyoto, Japan) equipped with a chromatographic column of fused silica (Agilent J&W DB-WAX 122-7032, Santa Clara, CA, USA) and an ionization detector flame. The parameters used in the program were: detector temperature (240°C), injector temperature (240°C) and column temperature with heating at 10°C min^-1 from 180 to 240°C, kept at this temperature for 10 min. Nitrogen was used as the carrier gas with a column flow of 0.6 mL min^-1 and a linear velocity of 14 cm s^-1, with a total flow of 52 mL min^-1 and a column pressure of 167 kPa, split 1:75. The chemical composition and lipid profile were performed at the Laboratory of Food Analysis of the Department of Animal Science and the Department of Nutrition of the UFV, respectively.

Statistical analyses were performed using software R, version 2.7.1 (São Paulo, SP, Brazil). Data were submitted to the Shapiro-Wilk test to verify the normality of the errors and to the Bartlett test to verify the homogeneity of the variances. The effects of dietary supplementing with different vegetable lipids sources were evaluated by analysis of variance (ANOVA) and by the Scott-Knott test, a procedure of means grouping, at 5% of significance.

The present study revealed that using different vegetable lipid sources in diets for giant trahira did not lead to significant changes in fish feed intake. Our results differ from those observed in other studies, which revealed a decrease in fish feed consumption when they were fed diets containing linseed oil exclusively as a lipid source. For example, silver catfish (Rhandia quelen) fingerlings fed diets containing 16.7 to 50 g kg^-1 of linseed oil showed a reduction of up to 26% in consumption, compared to fish fed diets containing corn or fish oil (Vargas et al., 2008). The same happened with juvenile silver barb (Puntius gonionotus), which showed a linear decrease in consumption as the levels of linseed oil in the diet increased from 30 to 90 g kg^-1 until reaching the total replacement of fish oil (Nayak et al., 2017). All these authors related that the reduction in fish feed intake is probably due to the less palatability of linseed oil. However, for giant trahira juveniles, even when fed diets containing 80 g kg^-1 of linseed oil, the fish did not show feed intake differences compared to fish fed with soybean, canola or olive oils. This evidences a prompt acceptance of diets by fish, indicating that the vegetable lipid sources evaluated did not influence the palatability of the diets. These results may explain the absence of cannibalism and aggressive behaviors commonly observed in carnivorous fish species.

The use of different vegetable lipids sources can cause different effects on lipid deposition in fish (Bell et al., 2003). It is possible that the higher and the lower level of whole-body lipids in fish fed dietary canola oil and linseed oil, respectively, was related to the regulation of gene expression and the enzyme activity involved in lipid metabolism (Castro et al., 2016b). The oxidation or deposition of lipids in tissues depends on each fatty acid’s dietary source and its function in the fish body. The lipogenesis and fatty acids bioconversion pathways are regulated by the dietary lipid source (Castro et al., 2016b). In the present study, fish fed dietary linseed oil showed a higher proportion of DHA deposition on fatty acid profiles. High proportions of PUFA in the lipid profile of fish are generally related to high rates of lipogenesis, which consequently leads to lower lipid deposition (Glencross, 2009; Hatlen et al., 2012).

The different vegetable lipids sources in the diet resulted in variation in the fatty acid profile of the fish, but all oils tested provided similar growth performance indices. Of note, as all diets met the nutritional requirements of energy and essential fatty acids (Veras et al., 2010; Faria et al., 2019); therefore, treatments were already expected not to affect fish growth performance. No difference in the growth performance parameters was neither observed for surubim (Pseudoplatystoma coruscans) (Martino et al., 2002) nor jundiá (Losekann et al., 2008; Vargas et al., 2008) fed different vegetable lipid sources, both Neotropical freshwater carnivorous species. Absence of growth performance effects was also observed for yellowfin seabream (Acanthopagrus latus) (Abbasi et al., 2020), rainbow trout (Shafaeipour et al., 2008), tilapia (Oreochromis niloticus) (Matsushita et al., 2006), tambaqui (Colossoma macropomum) (Paulino et al., 2018) and Cyprinus carpio (Graeff & Tomazelli, 2007) fed different vegetable lipid sources. On the other hand, the use of cod liver oil rich in PUFA n3 caused a decrease in the growth performance of South American catfish (Pseudoplatystoma fasciatum) (Arslan et al., 2008) and African catfish (Clarias gariepinus) (Ng et al., 2003). This effect is probably related to the high ratios of n3/n6 in these diets since the LNA (18:3 n3) and LA (18:2 n6) fatty acids are substrates of the same enzyme, Δ6 desaturase, and this enzyme has more affinity for the n3 precursors which can lead to more significant formation of PUFA n3 in relation PUFA n6 (Zheng et al., 2009). The n3/n6 ratio outside the acceptable range may promote adverse effects on fish development.

The fatty acid profiles of fish whole-body generally reflect the profile of the diets provided (Turchini et al., 2009; Olsen, 2011; Gomes et al., 2016). This was also confirmed for striped catfish (Pangasius hypophthalmus) (Asdari et al., 2011), rainbow trout (Yildiz et al., 2018), lambari (Astyanax altiparanae) (Pontes et al., 2019); largemouth bass (Micropterus salmoides) (Chen et al., 2020); gilthead seabream (Sparus aurata) (Ofori-Mensah et al., 2020). However, the fatty acid profiles of H. lacerdae showed distinct changes between the retention of some PUFA and the experimental fish diets, especially in the case of DHA. All vegetable oil sources tested did not present DHA. However, the presence of this fatty acid in the fish carcass suggests that H. lacerdae, like other freshwater fish species, can synthesize PUFA via desaturation and elongation of the LA and LNA (Tapiero et al., 2002; Tocher, 2003). Therefore, although fish fatty acid profiles are directly related to the fish diet, the fatty acid metabolism by fish can have a measurable effect (Emery et al., 2013), and this biosynthesis is one of the most targeted pathways under investigation (Tocher, 2015; Castro et al., 2016a).

Hoplias lacerdae juveniles fed diets containing canola oil had lower saturated fatty acid deposition than fish fed other lipid sources. This result was similar to that found for Murray cod (Maccullochella peelii peelii) when fed with canola oil compared to fish fed with fish oil and linseed oil (Francis et al., 2006; 2007). Therefore, these results indicate that the deposition of fatty acids in the fish whole-body did not correlate well with the composition of fatty acids in the diet since diets containing linseed oil, whose content of SFA was lower, caused, in the same way as diets with soybean oil, more significant deposition of these fatty acids. According to Turchini et al. (2003a,b), for some species, the deposition of SFA does not reflect the composition of the diet well, as these may not be used efficiently as an energy source and is preferably deposited in the whole-body of fish.

The highest values of LA in the fish whole-body fed diets supplemented with soybean oil, were due to the high proportion of this fatty acid in this diet. Similar results were observed for turbot (Psetta maxima) (Regost et al., 2003) and lambari (Pontes et al., 2019). Since diets supplemented with soybean, canola, and olive oil showed a higher proportion of LA in relation to LNA, ARA would be expected to be present in these fish since LA is the precursor to the synthesis of ARA. However, the presence of this acid was not detected in fish from all treatments. A DHA>EPA>ARA deposition ratio is common for lipid deposition, especially in carnivorous fish fed on vegetable oil sources (Fountoulaki et al., 2009; Yildiz et al., 2018). The low deposition of ARA in the fish muscle is explained by the fact that this fatty acid is preferentially deposited in other tissues, such as the liver. This result was similar to that of Pontes et al. (2019), who observed the absence of ARA in lambari fed with vegetable oil supplemented diets.

The higher LNA values found in fish fed a diet supplemented with linseed oil were also due to the high proportion of this fatty acid in this diet. For this reason, these fish were the only ones that presented EPA in the carcass and those that presented the highest deposition of DHA, which can be explained by the high levels of n3 PUFA in this oil source. Previous study attested that lambari fed linseed oil showed higher whole-body amounts of EPA and DHA than that fed soybean oil (Pontes et al., 2019). The higher retention of DHA to EPA is probably related to the fact that EPA has a faster rate of beta-oxidation in muscle and liver tissues (Herzberg et al., 1996; Madsen et al., 1998).

In fish fed the diet supplemented with olive oil, the proportions between the fatty acids were also kept compared with the diet. However, there were high amounts of monounsaturated fatty acids (MUFA) and lower amounts of LNA and, consequently, the DHA compared with the other diets. This result was similar to that observed in European sea bass (Dicentrarchus labrax L.) fed diets containing 60% olive oil in substitution for fish oil and compared with diets whose substitution was made with 60% soybean oil (Mourente et al., 2005).

The absence of significant differences in growth performance and the high deposition of PUFA in the carcass indicates that the evaluated vegetable lipid sources are suitable for the nutrition of giant trahira. Thus, the ability to modulate the fatty acid profile through the diet opens the possibility of producing fish with different fatty acid profiles for human consumption. Therefore, due to the higher deposition of DHA and EPA, in this study, linseed oil was considered a good vegetable lipid source for giant trahira.