The biofloc technology (BFT) system has become a promising alternative to promote the sustainability and biosecurity of shrimp culture by the utilisation of smaller volumes of water. Biofloc are aggregates of organic material and microorganisms, including bacteria, protozoa, and algae. As such, it is considered a supplementary food source for animals raised in this type of farming system (Avnimelech, 1999).

Marine shrimp feeds are still dependent on different degrees of fishmeal, due to its attractability and palatability proprieties and nutritional content, including amino acids and essential fatty acids, as well as vitamins and minerals (Amaya et al., 2007a,b; Suárez et al., 2009). However, a reduction or replacement of fishmeal for a suitable alternative ingredient would be desirable, preferably one that is cheaper and renewable.

Previous nutritional studies have attempted to find a renewable substitute with a high concentration of protein, adequate amino acid profile, and lower cost relative to fishmeal (Hardy, 2010; Ray et al., 2010; Sookying et al., 2013). Among soy derivatives, soybean protein concentrate (SPC) appears most promising because it has a better amino acid profile and nutritional balance, greater digestibility (energetic and protein), low carbohydrate and fibre levels, and lower levels of anti-nutritional factors compared with other protein sources of vegetable origin (Suárez et al., 2009; Gamboa-Delgado et al., 2013). However, compared with fishmeal, SPC does have a methionine deficit (Drew et al., 2007).

Paripatananont et al. (2001) and Bauer et al. (2012) evaluated diets with 350 and 400 g fishmeal/kg diet for Penaeus monodon and Litopenaeus vannamei, with 0, 25, 50, 75, and 100% of fishmeal replaced by SPC in clear water. Both studies indicated a significant potential for its use in the diet of marine shrimp. Although SPC has been considered a possible replacement for fishmeal, little information is available (Gatlin III et al., 2007; Sá et al., 2013) on its use in BFT for Pacific white shrimp. Therefore, the objective of this research was to evaluate the water quality, biofloc chemical composition, growth performance (survival, feed efficiency, protein efficiency ratio), and chemical composition of the Pacific white shrimp (L. vannamei) reared in a super-intensive BFT system with diets containing different levels of SPC as a replacement for fishmeal.

The study was conducted for 40 days between March and April 2013. The BFT system was implemented in Barra da Lagoa, Florianopolis (27° 58' S, 48 ° 44' W), SC, Southern Brazil.

Four different isocaloric diets were formulated (Table 1) with different levels of SPC as a replacement for fishmeal (0, 33, 66, and 100%). Diets were formulated with 300-330 g kg^-1 of crude protein (Jatobá et al., 2014) and similar amounts of marine-origin lipids (fish oil + fat contained in the fishmeal), aiming to provide similar fatty acid composition.

Diets were produced in Eusébio, CE, Brazil. The dry ingredients (fishmeal, soy protein concentrate, soybean, broken rice, and wheat flour) were first ground and sieved (500 μm mesh). After that, the microingredients were homogenised in a Y mixer for 10 min and then combined with the macroingredients for mixing in a food mixer for an additional 10 min. Soon after, fish oil, soybean oil, and soy lecithin were added, followed by 20% water. The resulting mixture was extruded at 90 °C in a micro-extruder with a capacity of 15 kg h^-1 (model EX Micro, EXTEEC Machines, São Paulo, Brazil) and dried to approximately 10% moisture in a drying oven (35 °C).

Shrimp used in this study were provided by Genearch Ltda. (Rio Grande do Norte, Brazil) and obtained from the breeding of a specific pathogen-free strain of mandatory notification (WSSV, IHHNV, TSV, IMNV, and YHV) as required by the Organization International des Epizooties (OIE). Shrimp were reared in a closed, intensive system with bioflocs until they reached the required body weight to begin the experiment.

Marine shrimp were grown under a super-intensive BFT system in a circular fibreglass tank with a capacity of 50 m³. Shrimp were sampled from three different locations in the tank and a total of 2,400 shrimp weighing 3.96±0.04 g (average ± standard deviation) were transferred to the experimental units.

Shrimp were maintained under constant aeration at a temperature of approximately 27 °C. On the day juveniles were transferred, the water in the rearing tanks measured 35 g L^-1 salinity, 27 °C temperature, 5.2 mg L^-1 dissolved oxygen, 434 mg L^-1 total suspended solids, 152 mg L^-1 alkalinity, 0.27 mg L^-1 ammonia, 3.06 mg L^-1 nitrite, and 7.42 pH. The biofloc in the tank formed after 30 days. Biofloc were in a heterotrophic stage, thus molasses was added at a rate of 50% of the feed amount.

Until transfer to the experimental units, shrimp were fed four times a day with a commercial diet containing 350 g kg^-1 crude protein (CP) (Guabi Vannamei 35 EXT, minimum of 350 g kg^-1 CP; minimum of 75 g kg^-1 ether extract; maximum of 100 g kg^-1 humidity; maximum of 50 g kg^-1 crude fibre; maximum of 130 g kg^-1 mineral material; 15 to 30 g kg^-1 calcium (Ca); minimum of 14.5 g kg^-1 phosphorus (P); maximum of 3 g kg^-1 magnesium (Mg); and minimum of 8 mg kg^-1 potassium (K), according to the manufacturer).

Twelve experimental units of polyethylene with 800 L of water (radius = 0.65 m, bottom area = 1.32 m², depth = 0.60 m) were independently supplied by pumping water from the biofloc culture system matrix tank. The units, which had an area of 4 m² (bottom and sides), were extended by another 2 m² by the addition of an artificial substrate (five rectangular substrates, 0.60 × 0.65 m) to comfortably accommodate the animals (Schveitzer et al., 2013a). Constant aeration was provided through diffusion hoses connected to a 7.5-hp blower and titanium heaters (800 W) with a thermostat were used in each tank to maintain the temperature at 29±0.5 °C.

The experimental units were distributed among four treatments (0, 33, 66, and 100% of fishmeal replacement for SPC) completely at random, in triplicate. In each tank, 200 shrimps were stocked, maintaining an initial density of 250 shrimp m^-3.

Shrimp were fed four times a day (8:00, 11:00, 14:00, and 17:00 h) with a monitoring program to confirm intake, as described by Jatobá et al. (2014). The first feeding supplied 4% of the initial stocked shrimp biomass, with total 90% of the meal broadcast over the water surface and the remainder in feeding trays. After 1 h, trays were checked to estimate the diet intake rate per experimental unit. When feed leftovers were observed after two consecutive feeding times, the amount of feed offered was reduced by 10%. When no feed leftovers were observed after two consecutive feeding times, the amount of feed offered was increased by 10%.

During the experiment, dissolved oxygen and temperature were measured two times a day. Measurements of pH, floc volume (Imhoff cone), total dissolved ammonia, nitrites and nitrates, alkalinity, total suspended solids (APHA, 2005), dissolved orthophosphates (Strickland and Parsons, 1972), and salinity (refractometer) were performed twice a week.

The amount of solids was maintained approximately between 400 and 600 mg L^-1 (Schveitzer et al., 2013b). When this value was exceeded, the excess was removed with the aid of sedimentation tanks with a capacity of 60 L (adapted from Ray et al., 2010).

Since the culture system was not equipped with a water exchange system, only the evaporated volume was replaced by fresh water. Calcium hydroxide was added when alkalinity fell below 120 mg L^-1 CaCO₃ and, when necessary, the dose was 10% of the daily ration. Molasses was used until the 29th day of the experiment to estimated carbon:nitrogen ratio in the water between 12:1 and 15:1, in accordance with Avnimelech (1999).

Samples were collected weekly to quantify shrimp growth. At the end of the experiment, the apparent feed efficiency, protein efficiency ratio, apparent feed intake, specific growth rate, final body weight, weight gain, survival, and yield were assessed.

Samples of all diets were sent to Santa Maria, RS, Brazil, for analysis. Twelve shrimp samples and 12 biofloc samples were sent to Campinas, SP, Brazil, to determine the amino acid profile through high-performance liquid chromatography.

All amino acids were measured using AOAC (2005) official method 994.12, except methionine and cystine that were measured using AOAC (2005) official method 985.28. Protein contents were also measured according to AOAC (2005) using the Dumas nitrogen combustion method. Crude lipids were quantified using the ether extraction method and energy was determined with a bomb calorimeter. Crude fibre, ash, and moisture were determined using AOAC (2005) official methods 978.10, 942.05, and 930.15, respectively. Cholesterol concentration was also analyzed in the shrimp samples and amino acid profile was determined through HPLC in the biofloc samples.

All data were first subjected to Bartlett's analysis to verify the homogeneity of variance and then subjected to one-way ANOVA. Significant differences among treatments were analysed using the Student-Newman- Keuls test (SNK). All tests were conducted at a 5% level of significance (Zar, 2010).

The diets had similar proximate composition. The most divergent essential amino acid values were observed in phenylalanine, lysine, methionine, methionine + cystine, glutamic acid and glycine (Table 2).

Shrimp fed the 0% and 33% replacement levels showed higher average final weight and weekly gain than those fed the 66 and 100% replacement. The yield and intake were higher in the treatment without replacement than with 100%. Survival, apparent feed efficiency, and protein efficiency ratio did not vary among treatments (Table 3).

Biofloc proximate composition showed more ash, arginine, methionine, alanine, proline, and glycine in treatment without fishmeal replacement than with 100%. Aspartic acid was lower in treatment with 33% replacement than with 100%. Other amino acid and proximate composition of biofloc (Table 4), as well as the variables of water quality (Table 5) and proximate composition of dried marine shrimp (Table 6) did not vary among treatments.

Experimental diets in this work showed proximate composition and amino acid profiles, with small differences among dietary treatments. Some amino acids such as arginine, phenylalanine, isoleucine, leucine, and glutamic acid increased their concentrations in diets with higher fishmeal replacement, while ash, histidine, methionine, methionine + cystine, and glycine reduced with the increase of fish meal replacement for SPC. No trend was observed for crude fibre, lysine, threonine, valine, alanine, aspartic acid; these nutrients did not vary according to the replacement levels; in addition, valine, proline, and serine maintained close concentrations, regardless of replacement levels of SPC for fishmeal (Table 2). These fluctuations in dietary amino acid concentration are normal when trying to perform a replacement of some ingredient (Kuhn et al., 2016; Xie et al., 2016).

Methionine concentrations, at 100% replacement level, were below the recommended 7.0-9.0 g kg^-1 (NRC, 2011). However, the sum of methionine + cystine may have compensated for these levels, because 50% of the requirement for methionine can be provided by this combination, thus justifying the growth performance obtained in this research (Lall and Dumas, 2015). In addition, in all diets, arginine and histidine concentrations were below the recommended 19.0 and 8.0 g kg^-1, respectively (NRC, 2011). Even though these values were below recommendation levels, the results might still be considered promising based on shrimp growth performance (Table 3). It is likely that the requirements for these amino acids were partially met by their availability in the bioflocs, which provided some of the nutrients (Table 4) that were otherwise below recommended levels. However, even the presence of bioflocs was insufficient to promote equitable growth performance among treatments.

Water quality parameters were similar among treatments and recorded small variations in some parameters such as pH and suspended solids, demonstrating that the system was stable (Table 5), remaining within the recommended limits for white shrimp culture. These results corroborated the results of Bauer et al. (2012) and Sá et al. (2013), who did not observe any significant differences in water quality parameters using fishmeal or SPC as protein source in clear water.

The replacement of fishmeal in the diets of marine shrimp has been proposed by many authors. However there are many contradictory results (Amaya et al., 2007a,b; Suárez et al., 2009; Tacon and Metian, 2008; Bauer et al., 2012; Gamboa-Delgado et al., 2013; Sá et al., 2013; Sookying et al., 2013; Yang et al., 2015; Carvalho et al., 2016; Kuhn et al., 2016; Xie et al., 2016). Ray et al. (2010) demonstrated the potential of soybean products for replacing fishmeal after culturing 460 shrimp m^-3 in a BFT. In the present study, shrimp fed diets containing 0 and 33% replacement of fishmeal for SPC had higher final body weights and weekly gains compared with shrimp fed diets having either 66 or 100% replacement (Table 2). This could be attributed to lower levels of essential nutrients (below recommendation) or unknown growth factor of fishmeal (Hardy, 2010; Ray et al., 2010; Sookying et al., 2013). These essential nutrients could include methionine, threonine, or histidine in the higher percentage replacement diets. For this reason, Gatlin III et al. (2007) suggested supplementation of methionine and threonine in diets with higher levels of replacement of fishmeal for SPC. Even without inclusion of synthetic amino acids or any other feed additives, in this study, 33% fishmeal replacement in a diet containing 33% crude protein was feasible to be used for shrimp culture under BFT.

Shrimp final was adequate for the species, with an average of over 77.0% (Table 3). The weekly gain was similar to that observed by Jatobá et al. (2014) for shrimp fed experimental diets containing fishmeal (1.79 g/week). For shrimp fed a diet with 100% replacement of fishmeal, the lowest weekly gain of 1.57 g was still higher than the growth reported in other studies with fishmeal-free diets (Ray et al., 2010; Scopel et al., 2011; Sookying and Davis, 2011; Bauer et al., 2012; Sá et al., 2013) or experimental and/or commercial diets with fishmeal (Ray et al., 2010; Scopel et al., 2011; Bauer et al., 2012; Sá et al., 2013; Schveitzer et al., 2013a; 2013b). These results demonstrated that the diet without fishmeal, even though deficient in essential amino acids, provided a satisfactory growth. However, to improve these results, it would be necessary to assess different diets, carbon sources, or any other inputs that enhance the nutritional value of the biofloc (Avnimelech, 2006; Michaud et al., 2006; Wasielesky et al., 2006; Ray et al., 2010).

White shrimp fed a diet with 100% replacement of fishmeal for SPC weighed an average of 13.84 g, 0.50 g lower than shrimp fed diets without fishmeal replacement, after ten weeks of culture in clear water (Sá et al., 2013). The same authors tested 31% fishmeal replacement for SPC with the inclusion of 20 g of fish oil per kg of feed. These data corroborated our results, indicating that replacement of 33% of the fishmeal in the diet with SPC, including 25.1 g of marine lipids added per kg feed, did not harm shrimp performance.

Apparent feed efficiency and protein efficiency ratio showed similar efficiency in 100% replacement (Table 3), which could be explained by a similar apparent crude protein digestibility in diets with different levels of fishmeal or SPC inclusion (Carvalho et al., 2016). This corroborates other studies with L. vannamei, in which SPC was used as the main protein source in the diet (Bauer et al., 2012; Sookying and Davis, 2011; Sá et al., 2013). A higher yield and apparent feed intake were observed in shrimp fed the diet without replacement (0%) compared with 100% replacement (Table 3).

Several factors may affect shrimp feed intake, including water quality (temperature, dissolved oxygen, and nitrogen compounds), diet composition, energy, and feed attractiveness and palatability (Kureshy and Davis, 2002; Cuzon et al., 2004; Nunes et al., 2006). All treatments resulted in the same apparent feed efficiency, suggesting a similar feed conversion in the tissues of animals. Therefore, the low final weight and weekly gain of shrimp fed a diet with 100% replacement of fishmeal may have resulted from lower food intake because high concentrations of plant ingredients reduce attractiveness and palatability (Nunes et al., 2006; Amaya et al., 2007a,b).

Bioflocs provide a constant source of food for farmed animals (Avnimelech, 1999; Burford et al., 2003; Avnimelech, 2006; Michaud et al., 2006; Wasielesky et al., 2006; Taw, 2010; Jatobá et al., 2014). In the proximate composition of bioflocs, a higher level of ash was observed in the tanks with shrimp fed diet without replacement of fishmeal (0%) in comparison with tanks with 100% replacement, whereas the 33 and 66% treatments were not significantly different. This result can be attributed to the higher concentration of ash in fishmeal compared with SPC. For crude protein, ether extract, nitrogen extract, crude fibre, or energy, no significant differences were seen among the dietary treatments (Table 5).

A higher concentration of alanine was observed in the biofloc of tanks without fishmeal replacement (0%) compared with tanks with 100% replacement. Treatments with 33 and 66% replacement did not differ from others. Higher concentrations of fibre, arginine, and methionine were detected in the biofloc of treatments with 0 and 66% replacement relative to arginine and methionine concentrations with 100% replacement. Aspartic acid concentration was higher in the bioflocs of tanks with shrimp fed 100% replacement compared with that in the bioflocs of tanks with shrimp fed 33% replacement. Finally, proline concentration was lower for treatment with 100% replacement compared with other treatments. No other amino acids showed significant differences among dietary treatments (Table 5).

Bioflocs can be used as food source containing potential “growth promoters” (Avnimelech, 1999, 2006; Burford et al., 2003; Michaud et al., 2006; Taw, 2010; Wasielesky et al., 2006). In this study, different trends were recorded between diet and biofloc compositions. For example, methionine, glycine, and alanine were decreased in high levels of replacement, probably because SPC contains lower level of these amino acids than fishmeal. Only arginine was higher in the diet with 100% fishmeal replacement, though its concentration was lower in the biofloc of the same treatment. These results showed no correlation between diets and biofloc proximate composition. However each nutrient may have a different fluctuation and these differences did not affect the survival or apparent feed efficiency of the shrimp or their composition. This can suggest that diets and biofloc contain sufficient nutrients for shrimp performance, probably because of a growth-promoting effect, which may be associated with trace minerals or other unknown nutrients (Sabry Neto et al., 2015).

It is well known that the nutritional value of bioflocs can be influenced by the developmental stage of microbial biofloc, whether autotrophic, heterotrophic, or chemoautotrophic, as well as the adopted management plan. Also, the heterotrophic phase provides more nutritional value (Avnimelech, 2006; Ebeling et al., 2006; Wasielesky et al., 2006; Taw, 2010) than the others. The levels of crude protein in the biofloc (275.9 g kg^-1) were higher than those (average 194.7 g kg^-1) observed by other authors (Chamberlain et al., 2001; Tacon et al., 2002; Williams et al., 2005; Wasielesky et al., 2006; Taw, 2010; Jatobá et al., 2014), who recorded values of crude protein below 25.0%.

These results could be attributed to the heterotrophic developmental stage of the bioflocs in the present study, as they required the addition of molasses and hydrated lime, but only at the end of the culture period, whereas Jatobá et al. (2014) used a “mature biofloc”, representing the chemoautotrophic stage of development.

In the present study, it was possible to replace 33% of fishmeal by SPC using a 33% crude protein diet with 21% fishmeal. Yang et al. (2015), using diet containing 40% crude protein and 30% fishmeal, found a 20% threshold above which shrimp would be harmed. The difference between our study and that of Yang et al. (2015) may result from biofloc presence in the water, since in BFS, biofloc is used as “natural” food source with some essential nutrients for shrimp, and the protein requirement of Pacific white shrimp, which is influenced by the culture system used (Jatobá et al., 2014).

No change in proximate body composition was observed among dietary treatments (Table 6) which demonstrates that all levels of fishmeal replacement did not interfere in shrimp tissue synthesis (Mente et al., 2002; Carter and Mente, 2014). This could be explained by the similarity among proximate compositions of experimental diets.

To enhance the results obtained in this study or to evaluate higher levels of fishmeal replacement for SPC, new formulations that include ingredients that increase the feed attractiveness are recommended, such as squid meal, krill meal (Sáchez et al., 2012), synthetic amino acids, such as methionine, lysine and/or threonine (Sookying et al., 2013), or taurine, free amino acids, non-essential amino acids, and food stimulants recommended for better growth performance of fish and shrimp (Coman et al., 1996; Martinez et al., 2004). Irrespective of fishmeal replacement level, shrimp in the present study exhibited the same apparent feed efficiency, demonstrating their ability to absorb nutrients from SPC.

It is viable to partially replace fishmeal by soy protein concentrate in diets for the Pacific white shrimp (L. vannamei) reared in biofloc technology systems.