TILAPIA CULTIVATED IN A LOW-SALINITY BIOFLOC SYSTEM SUPPLEMENTED WITH Chlorella vulgaris AND DIFFERENTS MOLASSES APPLICATION RATES

The aim of this study was to evaluate the effect of supplementation with Chlorella vulgaris and molasses application rates on water quality, zootechnical performance, proximate composition and health status of Nile tilapia (Oreochromis niloticus) fingerlings cultivated in low-salinity (10 g L-1) biofloc systems. Four treatments were tested in a factorial design (supplemented with microalgae and molasses application rates): BFT-C30 (Biofloc supplemented with C. vulgaris and molasses application rates of 30% of the total daily feed); BFT-30 (Biofloc with molasses application rates of 30% of the total daily feed); BFT-C50 (Biofloc supplemented with C. vulgaris and molasses application rates of 50% of the total daily feed) and BFT-50 (Biofloc with molasses application rates of 50% of the total daily feed), for 70 days. Fingerlings of O. niloticus (initial mean weight of 3.15 ± 0.5 g) were stocked at a density of 680 fish m-3 in experimental units (50L), where 50% of this volume was biofloc previously matured. Throughout the experiment, they were supplemented with C. vulgaris every five days at the concentration of 5x104 cells mL-1. A significant interaction between supplementation with C. vulgaris and molasses application rates for final weight and length, survival, feed conversion ratio, specific growth rate, water consumption, protein efficiency ratio, sedimentation time, planktonic community and hematological indices were observed. The results indicated that the high molasses application rates (50%) in the biofloc system affects the zootechnical performance, water consumption, sedimentation time and the hematological indices of the Nile Tilapia fingerlings, hampering their development. Therefore, molasses application rates of 30% of the total daily feed for the tilapia fingerlings culture in low-salinity biofloc system is recommended.


INTRODUCTION
In 2016, world aquaculture production reached 80 million tons excluding aquatic plants, with inland fish aquaculture production representing 59.3% (FAO, 2018), among which the main species of fresh water aquaculture were the Nile tilapia Oreochromis niloticus. This fish is one of the most important species (4.2 million tons, representing 8% fish aquaculture) (FAO, 2018) due to their environmental resistance, rapid growth and salinity tolerance between 0 and 25 g L -1 (El-Sayed, 2006;Pereira et al., 2016).
Tilapia culture in brackish water is an alternative to Brazilian semi-arid zones, where water resources are limited and saline groundwater is the unique source of water available, with salinities between 1.1 and 10.0 g L -1 (Andrade Júnior et al., 2006;Brasil, 2012). Moreover, the desalination technology used for brackish groundwater results in highly saline residual water, which has a potential use for aquaculture (Soares et al., 2006).
Although aquaculture is considered a suitable alternative system for the increasing demand for animal-based protein, this activity has limitations in relation to water and land use. Therefore, intensive culture systems, such as biofloc technology (BFT), is a promising alternative . BFT is based on carbon:nitrogen ratio management and minimum water exchange for production of microorganism biomass (Azim and Little, 2008;Avnimelech, 2012;Emerenciano et al., 2017), which results in higher productivity compared to conventional systems due to the elevated stocking density used (Avnimelech, 2015).
The carbon:nitrogen ratio is maintained through the addition of a source of organic carbon, which is used by bacteria to convert TAN (Total Ammoniacal Nitrogen) to bacterial biomass (Ebeling et al., 2006). Among the various sources of carbon, molasses is widely used because it stains water, reducing light penetration and associated algal growth, however, may present high level of impurities and content variability (Samocha et al., 2017). In addition, molasses is an inexpensive source of carbon used for various industrial fermentations (Miranda et al., 1996;Najafpour and Shan, 2003).
Some authors reported successfully culturing Nile tilapia in BFT (Long et al., 2015;Miranda-Baeza et al., 2017;Zapata-Lovera et al., 2017), including a study in brackish water (8 g L -1 ) (Brol et al., 2017;Lima et al., 2018). However, due to the high stocking densities and minimal water exchange, nutrients accumulate, mainly nitrogen and phosphorus, in the biofloc culture (Krummenauer et al., 2011). About 60% of the nitrogen and 65% of the phosphorus that enter in the system are not converted into shrimp biomass (Silva et al., 2013). An alternative used to overcome the disadvantages is the use of microalgae in this system, due to their highly efficient removal of nitrogen and phosphorus (80-100%) in aquaculture, livestock and industry as have been reported by Prajapati et al. (2014), Posadas et al. (2015), Kuo et al. (2016) and Singh et al. (2017).
In addition, although the use of bioflocs serves to feed aquatic animals, it has low lipid content and therefore requires supplementation with natural foods, such as microalgae (Jung et al., 2017;Marinho et al., 2017). Inoculation of C. vulgaris together with Scenedesmus obliquus, in the culture system without water exchange, may improve the survival and development of cultured organisms (Jung et al., 2017). When cultured under mixotrophic conditions, Chlorella sp. showed higher lipid productivity (67-144 mg L -1 day -1 ) (Yeh and Chang, 2012).
In this context, this study evaluated the effects of supplementation with C. vulgaris and two molasses application rates on planktonic community, zootechnical performance, proximate composition and hematological indices in low-salinity biofloc systems.

Experimental conditions
All procedures were previously approved by the Ethics Committee on Animal Use of UFRPE under license number 129/2016. A 70-day indoor trial was conducted at the Sustainable Mariculture Laboratory (LAMARSU) of the Department of Fisheries and Aquaculture (DEPAq) of the Federal Rural University at Pernambuco (UFRPE), Recife, Brazil. The experiment had a 2 × 2 factorial design (Supplemented with C. vulgaris and molasses application rates) with the following treatments: BFT-C30 (Biofloc supplemented with C. vulgaris and molasses application rates of 30% of total daily feed); BFT-30 (Biofloc with molasses application rates of 30% of total daily feed); BFT-C50 (Biofloc supplemented with C. vulgaris and molasses application rates of 50% of total daily feed) and BFT-50 (Biofloc with molasses application rates of 50% of total daily feed), all in triplicate.
Five days prior to fish stocking, water from an indoor biofloc matrix tank (TAN 0.19 mg L -1 , N-NO 2 0.12 mg L -1 , N-NO 3 0.89 mg L -1 , alkalinity 160 mg CaCO 3 L -1 , pH 7.93, orthophosphate 1.76 mg L -1 , TSS 366 mg L -1 , SS 25 mL L -1 and salinity 10 g L -1 ) was mixed and equally distributed to fill twelve experimental black-plastic rectangular tanks (50 L) up to ~50% of the volume, with the remaining volume filled with clean water (10 g L -1 salinity). The experimental units were maintained under constant aeration by using three cylindrical air stones (diameter 2.4 cm and length 2.6 cm) per tank. No water exchange was carried out during the experimental period, except for the addition of dechlorinated freshwater to compensate for evaporation losses. Light intensity was at ~ 1000 lux with a 12h light/12h dark regime.
Sugarcane molasses (30% organic carbon) was applied daily (10:00 a.m.) to each tank as a source of carbohydrates to promote the growth of heterotrophic bacteria. Molasses inputs were based on a percentage of the daily feed allotments (by weight) with application rates of 30% (C:N -12:1) and 50% (C:N -20:1) of the total daily feed. Calcium hydroxide (Ca(OH) 2 ) was added to maintain the alkalinity (> 100 mg L -1 ) and the pH (> 7.5) in all treatments.

Supplemented with C. vulgaris
The microalgae C. vulgaris were obtained from Live Food Production Laboratory of DEPAq-UFRPE and cultures in a Provasoli medium. The culture was maintained at 25 ± 1 °C, salinity of 10 g L -1 , pH 7.9 and light intensity of ~2000 lux using a fluorescent lamp with a 24-h light photoperiod. The microalgae were supplemented every five days in the BFT-C30 and BFT-C50 treatments at a concentration of 5x10 4 cell mL -1 , corresponding to an addition of approximately 250 mL of microalgae to the tanks, regardless of the waste from unconsumed C. vulgaris.

Microbial activity
The methods used to estimate microbial activity were those described by Vinatea et al. (2010). Once weekly, water from the tanks was collected at a depth of 10 cm and placed in triplicate sets of 200 mL glass bottles (clear or black). Initial and final oxygen concentrations were measured with an YSI 556 digital oxygen meter (Yellow Springs, Ohio, USA). Once initial oxygen was recorded, the bottles were sealed with glass stoppers held in place by plastic lids of the same color as the bottle.
The bottles were attached to a rotating table (QUIMIS Q225M) and incubated for 2 h, 2000 lux and at 200 rpm, a rotation sufficient to keep the particles of flocs in the water column. Gross primary production (GPP), net ecosystem production (NEP) and water column respiration (R) rates were recorded by the classic dark and light bottle method using the following formulae: gross primary production (GPP) (Strickland, 1960).

Cyanobacteria and phytoplankton monitoring
Vertical water sampling was performed at the start and weekly using plastic bottles of 500 mL for collection. The water was filtered through a cylindrical-conical 70 and 50 µm net mesh with the purpose of reducing the amount of suspended solids in the sample) and then it was filtered with a 15 µm mesh, for phytoplankton and cyanobacteria retention. The phytoplankton and cyanobacteria were fixed with formalin (4%), buffered with borax (1%) and stored in 2.5-mL plastic recipients. A Sedgewick-Rafter chamber and binocular optical microscope (Olympus CH30) with magnification of 400x were used for identification at the genus level, with the aid of identification keys (Hoek et al., 1995;Bicudo and Menezes, 2006), and quantification of the phytoplankton and cyanobacteria samples (Pereira-Neto et al., 2008).

Proximate composition
Analysis of crude protein, crude lipids, moisture content and ash contents whole body fish, biofloc samples and commercial feed were performed at the beginning and at the end of the experiment, in triplicate using standard methods (AOAC, 2012) at the Laboratory of Physical-Chemical Analysis of Foods of the UFRPE. Ten fish, at the beginning, and three fish in each tank, at the end, were randomly selected and macerated for the analyses. The biofloc samples (30g) were collected with cylindrical mesh net of 50 µm for retention of solids. The commercial feed (30g) were randomly selected and macerated for the analyses. Crude protein was determined by measuring nitrogen (N·x 6.25) using the Kjeldahl method and crude lipids was by the ether extraction method with the Soxhlet apparatus. The moisture content was determined by drying it at 105 °C for 18 hours until a stable weight was attained and the ash content was determined by incineration in a muffle at 550 °C.

Fish stocking, feeding and monitoring
The sex reversed male fingerlings of Nile tilapia O. niloticus (with 1.1 ± 0.4 g mean weight) were from commercial hatchery (Piscicultura Vale da Mina, Paulista, Pernambuco, Brazil), and stored in plastic bags with water for transportation to the laboratory. The fish were acclimatized for 5 days before being placed into experimental units. The fish were stocked in a 350 L tank with clean freshwater, at density of 625 fish m -3 and fed at 10% of fish wet biomass with commercial feed (36% crude protein) adjusted daily according to the estimated fish consumption. Each day 2 g L -1 of salinity was increased with marine water for acclimatization from freshwater until 10 g L -1 salinity within 5 days. After this acclimatization all fish were maintained at desired salinity for ten days.
The experimental units were stocked with Nile tilapia (3.15 ± 0.5g initial weight) at a density of 680 fish m -3 (34 shrimp per experimental unit). The fish fingerlings were fed four times a day (at 08:00 a.m., 11:00 a.m., 2:00 p.m. and 4:00 p.m.), with a commercial fish feed (36% crude protein, 4% crude fat, 5% crude fiber, 12% moisture and 3.276,1 kcal kg -1 of digestible energy). The daily feeding rate was 8% of body weight at the start of the experiment, gradually reduced to 5% of body weight at the end of the 70-day experiment based on the weekly biometrics.
Fish weight (BEL Engineering M503 -0.001g) and length (ichthyometer) were monitored weekly (30% of population) in each experimental unit to determine biomass and survival. All fish 4/11 were counted weekly in each experimental unit for available survival. At the end of the experiment, biomass gain, final mean weight (W), final length, survival, feed conversion ratio (FCR), specific growth rate (SGR), protein efficiency ratio (PER), nitrogen retaining, yield, water consumption (WC) and sedimentation time (ST) were calculated, based on the following equations: final body nitrogen g initial body nitrogen g Nitrogen retaining x total nitrogen consumed g

Hematological assays
At the end of the experiment, ten animals from each experimental unit were collected, anesthetized with eugenol (1.0 mL L -1 ) and blood samples were collected from the caudal vessels by EDTA treated syringes. The hematological assays measured: hematocrit level (Goldenfarb et al., 1971); red blood cell count and mean corpuscular volume (MCV) (Wintrobe, 1934); and blood glucose levels using ACCU-CHEK ACTIVE blood glucose test meter (Roche), performed in the Aquatic Animal Health Laboratory of DEPAq-UFRPE. Moreover, during the weekly biometrics, external symptoms such as injuries, infection and other abnormal condition of fish body (integument and gills) were evaluated.

Statistical analysis
A two-way analysis of variance (ANOVA) was used to determine the effect of supplementation with C. vulgaris and molasses application rates of 30% and 50% of total daily feed and their interaction, after confirming homoscedasticity (Cochran p < 0.05) and normality (Shapiro-Wilk p < 0.05). Water quality variables, microbial activity, cyanobacteria and planktonic community were analyzed by performing repeated ANOVA measures. Tukey's test was used when differences between factors and treatments were detected (p < 0.05). Data analyses were performed using Statistica 10 software.

Microbial activity
The GPP (-0.002-0.076 mg O 2 L -1 h -1 ), NEP (-0.264-0.334 mg O 2 L -1 h -1 ) and R (0.274-0.350 mg O 2 L -1 h -1 ) were not significantly affected (p > 0.05) by supplemented with microalgae and molasses application rates (Table 3). There were fluctuations for GPP, NEP and R, however, the NEP presented negative values in all the treatments, throughout the experimental period, indicating that the environment was dominated by bacteria, even in the treatments where microalgae were added.
The Chlorella sp. was the most abundant genera in all treatments, but due to its addition in BFT-C30 and BFT-C50 treatments, they had higher average levels of relative abundance, 55.99% and 57.54%, respectively, followed by BFT-30 and BFT-50 treatments, with 41.17% and 44.89%, respectively (Table 4). Although there was no significant difference between treatments, the supplementations with C. vulgaris added influenced the abundance of Chlorophyta (Chlorella sp., Chlorococcum sp. and Tetraedron sp.) and Cyanobacteria (Aphanocapsa sp., Chroococcus sp. and Oscilatoria sp.). The relative abundance of Cyanobacteria was higher in the treatments without supplemented with C. vulgaris, while the relative abundance of Chlorophyta was higher in the treatment with supplemented with C. vulgaris. The data correspond to the mean of thirty replicates ± standard deviation by treatments. Mean values on the same row with different superscript letters differ significantly (p < 0.05); C = supplemented with C. vulgaris; M = molasses application rates; CxM = supplemented with C. vulgaris and molasses application rates; ns = not-significant (p >0.05); *p <0.05; ¥ Results from split-plot two-way ANOVA and Tukey's test; DO = Dissolved oxygen; TAN = Total ammonia nitrogen; PO 4 3 = orthophosphate; SS = Settleable solids; TSS = Total suspended solids; VSS = Volatile suspended solids; BFT-C30 = Biofloc supplemented with C. vulgaris and molasses application rates of 30% of total daily feed; BFT-30 = Biofloc with molasses application rates of 30% of total daily feed; BFT-C50 = Biofloc supplemented with C. vulgaris and molasses application rates of 50% of total daily feed; and BFT-50 = Biofloc with molasses application rates of 30% of total daily feed, all in triplicate. The data correspond to the mean of thirty replicates ± standard deviation by treatments. Mean values on the same row with different superscript letters differ significantly (p < 0.05); C = supplemented with C. vulgaris; M = molasses application rates; CxM = supplemented with C. vulgaris and molasses application rates; ns = not-significant (p >0.05); ¥ Results from split-plot two-way ANOVA and Tukey's test. GPP = Gross primary production; NEP = Net ecosystem production; R = Respiratory activity. Treatments' abbreviations as in Table 1.

6/11
Proximate composition The moisture, crude protein and lipids content in fish were significantly affected (p < 0.05) by molasses application rates and its interaction with microalgae addition (Table 5). While for the biofloc content the crude protein was significantly affected (p < 0.05) by molasses application rates. The proximal composition found for commercial feed (8.38 ± 1.25% moisture, 35.56 ± 1.31% protein, 2.61 ± 0.36% lipid and 6.31 ± 0.68% of ash) was different from the information described on the label, especially lipids.

Tilapia zootechnical variables
All the zootechnical performance variables were significantly affected (p < 0.05) by molasses application rates and their interaction ( Table 2). In the final weight, final length, survival, SGR, PER, nitrogen retaining, yield in the molasses application rates of 30% of total daily feed were higher compared with molasses application rates of 50%. However, FCR was lower in molasses application rates of 30% compared with 50%.  The data correspond to the mean of three ± standard deviation by treatments. Mean values on the same row with different superscript letters differ significantly (p < 0.05); C = supplemented with C. vulgaris; M = molasses application rates; CxM = supplemented with C. vulgaris and molasses application rates; ns = not-significant (p >0.05); *p <0.05; ¥ Results from split-plot two-way ANOVA and Tukey's test. Treatments' abbreviations as in Table 1.

Hematological assays
The hematocrit levels and erythrocyte counts were significantly affected (p < 0.05) by molasses application rates and their interaction, and glucose was significantly affected (p < 0.05) by interaction between the factors; however MCV was not significantly affected (Table 6). In the molasses application rates of 30% of total daily feed the hematocrit levels and erythrocyte counts were higher compared with molasses application rates of 50%. From the fourth week of experiment, signs of stress were observed, such excess of superficial mucus and hemorrhage at the bases of the pectoral and caudal fins and mouth of the animals in the BFT-50 and BFT-C50 groups.

DISCUSSION
During the experiment, the variables of temperature and dissolved oxygen were maintained in ideal values for the cultivated Nile tilapia (El-Sayed, 2006). In heterotrophic systems, such as in the biofloc system, it is expected that the levels of inorganic nitrogen compounds would be kept low due to direct conversion into microbial biomass as found in another study (Ebeling et al., 2006). The daily addition of a carbohydrate source to stimulate the growth of heterotrophic bacteria contributed to the reduction of the nitrogen compounds of water , furthermore, fluctuations did not occur in nitrogen (TAN, N-NO 2 , N-NO 3 ) concentrations and may be related to the fact that the experiment started with a previously matured biofloc.
The orthophosphate levels increased throughout the rearing time and ranged from 1.08 to 1.45 mg L -1 . The low levels of orthophosphate compared with those found by Luo et al. (2014) (~12.3 mg L -1 ) using the biofloc system, may be related to the abundance of the genera Chlorella (41-57% phytoplankton community) in all treatments. According to Ruiz et al. (2011) and Singh et al. (2017), C. vulgaris has a high potential for reducing the phosphorus of effluents.
In relation to alkalinity, biofloc systems lose buffering capacity and require frequent corrections (Azim and Little, 2008). The alkalinity levels were affected (p=0.01) by molasses application rates and their interaction. The BFT-50 and BFT-C50 treatments had higher alkalinity levels compared with BFT-30 and BFT-C30, probably in higher molasses application rates there is more heterotrophic microbial biomass in relation to the nitrifying bacteria. According Ebeling et al. (2006) the nitrification process promotes a more intense alkalinity reduction compared with the conversion of ammonium nitrogen into heterotrophic microbial biomass. The pH and alkalinity remained within the range considered adequate for tilapia cultivated by Avnimelech (2015) due to the addition of calcium hydroxide.
The mean values of settleable solids varied between 43 and 55 mL L -1 and were affected (p=0.02) by molasses application rates and their interaction. The higher molasses application rates (50% of the total daily feed, C:N 20:1) presented settleable solids values higher than 50 mL L -1 , in addition to denser and larger diameter bioflocs. Due to the high levels of solids, the treatments with higher molasses addition required a longer sedimentation time. Due to the use of the settler chamber, the values of SS and TSS were maintained at ideal levels by Avnimelech (2012), between 5 and 50 mL L -1 for SS and 200 and 500 mg L -1 for TSS. However, an increase of these values was observed throughout the rearing time, similar to that found by Long et al. (2015). The high concentrations of SS and TSS can cause discomfort in fish such as the accumulation of organic matter in the gills, which affect oxygen diffusion and also growth (Hargreaves, 2006;Avnimelech, 2012;Arantes et al., 2017). In a study of Litopenaeus vannamei culture in biofloc, Arantes et al. (2017) found that the use of a settler chamber promoted improved the zootechnical performance of shrimp, resulting in higher productivity, survival and final weight.
Based on the results of the net productivity, even supplemented with C. vulgaris the system presented negative values from the beginning of the experimental period. According to Avnimelech (2012), when using a C:N ratio greater than 10:1, it is possible to promote the succession and dominance of bacteria on the microalgae, thus corroborating the findings in the present study using the C:N ratio of 12:1 and 20:1, in treatments with 30% The data correspond to the mean of three ± standard deviation by treatments. Mean values on the same row with different superscript letters differ significantly (p < 0.05); C = supplemented with C. vulgaris; M = molasses application rates; CxM = supplemented with C. vulgaris and molasses application rates; ns = not-significant (p >0.05); *p <0.05; ¥ Results from split-plot two-way ANOVA and Tukey's test. MCV-Mean corpuscle volume. Treatments' abbreviations as in Table 1.
(55 -57%) may have reduced the density of cyanobacteria in BFT-C30 and BFT-C50. This result is similar as observed by Turker et al. (2003) who verified that Chlorella sp. reduces the abundance of phytoplankton, especially cyanobacteria in Partitioned Aquaculture System (PAS). On the other hand, Malbrouck and Kestemont (2006), mentioned that the excess of cyanobacteria in the aquaculture system can cause problems to species. Moreover, Miranda-Baeza et al. (2017) observed that the incorporation of the genus Oscillatoria had a significant negative effect on the survival and growth performance (<90% and around 19g) of tilapia in the biofloc system, especially in treatments with greater abundance this genus. The fish crude protein and lipids were affected (p < 0.05) by molasses application rates and their interaction. Jung et al. (2017) identified an influence of supplementation with C. vulgaris and Scenedesmus obliquus on the nutritional values (protein and lipids) of the tilapia cultivated in a biofloc system. However, in our study, we did not find effects just by supplementing with microalgae. This result may be related to the concentration and or frequency of supplemented microalgae. The values for proximal composition (53-80% crude protein and 4-27% lipids of the dry weight) of tilapia were similar to those reported by Azim and Little (2008), Luo et al. (2014) and Jung et al. (2017). The biofloc can also be used as a supplementary food source contributing close to 50% of protein requirement for fish (Avnimelech, 2007). The values for proximal composition (34-38% crude protein of the dry weight), were close to the values of protein content suggested for the diet in rearing tilapia (25 and 40%) (Craig and Helfrich, 2002;El-Sayed, 2006). The lipid content of the biofloc 1.67 ± 0.33% of the dry matter was relatively low, as were the results found by Jung et al. (2017). Several studies have reported low lipid levels in biofloc from fish (1.27 to 3.16% of dry matter) (Azim and Little, 2008;Luo et al., 2014). Low lipids levels in bioflocs may be related to low lipid content of the commercial feed used in this experiment (2.61% crude lipids).
Regarding the effect of salinity, there is an increase in energy expenditure as a function of osmotic regulation, since in fish, osmoregulation requires a high demand of metabolic energy, ranging from 20 to 50% of total energy expenditure, reducing the energy available for growth (Boeuf and Payan, 2001). Moreover, the low nutritional content of the commercial feed used probably influenced the reduced growth and higher FCR (2.03 to 5.86), due to the low lipid content found in the commercial feed (2.61%), different from the information described in the label. Lima et al. (2018) found similar values to those of the present study at a lower time (42 days), from 12.40 to 18.99g, 96.82 to 100% of the survival, and lower values of FCR (1.24-1.40) when cultivating tilapia at different densities in biofloc with 36% protein and 4% lipid feed. However, the difference in the feed conversion factor between treatments with 30% and 50% addition of molasses is due to the fact that the animals of the treatments with the highest application of molasses showed signs of stress. Fish exposed to stress situations presents changes in homeostasis, inducing changes in their physiological responses (Furuya, 2010). Even so, the yield values in BFT-30 and BFT-C30 were within the range recommended by Avnimelech (2015), between 10 to 40 kg fish m -3 .
The Zootechnical performance of Nile tilapia fingerlings was affected (p < 0.05) by molasses application rates and their interaction. However, the addition of microalgae C. vulgaris had no influence, as reported by Araújo et al. (2019), when evaluating different densities of C. vulgaris (2.5, 5 and 10x10 4 cell mL -1 ) in the culture of tilapia in biofloc with stocking density lower than the present study (250 fish m -3 ), and found final weight of approximately 21 g and survival greater than 80%. The addition of C. vulgaris microalgae may not have influenced fish performance because it is already abundant in the culture system, as we found when evaluating the phytoplankton community, where C. vulgaris represented 55.99% (BFT-C30) and 57.54% (BFT-C50) in addition treatment, and 41.17% (BFT-30) and 44.89%  in treatments without addition.
The final weight and length, FCR, SGR, PER and yield were higher in BFT-30 and BFT-C30 as compared than BFT-50 and BFT-C50. Pérez-Fuentes et al. (2016), Zapata-Lovera et al. (2017) and Liu et al. (2018) observed a higher zootechnical performance of Nile tilapia fingerlings in C/N (10:1) ratios compared with higher C/N (20:1) ratios, as in the present study, where treatments with 50% addition of molasses presented lower final weight at the end of the cultivation. It is probable that the lower levels of solids (Pérez-Fuentes et al., 2016) and the dominance of the mix of microalgae, heterotrophic bacteria and autotrophic bacteria is more beneficial for fish and shrimp growth (Xu et al., 2016). Moreover, in high C:N ratios there may be consumption of large amounts of oxygen (Liu et al., 2018).
Due to the stress and reduction of feed intake observed in fish from treatments with 50% addition of molasses presented lower values of nitrogen retention, and consequently a lower rate of protein efficiency when compared to the other treatments (BFT-30 and BFT -C30). However, as nitrogen retention depends on the digestible energy content of feeds (Kaushik and Oliva Teles, 1985;Einen and Roem, 1997), the values found in the present study were lower than those reported by other studies with tilapia (>30%) (Furuya et al., 2005;Righetti et al., 2011).
The highest WC was from the BFT-50 treatment (93.28 ± 12.89 L kg -1 ), while in the other treatments the consumption varied between 47.19-55.22 L kg -1 , showing a significant effect (p < 0.05) by supplemented with C. vulgaris and molasses application rates and their interaction. A similar result was found by Jung et al. (2017) where WC decreased about 82% when supplementation with two microalgae species were used. The WC was similar (52.48 L kg -1 with 500 fish m -3 ) that reported by Lima et al. (2018) and lower compared with 750 to 1250 fish m -3 (76.32 to 101.54 L kg -1 ).
A good indicator of fish health may be hematological indices (Martins et al., 2017). In the present study, stress caused by molasses application rates of 50% (C/N 20:1) of total daily feed (BFT-50 and BFT-C50) caused significant changes in the hematocrit level and the erythrocyte counts of the fish, resulting in macrocytic anemia. The values found in the treatments with molasses application rates of 30% of total daily feed (C:N 12:1) (BFT-30 and BFT-C30) remained in the ideal range described by several authors (20 to 32% for hematocrit and 1.9 to 5.0 x 10 6 µL -1 for erythrocyte counts) (Ueda et al., 1997;Signor et al., 2010;Costa et al., 2014;Martins et al., 2017). However, the values for treatments molasses application rates of 50% of total daily feed (BFT-50 and BFT-C50) were below the recommended level (21-44% hematocrit level and 1.50 to 3.76 x 10 6 µL -1 erythrocyte count) (Tavares-Dias, 2015).
In addition, the values found for mean corpuscular volume (MCV) were higher than those described by Signor et al. (2010), Salvador et al. (2013) and Tavares-Dias (2015) for Nile tilapia, between 113.4 and 170 fL. This increase in MCV is the first response of fish in order to compensate for the decrease in the number of erythrocytes, by carrying larger amounts of hemoglobin (Tavares-Dias et al., 2002). Higher glycemia values were obtained in the treatments with molasses application rates of 50% of total daily feed and without supplemented microalgae. Although the values found were within the range described by Tavares-Dias (2015), between 14.1 and 92.1 mg dL -1 , the increase in glucose levels was reported in fish when associated with stressful conditions, with elevated levels of cortisol and high blood glucose due to the hyperglycemic characteristic of cortisol (Iwama et al., 1999).
The low quality of the feed used in the experiment influenced the performance of the fish, along with the high rate of addition of molasses. A high amount of molasses can cause serious physiological problems in fish (Azim and Little, 2008;De-Schryver et al., 2008), for these reasons, attention is needed when adding this carbon source to the system and using high C:N ratios and in the quality of the feed.

CONCLUSION
In summary, the data of the present study show that the interactions between supplementation with C. vulgaris and molasses application rates, influenced zootechnical performance, ST, WC, cyanobacteria density, fish proximal composition, hematocrit and erythrocyte levels of Nile tilapia fingerlings cultured in low-salinity biofloc system. However, the main factor was molasses application rates of 30% (C:N 12:1) of total daily feed. Future research should also address the use of different densities and/or the frequency of supplementation with microalgae, the species used and the nutritional role of phytoplankton in the tilapia biofloc culture system.