Effects of high wheat bran input on the performance of a biofloc system for Pacific white shrimp ( Litopenaeus vannamei )

Applying an external carbohydrate source to stimulate heterotrophic bacteria growth in biofloc systems is a common practice which could be simplified by combining the carbohydrate source with the feed into one pellet. In the current study, such an approach was tested in Pacific white shrimp, Litopenaeus vannamei . Wheat bran (WB) containing non-starch polysaccharides (NSP), a hardly digestible carbohydrate, was mixed with a commercial-like shrimp diet (control; CONdiet) to create a wheat bran rich diet (WBdiet). Shrimp were fed isonitrogenously, resulting in the WBdiet group receiving equal absolute amount of nitrogen but more carbon than the CONdiet group. A digestibility trial was performed in a clear-water recirculating aquaculture system for 35 days, using 6 replicates aquaria per diet. Compared to the CONdiet, shrimp fed with the WBdiet had a reduced overall digestibility and an increased faecal carbon to nitrogen ratio (C:N ratio) from 12 to 20. Following the digestibility trial, a second experiment was performed to monitor the growth performance of the shrimp reared in biofloc systems for 42 days, using three replicate tanks per diet. The dissolved nitrogen species concentration was lower in the tanks fed the WBdiet, while the dissolved carbon concentration was similar between the diets. From a system perspective, similar amounts of nitrogen accumulated across the system compartments with both diets, with maximum 3% total nitrogen loss reached at the end of the experiment. The percentage of carbon loss per kg feed was similar between diets. Overall, diluting the CONdiet with wheat bran and feeding both diets iso-nitrogenously did not reduce shrimp survival and had minor effect on growth (2% reduction of the specific growth rate). Therefore, incorporating NSP-rich ingredients, such as wheat bran applied as one pellet has a potential to simplify the carbon addition management in a biofloc system.


Introduction
Biofloc technology has been widely studied and applied in shrimp culture (Bossier and Ekasari, 2017;Panigrahi et al., 2020;Tinh et al., 2021b).In a biofloc system, the rearing tank is also an active sludge tank, in which metabolic wastes and uneaten feed are converted into biofloc.The principal actors in a biofloc system are heterotrophic bacteria, which break down organic matter for the synthesis of their own biomass (Khanjani et al., 2022).In this process, also potentially harmful inorganic nitrogen is converted into bacterial biomass, thus contributing to maintaining good water quality.These heterotrophic bacteria produce an extracellular polymeric substances (EPS) matrix, which aggregates microbiota into biofloc particles (Wilen et al., 2008;Osemwegie et al., 2020).Biofloc can be in situ consumed by the shrimp as an additional food source, enhancing production and lowering the feed conversion ratio (FCR) (Burford et al., 2004;Tacon and Metian, 2015;Ray et al., 2017).A biofloc system improves organic matter retention at system level in shrimp and biofloc biomass, as well as dissolved nutrients in the water (De Schryver et al., 2008;Cardona et al., 2015;Khanjani et al., 2023).In a biofloc system, carbon is required to assimilate nitrogen into bacterial biomass.The optimal carbon to nitrogen ratio (C:N ratio) in a biofloc system has been reported to be between 15 and 20 (Avnimelech, 2009;Wei et al., 2016).To achieve the optimal C:N ratio, the current common practice is to add a carbon source directly into the water besides the feed, which is labor intensive and prone to errors.Another approach can be to incorporate the carbon source directly into the pelleted feed.The feed then acts as a single-nutrient-input-package for both shrimp and biofloc, and as such simplifies biofloc management and reduces the workload of the farmer.
The type of carbon source, supplemented via the diet, may differently affect biofloc development and shrimp performance.A previous study showed that combining an easily digestible carbon source (e.g. corn starch) with feed into one pellet did not improve the growth and the nutritional content of biofloc, compared to feed without corn starch addition (Tinh et al., 2021c).Corn starch is easily digested by shrimp leading to quick carbon dioxide (CO 2 ) release (Rosas et al., 2001), resulting in limited carbon left in the shrimp faeces.Consequently, insufficient energy is available for heterotrophic respiration in the biofloc tank.Wheat bran, on the other hand, includes non-starch polysaccharides (NSP), which are complex carbohydrates that shrimp hardly digests (Kaushik et al., 2022).Therefore, dietary inclusion of wheat bran in the pelleted diet, could be an alternative way to provide organic carbon to the biofloc via the undigested fraction of NSP in the faeces.Dietary inclusion of NSP, however, can reduce the digestibility of other macronutrients, which was shown for Nile tilapia and rainbow trout (Sinha et al., 2011;Haidar et al., 2016;Staessen et al., 2019).To date, information on NSP digestibility in shrimp is still lacking, while knowledge on the effect of a high dietary NSP input on biofloc system performance is limited.
Biofloc studies performed during the last two decades have demonstrated that the type of carbohydrate influences biofloc formation and activity (Wang et al., 2016;Xu et al., 2016) to a large extend due to differences in the degradation velocity between types of carbon sources (Vilani et al., 2016).Molasses, which contains simple sugars, dissolves and mineralizes faster than rice bran, which contains more complex carbohydrates in the form of NSP (Serra et al., 2015).Variation in the degradation rates of the different types of carbon sources can induce non-simultaneous nutrient availability, thereby affecting bacterial growth.For instance, adding a carbon source which is slowly degraded may slow down the immobilization of inorganic, potentially harmful, nitrogen waste, while an easily degraded carbon source may lead to excess biofloc production and rapid fluctuations in water quality.Therefore, the types of carbon can affect biofloc production and water quality dynamics (Ekasari et al., 2014;Rajkumar et al., 2016).
Here, we hypothesized that NSP is indigestible by shrimp and therefore dietary wheat bran inclusion will increase the C:N ratio of the faeces.The higher availability of organic carbon relative to nitrogen entering the system through the faeces and branchiary and urinary loss by the shrimp will provide more energy for bacterial mineralization and thus stimulate biofloc formation.Stimulation of the biofloc production will raise natural food availability and enhance immobilization of inorganic (toxic) nitrogen species and help to maintain good water quality during culture.We mimicked the biofloc management practice in the field by giving additional carbon input to the system.However, we supplemented carbon through the feed, as opposed to the conventional approach which deliver it separately from feeding.Therefore, this study investigated the effects of increasing carbon input via dietary wheat bran supplementation in pelleted diet (wheat bran diet; WBdiet) on nutrient digestibility of Pacific white shrimp (Litopenaeus vannamei) and on system performance (shrimp, water quality, biofloc, and nutrient mass balances).Mature biofloc water was used to ensure that the system receiving no extra carbon (control diet; CONdiet) would not collapse when exposed to nitrogen waste from the shrimp excreta.Our study focused not only on animal growth, but also on the system's components, which included both biotic and abiotic factors, providing a comprehensive understanding of the system's dynamics.While the CONdiet might appear to be better for shrimp, the WBdiet would be better in terms of system performance.

Materials and methods
The experiment followed the protocol from Carus Animal Research Facilities of Wageningen University and Research (WUR), which assured the animal welfare.The stocking density and water quality conditions were maintained according to water quality limits for shrimp (Mohanty et al., 2018).Two experiments were executed: a digestibility experiment with daily faeces collection in clear water aquaria without biofloc and a growth performance experiment in biofloc mesocosm tanks.The treatment consisted of wheat bran-rich diet (WBdiet), which was compared to a control diet (CONdiet).The digestibility experiment used aquaria as experimental units with 6 replicates per treatment, while the growth experiment involved mesocosm tanks with 3 replicates for each treatment.Both experiments used completely randomized designs.

Experimental animals and diets
Pacific white shrimp were obtained from Crevetec, Ternat, Belgium.Two diets were formulated (Table 1).One diet served as the control diet with a composition representative for a commercial shrimp diet, further referred to as the CONdiet.The second diet was formulated diluting the ingredient mix of the CONdiet with wheat bran, an ingredient rich in NSP.This diet is further referred to as the WBdiet (Table 1).The percentages of soya lecithin, monocalcium phosphate (Ca(H 2 PO 4 ) 2 ), calcium carbonate (CaCO 3 ), cholesterol, vitamin and mineral-mix (premix), and yttrium (Y) in the WBdiet were kept similar to the concentrations in the CONdiet to make sure the nutrient requirement of Pacific white shrimp was also met with the WBdiet (Table 1).Both diets contained 0.02% yttrium oxide (Y 2 O 3 ) as an inert marker for digestibility measurement.
Shrimp were fed daily continuously using mechanical belt feeders for 10-12 h starting at 16.00.In the WBdiet treatment, the daily feeding rate gradually declined as the shrimp grew from 11.9% to 3.3% body weight (BW), assuming that the FCR gradually increased from 0.6 to 1.1 during the experiment.Diluting the CONdiet with wheat bran reduced the diet protein content.The amount of CONdiet fed daily to the shrimp was calculated aiming to feed the same absolute amount of protein as with the WBdiet.Consequently, the dietary input of protein was similar, but the input of carbohydrate was higher in the WBdiet.

Digestibility experiment
Determining digestibility is the first step in evaluating the usefulness of feed ingredients in formulated diets for target species (Ayisi et al., 2017).Apparent digestibility was determined as first described by (Cho et al., 1982).The feed digestibility experiment was executed in clear-water recirculating aquaculture system (RAS) in absence of biofloc.Shrimp were grown in 120-L aquaria connected to the water purification unit of one RAS, with 6 replicate aquaria per diet.Per aquarium, 25-30 shrimp of 13 ± 2 g were stocked.Each aquarium was continuously aerated to maintain the dissolved oxygen concentration above 6.5 mg O 2 L − 1 and the water temperature at 27 ± 1.0 0 C. Shrimp were fed overnight as explained in Section 2.1.To collect faeces, each morning per aquarium 1-1.5 g extra feed was administrated, observing a 10-minute time interval between the aquaria.One hour after feeding, all uneaten feed, faeces and carapaces were removed from each aquarium by siphoning.Again, one hour later, the shrimp faeces in each aquarium were collected and stored in a collection bottle that was put on ice in a bucket, to slow down deterioration until the faeces from all tanks were collected.De-ionized water was added to the bottle to dilute remaining culture water containing salt from siphoning process, and to rinse off salt that was attached to the faeces.The de-ionized water was then removed from the bottle, eliminating the salt from the faeces sample.Per treatment, the rinsed faeces were pooled into one aluminum tray and stored at − 20 0 C before drying at 70 0 C for 72 h.The dried faeces were then manually ground using a porcelain mortar, and stored in a plastic jar at room temperature, until proximate analysis.The experiment lasted 35 days.

Growth experiment
Biofloc inoculum water was taken from the previous culture cycle.Prior the stocking day, inoculum was pooled in a large tank, mixed, and kept in suspension through aeration.Subsequently, from the large tank filled with mature biofloc, 450-L water was transferred to each of 6 mesocosm tanks.The water volume in each tank was increased by adding a mix of salt and fresh water until reaching a volume of 750 L and a salinity of 23 ± 1.0 ppt.The shrimp were stocked the next day at stocking density of 100 ind/mesocosm tank.Besides the diet, no external carbon input was added to the mesocosm tanks.Freshwater was added every week to each tank to compensate for evaporation and splashing of small water droplets due to aeration (84 ± 7 L / week).During the experiment, water quality was checked daily and maintained at 27.2 ± 0.6 0 C, 7.0 ± 0.1 mg O 2 L − 1 , 8.0 ± 0.1 pH, 23.2 ± 1.0 ppt salinity and 36.7 ± 1.5 µS cm − 1 conductivity.These water quality parameters were measured using a multi-parameter electronic meter (WTW Multi3630-IDS™).In addition, each day in each biofloc tank the ammonium (N-NH 4 ) and nitrite (N-NO 2 ) concentrations were measured using colorimetric test kits (mg/L, MColortest™, EMD Millipore), while the nitrate (N-NO 3 ) concentration was checked with colorimetric test strips (mg/L, MQuant®, EMD Millipore).

Sample collection and preparation
In each biofloc tank, feed, shrimp, biofloc (particulate matter retained on a 1.5 µm glass fiber filter) and water that passed through the glass fiber filter were sampled.Samples were collected at the beginning of day-1 (D1), at the end of day-21 (D21) and at the end of day-42 (D42).To avoid stressing the shrimp, no shrimp samples were taken at D21.
To sample biofloc and water, 2-L closed jars was positioned in the middle of the tank, opened to fill and closed when full.Per biofloc tank, ten 2-L jars were filled, and subsequently pooled in a 50-L plastic bucket, and mixed at 300 rpm.While still mixing, four 2-L jars were filled with mixed sample.Of these, the water from 2 jars was filtered in batches of 100 mL through a 1.5 µm pore size filter to collect biofloc.The 100 mL samples were collected while continuously mixing the water in the jar with a magnetic stirrer (Heidolph Mr Hei-Mix L, Heidolph Instruments, Germany) at 300 rpm.Once the biofloc-water was filtered, 100 mL of demi water was poured into the filtration chamber and filtered to dilute and remove the salt from the filter.When no more water in the filter could be removed by the vacuum pump, the filtration was stopped.The filter having the biofloc was folded and stored in a clean, dry tube.Samples were then kept in − 20 0 C freezer.Biofloc accumulated on the pore filters was used to determine dry matter (DM), ash, calcium (Ca), magnesium (Mg), total phosphorus (TP), total suspended solid (TSS) and volatile suspended solid (VSS), protein, energy, C:N ratio, chemical oxygen demand (COD) and chlorophyll-a (Chl-a).In addition, for Chl-a analysis, also 3 unfiltered water samples were collected.Each analysis was performed in triplicate.Filters containing biofloc were dried in the oven at 70 0 C for 3 days before being analyzed.For COD and Chl-a parameters, biofloc-containing filters were not dried prior the analysis.
The filtrate after sampling was pooled and thoroughly mixed before collecting a 50-mL sample that was acidified with 3 N hydrogen chloride (HCl) to reduce the pH to less than 3.A volume of 12 mL of acidified water was then filtered using 0.45 µm pore size syringe filter.The final filtrate was then stored in a plastic tube with cap and put aside until analysis within the next 24 h.Of the remaining two jars, one jar was acidified with 3 N HCl, while the other jar was kept neutral.They were stored at − 20 0 C as back-up sample sources.
A representative 100 g shrimp sample was collected from the base population at the start of the experiment (D1), while at the end of experiment (D42), all shrimp in each biofloc tank were collected, counted and batch weighed.Shrimp samples were kept at − 20 0 C until being freeze-dried for 5 days and then ground for the proximate analysis (AOAC, 2020).Twenty g of the CONdiet and WBdiet was added weekly to 1 large container per diet and used to determine the proximate composition of the feed at the end of the experiment.Sample of dried faeces, shrimp and feed, were stored in plastic containers at room temperature until further analysis.

Sample analysis
The proximate composition (DM, ash, protein, fat, and energy) of shrimp, feed, faeces and biofloc were determined by the Aquaculture and Fisheries (AFI) laboratory of WUR.The DM concentration was calculated after oven-drying each sample at 103 0 C for at least 4 h until constant weight following ISO (1999).After DM determination, the samples were incinerated at 550 0 C for at least 4 h until constant weight following ISO (2002).Ash-free dry matter (AFDM) is calculated by subtracting ash from DM.The TSS concentration was the amount of DM per liter culture water, while the VSS concentration was the weight loss after incineration of the TSS per liter of culture water (APHA, 1995).The ash was analyzed for total phosphorus, calcium, magnesium, and yttrium using plasma-mass spectrophotometry (ICP OES) following the NEN 15510 procedure.The crude protein analysis was determined by the Kjeldahl method according to ISO (2005), while energy content was determined using bomb calorimetry by direct combustion (IKA® werke C7000; IKA Analysentechnik, Weitershem Germany) (ISO 1998).The fat analysis was carried by Soxhlet (ISO 1999).The carbon (C) and nitrogen (N) content were determined using DUMAS analyzer (Leco CN 628, Leco Instrumente GmbH., Germany).The C:N ratio (mass) was calculated by dividing the C content with and N content.
Water samples were analyzed using a segmented flow analyzer (SAN++, Skalar Analytical B.V., The Netherlands), measuring total organic carbon (TOC), total inorganic carbon (TIC), total carbon (TC), ammonium N-(NH 4 ), nitrate+nitrite N-(NO 3 +NO 2 ), total organic nitrogen (TON), total nitrogen (TN), and orthophosphate (P-PO 4 ).The biological oxygen demand (BOD 5 ) analysis was performed according to A. Vinasyiam et al.APHA (1995) with a modification of the incubation time.A dark-colored glass bottle was filled with 500 mL biofloc water from the culture tank.A magnetic stir rod was put into each glass bottle.The bottle was then incubated on an underwater magnetic stir plate in an open water bath at 27 ± 1 0 C, mimicking the temperature in the culture tank.Water in the bottle was continuously stirred to maintain the biofloc in suspension.During preliminary incubation runs, all dissolved oxygen (DO) in the bottle was consumed in less than 5 days.Diluting the biofloc water with autoclaved-artificial saltwater did not result in consistent and logical DO concentration values, probably due to oxygen consumption by salt.Therefore, the analysis was done in duplicate, filling two 500 mL DO bottles: one bottle (replicate A; replicate A) was kept unopened until the end of the incubation, and the other (replicate B; Rep B) was opened and closed every hour to measure the DO concentration.Before closing bottles of Rep A and Rep B at the start of the incubation, the DO concentration was recorded.When the DO concentration in Rep B dropped below 2, the incubation of Rep A was terminated, and the DO concentration was measured.The time period between the start to end of incubation of each sample was recorded.Only the dissolved oxygen consumption in Rep A was considered to calculate the BOD 5 by extrapolating the observed DO consumption to a 5-day period.
The COD concentration was analyzed based on protocol of ISO (1989), meanwhile the Chl-a was measured according to the standard protocol 10200 H issued by APHA (1995).The biodegradability index (BI) is calculated as BOD 5 /COD (Rojas-Tirado et al., 2018).The microbial activity was determined using hydrogen peroxide (H 2 O 2 ) degradation assay according to Pedersen et al. (2019).

Calculation and data analysis
Parameters of shrimp's performances, such as the apparent digestibility coefficient (ADC), the specific growth rate (SGR), shrimp production, feed conversion ratio (FCR) with and without wheat bran (FCR carbon excl.), protein efficiency ratio (PER), as well as protein and energy retention efficiency, were calculated as follows: (3) The energy retention efficiency (%) = (retained energy (kJ) / energy input (kJ) ) × 100 Nut faeces (g/kg) and Nut diet (g/kg) are the nutrient concentrations contained in faeces and diet, in g/kg; Yttr diet (g/kg) / Yttr faeces (g/kg) are the Yttrium concentrations contained in faeces and diet, in g/kg; W day-42 (g) and W day-1 (g) are the biomass of shrimp at the end of day-42 and at the beginning of day-1 in each aquarium or biofloc tank, in g (wet); water volume ( m 3 ) is the total water volume in the biofloc tank in m 3 , diet input (g) and wheat bran input (g) are the total absolute amount of diet and additional wheat bran given per tank within 42 days culture period, in g (wet); retained protein (g) and retained energy (kJ) are the protein and energy retained in shrimp within 42 days culture period, in g (dry) and kJ, respectively; protein input and energy input are the total absolute amount of protein and energy from diet given per tank during the 42 days culture period, in g (dry) and kJ, respectively.
Parameters of system's nutrient balance, such as total nutrient input and nutrient loss were calculated using the formulas: Nutrient loss (g) = Total nutrient input (g) -(Nut shrimp day-42 (g) + Nut biofloc day-42 (g) + Nut water day-42 (g) ) (10) Nut diet (g) is total nutrient in the diet; Nut shrimp day-1 (g) , Nut biofloc day-1 (g) , and Nut water day-1 (g) are the nutrient present in the shrimp, biofloc, and water, at the beginning of day-1, respectively; Nut shrimp day-42 (g) , Nut biofloc day-42 (g) , and Nut water day-42 (g) are the nutrient present in shrimp, biofloc, and water, at the end of day-42, respectively.Statistical analyses were performed using the IBM Statistics software (IBM Corporation, NY, USA).The effect of diet on ADC, faecal composition, water quality, as well as shrimp growth and body composition were analyzed using one-way ANOVA (GLM).The biofloc performance, including quality, quantity and activity, were analyzed using repeated-measure ANOVA (GLM).The nutrient distribution between each compartment in each biofloc tank at each sampling time was compared between the diets with one-way ANOVA (GLM).The normality and homoscedasticity assumptions were checked using the Shapiro-Wilk and Levene's tests.If the assumptions of the ANOVA were not met, then the data were analyzed nonparametrically using a ranking test.However, ranking test done showed similar results as those obtained using ANOVA.When there was significant interaction (P < 0.05) in the result of the repeated measure ANOVA, a one-way ANOVA was performed combining all possible factor combinations followed by a post-hoc Tukey analysis.

Digestibility experiment
The ADC and the proximate composition of the faeces of shrimp reared in biofloc tanks fed the CONdiet and WBdiet are listed in Table and Table 3, respectively.The ADCs of all measured parameters including protein, fat, energy, carbohydrate, phosphorus, calcium, magnesium and carbon, were lower in shrimp fed the WBdiet compared to shrimp fed the CONdiet (P < 0.05) (Table 2).In addition, the faecal proximate composition of shrimp fed the WBdiet was different from shrimp fed the CONdiet (P < 0.05), except for energy and carbon (P > 0.05).The concentrations of crude protein, phosphorus, calcium and magnesium were higher in the faeces of shrimp fed the CONdiet than in faeces of shrimp fed the WBdiet (P < 0.05), while the concentrations of fat and carbohydrate in faeces were lower (P < 0.05) (Table 3).

Table 2
Effect of wheat bran addition in the diet on the apparent digestibility coefficient (ADC) of nutrient in the diet of Litopenaeus vannamei.Values are the mean and the standard deviation (sd) of each diet (CON = control diet, WBdiet = wheat bran diet).The data was observed from digestibility trial.P-value = probability value.*Calculated by subtracting ash, crude protein and fat from dry matter.

Shrimp
The shrimp's growth performance and proximate composition are summarized in Tables 4 and 5, respectively.The amounts of feed and protein fed were controlled; therefore, no statistical analysis was performed.No significant differences were observed in the final individual weight, individual weight gain, biomass gain and production between diets (P > 0.05).The specific growth rate was higher in CONdiet (P = 0.05), even though the difference with the CONdiet was only 2%.No differences in protein efficiency ratio, survival and protein retention efficiency were observed between the diets (P > 0.05).The FCR of the WBdiet (1.70) was 44% higher than the CONdiet FCR (1.18) (P < 0.05).The shrimp's body composition was also similar between the dietary treatments.Exceptions were the crude protein body content of shrimp, which was 3.4% higher in shrimp fed the CONdiet, and the phosphorus body content which was 7.9% higher in shrimp fed the WBdiet (P < 0.05).

Biofloc
The nutritional quality of biofloc in terms of AFDM, crude protein, energy, carbon content and C:N ratio, were similar in tanks fed either the CONdiet or WBdiet (P > 0.05, Table 6).Diet also did not affect the TSS and VSS concentration in rearing tanks (P > 0.05, Table 6 and Fig. 1).However, TSS and VSS concentrations changed over time (P < 0.05).The TSS concentration decreased between D1 and D21 and remained similar thereafter (Table 6).There was a significant diet by time interaction for VSS, with a higher concentration in tanks fed the WBdiet than in tanks receiving the CONdiet at the end of the experiment on D42, while differences between the diets were less pronounced during the first half of the experiment (P < 0.05) (Fig. 1).The BOD 5 measurement on D1 samples were not successful.Based on data collected in D21 and D42, the BOD 5 was higher in mesocosms tanks fed the WBdiet (P < 0.05).Organic matter degradability, as indicated by the biodegradability index, was similar between tanks fed the CONdiet and WBdiet (P < 0.05).Similarly, no differences in Chl-a concentrations were observed between the CONdiet and WBdiet (P > 0.05).This was also the case for microbial activity in the biofloc tanks (P > 0.05), even though the microbial activity decreased between D1 and D21 and subsequently increased between D21 and D42 (P < 0.05) (Table 6).

Water quality
No impact of the diet was observed on the concentrations of TOC, TIC and TC in water (P > 0.05), while time had a significant effect on the concentration of TIC, the latter increasing between D1 and D21 and subsequently being lowest on D42 (P < 0.05, Table 7).The N-NH 4 Values are the mean and the standard deviation (sd) of each diet (CONdiet = control diet, WBdiet = wheat bran diet).The data was observed from digestibility trial.CN ratio = carbon to nitrogen ratio, P-value = probability value.*Calculated by subtracting ash, crude protein and fat from dry matter.

Table 4
Dietary input and effect of wheat bran addition in the diet on the growth performance of the Litopenaeus vannamei at biofloc system level.Values are the mean and the standard deviation (sd) of each diet (CON = control diet, WBdiet = wheat bran diet).CN ratio = carbon to nitrogen ratio, P-value = probability value; *Calculated by subtracting ash, crude protein and fat from dry matter.
concentration in the water was unaffected by diet (P > 0.05) but decreased over time (P < 0.05).In contrast, the concentrations of TON, TIN, TN and N-(NO 3 +NO 2 ) were lower in tanks fed the WBdiet, as compared to tanks fed the CONdiet (P < 0.05).In addition, the concentrations of all dissolved nitrogen species increased over time (P < 0.05).The dynamics of TON, TIN, and TN are displayed in Fig. 2. A diet by time interaction was only observed for the P-PO 4 concentration.Data show that the difference between the two diets gradually increased over time (P < 0.05; Table 7).

Nutrient balances (carbon, nitrogen and phosphorus)
The amounts of nitrogen, phosphorus and carbon present in shrimp, biofloc, and water in the biofloc rearing tanks on D1, D21 and D42 are shown in Fig. 3.The dietary nitrogen input was similar between the CONdiet and WBdiet, whereas more phosphorus and carbon were administered with the WBdiet.At the start of the experiment on D1, similar concentrations of nitrogen, phosphorus and carbon were present in shrimp, biofloc and water in the biofloc tanks of both dietary treatments (P > 0.05).This was also the case for nitrogen on D21 and D42 (P > 0.05), and no differences in nitrogen input and nitrogen loss were observed during the experiment (Fig. 3a).Meanwhile, more phosphorus accumulated in biofloc and water in tanks fed the WBdiet (P < 0.05, Fig. 3b).The unaccounted phosphorus refers to phosphorous species other than PO 4 -P, as only the latter was measured in filtrate water.The amounts of carbon present in shrimp, biofloc and water on each sampling date were always similar between diets (P > 0.05, Fig. 3c), but considering the whole experimental period, more carbon was lost from biofloc tanks fed the WBdiet than from tanks fed the CONdiet (P < 0.05).
The distribution of nitrogen, phosphorus and carbon, as percentage of the total amount present at each sampling date in tanks fed the CONdiet and WBdiet, is shown in Fig. 4.During the experiment, most nitrogen was present in the water (Fig. 4a), whereas phosphorus was mainly present in biofloc (Fig. 4b).For carbon, the highest fraction was lost (Fig. 4c).The percentages of nitrogen, phosphorus and carbon retained in the biofloc tanks on D42 were 97-101%, 78-82%, and 39-44% respectively, concurring with losses of 0-3% for nitrogen, 18-22% for phosphorus and 56-61% for carbon (Fig. 4).

Effect of high wheat bran input on nutrient digestibility
In this study, the aim of increasing the carbon input through dietary wheat bran inclusion was to raise the C:N ratio in the shrimp faeces and the availability of organic carbon to stimulate biofloc formation and activity.Adding wheat bran in the diet indeed decreased the overall ADC of nutrients (Table 2) and altered the shrimp faeces composition (Table 3).This was suggested to be related with their NSP contents.Wheat bran contains 56% NSP (dry), mostly in the form of arabinoxylans (70%) and cellulose (24%) (Maes and Delcour, 2002;Stevenson et al., 2012;CVB, 2022).Including 40% wheat bran into the CONdiet increased the dietary NSP content by 22%, and this resulted in a 28% reduction of ADC of AFDM, 11% of crude protein, 22% of fat, 28% of energy and 41% of carbohydrate (Table 2).Due to these differences in  ADC, the C:N ratio in the faeces increased from 12 in the CONdiet to 20 in the WBdiet (Table 3), which was also what this inclusion was aiming for.Negative values for the ADC of phosphorus, calcium and magnesium might be due to salt contamination of the collected faecal samples even after rinsing with de-ionized water.
Based on earlier studies in fish, NSP affects digestion, absorption, and metabolic processes by altering digesta viscosity, gastric emptying rate, gut morphology and the intestinal microbiota composition (Sinha et al., 2011).In addition, in fish the production of NSP-digestive enzymes such as β-glucanases or β-xylanases is low, and NSP degradation is assumed to be mainly due to microbial fermentation in the gut (Sinha et al., 2011;Maas et al., 2020).It remains, however, unknown whether NSP degrading enzymes are synthesized in shrimp and if NSP fermentation is taking place.A decrease in the ADC of shrimp fed with diets containing NSP-rich ingredients such as lupin kernel meal (Molina-Poveda et al., 2013), high protein distiller's dried grains (Qiu et al., 2017) and soyabean meal (Cruz-Suárez et al., 2009;Fang et al., 2016) has been reported, although the magnitude to which NSP reduces nutrient digestibility varies between studies.This may be affected by NSP type, NSP dietary inclusion level, the pre-treatment method applied on NSP-rich dietary ingredients, or by a combination of these factors (Cruz-Suárez et al., 2009;Sinha et al., 2011).Further research is needed to understand the mechanism how WBdiet reduced the ADC in Pacific white shrimp.

Shrimp
Shrimp in both treatments were fed equal amounts of protein, and showed a lower protein digestibility when fed the WBdiet, the similar protein retention in shrimp most likely was due to shrimp grazing on biofloc.The specific growth rate of shrimp fed the WBdiet was statistically lower than the growth of shrimp fed the CONdiet, although in absolute terms the difference was small (1.8%; Table 4).For the other performance parameters (biomass gain, production, and growth rate), no significant differences were observed.This suggested that supplementing a high input of complex carbon source did not give major impact on the shrimp growth, when protein and phosphorus input did not limit the shrimp growth (Prabhu et al., 2013).Eating biofloc as an additional food might compensate for potentially reduced shrimp growth due to wheat bran inclusion in the diet.Previous studies suggested that biofloc can improve feed utilization in shrimp by either stimulating synthesis of endogenous digestive enzymes or by using exogenous digestive enzymes obtained by eating biofloc (Wang et al., 2016;Panigrahi et al., 2021).In this study, biofloc consumption by shrimp resulted in similar production and protein retention efficiency between both diets (Table 4).The level of biofloc contribution to shrimp growth, however, could not be calculated in this study.To carry out such analysis, ones can compare the growth of shrimp reared in biofloc and clear water systems or conduct a nutrient isotope study.Another limitation of this study was that the nutrient digestibility of the feed might have been underestimated because the digestibility trial was done in clear water system with no biofloc consumption by shrimp.Values are the mean of three sampling times of each diet (CON = control diet, and WBdiet = wheat bran diet) and the mean of two diets of each sampling times (D1 = day-1, D21 = day-21, and D42 = day-42).TON = total organic nitrogen; TIN = total inorganic nitrogen; TN = total nitrogen; IC = inorganic carbon; TOC = total organic carbon; TIC = total inorganic carbon; TC = total carbon, SEM = standard error of the mean, P-value = probability value.For each factor (diet or time), different letters in bold show significant difference (P < 0.05).7.

Water quality and biofloc
We hypothesized that the higher input of carbon-rich faeces coming from the WBdiet would stimulate a higher biofloc formation, in terms of TSS and VSS, compared to the CONdiet.A decline of TSS observed after the start of the experiment in both treatments may be related to a reduction in the organic load via lower feed load in the systems compared to where the biofloc were maturated before.Another explanation might be owing to the age of the biofloc, where mature biofloc aggregation could degrade after a period of time (Martins et al., 2020).At the start of the experiment, the unforeseen influx of organic carbon from decaying biofloc and photosynthetic carbon assimilation were larger than the extra influx through shrimp faeces in tanks fed the WBdiet, which could have masked the effect of the latter (Tinh et al., 2021c).This could also explain why the WBdiet did not increase the final TSS concentration or lead to differences in biofloc composition compared to the CONdiet (Table 6).Mature biofloc had the in situ carbon source to maintain the biofloc needs (Samocha et al., 2007;Xu et al., 2016;Martins et al., 2020).However, we hypothesized that this independence would only remain shortly without external carbon supplies, as the part of the in situ carbon loss out of the system in form of ).Loss = (total nutrient inputtotal nutrient present in the system) at the particular day.The total nutrient input = total nutrient from feed + total initial nutrient present in the system.Feed input = total nutrient input from feed given in 42-day experiment, unacc.phosphorus = unaccounted phosphorus in the water, in the form other than orthophosphate.Different letters on top of the bar showed significant difference (P < 0.05) between the diet.The feed input has no variation within diet treatment, and therefore was not statistically analyzed.
CO 2 via microbial respiration.This was supported by our finding, the VSS concentration started to be higher at the end of experiment (D42) in tanks fed the WBdiet compared to the CONdiet (Fig. 1).In CONdiet, with a low external carbon input, the biofloc decaying rate outcompeted its development rate.Meanwhile, in WBdiet tanks, with higher external carbon input, the opposite occurred.This, therefore, could explain why biofloc VSS concentration was higher on D42 in tanks fed the WBdiet compared to the CONdiet (Fig. 1).
At the end of the experiment on D42 the N-content in biofloc was 41% and 75% higher in the CONdiet treatment and WBdiet treatment, respectively, than on D1 (Table 6).This concurs with the higher organic matter accumulation shown by BOD 5 and a trend (P < 0.1) for higher microbial activity in the WBdiet than in the CONdiet tanks (Table 6).An increase in organic matter correlates positively with an increase in microbial biomass, consisting of mainly heterotrophic bacteria in the biofloc (Ray and Lotz, 2014;Xu et al., 2018).At the end of the experiment, the higher organic matter accumulation in WBdiet tanks suggested greater biomass of heterotrophic bacteria, compared to the CONdiet tanks.In addition, there was an indication of more intensive heterotrophic mobilization of nitrogen in the WBdiet tanks, as seen from the lower final TON concentration, despite the higher faecal nitrogen input (due to lower dietary protein ADC).Despite these differences, the concentrations of TON and ammonium were maintained low in both treatments, throughout the experiment.
Like heterotrophic bacteria, autotrophic bacteria also contributed to the water quality maintenance, but it is expected that this group has a low biomass consisting of only a minor fraction of the total biofloc (Ebeling et al., 2006).An increase in the TIN concentration in both treatments during the experiment, mainly as nitrate, the final product of nitrification, indicated the presence of active autotrophic bacteria in both treatments.This finding confirms previous studies that showed active nitrifying bacteria in mature biofloc (Krummenauer et al., 2014;Emerenciano et al., 2017).Adding external carbon input to a system with mature biofloc, the expectation is that heterotrophic bacteria would gradually reduce the contribution of autotrophic bacteria in controlling the ammonia while producing nitrite and nitrate (Xu et al., 2016).This actually happened in our study, as seen from the lower accumulation of N-(NO 3 +NO 2 )(Table 7) in the WBdiet tanks compared to the CONdiet.However, it was slower than expected as the difference was small (7%, Table 7).Adding more carbon to the diet to make a major change in the nitrite and nitrate concentrations may risk the shrimp performances.Further research is needed to investigate an adequate C:N ratio of the WBdiet for both shrimp and system performances.
In biofloc, algae can produce in-situ organic carbon and immobilize nutrients (Emerenciano et al., 2017).In this study, we measured Chl-a to get an indication of algae development in the biofloc tanks.The total Chl-a concentrations were statistically similar between the dietary treatments (Table 6).However, the trend showed that diet affected the Chl-a distribution within the system.In the WBdiet system, Chl-a tended to be more profound in the biofloc aggregate than in the water, contrasting the condition observed in the CONdiet (Table 6).This could be because WBdiet had more abundant EPS matrix due to the increased number of heterotrophic bacteria or that diets generated different types of algae in terms of size and aggregation ability.Further in-depth analysis of the algal composition is necessary to confirm the second notion.
Our study was the first to apply the H 2 O 2 degradation analysis for microbial activity analysis in the biofloc water sample.It measures the enzyme activity of bacteria, algae, and protozoans (Rojas-Tirado et al., 2018;Pedersen et al., 2019), and shows it to be a promising technique to measure microbial activity in biofloc systems.The measurement is significantly considerably less time-consuming than BOD 5 analysis (half day vs 5 days).More trials were still needed to establish a standardized protocol for biofloc sample.

Nutrient mass balances
By feeding iso-nitrogenous, similar amounts of nitrogen were fed to all biofloc tanks during the experiment.Nevertheless, with the two diets, similar amounts of nitrogen were present in all tank compartments during the experiment.The overall nitrogen retention relatively to the nitrogen input in the system in both treatments was 97-100% (Fig. 4a).This retention was higher compared to the 77-87% reported in previous studies (Tinh et al., 2021a;Tinh et al., 2021c).The high retention efficiency of nitrogen could be due to the low ammonia-N concentrations, keeping ammonia volatilization low.In addition, the dissolved oxygen concentrations remained high throughout the study, reducing the room for anoxic and anaerobic conditions, favorable to the loss of nitrous oxides from the biofloc tanks, to develop (Ebeling et al., 2006).
Based on previous study, the phytic acid content in wheat bran, fluctuates between 3.0 and 8.5 mg/g (Pramitha et al., 2021), which is higher than for most other dietary ingredients, causing the P-PO 4 content in the WBdiet to be higher than in the CONdiet (Table 7).Two grams more phosphorus was fed per kg feed on a dry matter basis with the WBdiet than with the CONdiet.Unfortunately, because total ).Loss = percentage of absolute loss relatively to total nutrient present in the system, at the particular day.Unacc.phosphorus = unaccounted phosphorus in the water, in the form other than orthophosphate.
phosphorus was not measured during this experiment, the phosphorus loss (P-loss) from the system could not be accurately measured and is therefore referred to as unaccounted phosphorus (Fig. 3b).
For carbon, the amount lost was higher in tanks fed the WBdiet, yet the percentage loss was similar between diets (56-61%).This makes it likely that NSP was biodegraded by the microbiota resulting in a similar biodegradability index (0.41), observed for both diets (Table 6).The 0.41 biodegradability observed in this experiment is in the same range as reported by Rojas-Tirado et al. (2017).A biodegradability index ranging between 0.3 and 0.6 indicates that the biodegradability of the organic matter in the biofloc tanks was average (Srinivas, 2008).

Conclusions
Incorporating wheat bran in the diet and feeding isonitrogenously decreased the overall nutrient digestibility and increased the faecal C:N ratio in Pacific white shrimp.However, this study concluded that no major differences were observed in biofloc formation or shrimp performance during a 42-day culture period.From this perspective, the WBdiet treatment was not an improvement, but it could serve as alternative and simplified method for biofloc management practices.Nevertheless, it is worth noting that the organic matter accumulation in biofloc fed WBdiet started to surpass that in the CONdiet at the end of culture period, suggesting a potential long-term effect of WBdiet on biofloc formation.Additionally, in situations where new biofloc develops during the culture period and no algae are present, the WBdiet may perform better than the CONdiet in biofloc formation.It would lead to a higher availability of biofloc as additional food for shrimp growth.No difference in protein retention efficiency was observed in this study between the dietary treatments, suggesting that biofloc contributed to shrimp nutrition.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 1 .
Fig.1.Effect of wheat bran addition in the diet on volatile suspended solid (VSS) concentration in the water.Values are the mean and the standard deviation of each diet (CONdiet = control diet, and WBdiet = wheat bran diet) at each sampling times (D1 = day-1, D21 = day-21, and D42 = day-42).Data were analyzed with 2-way ANOVA, with diet as main factor and time as repeated measure: P-value of diet= 0.041, time= 0.000 and inter-action= 0.003.Different letters on top of each bar indicate significant difference (P < 0.05) based on one-way ANOVA considering the 6 diet x time combinations, followed by Tukey's test.

Fig. 2 .
Fig. 2. Effect of wheat bran addition in the diet on concentrations of dissolved nitrogen: total organic nitrogen (TON) (a), total inorganic nitrogen (TIN) (b), and total nitrogen (TN) (c).Values are the mean and the standard deviation of each diet (CONdiet = control diet, and WBdiet = wheat bran diet) at each sampling times (D1 = day 1, D21 = day 21, and D42 = day 42).No significant differences between treatments are shown in the figure because there was no interaction between the dietary treatments x time.P-values for the factors diet and time and their interactions (diet x time) are given in Table7.

Fig. 3 .
Fig.3.Effect of wheat bran addition in the diet on nitrogen (a), carbon (b) and phosphorous (c) distribution in the biofloc system (g absolute amount / tank).Values are the mean and the standard deviation of each diet (CONdiet = control diet, and WBdiet = wheat bran diet) at each sampling times (D1 = day-1, D21 = day-21, and D42 = day-42).Loss = (total nutrient inputtotal nutrient present in the system) at the particular day.The total nutrient input = total nutrient from feed + total initial nutrient present in the system.Feed input = total nutrient input from feed given in 42-day experiment, unacc.phosphorus = unaccounted phosphorus in the water, in the form other than orthophosphate.Different letters on top of the bar showed significant difference (P < 0.05) between the diet.The feed input has no variation within diet treatment, and therefore was not statistically analyzed.

Fig. 4 .
Fig. 4.The effect of wheat bran addition in the diet on the distribution of nitrogen (a), phosphorous (c), and carbon (c) in the biofloc system (in percentage of total particular nutrient at each sampling times).Values are the mean of each diet (CON = control diet, and WBdiet = wheat bran diet) at each sampling times (D1 = day-1, D21 = day-21, and D42 = day-42).Loss = percentage of absolute loss relatively to total nutrient present in the system, at the particular day.Unacc.phosphorus = unaccounted phosphorus in the water, in the form other than orthophosphate.

Table 1
Diet formulation used in this experiment.
Values are the mean of each diet (CONdiet = control diet, WBdiet = wheat bran diet.CN ratio = carbon to nitrogen ratio.*Calculatedbysubtracting ash, crude protein and fat from dry matter.**AnalyzedusingDUMAS analyzer.A.Vinasyiam et al.

Table 3
Effect of wheat bran addition in the diet on the proximate quality of feces of the Litopenaeus vannamei.

Table 5
Effect of wheat bran addition in the diet on the body composition of the Litopenaeus vannamei in biofloc system.

Table 6
Effect of wheat bran addition in the diet on biofloc quality, quantity and activity during the experiment.

Table 7
Effect of wheat bran addition in the diet on the water quality.