A Simultaneous Denitrification and Fermentation Process for Nitrate Removal in a Whiteleg Shrimp (Litopenaeus vannamei) Recirculating Aquaculture System: Using Endogenous Carbon from a Biofilm

Abstract

During the process of circulating aquaculture, high concentrations of nitrate will accumulate. A simultaneous denitrification and fermentation process is described to remove nitrate from a recirculating aquaculture system using endogenous carbon on a biofilm. ¹⁵N isotope technology was used to assess the nitrate removal, mainly through heterotrophic denitrification. The nitrate removal rate could be as high as 98.97%, with a final concentration of nitrate below 1 mg/L. The denitrification process obeys a Michaelis–Menten-type enzyme kinetic model, with a half saturation constant of 99.91 mg/L and a maximum nitrate removal rate of 0.39 mg L⁻¹ h⁻¹ at 28 °C. The functional genes narG and narH for nitrate removal were obtained from Nitrospirae spp. at proportions of 39.13% and 26.16%, respectively. The acetate, propionate and iso-valerate produced by anaerobic fermentation provided the principal electron donors for denitrification.

1. Introduction

The whiteleg shrimp Litopenaeus vannamei constitutes 83% of global shrimp production [1]. A recirculating aquaculture system (RAS) can provide ideal environmental and management conditions for the high-density production of L. vannamei or fish [2]. Currently, nitrate usually accumulates in large amounts in RASs [3]. Long-term exposure to high concentrations of nitrate can lead to growth stress and even the death of L. vannamei [4]. The nitrate safety level for juvenile L. vannamei has been reported to be 60.05 and 127.61 mg/L, under salinities of 5 g/L and 10 g/L, respectively [5]. Finding simple, economical, and acceptable treatment strategies to solve the accumulation of wastewater nitrate has become one of the bottlenecks restricting the further development of RASs.

Biofilms are three-dimensional structures, and aerobic and anaerobic bacteria can colonize their aerobic surface regions and anaerobic inner regions, respectively [6]. In the past, aerobic nitrifying bacteria on biofilms have been commonly used to convert ammonia to nitrate in RASs [7,8], but the action of anaerobic microbes on biofilms has often been overlooked, and the endogenous carbon in biofilms has not been utilized at all. Recent studies have shown that RAS biofilms can effectively remove nitrate from aquaculture water, even at low COD/NO₃⁻-N (C/N) concentrations. For example, Lu et al. [9] developed a biofilm reactor to treat wastewater from an RAS, and found that ammonia, nitrite, and nitrate concentrations could be kept below 1 mg/L, with removal rates of 67%, 100%, and 100%, respectively, without adding extra organic carbon to the system. Feng et al. [10] used a biofilm reactor to treat RAS wastewater; under a C/N ratio of 1.61, the total nitrogen removal rate was 34.11%. Our previous study showed that biofilms could remove 98.07% of nitrate from L. vannamei aquacultural wastewater at an initial C/N ratio of 2.70 [11]. Lu et al. [9] found large amounts of anaerobic ammonia oxidizing bacteria (AnAOB) present in biofilms and postulated that anammox played an important role in the removal of nitrate from RASs, because the low C/N ratio was suitable for this process. Unfortunately, this study failed to provide direct evidence for the contribution of anammox using ¹⁵N isotope technologies [12]. In our previous study, predictions regarding the function of 16S rRNA showed that heterotrophic denitrification and fermentation metabolism, rather than anammox, occurred in RAS biofilms [13]. Making the best use of endogenous organic carbon from biofilms to remove nitrate from RASs is a more cost-effective solution and produces less exfoliated biomass. The actual mechanisms by which biofilms remove nitrate in aquacultural wastewater are currently unclear, and further exploration is needed.

Nitrate is a substrate of denitrification, and its concentration has a positive effect on the denitrification process. Nitrate reduction rates have been reported to increase with increasing nitrate concentrations in woodchip bioreactors, and become saturated when the nitrate concentration increases to 30–50 mg/L [14]. A similar phenomenon was found in a biofilm reactor; the rate of denitrification improved almost 1.5-fold when the nitrate concentration rose from 8.0 ± 0.5 mg/L to 12.5 ± 0.5 mg/L [15]. Generally, denitrification obeys the Michaelis–Menten model; few studies have determined the key parameters of this kinetic model for nitrate removal in biofilms in RASs.

The aims of this study were to (1) illustrate the mechanism of nitrate removal using endogenous carbon in biofilms; (2) explore the contribution of heterotrophic denitrification to nitrate removal in biofilms by using ¹⁵N isotope-tracing technology; (3) explore the influence of nitrate concentration on nitrate removal; (4) analyze the fermentation products produced during the nitrate removal process; (5) explore the coupling relationship between anaerobic fermentation and heterotrophic denitrification in biofilms; (6) identify the composition of the microbial communities on RAS biofilms; and (7) predict the potential simultaneous denitrification and fermentation metabolic pathways using metagenomics sequencing technology. These results could provide new strategies to solve the insufficient organic carbon during nitrate removal in RAS wastewater.

2. Materials and Methods

2.1. Sample Collection

The suspended carrier and aquaculture water used this study were derived over a long period at the Pond Ecological Engineering Research Center of the Chinese Academy of Fishery Sciences (30°95′ N, 121°16′ E), Songjiang District, Shanghai, China. The RAS consisted of a culture tank (7 m diameter × 1.5 m height, with a water volume of 45 m³), an in situ moving bed bioreactor (MBBR; working volume: 2 m³), and settling tank, yielding 250 kg of shrimps per batch. Solid waste from the system was removed via an inclined plate in the settling tank, after which the water was returned to the culture tank. The in situ MBBR was filled with high-density polyethylene suspended carriers (25 mm diameter × 4 mm height) covered with a biofilm responsible for the aerobic nitrification process. The water used for the shrimp culture was maintained at the following conditions: temperature, 28–30 °C; salinity, 5–7 g/L; and dissolved oxygen (DO) > 5 mg/L. The aquaculture water was not changed during the whole experimental period, with only small amounts of water being regularly replenished to make up for losses due to solid waste discharge and evaporation.

In the shrimp culture metaphase, the suspended carriers and aquaculture water were collected, stored at 4 °C, and immediately transported back to the laboratory. Laboratory-scale experiments on samples of the suspended carriers and aquaculture water were conducted immediately. Some of the suspended carriers were frozen at −20 °C for a subsequent metagenomic analysis. A 50 mL water sample was filtered through a 0.22 μm filtration membrane, then stored at −20 °C and sent for analysis at Majorbio (Shanghai, China) together with the suspended carrier samples.

2.2. Analysis of Potential Rates of NO₃⁻-N Reduction Using ¹⁵N-Tracer Technique

Suspended carrier samples (approximately 30 g each) were transferred to several 100 mL glass vials filled with He-purged 5‰ salinity ddH₂O. Three parallel assays were conducted for each suspended carrier sample. The mixture was first pre-incubated under anaerobic conditions for 24 h at 30 °C to deplete the NO_x⁻ and O₂ in the vials. Next, ¹⁵NaNO₃ was injected to each vial to reach 100 μM. Three initial samples were immediately added to 7 M ZnCl₂ (to produce a final concentration of 200 μM). Residual samples were anaerobically incubated at 30 °C and stopped after 8 h by adding 7 M ZnCl₂ (to produce a final concentration of 200 μM). The potential rate of heterotrophic denitrification to ammonia (DNRA) was assessed by tracking the rate of ¹⁵NH₄⁺ production, measured after conversion to N₂. Alkaline hypobromite (1 mL) was used to completely oxidize the ¹⁵NH₄⁺ to ³⁰N₂ during the 8 h incubation. For the anammox and denitrification processes, based on the amounts of ²⁹N₂ and ³⁰N₂ produced, the potential rates could be calculated. The ²⁹N₂ or ³⁰N₂ produced were measured using membrane inlet mass spectrometry. The formulas used to calculate denitrification, anammox, and DNRA were consistent with those described in [16].

2.3. Effect of NO₃⁻-N Concentration on Denitrification Rate

The denitrification performance of the biofilms under different NO₃⁻-N concentrations was investigated; samples of suspended carriers were added to 800 mL of aquacultural wastewater to make 1 L in total. The water quality indicators of the actual aquaculture wastewater are shown in Table S1. Under the action of aerobic microorganisms, the DO quickly dropped from over 5 mg/L to below 2 mg/L and remained stable. The initial NO₃⁻-N concentrations were set to 40, 50, 80, and 150 mg/L by adding KNO₃ (Sinopharm, Beijing, China). To evaluate the wastewater denitrification potential, the control test was set without adding biofilms, and N₂ was used to maintain the DO below 2 mg/L. All of the measurements were performed in triplicate. When the NO₃⁻-N concentration did not fall in the 40 mg/L group, the samples were aerated for 24 h (ensuring that the DO level was over 5 mg/L), so that the NH₄⁺-N and NO₂⁻-N produced in the anaerobic process through nitrification could re-oxidated to NO₃⁻-N. Subsequently, the aeration was halted, causing the DO to rapidly drop below 2 mg/L within 30 min, enabling the continued removal of the remaining NO₃⁻-N in the wastewater. This anoxic–aerobic process was repeated until there was no further reduction in the NO₃⁻-N in the water. The experimental temperature in the stationary culture was set at 28 °C; the concentrations of NO₃⁻-N, NO₂⁻-N, and NH₄⁺-N were measured every day, and the COD was measured every five days. During the experiment, the nitrate removal efficiency of the biofilms under each condition were calculated following Equation (1) and using a zero-order model to fit the rate of nitrate reduction during the first anoxic stage following Equation (2).

$$\mathrm{R}=\frac{{\mathrm{C}}_{\mathrm{n}} - {\mathrm{C}}_0}{{\mathrm{C}}_0} \times 100\%$$

(1)

$$C_t=-k_{0}t+{\mathrm{C}}_0$$

(2)

where R represents the nitrate removal efficiency (%); C_n and C₀ represent the NO₃⁻-N concentrations at the end and start of the experiment, respectively; C_t is the concentration of NO₃⁻-N (mg/L) at time (h); k₀ is the zero-order rate constant (mg L⁻¹ h⁻¹); and C₀ is a constant representing the concentration of nitrogen before reduction.

2.4. The Biofilm Fermentative Test

In order to further show whether organic carbon on the biofilm surfaces could provide electrons for heterotrophic denitrification under anoxic conditions, the following fermentation test was performed: The biofilms were washed with ultrapure water and then transferred to 1 L flasks. The filling ratio of the biofilms and ultrapure water were the same as in the denitrification batch tests described above, but without the added KNO₃. The experimental temperature was set at 28 °C in a stationary culture. Samples were taken each day and filtered through a filter (Whatman GF/C) to determine the NO₃⁻-N, NO₂⁻-N, NH₄⁺-N, and volatile fatty acid (VFA) levels.

2.5. Metagenomic Sequencing and Annotation

To gain a deeper understanding of the functional bacteria and metabolic pathways involved in the simultaneous denitrification and fermentation process, DNA was extracted from the biofilms and water for a metagenomic sequencing analysis by Majorbio Co., Ltd. (Shanghai, China). The extracted DNA samples were fragmented to an average size of about 300 bp to construct a pair-end sequencing library, and a sequencing analysis was performed according to Illumina’s (Illumina Inc., San Diego, CA, USA) standard procedures. The raw sequence data reported here were deposited in the NCBI Sequence Read Archive with accession number PRJNA1029501. The quality control, assembly, and gene prediction of the raw reads were conducted as described in [17]. The DIAMOND software package (http://www.diamondsearch.org/index.php (accessed on 1 June 2021)) was used for species taxonomy classification with an e-value cutoff of 1 × 10⁻⁵ to blast the unigenes from the NCBI NR database. The NCBI and Kyoto Encyclopedia of Genes and Genomes functional databases was used for functional annotation.

2.6. Measurements and Data Analysis

After filtration through 0.45 filters, the concentrations of NH₄⁺-N, NO₂⁻-N, and NO₃⁻-N were measured with standard methods [18]. The COD was measured using a Hach COD test kit (Hach Company, Loveland, CO, USA; low range 3–150 mg/L). A GC6890 gas chromatograph (Agilent Technologies, Santa Clara, CA, USA) equipped with a flame ionization detector was used for VFA detection following the protocols described by Wen et al. [19]. The temperature and DO were monitored using a Water Quality Meter AZ86031 (Hengxin, Taiwan, China).

The data were recorded as mean values with standard deviations, and all tests were conducted in triplicate. Office 2016 (Microsoft, Redmond, WA, USA) was used for data comparison and analysis. A variance analysis was used to evaluate the significance of the results using R (https://www.r-project.org/ (accessed on 21 August 2021)). The figures were drawn using the Origin 2018 software (https://www.originlab.com/ (accessed on 15 April 2021)).

3. Results and Discussion

3.1. Contribution of Different Nitrate Reduction Pathways

In order to identify the main metabolic pathway for nitrate removal in the biofilm of the L. vannamei RAS, we explored the contributions of heterotrophic denitrification, anammox, and DNRA with ¹⁵N isotope-tracing technology. The results indicated that the RAS biofilm exhibited these three nitrate removal processes simultaneously. Among them, the heterotrophic denitrification rate was 5.24 ± 0.28 mg N/m² d, and contributed 99% to the nitrate removal. The anammox and DNRA rates were both below 0.2 μmol/kg h (Figure 1), suggesting that heterotrophic denitrification was mainly responsible for the nitrate removal. During the ¹⁵N isotope-tracing test, the biofilms were exposed to ultrapure water without any additional source of organic carbon; the ¹⁵NO₃⁻ concentration was 100 μM. Thus, it can be inferred that the electrons required for heterotrophic denitrification in the RAS came entirely from the endogenous carbon in the biofilm. Similar heterotrophic denitrification (4.2 mg N/m² d) and DNRA rates (near to zero) in the RAS biofilms were detected using ¹⁵N isotope-tracing technology without the addition of organic carbon [20]. Aalto et al. [21] used a woodchip bioreactor for nitrate removal in an Oncorhynchus mykiss RAS, and similarly found that denitrification contributed to 77–95% of the nitrate removal. These results indicate that denitrification is a common phenomenon in RASs and is not only limited to the L. vannamei RAS that we studied.

3.2. Effect of Nitrate Concentration on Denitrification Performance

Nitrate is a heterotrophic denitrification substrate, and we set out to clarify the relationship between the initial nitrate concentration and nitrate reduction rate. Under different initial NO₃⁻-N concentrations ranging from 40 to 150 mg/L, the nitrate reduction rates of the biofilms were tested. The changes in the NH₄⁺-N, NO₂⁻-N, NO₃⁻-N, and COD levels under the different NO₃⁻-N concentrations are presented in Figure 2a–d. In general, the NH₄⁺-N, NO₂⁻-N, and NO₃⁻-N concentrations followed the same trend during the first anoxic phase (0~264 h). The NO₃⁻-N concentration decreased gradually over time from an initial concentration of 40.06 ± 1.22,47.81 ± 0.74, 74.04± 1.03, and 147.66 ± 0.52 mg/L, in the three treatments, respectively. At the same time, the NO₂⁻-N concentration rose at first to 2.38–6.98 mg/L, and then decreased to 0.02–0.16 mg/L. The NH₄⁺-N concentration increased to 5.02–6.80 mg/L. At the end of the test, the concentration of NO₃⁻-N eventually dropped below 1 mg/L in the three treatment groups with relatively low nitrate concentrations, and the nitrate removal rates were 98.07%, 98.29%, and 98.97%, respectively (Figure 2a–c). The NO₃⁻-N removal efficiency was only 62.51% in the highest nitrate concentration group (Figure 2d), although a COD still existed, resulting from the release of an endogenous carbon source, with the organic matter present being difficult to use for denitrification [13].

In this study, when the initial NO₃⁻-N concentrations were 40.06 ± 1.22, 47.81 ± 0.78, 74.04 ± 1.03, and 147.66 ± 0.52 mg/L, and the nitrate removal rates were 0.115, 0.135, 0.155, and 0.238 mg L⁻¹ h⁻¹, respectively, the correlation coefficients (R²) of the zero-order model were all greater than 0.9 (Figure 2e). When the initial nitrate concentration rose from 40.06 ± 1.22 to 147.66 ± 0.52 mg/L, the nitrate reduction rate increased about two-fold. This suggests that a higher initial nitrate concentration could promote nitrate removal. To better characterize the biofilm denitrification rate, the relationship between the initial nitrate concentration and denitrification rate was described using the Michaelis–Menten model (Figure 2f). The maximum nitrate removal rate (V_max) for biofilms under 28 °C was 0.39 mg L⁻¹ h⁻¹, and the half saturation constant (K_M) was 99.91 mg/L. The K_M value in this experiment was similar to that found by García-Martínez et al. [22] in a biofilm reactor (about 100 mg/L), but much higher than the K_M value found in a woodchip bioreactor [23]. The higher K_M value was due to biofilm formation in a high-flow/high-shear-stress environment, which led to poor diffusion and high mass transfer resistance in the static condition [24]. As the substrate concentration increases, the resistance to substrate diffusion decreases, leading to enhanced substrate utilization and the improved metabolism of microorganisms.

In this study, when the ratio of wastewater and biofilm carrier was 1:1, heterotrophic denitrification could remove 80 mg/L of nitrate from wastewater without an additional organic carbon source. This suggests that biofilm denitrification may have valuable applications in production practice.

To further prove the value of applying the denitrification properties of biofilms in an RAS, a 12-day L. vannamei culture experiment was conducted. The culture systems used in this experiment are shown in Figure S1. During the test, the concentration of NO₃⁻-N in the control group increased from 50.77 ± 1.53 to 98.78 ± 2.55 mg/L without denitrification equipment. In the experimental group, equipped with a denitrification reactor, the NO₃⁻-N concentration only increased from 50.36 ± 0.25 to 73.63 ± 2.64 mg/L (Figure S2). At the end of the culturing, the shrimps in the treatment control group began to die, with a survival rate of 78.95%, lower than the 97.83% of the experiment group. These results are similar to those reported previously: Neto et al. [5] calculated that the safety levels under salinities of 5 and 10 g/L for L. vannamei cultivation are 60.05 mg/L and 127.61 mg/L of NO₃⁻-N concentration, respectively. This experiment clearly demonstrated that nitrate could be effectively removed via biofilm denitrification processes without additional organic carbon during the L. vannamei culture in the RAS studied, therefore alleviating the stress effect of high concentrations of NO₃⁻-N on L. vannamei survival. In this study, the nitrate removal rate was still at a low level, possibly caused by the slow release of endogenous carbon from the biofilm.

3.3. Simultaneous Denitrification and Fermentation on the Biofilm

From the above information, it can be concluded that the electron donors required for heterotrophic denitrification mainly came from the biofilm. To further decipher the mechanism of heterotrophic denitrification under a low COD, a fermentation test was conducted to investigate the production of VFAs. In the group without added NO₃⁻-N (Figure 3a), acetate (1.95–13.69 mg/L), propionate (0–4.40 mg/L), and iso-valerate (0–3.45 mg/L) were detected, but no other VFAs were found. When NO₃⁻-N was added, the acetate was kept at a low concentration (0–2.74 mg/L) in the reaction system, and the other two VFAs were not detected (Figure 3b). Moreover, the concentration of NO₃⁻-N decreased from 36.85 ± 0.19 mg/L to 29.07 ± 0.06 mg/L during the experiment. This suggested that the fermentation was accompanied by heterotrophic denitrification. The electrons required for heterotrophic denitrification came from the VFAs produced by the fermentation processes of the biofilms. A similar distribution pattern of VFAs, with a high proportion of acetate, was also found by Ji et al. [25] in a sludge reactor. The temporary accumulation of nitrite during denitrification is not only related to the low C/N ratio in the water, but also to the distribution pattern of VFAs [26]. On the one hand, the higher activation energy needed for NO₂⁻-N reductase leads to a temporary inhibition during electron competition with NO₃⁻-N reductase in a COD-limited environment. This temporary inhibition dissipates at low NO³⁻-N concentrations [27]. On the other hand, the metabolic difference among the VFAs leads to differences in electron transfer routes, in turn leading to different nitrate and nitrite reduction rates [28]. Therefore, in future, the fermentation efficiency of endogenous carbon in biofilms should be further improved, and the distribution pattern of VFAs should be adjusted to improve the removal efficiency of NO₃⁻-N and reduce the temporary accumulation of intermediate NO₂⁻-N.

3.4. Mechanism of a Simultaneous Denitrification and Fermentation System

3.4.1. Functional Bacterial Community in a Simultaneous Denitrification and Fermentation System

To deeply understand the mechanism of a simultaneous denitrification and fermentation system, we analyzed the functional genes involved in the nitrate removal process (Figure 4). During the process of heterotrophic denitrification, nitrate was gradually reduced to nitrogen, catalyzed by nitrate reductase (narGHI and napAB encoding), nitrite reductase (nirSK encoding), nitric oxide reductase (norBC encoding), and nitrous oxide reductase (nosZ encoding). In general, the denitrification functional genes’ abundance in the biofilm was higher than in the water, which was consistent with the comparison between the denitrification rates in the biofilm and the aquacultural water (only 0.035 mg L⁻¹ h⁻¹) (Figure S3). As shown in Figure 4, during the process of nitrate reduction to nitrite in the biofilm, the relative abundance of narGHI was higher than that of napAB, suggesting that the narGHI gene was mainly responsible for the process. The results of the metagenomics analysis indicated that the narGHI gene mainly originated from Nitrospira (narG 10.80%, narH 26.16%), g_unclassified_f_Nitrospiraceae (narG 28.33%), and g_unclassified_c_Alphaproteobacteria (narG 12.45%, narH 14.31%). The contribution of Nitrospirae spp. to the narG and narH genes was as high as 39.13% and 26.16%, respectively. In addition, Nitrospirae spp. may also be crucial during the process of nitrite reduction to NO, because their contribution to the nirK gene was 23.71%. These results indicated that Nitrospira spp. could well be the main contributors to nitrate reduction under anaerobic conditions. This could be explained by the mixotrophic lifestyle of Nitrospira in anaerobic environments; compared with formic (1.0%) and propionate (2.1%) environments, the relative abundance of Nitrospira is relatively high in the presence of acetate (5.3%), and these organics can also donate electrons for nitrate reduction [29]. Additionally, in long-term anaerobic conditions (DO: 0.17 ± 0.08 mg/L), there was an observed increase in the copy numbers of the 16S rRNA gene. These numbers rose from 2.61 × 10⁸ to 1.67 × 10¹⁰ (copies/g of dry sludge) within a single-stage partial nitrification–anammox reactor [30]. It was demonstrated that low DO and organic material levels favored the growth of Nitrospira in full-scale municipal wastewater treatment plants [31]. Therefore, we hypothesized that Nitrospira in the RAS biofilm could not only oxidize nitrite to nitrate under aerobic conditions, but also reduce nitrate to NO under low DO and C/N ratios.

In this research, Nitrospira spp. did not contain a complete suite of heterotrophic denitrification functional genes; thus, cooperation with other microorganisms in the biofilm was needed to further reduce NO to N₂. During the process of NO reduction to N₂O, g_unclassfied_p_Bacteroidetes (norB 17.99%, norC 40.16%), g_unclassfied_p_Acidobacteria (norB 15.81%), and g_unclassfied_p_Planctomycetes (norB 12.29%) played an important role. NosZ mainly originated from g_unclassfied_p_Bacteroidetes (13.95%) and g_unclassfied_p_Chloroflexi (15.01%). These microorganisms also provided annotated denitrification functional genes in a fermentation reactor [17]. G_unclassified_p_Chloroflexi, g_norank_f__Anaerolineaceae, g_norank_f__Saprospiraceae, g_unclassified_f__Comamonadaceae, and g_Streptomyces, which have been reported as closely related to fermentation, were identified with abundances of 7.10%, 1.50%, 0.58%, 0.16%, and 0.12%, respectively (Figure S4), [25,32]. These results suggest that denitrification and fermentation could be coupled by these microorganisms in anaerobic environments.

3.4.2. Utilization of Endogenous Organic Carbon from Biofilms

The potential metabolic mechanism in simultaneous denitrification and fermentation is depicted in Figure 5. Within biofilms, phosphohexokinase (EC 2.7.1.11, 23,250–25,672 reads), D-amino acid dehydrogenase (EC 1.4.5.1, 7962–8572 reads), and acyl-CoA dehydrogenase (Acyl-CoA, EC 1.3.8.7, 25,024–26,976 reads) were notably enriched. These enzymes play a role in converting glucose, amino acids, and fatty acids into pyruvate, suggesting their strong potential for pyruvate production. Pyruvate serves as a crucial precursor substrate of the tricarboxylic acid (TCA) cycle, generating a surplus of NADH during this process. In addition, pyruvate converts to VFAs through a series of biochemical reactions, also producing NADH [33]. This excess NADH can be utilized as a direct electron donor for denitrification. During the denitrification process, NADH is oxidized to NAD⁺ to provide electrons for the reaction. The produced electrons travel along the electron transport chain to the nitrate reductase (NAR), nitrite reductase (NIR), nitric oxide reductase (NOR), and nitrous oxide reductase (NOS), which gradually reduce nitrate to nitrogen.

The key enzyme for acetate production is acetate kinase (EC 2.7.2.1, 5220–5776 reads). It was present at a higher level than propionate CoA-transferase (EC 2.8.3.1, 1120–1426 reads), which converts propionyl-CoA to propionate. This result was consistent with the higher acetate production in the fermentation test. Butyrate kinase (EC 2.7.2.7, 6–18 reads) catalyzes butyrate generation and was present at low levels, which could explain the absence of butyrate during fermentation. In addition, the enzyme 2,3-dihydroxyisovalerate dehydratase (EC 4.2.1.9, 13,342–14,226), related to iso-valerate production, was also abundant. The active valerate production pathway could be explained by the large amount of high-protein feed (more than 40%) put into the RAS every day. It was found that higher levels of protein resulted in higher valerate production [34]. However, during the fermentation tests, more acetate was produced than iso-valerate, possibly as a result of the complex acetate generation process. This acetate was produced simultaneously during valerate production [35].

The coenzymes F420 hydrogenase (EC 1.12.98.1, 1440–1478 reads) and tetrahydromethanopterin S-methyltransferase (EC 2.1.1.86, 606–772 reads) were in low abundance during the process of methanogenesis. There were indications that methanogenesis was weaker than acidogenesis during anaerobic fermentation. The main aim of the simultaneous denitrification and fermentation process is to achieve a balance between maximizing the production of VFAs while avoiding methanogenesis. Our results indicated that the biofilm metabolic pathway is conductive to denitrification [25,32], a result which has valuable applications in the future.

4. Conclusions

Through a process of simultaneous denitrification and fermentation, a biofilm using endogenous carbon could remove 98.97% of the nitrate in aquacultural wastewater, with a final nitrate concentration below 1 mg/L. Heterotrophic denitrification contributes more than 99% to nitrate removal. The denitrification process obeyed the Michaelis–Menten enzyme kinetic model, with a V_max of 0.39 mg L⁻¹ h⁻¹ and a K_M of 99.91 mg/L, at 28 °C. The acetate, propionate, and iso-valerate produced by anaerobic fermentation were the principal electron donors for denitrification. The heterotrophic denitrification functional genes mainly originated from Nitrospira, g_unclassified_f_Nitrospiraceae, g_unclassfied_p_Bacteroidetes, and g_unclassfied_p_Acidobacteria. By analyzing the enzymes related to fermentation, the fermentation mainly stopped at acidogenesis, which could drive nitrate removal more effective using an endogenous carbon source.