Two mining companies, 1) a platinum mine situated in South Africa (Pt-SA) and 2) a diamond mine in Lesotho (D-Lst) with contamination alerts of high nitrate concentrations (over 100 mg/L-NO₃ ^-) in their wastewaters, were used as study sites for the project. The names and precise locations of the companies cannot be revealed due to a non-disclosure agreement (NDA) that is in place. Site 1: Pt-SA is an underground platinum (Pt) mine situated in the Northwest Province of South Africa with water around the mine reported to contain elevated NO₃ ⁻ concentrations from the nearby mine wastewater catchment and sewage drainages. The mine has a wastewater catchment for potential purification and reuse in the mining processes; however, the water has not been successfully treated, hence its storage. Site 2: D-Lst is a diamond mine situated on the north-eastern side of the Kingdom of Lesotho, a landlocked country within South Africa. The mine is an open cast mine with water catchments near the river for which if it overflows during heavy rains, the water will easily overflow into the river causing a potential threat to the community using water from the river. Both mines utilize tons of ammonium-based explosives, which are most likely to be the source of elevated NO₃ ⁻ concentration in their wastewater catchments.

Water samples were collected using sterile 25 L polyethylene drums. The drums were rinsed three times with site water before collection. A generator-powered water pump connected to a two-way hose pipe was used to transfer water from the catchments into the drums. The collecting end of the hose pipe was immersed 50 cm vertically into the water source before pumping into the sampling drum. The following chemical parameters were measured on site: pH and total dissolved solids (TDS) using an ExStik^®II probe; Model EC500 (EXTECH Instruments). On arrival at the laboratory, the working samples were harvested from the drums and hydrogeochemical analysis was performed within 24 h of sample storage for further water sample characterization.

The working samples were subjected to complete chemical analysis to determine their physicochemical characteristics. The concentration of dissolved anions, such as NO₃ ⁻, SO₄ ^2-, and Cl⁻, was determined using ion chromatography (IC), while total organic carbon (TOC) and CaCo₃ were determined using a TOC analyzer. Concentrations of elemental cations in the wastewater samples (i.e., Ca, Mg, K, P, Na, Fe, Al, Mn, Ni, and Zn) were determined by inductively coupled plasma-optical emission spectroscopy (Prodigy 7 High Dispersion ICP-OES). For the ICP-OES sample preparation, 200 mL of each wastewater sample was acidified to a pH of less than 2 with HNO₃ (2%) and filtered through 0.22 µm pore-sized filters. The samples were then stored at 4°C until they were sent to the Institute of Groundwater Studies (IGS, RSA) at the University of the Free State (UFS, RSA) for the abovementioned chemical analysis.

Molecular analysis of the bacterial content in the water was conducted by concentrating bacterial cells through centrifugation (Eppendorf-Centrifuge 5810R) of 400 mL of the two wastewater samples (6000 RCF, 10 min) in 50 mL portions. The pellets were combined for each sample and used for genomic DNA (gDNA) extraction using the NucleoSpin^® Soil kit (Macherey-Nagel, Germany) according to the manufacturer’s instructions. The quality (pure dsDNA concentration and ds/ssRNA contamination concentration) and integrity (DNA sheering) of the DNA were determined using a NanoDrop 3300 fluorospectrometer (Thermo Scientific, Waltham, MA, United States) and 1% agarose gel stained with ethidium bromide (EtBr), respectively. Agarose gels were visualized under UV illumination with a Chem-Doc XRS (Bio-Rad Laboratories) gel documentation system as a quality control measure.

To explore microbial communities in the wastewater samples, the extracted gDNA samples were subjected to Illumina MiSeq sequencing analysis at the Centre for Proteomic Genomic Research (CPGR), Cape Town, South Africa. The sequence libraries were prepared by amplifying approximately the 460 bp region located in the hyper-variable V3/4 region of the 16S rRNA gene with overhang Illumina adapter overhang nucleotide sequences (Klindworth et al., 2013). A polymerase chain reaction (PCR) was performed with a broad-spectrum 16S rRNA primer set: forward primer: 5′-TCG TCG GCA GCG TCA GAT GTG TAT AAG AGA CAG CCT ACG GGN GGC WGC AG-3′ and reverse primer: 5′-GTC TCG TGG GCT CGG AGA TGT GTA TAA GAG ACA GGA CTA CHV GGG TAT CTA ATC C-3’ (Fadeev et al., 2021). The 16S V3/4 amplicons were purified using the Agencourt-AMPure XP bead clean-up kit (Beckman Coulter Genomics, Danvers, MA, United States), followed by a second amplification to attach dual indices and Illumina sequencing adapters using the Nextera XT Index kit (Illumina, San Diego, CA, United States). Final purification using the AMPure XP bead clean-up kit (Beckman Coulter Genomics) was carried out, followed by library quantification, normalization, pooling, and denaturing before being subjected to a 2 × 301 cycle sequencing on the Illumina MiSeq using the MiSeq v3 reagent kit (Illumina). The obtained 16S rRNA gene sequence data were analyzed using QIIME v1.9.1 as described by Caparoso et al. (2010) with adopted modifications by Cason et al. (2020). Operational taxonomic unit (OTU) tables were rarefied to 39,787 reads per sample. Metabolic activities of OTUs detected in the wastewater samples were predicted using Functional Annotation of Prokaryotic Taxa (FAPROTAX) v1.2.2. Analyses of the abundance tables were carried out using R v3.6.1 (www.r-project) (R Development Core Team, 2011) and the phyloseq package (McMurdie and Holmes, 2013).

In all dynamic and optimization experiments, monitoring of parameters was conducted by regular sampling from batch cultivations with defined media, incubation periods, and conditions. Cultivations were carried out in 100 mL volumes in 150 mL serum vials for anaerobic cultivations and 250 mL Erlenmeyer flasks for aerobic cultivations unless otherwise reported. Samples were collected using sterile syringes into Eppendorf tubes, and the analysis was conducted within 1 h after collection using standardized methods described as follows based on the parameters monitored. All batch experiments were incubated at a constant temperature of 30°C in a shacking incubator rotating at 150 rpm over a defined period according to each experiment. Microbial growth in each culture was monitored by measuring the optical density at 600 nm (OD_600nm) using a Spectronic^® GENESYSTM 5 spectrophotometer. Fresh medium was used to blank the spectrophotometer prior to analyzing samples. The concentrations of dissolved N-species, NO₂ ⁻ and NO₃ ⁻, were determined using the Griess assay adopted from Bhusal and Muriana (2021). Fresh reagents used in the assay were prepared every 14 days during the analysis with the validation of their efficacy tested using stand solutions. The amount of biogenic gaseous N-species, N₂O and N₂, produced from denitrification experiments was determined by analyzing collected gas samples from vials using gas chromatography (GC) (SHIMADZU Scientific Instruments). Two different columns were used to analyze and verify N₂ and N₂O percentages due to the interference of Ar(g) used as an inert gas in the vials. Helium was used as a carrier gas in both analyses in which N₂ gas was analyzed using the Molsieve GC capillary column (Agilent J&W CP-Molsieve 5Å) and N₂O gas was analyzed using the Shincarbon-ST column (Okada et al., 2005). The TCD injector, oven, and detector were preheated to 200°C prior to the manual gas injection. Owing to the instability of NO that spontaneously and rapidly oxidizes to NH₄ ⁺ or NO₂ ⁻ and eventually NO₃ ⁻, it was not analyzed but accounted for when analyzing NO₃ ⁻ in the medium. Concentrations of carbon sources used in the experiments: glucose and glycerol, were determined by high-performance liquid chromatography (HPLC) with a Shimadzu Prominence Liquid Chromatograph fitted with a refractive index detector (RID) and a BioRad Aminex HPX 87H (300 mm × 7.8 mm) analytical column. The RID cell temperature was set at 40 °C, and the column was held at 65 °C. Sulfuric acid (0.5 mM) was employed as an eluent at 0.6 mL/min. Analytical-grade glucose in a series of concentrations from 0.1 to 5 g/L was employed to construct a calibration curve. Shimadzu LabSolutions were employed for instrument control, and data analysis was conducted with Microsoft Excel. The final selected consortia were subjected to molecular analysis to identify and characterize the denitrifying bacterial groups in the culture. The culture was centrifuged to pellet the cells from a 100 mL culture, and genomic DNA was extracted from the cells as previously described. The extracted DNA was sent to Illumina MiSeq sequencing analysis similar to that carried out for wastewater samples.

Enrichment of denitrifying bacteria was conducted in a nutrient-rich nitrate-reducing medium (NRM) with medium contents indicated in Supplementary Table S1. The medium was prepared with key ingredients of 1 g of nitrate (NO₃ ⁻) as the main electron acceptor and 1 g of glucose as the sole electron donor with the pH adjusted to 6.5 (Ogden et al., 2002; Romano et al., 2019; Böllmann and Martienssen, 2020). Medium (90 mL) was aliquoted into 150 mL serum vials, deoxygenated with Ar(g), autoclaved, and then, inoculated with 10% (v/v) of the wastewater samples to make up 100 mL cultures (Calderer et al., 2010). After inoculation, the cultures were immediately sampled and then incubated at 30°C in an automated shaker with a rotating speed of 150 rpm for 120 h (5 days). From each culture, 2 mL of samples were collected every 24 h for analyzing the following parameters: microbial growth analysis by measuring OD at 600 nm, denitrification dynamics by quantifying reduced NO₃ ⁻, and NO₂ ⁻ produced and reduced using a Griess assay as described previously. For all cultivations, each batch was conducted in triplicates for all modified conditions during the enrichment and optimization experiments. The natural selection of denitrifying bacteria from the enriched consortia was induced by sub-culturing the primary enrichment to subsequent generations in the NRM with conditions that promoted the proliferation and dominance of nitrate-reducing bacteria in anaerobic conditions. The 10 mL cultures were passaged from one generation as inoculum into 90 mL of fresh NRM until the third (tertiary) generation (Brűck et al., 2002). Among the cultures that were attained, those that attained less than 80% NO₃ ⁻ reduction were discontinued from the enrichment cycle, while cultures with over 95% NO₃ ⁻ and NO₂ ⁻ reduction were subjected to further denitrification optimization experiments.

The selected consortia from the enrichments were subjected to complete denitrification tests with a focus on consortia that could reduce NO₃ ⁻, NO₂ ⁻, and N₂O to produce N₂ gas as a terminal byproduct. The consortia were cultivated and maintained in the NRM for growth optimization and to promote denitrification over other metabolic processes (Green et al., 2010; Fan et al., 2018). All cultures were inoculated at 10% (v/v) inoculum to medium ratio with an inoculum of not more than 7 days old for maintenance and all optimization batch experiments below unless otherwise stated (Brűck et al., 2002). Glycerol was used as the sole carbon source in all batch experiments applied at 1:1 ratio with nitrates (C/N) to benchmark feasible denitrification with an economical carbon source used in wastewater treatment plants as opposed to glucose or lactate (Fountoulakis et al., 2010; Schroeder et al., 2020; PlÖhn et al., 2022).

Physicochemical parameters of the wastewater samples (Pt-SA and D-Lst) were determined, and the obtained results are recoded in Table 1. Owing to potential sewage spillages in Pt-SA, the introduction of fecal microorganisms in water can alter the water chemistry, including the elevation of NO₃ ⁻ concentrations (Sharma and Bhattacharya, 2017; Olds et al., 2018). The pH of both samples was higher than 5 with the evidence that it could have resulted in low metal leaching from tailings, thus limiting their concentrations in water (Król et al., 2020; Matodzi et al., 2020; Park et al., 2020). Chemical compositions of industrial wastewaters, such as mining, differ in dissolved metal concentrations due to the type of rocks associated with the mineral being mined (McCarthy, 2011; Guo et al., 2013). The Pt-SA sample is collected from an active platinum mine where the geological landscape is composed of mainly mafic and ultra-mafic igneous rocks from the crystallization of magma, which is less soluble and resistant to erosion predominantly forming reefs. In contrast, the D-LSt sample was collected from an active diamond mine where the geology of the environment is predominantly Kimberlite. The Kimberlite rock is composed of various chemicals and harbors not only diamonds as minerals but also Olivine [(Mg²⁺, Fe²⁺)2SiO₄], which is mainly magnesium and iron (Robles-Cruz et al., 2012). Overall, water chemistry revealed close similarities of the physicochemical parameters of the wastewaters with low dissolved metal concentrations. Amongst parameters that exceeded the SANS limits, NO₃ ⁻ was found to be over 100 mg/L in both samples.

TDS of 2,016 mg/L determined in the Pt-SA sample can be attributed to the high dissolved inorganic salts and organic particles (Table 1), most likely due to the influx of sewage into the water source (Butler and Ford, 2018). Comparatively, the D-Lst wastewater sample had a lower TDS of 1,192 mg/L with the lowest metal concentrations. Although below the SANS, Fe concentrations were higher in sample D-Lst at 0.117 mg/L compared to 0.027 mg/L of sample Pt-SA. In both samples, three heavy metal cations, arsenic (As), chromium (Cr), and lead (Pb), had concentrations below the detection limit of the ICP, which further corroborates that the wastewater samples do not have typical acid mine wastewater characteristics (McCarthy, 2011; Egashira et al., 2012; Modoi et al., 2014). It can further be suggested that based on these physicochemical parameters, the wastewater samples present an oligotrophic environment rich in nutrients, specifically nitrogen with a slightly acidic to neutral pH, and reducing micro-toxic conditions suitable for facultative anaerobes, including most denitrifying bacteria (Ponce-Soto et al., 2015). The majority of the detected anions (Table 1) were within the SANS limits except for NO₃ ⁻, SO₄ ^2-, and PO₄ ^3-, specifically in Pt-SA. The wastewaters had high Mg²⁺ and CaCO₃ concentrations, which is indicative of water hardness (Ahn et al., 2018). Ordinarily, sulfates (SO₄ ^2-) and carbonates (CaCO₃) occur in higher concentrations in mine wastewaters due to their abundance in the earth’s crust and are hence associated with most mineral-bearing rocks that get mined for mineral extraction (Fang et al., 2018). Although not alarmingly high, SO₄ ²⁺ appeared to be the second dominant contaminant also above the SANS limits in both samples. Pt-SA, with sewage contamination, had higher anion concentrations as compared to the D-Lst sample with more water hardness containing elevated Mg²⁺, Ca²⁺, and CaCO₃ concentrations (Grochowska, 2020). According to the Department of Water Affairs and Forestry (DWAF) regulations in South Africa, the value of sodium adsorption ratio (SAR) above 1.5 in irrigation water is considered toxic for the environment and both samples’ higher SAR values of 4.2 in Pt-SA and 2.41 in D-Lst indicate their toxicity, even for farming activities (DWAF, 1996; Kgopa et al., 2018).

Nitrate concentrations in the Pt-SA and D-Lst samples were 177 and 127 mg/L, respectively, which were both over 10 fold higher than the maximum acceptable limit of <11 mg/L according to the SANS. Most mining explosives contain NH₄NO₃, which can be oxidized to NO₃ ⁻, which completely dissolves in water, resulting in their accumulation over time (Cyplik et al., 2012; Degnan et al., 2016; Feng et al., 2020). Water contamination poses environmental and health threats that need an urgent remediation strategy to mitigate the impacts incurred by high NO₃ ⁻ concentrations (Moloantoa et al., 2022). With such wastewater physicochemical characteristics, the samples provide a suitable growth environment for nitrogen-metabolizing microorganisms that can be isolated, enriched, and applied in a NO₃ ⁻ bioremediation system.

Microbial diversity was assessed using 16S rRNA-targeted metagenomic sequencing, and the data acquired revealed 3,938 and 2,931 OTUs in samples PT-SA and D-Lst, respectively. Sample PT-SA seemed to be the most diverse relative to D-Lst with a total of nine against five dominant phyla detected, respectively (Figure 1A). The most abundant bacterial phyla in both samples were Proteobacteria, Bacteroidetes, Verrucomicrobia, Actinobacteria, and Planctomycetes, which are common in most mine wastewaters (Dadrasnia et al., 2017; Kisková et al., 2018; Osunmakinde et al., 2019). The remaining four phyla, Acidobacteria, Chlamydiae, Cyanobacteria, and Firmicutes, were exclusively found in sample Pt-SA, which had a slightly acidic to neutral pH of 5.01 and sewage flowing into the water promoting the dominance of various microorganisms that use different metabolites in the wastewaters (Cai et al., 2014). Furthermore, Proteobacteria was the most dominant phylum in both mine wastewater samples, accounting for 42.8% and 47.7% in Pt-SA and D-Lst, respectively, which is common in most mine wastewater samples (Ma et al., 2015; Kisková et al., 2018). Likewise, Zhang et al. (2018) reported the dominance of Proteobacteria as compared to other phyla in lakes with high NO₃ ⁻ concentrations. The second dominant bacterial phylum was Bacteroidetes, accounting for 15.8% and 15.7% in Pt-SA and D-Lst, respectively. The phylum Bacteroidetes has been regarded as the human-associated bacterial indicator in wastewater assessments, and its presence in both samples occurring in similar percentages of dominance indicates human-associated contamination in the samples (Cai et al., 2014; Olds et al., 2018). The presence of these two bacterial groups, Proteobacteria and Bacteroidetes, is common in NO₃ ⁻-contaminated water with a near-neutral pH, high NO₃ ⁻ concentration, and denitrification bioreactors in which their dominance in the wastewater samples can serve as a source of bacteria for bioremediation purposes (Safonov et al., 2018; Jinxiang et al., 2020).

In addition to sewage being the source of various bacterial groups in the Pt-SA sample, the near-neutral pH of 6.92 in sample D-Lst could have also been the reason why Acidobacteria were not dominant in the wastewater but prevalent in the Pt-SA sample with a pH of 5.01. For example, Sait et al. (2006) isolated more colonies of Acidobacteria at pH 5.5 than at pH 7, which further substantiated the relevance of lower pH in promoting the prevalence of Acidobacteria, even in environmental samples. The presence of specific bacterial communities in water systems, such as dams, lakes, and rivers, is strongly linked to pH and chemical composition that become selective to the microbiome of the environment (Sait et al., 2006; Wu et al., 2017; Paruch et al., 2019). The presence of Cyanobacteria in the Pt-SA wastewater sample serves as an indicative bacterial group for an environment that is stable and nutrient-rich, while in this case, sewage could have been the main source of excess nutrients as the phylum was abundant only in Pt-SA (Romanis et al., 2021). Among the phyla that were exclusively dominant in Pt-SA, Firmicutes was present and commonly known to dominate anaerobic digestors degrading polymers such as polysaccharides, proteins, and lipids, which are present in high concentrations in sewage, hence their absence in the D-Lst sample (Jabari et al., 2016; Nascimento et al., 2018). Sewage is mainly composed of human waste that includes pathogenic and gut microbiota associated with humans, which could be the result of Chlamydiae as a known human pathogenic bacteria group becoming prevalent in the Pt-SA sample (Collingro et al., 2020). A taxonomic comparison of the two wastewater samples at the genus level revealed 54 dominant genera and characterized their percentage of abundance in the samples; 39 were beyond 1% in dominance as annotated in Figure 1B. Among the detected genera, well-studied and reported denitrifiers, such as Pseudomonas, Citrobacter, and Clostridium, were among the dominant groups that drive the N-cycle in the wastewater systems, which aids the quest to enrich some known and novel denitrifying bacterial groups in the samples (Huang and Tseng, 2001; Kim et al., 2005; Lv et al., 2017; Rajta et al., 2019).

FAPROTAX analysis in Figure 1C revealed a total of 26 inferred metabolic pathways by the bacterial communities in the wastewater samples. Functional metabolic predictions in the D-Lst sample revealed that over 50% of the metabolic network consisted of chemo-heterotrophy, which is carried out by both the organotrophs and lithotrophs in the environment (Rajta et al., 2019). With high TOC (6.82), excess amounts of NO₃ ⁻ (127 mg/L) and SO₄ ^2- (256 mg/L) in the D-Lst sample relatively concurred with the presence of oligotrophs and lithotrophs utilizing chemical compounds, such as NH₃ ⁺, NO₃ ⁻, and SO₄ ^2-, as metabolites. The most abundant predicted metabolic pathways in Pt-SA were the ones facilitated by the Cyanobacteria that accounted for over 40% of the total functional groups. In a nutrient- and carbon-rich environment, such as the one presented by Pt-SA, photoautotrophic (Woo, 2018) and chloroplast-aided photosynthetic (Tang et al., 2011) metabolisms facilitated by the Cyanobacteria tend to be more prevalent. Among the common metabolisms predicted in both samples, there was nitrate reduction and nitrogen and nitrate respiration, which play roles in N-cycling in the environment. Genera of bacteria responsible for nitrate metabolisms, such as Bacillus, Clostridium, and Citrobacter, were concurrently detected in the samples, and the metabolic predictions further validate their presence in the mine wastewater samples (Kisková et al., 2018; Tiburcio et al., 2020). Nitrification and ammonium oxidation metabolic pathways both result in the formation of NO₃ ⁻ and were exclusively predicted in the Pt-SA sample that had the highest nitrate concentrations (177 mg/L), which strongly suggests that nitrifying bacteria converts NH₄ ⁺ and NO₂ ⁻ to NO₃ ⁻ (Spangler and Gilmour, 1966; Kim et al., 2005). In addition, the Pt-SA sample had more human-associated and mammal gut-related metabolisms, as reported previously (Cai et al., 2014; Kho and Lal, 2018; Collingro et al., 2020).

The composition of bacterial species and their functional potential can be influenced by geochemical parameters, such as pH, metal concentrations, and electron acceptors/donors, in the wastewater. For instance, wastewater samples and treatment plants can serve as excellent sources of indigenous bacteria capable of bioremediation of contaminated wastewater (Correa-Galeote et al., 2013; Nascimento et al., 2018). In fact, several authors have successfully enriched NO₃ ⁻-reducing bacteria from a wastewater treatment plant and used the enrichment cultures as inoculum for bioremediation research studies (Liu et al., 2016; Yao et al., 2020). The readily available NO₃ ⁻ and SO₄ ^2- in the samples serve as terminal electron acceptors in the metabolic activities of large groups of nitrogen- and sulfur-metabolizing bacteria. The presence of ammonium oxidizing microorganisms, which contain the amoA gene, indicates that the enrichments of all microbial groups without selection could lead to the conversion of NH₃ ⁺ to NO₃ ⁻ for denitrification to take place when oxygen levels become lower and conducive for denitrification (Rajta et al., 2019). In contrast, the same nitrifying microorganisms could also cause a reversal effect of NO₃ ⁻ production in a system aimed to reduce the contaminant. Indeed, the difference in the microbial diversities of the samples signified the importance of chemical differences and sources of contamination in wastewater samples.

As FAPROTAX only infers metabolic potential from the taxonomy data, further validation was required to identify the presence of nitrate-reducing genes in the samples analyzed. The cultivation and enrichments of bacterial species containing these genes could result in their application for NO₃ ⁻bioremediation (Huang et al., 2011; Levy-Booth et al., 2014; Rossi et al., 2015). Viable bacteria in the samples were successfully enriched in the NRM with glycerol as an electron donor and other components listed in Supplementary Table S1 (Liu et al., 2011). In general, nitrate-reducing bacteria are expected to be present in high-NO₃ ⁻-contaminated environments as they use oxidized nitrogen compounds as electron acceptors. However, their presence does not equate to their activity due to metabolic limitations imposed by high metal concentrations (Gang et al., 2013; He et al., 2020) and environmental conditions, such as low temperature, low pH, and high dissolved oxygen and carbon source availability (Saleh-Lakha et al., 2009; Igiri et al., 2018). The utilization of nitrate as the terminal electron acceptor under anoxic conditions served as a force of selection in growing facultative and obligate anaerobic nitrate-reducing bacteria (Melton et al., 2014). A total of 30 consortia were characterized based on their NO₃ ⁻ and NO₂ ⁻ reduction capacities as well as the volumetric measure of biogenic gas produced over time. A cut-off limit of 95% NO₃ ⁻ removal from an initial 1,000 mg/L of NO₃ ⁻ was set for the selection of consortia to characterize and optimize for complete denitrification. Overall, three of the 30 consortia attained 100% NO₃ ⁻ reduction within 48 h.

The three selected consortia (C1, C2, and C3) that attained 100% NO₃ ⁻ reduction were further subjected to denitrification tests in batch experiments to determine their potential and capacity to completely remove dissolved NO₃ ⁻ and NO₂ ⁻ at elevated concentrations. A minimum of 1,000 mg/L was applied during enrichment and a maximum of 10,000 mg/L was applied during tolerance studies by Pinar et al. (1997).

From the obtained results, all consortia attained a 100% reduction of 1,000 mg/L NO₃ ⁻ and all biogenic NO₂ ⁻ that formed subsequently over a 7-day incubation period (168 h). Bacterial tolerance to NO₃ ⁻ and NO₂ ⁻ was evaluated in experiments with up to 10,000 mg/L of initial NO₃ ⁻ and biogenic NO₂ ⁻ forming from nitrate reduction. Remarkably, microbial proliferation for all concentrations was not affected by the NO₃ ⁻ concentrations from 1,000 to 10,000 mg/L (Figures 2A–C). It is worth noting that the lag and exponential phases of the consortia cultures were all within 12 h. The maximum bacterial growth rate (μmax) above 0.36 h in all cultures was attained with their doubling time (Td) of less than 2 h except by C1 at NO₃ ⁻ concentrations of 5,000 and 10,000 mg/L, which had a Td of over 2 h. Although not uniform for all cultures, C1 and C2 had the same trend of lag phases that merged with the exponential phases indicating favorable growth and denitrifying conditions (Miyahara et al., 2010). For C3, there was a dip at the beginning of the lag phase indicating an adaptation period between 0 and 4 h followed by the sharp exponential phase. These results demonstrated a similar trend of dipping lag phases as observed by Ling and Juanjuan, (2018) during their biological NO₃ ⁻ removal studies with a consortium from a sewage plant. The differences in the lag and exponential phases for the three consortia could be due to the different bacterial compositions growing at different rates. Among all three consortia, C2 had the shortest doubling time (Td = 1.3) and the highest growth rate of more than 0.5, indicating favorable growth conditions for the consortium at 2,000 mg/L (Cua and Stein, 2014). Cultures cultivated in the lowest NO₃ ⁻ concentrations (1,000 mg/L) were the first to reach deceleration phases, most likely due to the depleted electron acceptor (NO₃ ⁻) in the medium. This was indicated by the gradual decline of the growth curves after 40 h of incubation (Pinar et al., 1997; Miyahara et al., 2010). In contrast, cultures in higher NO₃ ⁻ concentrations (2,000, 5,000, and 10,000 mg/L) maintained their stationary phases for up to 168 h. Evidently, the enriched consortia demonstrated great tolerance to high NO₃ ⁻ concentrations as the concentrations tested did not hinder or drastically affect microbial growth.

Notwithstanding the tolerance toward elevated concentrations of NO₃ ⁻, the consortia could not attain 100% NO₃ ⁻ and NO₂ ⁻ removal at higher NO₃ ⁻ concentrations (from 5,000 to 10,000 mg/L), probably due to NO₂ ⁻ accumulation and toxicity (Figures 2D–I). Regardless of the incomplete NO₃ ⁻ removal in the cultures with the highest NO₃ ⁻ concentrations (10,000 mg/L), over 4,000 mg/L of NO₃ ⁻ was removed, which highlights the tolerance and great capacity of the consortia to reduce elevated amounts of NO₃ ⁻. Ideally, the consortia had to be continuously fed with carbon source to combat NO₂ ⁻ accumulation as carried out in continuous cultivations (Sun et al., 2009), while in this study, no additional carbon source was supplied so as to notice the effect of NO₂ ⁻ on the consortia. The reduction of NO₃ ⁻ by the enriched consortium and pure cultures at such high concentrations is rarely obtained (Vijaya-Chitra and Lakshmanaperumalsamy, 2006). In C1 and C2, 100% NO₃ ⁻ reduction was attained at 18 h for both 1,000 and 2,000 mg/L NO₃ ⁻ attaining cultures at 55.55 and 111.11 mg/L per hour, respectively. The fastest NO₃ ⁻ reduction of 1,000 mg/L among the three consortia was 83.33 mg/L^-1 attained by C3 achieving 100% NO₃ ⁻ reduction within 12 h. For the consortia incubated in higher NO₃ ⁻ concentrations of 5,000 mg/L, the complete reduction was attained in all cultures at 36, 24, and 18 h by C1, C2, and C3, respectively, with C3 exhibiting the fastest reduction rate, removing 277.78 mg/L^-1. From all the reduction capacities displayed by the three consortia, it was evident that the enriched bacteria could reduce high NO₃ ⁻ concentrations beyond the 3,000 mg/L of NO₃ ⁻ reported by Cyplik et al. (2012). The biological denitrification rates and effectiveness in most remediation systems are determined solely by the removal of NO₃ ⁻. However, other intermediate nitrogen species, such NO₂ ⁻, NO, NH₄ ⁺, and N₂O besides N₂ gas, serve as an indication of incomplete denitrification (Krishna Reddy et al., 2017; Kong et al., 2018). In cultures with higher NO₃ ⁻ concentrations, the accumulation of more than 2,000 mg/L of NO₂ ⁻ was observed mostly due to the formation of reactive oxygen species (ROS) from molecular oxygen and nitrogen species (RNS), such as NO from the accumulation of NO₂ ⁻ and NO as NO₃ ⁻ (Durzan and Pedroso, 2002; Poole, 2005; Damiani et al., 2016; Yao et al., 2020). For consortia grown on 2,000 mg/L, C1 and C2 reduced over 90% and 50% of the formed NO₂ ⁻, respectively, while C3 had low NO₂ ⁻ production and a lower reduction capacity of NO₂ ⁻. The ability of consortium C3 to rapidly reduce high NO₃ ⁻ concentration and its lack of ability to reduce the NO₂ ⁻ could be attributed to the presence of nitrate-respiring bacteria that used NO₃ ⁻ for their metabolic activities but lack nitrite reductase genes that transcribe genes that facilitate NO₂ ⁻ reduction, hence its accumulation (Nogales et al., 2002; Philips et al., 2002). Owing to the overall performances of the consortia for NO₃ ⁻ reduction and denitrification under all concentrations, 2,000 mg/L of NO₃ ⁻ was deemed as the benchmark target for the downstream experiments to optimize and evaluate complete denitrification capacities of the selected consortia.

Heavy metals such as copper (Cu), iron (Fe), and molybdenum (Mo) have been reported as co-factors in protein activity, including those involved in denitrification (Glass and Orphan, 2012; Wang et al., 2013; Black et al., 2016). Both iron (Fe²⁺) and copper (Cu²⁺) have the capacity to promote enzymatic activities of various denitrification steps and have been successfully explored in studies to enhance NO₃ ⁻ reduction by nitrate reductases (narG) (Schaedler et al., 2018), NO₂ ⁻ reduction by nitrite reductases (NirK) (Green et al., 2010), and N₂O reduction by nitrous oxide reductases (nosZ) (Sanford et al., 2012). In this study, both Cu²⁺ and Fe²⁺ added at 5 mg/L (Lee et al., 2009) were used as standardized test concentrations to explore their dynamic effects on the reduction rates of dissolved NO₃ ⁻ with the initial concentration of 2,000 mg/L, while monitoring concentrations of biogenic NO₂ ⁻ and the production of N₂ as the terminal product for complete denitrification.

The obtained results recorded in Table 2 revealed that cultures behaved differently when exposed to different conditions; for instance, four of the nine (three consortia: C1, C2, and C3 conducted in triplicates) metal-supplemented consortia, C3:Cu-NO₃; C3:Cu and Fe-NO₃; C2:Cu and Fe-NO₃; and C2:Fe-NO₃, attained 100% NO₃ ⁻ reduction within 6 h followed by the other three consortia, C1:Cu-NO₃; C2:Cu-NO₃, and C1:Cu and Fe-NO₃, attaining 100% NO₃ ⁻ reduction at 12 h. All these cultures besides C3:Fe-NO₃ contained Cu²⁺ in the medium, which indicates the promoting effect Cu²⁺ (5 mg/L) had on NO₃ ⁻ reduction. In consortia supplemented with only Fe²⁺, the cultures demonstrated slow NO₃ ⁻ removal, with C2:Fe-NO₃ attaining 100% NO₃ ⁻ reduction after 24 h, while C1:Fe-NO3 attained less than 50% NO₃ ⁻ reduction after 36 h. There have been reports showing that Fe promotes NO₃ ⁻ reduction, but in this study, it seemed not to be the case (Glass and Orphan, 2012). Herein, our findings with the selected consortia supported the use of Cu²⁺ instead of Fe²⁺ to optimize NO₃ ⁻ reduction through metal co-factor supplementing.

Biogenic NO₂ ⁻ concentrations in the Fe²⁺-supplemented cultures C1:Fe-NO₃ and C2:Fe-NO₃ were the lowest, which correlated with the low amount of NO₃ ⁻ reduced. In other cultures, NO₂ ⁻ was observed to start accumulating after the third hour of incubation and became stagnant until all the NO₃ ⁻ was reduced. The same effects of poor NO₂ ⁻ formation and its removal was observed in cultures with Fe²⁺ as a sole metal cofactor, and a similar observation was noticed by Giannopoulos et al. (2020) when assessing the effects of different heavy metals on denitrification. The sole Fe²⁺-containing cultures performing poorly with respect to reducing NO₃ ⁻ could be due to NO₂ ⁻ toxicity that effects the growing bacteria needing Cu²⁺ to promote the catalytic effect on nitrite reductases to reduce NO₂ ⁻ as nirK genes are induced by Cu²⁺ ions to transcribe the nitrate reductase enzyme and promote its functionality (Smith et al., 2007). The biogenic gas analysis by GC revealed that most N₂ gas formed in the cultures containing Cu (C1:Cu-NO₃ and C2:Cu and Fe-NO₃), which further supported the findings reported above on Cu²⁺ inducing NO₃ ⁻ and NO₂ ⁻ reduction. High N₂ percentages in Cu²⁺ containing cultures, as opposed to the non-metal supplemented cultures, indicate the potential of the enriched consortia to have the capacity to completely reduce NO₃ ⁻ to N₂ when supplemented with Cu²⁺ (Sullivan et al., 2013; Zhang et al., 2019; Giannopoulos et al., 2020; He et al., 2020). Copper toxicity in N metabolism is known for forming lipid peroxidation and protein damage in most Cu²⁺ intolerant bacteria, including denitrifiers (Dupont et al., 2011; Ochoa-Herrera et al., 2011). The accumulation of NO₂ ⁻ beyond 1,000 mg/L has been reported to form nitrous acid (HNO₂), which is toxic to bacteria and has the potential to halt denitrification (Glass et al., 1997). To uncover and validate the possibility of NO₂ ⁻ accumulation being the reason for the truncation of complete denitrification, more optimization tests to bypass the hurdle were conducted, with the three consortia subjected to complete denitrification optimization tests to choose one consortium for downstream optimization experiments.

The inducing effects of Cu²⁺ on complete denitrification as a fundamental metal cofactor of denitrifying enzymes was further assessed on the three consortia in the NRM with 2,000 mg/L of NO₃ ⁻ with a two-fold increase (10 mg/L from 5 mg/L, as previously tested). Bacterial growth curves (Figure 3A) showed that C1 and C2 had a similar growth trend as opposed to C3, which had a longer lag phase. In this set of experiments, C1 and C2 attained stationary phases much faster (within 6 h) than C3 (within 12 h). Bacterial growth in all three cultures was not affected by the presence of 10 mg/L of Cu²⁺ as it was expected to negatively affect the bacteria given the metal toxicity at higher concentrations (Ochoa-Herrera et al., 2011). However, the Td and μmax of the cultures differed as C1 was observed to have the shortest Td (0.88 h) and the fastest μmax (0.788 h) when compared to C2 and C3 (Td of 0.934 and 1.193 h, respectively). Denitrification dynamics revealed unique effects of Cu²⁺ on all three consortia in which C2 and C3 had the fastest NO₃ ⁻ and NO₂ ⁻ removal rates, attaining 100% NO₃ ⁻ reduction simultaneously within 5 h as compared to C1, which attained complete NO₃ ⁻ removal after 72 h of incubation (Figure 3B). The presence of Cu²⁺ seemed to have boosted the metabolic activity of bacteria in C2 and C3 cultures to reduce 100% of the produced NO₂ ⁻ at 18 h and 30 h, respectively, which corroborates with the results obtained by Wagner et al. (2018), who observed an increase in the microbial activity through Cu²⁺ dosing that promoted bacterial growth and denitrification. The inability to reduce NO₃ ⁻ by C1 was observed and this could be due to the different denitrifying bacteria that do not positively respond to either 5 or 10 mg/L of the added Cu²⁺ as reported by Ochoa-Herrera et al. (2011). The rapid reduction of NO₃ ⁻, formation of NO₂ ⁻, and its subsequent reduction in C2 and C3 (Figure 3C) indicated a positive response of the microorganisms to the two-fold increase of the added Cu²⁺. Theoretically, copper addition would also assist in the reduction of N₂O to N₂ as Cu²⁺ is known to determine the transcription and activity rate of nitrous oxide reductases (Giannopoulos et al., 2020; Shen et al., 2020). After 7 days (168 h) of monitoring the denitrification dynamics, only C1 still failed to reduce 100% of NO₂ ⁻, indicating the consortium’s poor ability to thrive in the presence of the dosed metal cofactor.

Gas chromatography revealed that the lowest N₂O gas (Figure 3D) was detected in C2 cultures with an average of approximately 0.01% followed by C1 with N₂O levels as low as 0.04% of the biogenic gas with both consortia producing similar amounts of N₂ gas. C1, which was unable to reduce most of the biogenic NO₂ ⁻, also accumulated N₂O. C3 had the highest N₂O quantities produced by the bacteria with the lowest N₂ quantities, indicating the hindrance of N₂O reduction to N₂, which further indicates poor utilization of Cu²⁺. Another possible explanation could be the inherent toxic effect of Cu²⁺ lowering N₂O reduction capacities where the bacteria are unable to tolerate the toxicity of the metal, hence their inability to perform complete denitrification (Granger and Ward, 2003). This inability to reduce N₂O to N₂ could also be due to the absence of complete denitrifiers that expresses nitrous oxide reductases (nosZ gene) (Nogales et al., 2002). With low N₂O quantities in C2 cultures, it was expected that the N₂ quantities would have also increased beyond the N₂ quantities of C1, but this was not the case probably due to the incomplete reduction of N₂O that accounted for over 200 mg/L of residual NO₂ ⁻ that is not converted to gaseous nitrogen species.

With C1 being unable to reduce NO₂ ⁻ completely and C3 responding negatively to the Cu²⁺ supplement noticeable by high N₂O and low N₂ quantities, C2 was selected for downstream experiments for further denitrification optimization. The results further prove and support that Cu²⁺ in conjunction with the copper-tolerant denitrifiers in the consortium can promote N₂O reduction to N₂, rendering the consortium capable of complete denitrification.

To mimic the environmental conditions where the wastewater samples were collected, an SWW medium was prepared (Table 1) with the modifications reported by Foglar et al. (2005). The selected consortium (C2) was tested with the aim of reducing 2,000 mg/L of NO₃ ⁻ within 24 h under varied conditions monitored for a maximum of 120 h. Microbial growth curves under six different test conditions (Figure 4A) revealed that the variations imposed had different effects on bacterial growth. Compared to the rapid bacterial proliferation observed in NRM, it was evident that the bacterial growth rate decreased with the increasing Td when cultivated on the SWW. This was also observed by Delgadillo-Mirquezand et al. (2016) when switching media compositions during the optimization studies to remove nitrogen and phosphate from wastewater. Nevertheless, C2 was still able to achieve 100% removal of the 2,000 mg/L. Bacterial growth dynamics were not noticeably different when C2-cultivated glycerol was used as a sole carbon source with no metals, with Cu, Fe, and Cu and Fe combined. When glycerol was used as the sole C-source, an average Td of 7.789 h, with the slowest growth rate of 0.089, was observed, which was in contrast to when glucose was used in which Td decreased to 3.081 h. Similar effects were observed in the literature when glucose produced more bacterial biomass than glycerol in studies conducted by Liu et al. (2011), comparing bacterial fermentation processes in which glycerol and glucose were used independently and as co-substrates. In addition to glucose-shortening microbial Td in SWW, the biomass yields also appeared to be greater when glucose was used, resulting in a growth peak above all other experimental culture sets, validating the poor energy supply from glycerol as compared to glucose (Yao et al., 2016; Zhang et al., 2020). When using glucose as the C-source and different nitrogen sources, NO₃ ⁻ and NO₂ ⁻, microbial growth rates differed by 0.225 and 0.127 with Td of 3.081 and 5.458 h. respectively. Low growth rate and high Td served as an indication of NO₃ ⁻ preference over NO₂ ⁻ as the sole nitrogen source. Sang et al. (2020) observed similar results when they characterized the use of NO₃ ⁻, NO₂ ⁻, and NH₄ ⁺, whereby NO₂ ⁻ yielded the lowest microbial biomass. From the abovementioned bacterial growth dynamics, glucose was observed to promote higher bacterial proliferation, which may suggest modifications need to be adopted when treating wastewater with NO₃ ⁻ concentrations beyond 2,000 mg/L.

Fluctuations of pH were monitored for 24 h in three no metal-containing cultures with varying C and N sources (Figure 4D): 1. C2 in SWW with glycerol and NO₃ ⁻, 2. C2 in SWW with glucose and NO₃ ⁻, and 3. C2 in SWW with glucose and NO₂ ⁻. In glucose-containing cultures with both NO₃ ⁻ and NO₂ ⁻ as sole N-sources, pH lowered to below 5 from 6.5 within 6 h, while in the glycerol- and NO₂ ⁻-containing cultures, pH increased to over neutral lowered below pH 7 after 6 h. The noticeable pH fluctuations occurred between time 0 and 12 h where microbial growth was at its exponential phase, indicating high metabolic activities, hence the change in pH (Sánchez-Clemente et al., 2018). Biochemical oxidation of glycerol results in more proton release during the reduction of an electron acceptor, hence the increase in pH from 6.5 to beyond neutral pH between 0 and 6 h, which remained higher than the pH in cultures with glucose (Baeseman et al., 2006). The oxidation of glucose continuously releases hydrogen ions, hence lowering the pH and also contributing to the pH changes (Liu et al., 2011).

Nitrate and nitrite removal dynamics in all six experimental conditions were observed to be different for each optimization test (Figures 4B, C). Complete NO₃ ⁻ removal was achieved in all five tests but at different rates excluding the one supplied with NO₂ ⁻ as a sole N-source. The first culture to attain 100% NO₃ ⁻ removal with 9 h was the one supplied with glucose without metals, while the other supplied with glycerol without metals attained 100% NO₃ ⁻ removal at 18 h. As previously observed, Fe²⁺ promoted faster NO₃ ⁻ when supplemented solely, 100% NO₃ ⁻ removal was achieved at 12 h compared to Cu²⁺, and when Fe and Cu were combined, both tests achieved 100% NO₃ ⁻ removal at 18 h.

Biogenic NO₂ ⁻ accumulated (beyond 1500 mg/L) with a same trend for all culture with Fe²⁺ containing consortia producing the highest concentrations among all of them. The no metal-containing culture cultivated on glycerol also followed the same trend as the metal tests. The glucose-containing cultures without metals were also observed to accumulate NO₂ ⁻ but only up to a maximum of over 1,000 mg/L at the sixth hour and interestingly started getting reduced. Interestingly, up to 70% of the biogenic NO₂ ⁻ was removed, which was good for the ongoing denitrification process with glucose supplied as a sole C-source (Srinandan et al., 2012). The cultures containing NO₂ ⁻ and glucose as sole N- and C-sources, respectively, also demonstrated the gradual removal of NO₂ ⁻ from 2,500 mg/L to 548 mg/L (78% removal), which concurs with reports made by Yao et al. (2020) that denitrifiers can reduce NO₂ ⁻ directly and not as a biogenic product of NO₃ ⁻ reduction. In denitrification studies in which glycerol was used as a cheaper carbon source from the biodiesel production plant, NO₂ ⁻ accumulated as a terminal byproduct of the denitrification process (Baideme et al., 2017), and for further removal of NO₂ ⁻ by denitrification, C/N ration was re-adjusted to circumvent NO₂ ⁻ accumulation from 1.8 to 3.5, as was performed in studies conducted by Carneiro and Foresti (2017) when running up-flow bioreactors with glycerol as a sole C-source. Biochemical utilization of the three carbon (glycerol) and six carbon (glucose) molecules as energy sources depends on the bacteria using them for denitrification and their energy demands as the two molecules get biodegraded at different rates. Bacteria such as P. stuzeri can degrade glucose much faster than glycerol, promoting the rapid removal of NO₃ ⁻ and NO₂ ⁻ during denitrification (Akunna et al., 1993; Srinandan et al., 2012). It was reported in studies conducted by Bowles et al. (2012) that sulfates and sulfides have denitrification hindrance capacities, which could have been one of the contributing factors resulting in incomplete denitrification in our study as SO₄ formed part of the metal compound (CuSO₄ and FeSO₄) supplements (Yu and Bishop, 2001; Marietou, 2016).

The NO₂ ⁻ reduction dynamics could not be completely relied upon to determine the efficiency of the metals in boosting denitrification; hence, biogenic N₂O quantities in each culture were analyzed (Figure 4E) as it is a sensitive indicator of progressive denitrification (de Catanzaro and Beauchamp, 1985). The highest N₂O percentages were detected in the culture with Fe²⁺ as the sole metal co-factor, which indicates the positive boost to nitrite reductases that proceeded with trace quantities of N-metabolisms beyond NO₂ ⁻ reduction. A higher N₂O percentage (3.8%) in the Fe²⁺-containing cultures could be due to the absence of a metal co-factor (Cu²⁺) that boosts N₂O removal to N₂ (Baeseman et al., 2006). The culture with the lowest N₂O percentages was the non-metal-containing culture, which produced 2.3% of N₂O, which could be due to the non-boosting effects of nitrite reductases (Granger and Ward, 2003). The metal co-factor effects were noticeable with varied N₂O quantities but with less to no significant NO₂ ⁻ reduction by the consortium growing on glycerol as a sole C-source. The results obtained corroborate those attained by Sun et al. (2009) in which NO₂ ⁻ accumulated from the reduction of NO₃ ⁻ in cultures with four different carbon sources, methanol, ethanol, sodium acetate, and sodium propionate, when glucose was used and no NO₃ ⁻ or NO₂ ⁻ accumulation was observed. With the results obtained in these experiments, it was evident that the removal of dissolved N-species is possible with a change of growth parameters to optimize complete denitrification. For downstream experiments in the optimization of NO₂ ⁻ and N₂O reduction to N₂, the SWW medium with glucose as the sole C-source and copper as a metal co-factor for N₂O removal were conducted.

The effect of supplementing Cu²⁺ to assess complete denitrification through its catalytic effect to the nosZ gene was explored by adding increasing Cu²⁺ concentrations (0, 10, 25, 50, and 75 mg/L) to the consortium with 1:1 and 1:2 ratios of C:N (glucose:nitrate). Obtained GC results (Table 3) revealed that biogenic N₂ and N₂O percentages were noticeably lower than when there was no added Cu²⁺. Similar findings were noticed in denitrification studies by Sullivan et al. (2013), highlighting how Cu²⁺ promotes N₂O reduction to N₂ gas. According to the residual glucose concentrations determined by HPLC (Table 3), cultures that contained two-fold of glucose yielded more CO₂ than the cultures that had a 1:1 ratio with NO₃ ⁻. It is also noticeable that the cultures without Cu²⁺ and the one with the lowest Cu²⁺ concentrations used up all the glucose as the residual concentrations were found to be below the detection limit, indicating the complete utilization of the C-source without any hindrance of metal toxicity (He et al., 2020).

The most biogenic N₂ produced in these experiments accounted for over 54% of the total gas observed in cultures containing 50 mg/L of Cu²⁺ with a C:N ratio of 1:1. The highest Cu²⁺ concentration (75 mg/L) yielded 7.8% and 20.9% of the total gas in cultures with C/N ratios of 2:1 and 1:1, respectively, which is a sign of metal toxicity in all cultures with 75 mg/L of Cu²⁺. The bioavailability of Cu²⁺ in the denitrification system is dependent of the metal dissolution state, which is highly dependent on the pH of the medium (Król et al., 2020). Copper (II) has been reported to dissociate easily in acidic environments and precipitate in neutral to alkaline pH, thus reducing its bioavailability (Cuppett et al., 2006). In this study, the pH of the SWW medium was adjusted to 6.5, like the pH used during bacterial enrichments that yielded the best denitrification results. Owing to the precipitation factor of Cu²⁺ at a higher pH close to 7, the concentrations used in these experiments could have precipitated reducing the bioavailability of the required Cu²⁺ ions needed for the biochemical process in denitrification (Cuppett et al., 2006; Król et al., 2020). Alternatively, Cu²⁺ concentrations below 50 mg/L could have precipitated, thus showing no significant impacts on denitrification, as opposed to concentrations above 50 mg/L in which most of it could have dissolved beyond the limit that the consortium can tolerate.

Copper toxicity at 75 mg/L was noticed with low biogenic nitrogenous gas production in which only 7.8% of N₂ and 1.78% of N₂O were produced even when the C:N ratio was 2:1. The residual glucose concentrations were also the highest in the culture with the highest Cu²⁺ concentrations (75 mg/L), which again indicates poor bacterial proliferation with low utilization of glucose. The summed-up data for the cultures cultivated with Cu²⁺ revealed that 50 mg/L of Cu²⁺ promoted the best denitrification results with biogenic N₂ gas of over 55% and less than 1% of N₂O in the cultures after 48 h. Furthermore, the results further validate that supplementing denitrification cultures with Cu²⁺ does promote N₂O reduction to N₂ and revealed the toxicity levels of inorganic Cu²⁺ to be evident at concentrations over 75 mg/L. Owing to the abovementioned details, Cu²⁺ supplemented at 50 mg/L benchmarks the maximum concentrations that promotes the complete denitrification of over 2,000 mg/L of NO₃ ⁻. Finally, the abovementioned results provide successful establishment of optimization conditions that can be applied in a bioremediation system to obtain the best complete denitrification results using an indigenous bacterial consortium.

Based on the Illumina MiSeq data analysis (Figure 5) of the C2 consortium, it is evident that bacterial groups detected in the wastewater samples were successfully enriched. The selection in the NRM with high concentrations of NO₃ ⁻ as selective pressure resulted in a consortium comprising two main phyla, Proteobacteria and Firmicutes, which accounted for 84% and 16% of the bacterial community, respectively (Figure 5A). The phylum Proteobacteria has been widely reported as the primary bacterial group in remediation systems comprising most bacteria used in wastewater treatments (Cydzik-Kwiatkowska and Zielińska, 2016). The most abundant detected genera were Asaccharospora, Clostridium_sensu_stricto 18, Paraclostridium, Tissierella, Clostridioides, Escherichia-Shigella, and Tepidimicrobium (Figure 5B). It is worth noting that Paraclostridium accounted for over 50% of the whole microbial population in the selected consortium, whereas in the original water sample, it constituted less than 1% of the reads. It has been reported that Paraclostridium, an obligate anaerobic Firmicute, has the capacity to carry out complete denitrification, reducing NO₃ ⁻ to N₂ gas (Kutsuna et al., 2019). Among the dominant bacterial genera, Clostridium and Clostridiales were present and are known to facilitate dissimilatory nitrate reduction to ammonium (DNRA) and fermentative denitrification, respectively. In closed systems, the two bacterial genera can synergistically metabolize N by reducing NO₃ ⁻ to NH₄ ⁺ and then NH₄ ⁺ to N₂ for Clostridium and Clostridiales, respectively, which makes them key bacteria in the denitrifying consortium used in this study (Ahmad et al., 2021; van den Berg et al., 2017). Interestingly, the consortium attained the best denitrification dynamics without the dominance of any benchmarked known complete denitrifiers, such as Paracoccus and Pseudomonas, commonly used in denitrification kinetics and dynamics studies (Yao et al., 2020). Therefore, it can be concluded that the bacterial groups detected in the enriched consortium possess the synergistic metabolic relationship permitting them to perform complete denitrification, as most are known to be incomplete denitrifiers. Thus, the results suggest that bacteria capable of metabolizing different N-species were co-cultured and acted synergistically to perform complete denitrification as a complete denitrifier would when cultured in a suitable environment with all growth factors.