Supplementation of plant fermentation carbon sources significantly enhances the abundance of ammonia-oxidizing genes and increases nitrogen removal in a municipal wastewater treatment plant

Abstract

This paper deals with the effects of a cost-effective plant fermentation carbon source (PFCS) on nitrogen removal in a municipal wastewater treatment plant (WWTP). Carbon source and electron donor shortage remained one of the limiting factors for denitrification and nitrogen removal in the WWTPs of China. The PFCS was supplemented to increase the influent carbon/nitrogen (C/N) ratio from the original 9:1 to 82:1. The effluent ammonium (NH₄⁺-N), total nitrogen (TN), and total phosphorus (TP) concentrations from the high C/N ratio group (group A) were significantly lower than those of the low C/N ratio control group (CK). Both Illumina sequencing and nitrogen metabolism gene copy number quantification demonstrated a notable increase in the abundance of aerobic denitrifying bacteria and functional denitrification genes at higher C/N ratios. Additionally, the optimal C/N ratios were investigated by using response surface methodology. Overall, high C/N ratios significantly improved the performance of municipal WWTPs in nitrogen and phosphorus removal. Moreover, inexpensive PFCS may provide a cost-effective strategy for improving wastewater treatment capacity, although excessive sludge increased. Utilization of PFCS could reuse and recycle the high-fiber solid waste of crops and timbers generated from agricultural, forestry, and wetland plants, which could contribute to global sustainable development and environment protection.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12866-025-04093-2.

Introduction

Accelerated eutrophication of natural water bodies due to excessive nitrogen and phosphorus discharges has led to the degradation of aquatic ecosystems and ensuing water blooms and red tides [1]. One measure to prevent eutrophication of natural water bodies is the effective removal of nitrogen and phosphorus from municipal sewage and industrial wastewater by wastewater treatment plants (WWTPs) [2]. The anaerobic/anoxic/aerobic (AAO) process, used extensively in WWTPs, comprises three chambers: anaerobic, anoxic, and aerobic tanks, which operate based on microbial nitrification and denitrification pathways and the synthesis and depolymerization of cellular polyphosphates for enhanced nitrogen and phosphorus removal. However, municipal sewage in developing countries such as China is characterized by an imbalance of carbon and nitrogen ratios (C/N ratio by the ratio of chemical oxygen demand (COD) to total nitrogen (TN)) and low carbon sources [3, 4]. WWTPs with influents of low C/N ratios often require the addition of carbon sources to ensure effective nitrogen and phosphorus removal in the AAO process [4].

In previous studies, various carbon sources, including methanol, ethanol, sodium acetate, citrate, glucose, and sucrose, have been supplemented to enhance C/N ratios and remove nitrogen and phosphorus nutrients [5]. While carbon supplementation strategies enhance wastewater treatment efficiency, they may induce a significant sludge increment effect (12–18%), resulting in sludge management costs accounting for 25–60% of municipal wastewater treatment plant (WWTP) operational expenditures [6–8]. Studies indicate that exogenous carbon sources such as methanol and sodium acetate, despite optimizing C/N ratios and improving nutrient removal, frequently suffer from high costs, intricate dosage control, and effluent quality instability [5]. These findings underscore the imperative to achieve dynamic equilibrium between treatment performance optimization and circular economy objectives in carbon supplementation design [9].

Quantitative polymerase chain reaction (qPCR) techniques have been previously used as a definitive technique for quantifying differences in gene expression levels between samples [10]. qPCR has been utilized to quantify nitrification and denitrification genes encoding ammonia and nitrite oxidases, as well as nitrate, nitrite, nitric oxide, and nitrous oxide reductases [11]. The ammonia monooxygenase genes (Comamoxe amoA, AOB amoA), nitrate reductase genes (narG, napA), nitrite reductase gene (nxrB) and nitrous oxide reductase genes (nirS) were used as functional marker for the characterization of nitrifying and denitrifying bacterial communities [12, 13]. Copy numbers of different genes involved in denitrification pathways and their relative abundance could provide evidence for developing different denitrify communities under primary succession [11]. Changes in nutrients, environmental conditions, and environmental pollution put pressure on microbial communities, leading to changes in community structure accordingly. High-throughput sequencing of metagenomic DNA marker genes such as the 16 S ribosomal RNA (rRNA) gene could reveal the diversity and succession of bacterial communities. Furthermore, this approach can be employed to gain insight into the functional and redundant concepts in microbial communities, aiming to identify microorganisms with functional activity [14].

Plant carbon sources (PCS), comprising macrophytes, agricultural wastes, and forestry residues, have significant advantages due to their low cost and wide availability. PCS mainly consists of cellulose, hemicellulose, and lignin, which can be converted to soluble and micro-molecular substrates to a certain degree. PCS has been used to enhance the denitrification process in artificial wetlands due to its prolonged duration of carbon release and favourable utilization by denitrifying bacteria after microbial biofilm attachment [15]. However, due to engineering limitations, adding the solid plant carbon source directly into the reactor in WWTPs is impossible. The plant fermentation carbon source (PFCS) utilized in this study is the fermentation liquid produced by the fermentation treatment of PCS. The addition of carbon sources derived from plant fermentation to the reactor in liquid form results in the release of small molecules, including volatile fatty acids and sugars, which serve as supplementary carbon sources and electron donors for several processes, including denitrification, heterogeneous nitrate reduction to ammonium, anaerobic fermentation, and sulfate reduction processes. The denitrification rate can be enhanced by adding PFCS into the anoxic tanks, thereby supplementing the carbon source. However, if PFCS is excessive or inadequately added, it not only increases the system’s operational cost but also elevates the risk of COD exceeding the standard at the wastewater treatment plant. The appropriate utilisation of an efficacious carbon source can avert the exceedance of the COD value of the wastewater.

In this study, the effects of inexpensive PFCS supplementation on the nitrogen and phosphorus removal from municipal wastewater were analyzed. We used quantitative PCR (qPCR) and high-throughput sequencing technologies to investigate the changes in the functional genes (e.g. AOB amoA, narG, napA, nirS) for nitrogen removal and microbial community structure in the AAO process at different C/N ratios. We also explored the optimal application parameters of plant carbon sources using response surface methodology. This study could provide a theoretical reference for the feasibility of improving the efficiency of the AAO process for nitrogen removal by adding inexpensive PFCS.

Materials and methods

Experimental set-up

The experiment was set up with control and experimental groups, CK and A, respectively. The experiments were conducted using the AAO process of a municipal wastewater treatment plant (WWTP) in good operation. The WWTP was located in south-central China and had two parallel AAO processes. Each AAO process consisted of an anaerobic, an anoxic, and an aerobic tank, a secondary sedimentation tank, and a sludge thickening tank. The WWTP had an operating capacity of 109,400 m³/d. Activated sludge samples were collected from three sampling spots in the biochemical tank. The experiment was conducted 30 days from 21 October 2021 to 21 November 2021.

Effects of additional plant-derived carbon source on the water quality of wastewater

The plant fermentation carbon source (PFCS) was purchased from Hunan Zhengtai Water Services Ltd. The density, pH, and COD of PFCSs were 1.03 mg/cm3, 6.4, and 206,000 mg/L, respectively. Group A added the PFCS to the anoxia tank of the AAO reactors at a concentration of 5 mg/L, and group CK was not treatedz (Fig. 1). The PFCS was added at five points in time: October 21 st, 23rd, 29 th, 30 th, and 31 st. Activated sludge was sampled from three points in the anoxic tanks of two groups, referred to as Group A and Group CK. Samples were collected in 50 mL sterile centrifuge tubes. Key operating parameters, including chemical oxygen demand [16], ammonium (NH₄⁺-N), total nitrogen (TN), total phosphorus (TP), and total suspended solids (SS), were monitored daily in each tank. Determination of COD, TN, NH₄⁺-N, and TP concentrations in influent and effluent samples was made according to standard methods (APHA, 1998).

Fig. 1.

Schematic diagram of the WWTP process and PFCS addition sites

Copy numbers of nitrogen metabolism functional genes

The copy numbers of eight functional genes (including nirS, nxrB, narG, napA, comammoxe amoA, Anammox 16 S rRNA, and 16 S rRNA genes) were amplified by qPCR with the primers (Supplementary Table S1). 16 S rRNA genes of group CK and group A were used as controls for normalization between samples. Municipal wastewater samples were collected for centrifugation, and then genomic DNA was extracted from each sample using a UniversalGen DNA kit (Cwbio, China) according to the manufacturer’s instructions. QPCR reactions were performed using the CFX96 TM Touch Real-Time PCR Detection System (Bio-Rad, USA).

Microbial community structure analysis

Six activated sludge samples were collected from the biochemical tank of each AAO process. According to the manufacturer, the FastDNA@SPIN Kit for soil (MP Biomedicals) was used to extract total DNA from sludge samples. Amplification was performed using primers (515 F, 806R) targeting the V3 and V4 regions of the bacterial 16 S rRNA gene [17]. High-throughput sequencing was performed using the Illumina NovaSeq platform (Novogene Co., Ltd. (Beijing, China)) and yielded 250 bp paired-end reads.

All high-quality sequences were clustered into operational taxonomic units (OTUs) and OTUs with 97% similarity were subjected to bioinformatic statistical analysis [18]. The microbial alpha and beta diversity levels were estimated according to OTUs [19]. Microbial alpha diversity indexes (including Shannon, Observed Species, Simpson and Chao 1) were calculated by Mothur package. Network relationships were analyzed to identify the relationship between core functional microorganisms and environmental factors responsible for additional PFCS variation.

Box-Behnken method optimization experiment

Total nitrogen removal rate (TNR, Y) was used as the response value, HRT (A), C/N ratio (B), and temperature (T, C) as three experimental factors. The Box-Behnken test was performed on three factors and at three levels. Establish a regression model and test the levels and codes of each factor as shown in Supplementary Table S2.

Statistical analysis

All experiments were completed with triplicate samples. Data of water quality index were analyzed by SPSS 22.0 and OriginPro 2021. ANOVA analyzed statistical significance between and within groups. The Design-Expert 13 software system was used for Box-Behnken analysis.

Results and discussion

The in-situ addition of PFCS enhanced the removal performance of COD, TN, NH4+-N, and TP from wastewater

The influent and effluent indicators for COD, NH₄⁺-N, TN, and TP concentrations of the group CK and group A are shown in Table 1; Fig. 2. Statistical analysis revealed significant differences in contaminant removal between groups (p < 0.05). For TN, Group A demonstrated both statistical and practical superiority, achieving 72.22% removal (effluent: 6.88 ± 0.71 mg/L) compared to 61.6% (9.65 ± 0.90 mg/L) in Group CK, a meaningful 10.62% improvement supported by absolute concentration reductions. NH₄⁺-N showed statistical but not practical improvement (1.14% increase, within ± 5% error range), while TP’s 3.52% enhancement (p = 0.03) reflected higher initial loading (4.56 ± 1.97 vs. 2.27 ± 0.45 mg/L) despite lower absolute effluent concentrations (0.07 ± 0.03 vs. 0.12 ± 0.06 mg/L). Thus, while statistically significant, the practical superiority of Group A is more robust for TN than for NH₄⁺-N or TP.

Table 1.

Mean values of water quality indicators for groups A and CK

Indicator	Group CK			Group A
	Influent (mg/L)	Effluent (mg/L)	Removal rate (%)	Influent (mg/L)	Effluent (mg/L)	Removal rate (%)
COD	233.38 ± 74.16	8.94 ± 2.00b	95.83	211.25 ± 70.18	6.16 ± 1.94a	96.71
Ammonia (NH4+-N)	15.13 ± 2.66	0.25 ± 0.22b	98.21	19.23 ± 2.10	0.12 ± 0.11a	99.35
Total Nitrogen (TN)	25.77 ± 4.88	9.65 ± 0.90b	61.60	24.94 ± 2.44	6.88 ± 0.71a	72.22
Total phosphorus (TP)	2.27 ± 0.45	0.12 ± 0.06b	94.67	4.56 ± 1.97	0.07 ± 0.027 a	98.19

Group A added a PFCS to the anoxic tank of the AAO reactors at a concentration of 5 mg/L, and Group CK was not treated. The PFCSs were added at five time points: 21 October, 23 October, 29 October, 30 October, and 31 October. As a result, the average COD/TN ratios for Groups CK and A were 9.06:1 and 82.38:1, respectively. The influent and effluent values in the table correspond to the entire Anaerobic-Anoxic-Oxic (AAO) process

Fig. 2.

30 days of wastewater treatment plant performance

PFCS significantly upregulated the number of copies of genes related to ammonia-oxidizing microorganisms

The copy number of relevant nitrogen metabolism genes was analyzed in two groups of samples using absolute quantitative PCR. Figure 3A presented the copy numbers of nitrogen metabolism genes in the two groups of samples. The copy numbers of all genes in group A were higher than those in group CK, except for the napA and nirS. Apart from nirS, all functional genes in group A reached highly significant compared to the group CK (p < 0.05).

Fig. 3.

The number of copies of genes related to nitrogen metabolism was significantly upregulated by PFCS (A). The abundance of AOB/AOA and comammox (B)

The gene copies of AOB 16 S rRNA and AOB amoA in group A (1.76 × 10⁶ copies/µL, and 2.71 × 10⁸ copies/µL) were higher than those in group CK (1.37 × 10⁶ copies/µL, and 1.86 × 10⁶ copies/µL). Moreover, the gene copies of the AOB amoA was 145.4 folds higher than that of the group CK. AOB diversity and population abundance in activated sludge are the main factors affecting nitrification efficiency in activated sludge systems, and an increase in AOB diversity and population density can improve nitrification intensity [20].

The gene copies of narG in group A (7.10 × 10⁶ copies/µL) were higher than that in group CK (9.67 × 10⁵ copies/µL). However, napA in group A (9.93 × 10⁴ copies/µL) was lower than that in group CK (1.30 × 10⁶ copies/µL). It has been reported that the narG and napA enzymes had similar functions as nitrate reductases [21]. The physiological function of the narG in the membrane was to perform nitrate respiration under anaerobic and hypoxic conditions [22]. The periplasmic nitrate reductase napA is in the cytoplasm, and its function may be related to aerobic denitrification [23]. Due to the addition of PFCS, the microbial community in group A grew and multiplied faster with abundant carbon sources, and the growth and multiplication of bacteria require sufficient oxygen. Therefore, the dissolved oxygen concentration in the effluent was lower than in the CK group, and a hypoxic environment may occur. Therefore, the gene copy number of napA in group A was lower than that of napA in group CK, but because napA and narG were in the same ecological niche and had the same function, and narG compensated for the role of napA, the gene copy number of narG in group A was significantly higher than that of narG in group CK, and the denitrification effect of group A was stronger than that of group CK.

In addition, the gene copies of the nxrB in group A was 1.43 folds higher than that of the group CK. The gene nxrB, which encodes the beta subunit of nitrite oxidoreductase, is a functional and phylogenetic marker gene for nitrifying bacteria [24]. These results indicated that the activity of enzymes involved in the interconversion of nitrate and nitrite was significantly increased in group A compared to group CK. NirS is a pivotal gene in the denitrification process, catalysing the production of NO and N₂O using nitrite as a substrate [25, 26]. Despite the absence of a statistically significant disparity in the copy number of the nirS gene between the two groups, a 1.47-fold and 1.60-fold increase was observed in the copy numbers of AOB 16 S and Anammox 16 S, respectively, in group A compared with group CK. This finding indicates that the anaerobic ammonia oxidation pathway in group A may facilitate electron transfer from nitrite to ammonia more efficiently than in group CK.

Although nirS gene copies did not change significantly between the two groups, the Anammox 16 S copy numbers were 1.47 folds and 1.60 folds higher in group A than in group CK, respectively. The nirS is an extremely important gene in the denitrification process, catalyzing the production of NO and N₂O with nitrite as a substrate.

We also examined the abundance of AOB/AOA and comammox (Fig. 3B). It was found that the relative abundance of these standard ammonia-oxidizing microorganisms increased to varying degrees in all genera except Nitrosospira, whose abundance decreased. This result is consistent with the absolute quantitative PCR results.

PFCS significantly altered the bacterial community and their potential functions

A total of 742,616 high-quality sequences were obtained from 12 samples ranging in size from 48,829 to 71,929. A Venn diagram (Fig. 4) showed that the total number of OTUs for the two groups of samples were 6524, with 41.85% of OTUs being shared between the two groups, indicating that the bacterial communities changed in response to the increase in PFCSs. The entire coverage index for each sample reached 0.99, indicating that most bacteria were sequencing-detected in all samples. The Chao 1 index ranged from 2692.31 to 3370.41, indicating community richness (Supplementary Table S3). Compared with group CK, group A exhibited higher Observed Species and Chao1 indexes. It was demonstrated that group A with PFCSs showed higher community diversity and richness.

Fig. 4.

Venn analysis

A total of 51 phyla were detected in all samples (Fig. 5 A) and (Fig. 5B) showed the relative abundance of major bacterial groups at the phylum and genus levels, respectively. The two main phyla, Proteobacteria and Bacteroidota, account for about 27.01% and 24.58% in group A, 34.53% and 23.76% in group CK (Fig. 5A). The third dominant phyla in group CK was Myxococcota, accounting for about 11.90%. However, the third dominant phylum in group A was Firmicutes (11.02%). In addition, Nitrospirota was the fourth most dominant phylum in Group A (6.88%), which was 2.38 folds than that in Group CK (2.89%). The relative abundance of five phyla was significantly higher in group A compared to group CK, including the phyla Firmicutes, Acidobacteriota, Nitrospirota, Actinobacteriota, and Bacteroidota. The major bacterial phyla in wastewater samples were Proteobacteria, Firmicutes, Actinobacteria and Bacteroidetes [27]. Proteobacteria and Bacteroidota have been shown to perform important roles in water treatment systems [28]. Studies have demonstrated that Bacteroidota plays an essential role in the initial phase of biofilm formation [29]. It has been reported that Firmicutes and Bacteroidetes played a key role in the degradation of pollutants such as COD, NH₄⁺-N, and TN [30]. Genome annotation and metabolic reconstruction of Acidobacteriota suggested a high degree of metabolic diversity and possible involvement in denitrification, phosphorus removal, and iron reduction [31]. Proteobacteria, Bacteroidetes, Chloroflexi, Firmicutes, Planctomycetia, Acidobacteria, and Actinobacteria at the phylum level, account for more than 80.00% of the total bacterial population [32]. At the phylum level, denitrifying bacteria are mainly classified as Proteobacteria, Actinomycetes, Aquifaceae, Bacteroides, and Firmicutes, which include at least one strain with denitrification ability [32].

Fig. 5.

PFCS significantly altered the bacterial community and their potential functions. Microbial community structure of groups CK and A at the phylum and genus levels (A-B). Schematic correlation between different C/N ratios and microbial nitrogen metabolism (C)

There were significant differences between the top 20 genera of the two groups (Fig. 5B). The top five dominant genera in group A were Nitrospira, g_11–24, Bacteroidetes_BD2-2, Saccharibacteria, and Ferruginibacter, accounting for 6.9%, 3.0%, 2.3%, 2.2% and 1.9%, respectively. And the top five genus in group CK were Haliangium, Nitrospira, OM27_clade, C10-SB1 A and Sulfuritalea, accounting for 6.7%, 2.9%, 2.6%, 2.2%, 2.2% respectively.

In the present study, the average C/N ratios for Group CK and Group A were 9.06:1 and 82.38:1, respectively. The addition of PFCSs significantly increased the C/N ratio in the inlet water. Prior studies have demonstrated that DNRA is advantageous at elevated ambient C/N ratios, whereas denitrification is enhanced at reduced C/N ratios. Additionally, the presence of two prevalent cyclic adenosine monophosphate receptor proteins (designated CRP1) and their homolog CRP2 both be implicated in the modulation of the two competing anisotropic nitrate reduction pathways DNRA and denitrification [33]. We analyzed the significantly different genera in combination with nitrogen metabolism-related genes, as shown in (Fig. 5C). Nitrospira plays a critical role as an aerobic chemolithoautotrophic nitrite-oxidizing bacterium in nitrification processes [34]. Complete ammonia oxidizers (comammox), the most diverse and widespread nitrite-oxidizing bacteria (NOB), oxidize ammonia to nitrate and represent the primary biological source of nitrate [35]. The abundance of Saccharibacteria, Ferruginibacter, and Nitrospira was positively correlated with COD, nitrogen and phosphorus removal [36]. It has been reported that Paracoccus and Thauera were nirS-type denitrifiers, Pseudomonas was nirK-type and nirS-type denitrifier [4]. In this study, the abundance of Paracoccus and Thauera in group CK was higher than those in group A. However, the abundance of Pseudomonas in group CK was significantly lower than that in group A. These results were consistent with the gene copies of nirS. It is also noteworthy that Zoogloea, one of the 28 core populations in the WWTPs [37], was ranked eighth in abundance with 1.50% in the CK group, whereas it was significantly reduced to 0.31% in the A group. It has been shown that Zoogloea can only utilize small molecule carbon sources such as short chain fatty acids [38] and amino acids, and after the addition of large molecule carbon sources from plant polysaccharides and their degradation products oligosaccharides, trisaccharides, disaccharides and monosaccharides, Zoogloea cannot utilize these carbon sources and the population cannot grow [39].

To further improve the nitrogen removal, a liquid carbon source made from plant-derived materials, other than the commonly used lactate, acetate, ethanol and other small molecule carbon sources, was used to address the issue of low carbon/nitrogen ratio (electron donor deficiency) and to meet effluent water standards. Interestingly, some bacteria belonging to the classes of Bacteroidia and Clostridia had been enriched significantly in A, and most of them, including Ruminococcus and Ruminococcus-like genera, Prevotella [40], Saccharoferm [41] entans [41, 42], and Acetitomaculum, were usually found in the animal rumen and gut [43, 44]. These genera exhibited high abundance and biodiversity in terms of OTU numbers. Other enriched gut microbes included R-7 group of Christensenellaceae (Clostridia, Christensenellales), the RC9 gut group of Rikenellaceae (Bacteroidia, Bacteroidales), the NK3 A20 group of Lachnospiraceae (Clostridia, Lachnospirales) and the NK4 A214 group of Oscillospiraceae (Clostridia, Oscillospirales). In addition, another genus of Bacteroidetes, Ferruginibacter (Bacteroidia, Chitinophagales, Chitinophagaceae), was also enriched remarkably by supplement of PFCS and one of the Ferruginibacter species, accounting for more than 1%, was previously found in the freshwater sediment and activated sludge [45]. These bacteria, particularly Bacteroidetes, such as Prevotella ruminicola, a major amylolytic and proteolytic bacterium in the rumen [46], and saccharolytic bacteria of Clostridia, such as cellulolytic Ruminococcus flavefaciens and R. albus, could decompose and utilize structurally complex plant polysaccharides, including cellulose, xylan and pectin, and other biopolymers, and in the meantime the nitrogen and phosphorus nutrients could be uptake and assimilate for bacterial biosynthesis of biomacromolecules and growth. Some other enriched bacteria such as Lactobacillus vini could utilize pentoses [47]. On the other hand, the fermentation products such as lactate, acetate and ethanol [48], could be used as electron donors, and nitrate and nitrite as electron acceptors by denitrifying bacterial. Our findings align with studies using lignocellulosic carbon sources (e.g., pistachio shells) to enrich denitrifiers (Thauera, Pseudomonas) and nitrifiers (Nitrosomonas) [49], and mirror observations in vertical flow constructed wetlands where composite plant-based carbons enhance denitrification (nirS/nosZ↑ 1.8–2.5folds) while sustaining phosphorus removal [15]. However, our work diverges by revealing increase in ammonia-oxidizing bacteria (AOB) (Nitrosomonas), critical for initiating nitrification-a contrast to systems using simple carbons (e.g., methanol) where heterotrophs suppress AOB [34]. This unique trait stems from PFCS’s slow-release carbon properties, balancing heterotrophic and autotrophic activity to enable coupled nitrification-denitrification under low C/N, unlike single-pathway approaches.

Furthermore, all samples formed two clusters according to PCoA (Fig. 6). PCoA 1 and PCoA 2 explained 62.35% and 10.72% of the variability in ASVs, respectively. The Adonis test revealed that adding PFCSs significantly changed the bacterial community structure in groups A and CK.

Fig. 6.

Unveil the microbial community similarities OTU at 97%

Bacterial metabolic function was predicted for two groups by standardizing 16 S rRNA OTUs using PICRUSt and subsequently comparing them with the Kyoto Encyclopedia of Genes and Genomes (KEGG) database [50]. A total of 176 KEGG pathways involved in the ecosystem function of bacterial communities were obtained based for predicting the response of microbial metabolism to the addition of PFCS (Fig. 7). The three main categories of significant functional differences were metabolism, genetic information processing and cellular processes, with metabolism accounting for 77.7%. Based on KEGG abundance statistics, 14 pathways were significantly different at the 0.01 level. The relative abundance of bacterial chemotaxis, peroxisome, fatty acid biosynthesis, and synthesis and degradation of ketone bodies were higher in group CK than those in group A, while all other metabolic pathways were lower than those in group A. This suggested that the added PFCS significantly affected on promoting genetic information processing, biosynthesis of other secondary metabolites, metabolism of cofactors and vitamins, and amino acid metabolism. Genetic information processing and biosynthesis of other secondary metabolites were crucial for microbial metabolism, and it influenced the abundance of microbial diversity.

Fig. 7.

Three levels pathways of variation of bacterial metabolic function investigated using PICRUSt based on the KEGG database between groups A and CK

Ten genera were positively correlated with COD, TP, and TN, but negatively correlated with NH4⁺-N

The relationship between environmental factors and the top 20 dominant bacterial genera was investigated using Spearman rank correlation, and the correlation and significance p-value between the two were obtained. The heatmap was shown as follows (Fig. 8). Nine genera, including Nitrospira, Ellin6067, Bacteroidetes_BD2_2, Saccharibacteria, and Christensenellaceae_R_7_group showed significantly positive correlations with NH₄⁺-N, but negative correlations with COD, TP and TN indicators. Similarly, Sulfuritalea and Rhodoferax showed significant positive correlations with COD, TP, and TN but significant negative correlations with NH₄⁺-N. In addition, ten genera were positively correlated with COD, TP, and TN but negatively correlated with NH₄⁺-N.

Fig. 8.

Heatmap for the top 20 bacterial genera and environmental factors

The addition of PFCSs significantly increased the positive correlation between bacterial genera

The ecological network relationships (P < 0.5) between genera in the microbial communities of groups A and CK were demonstrated by the symbiotic network of microorganisms (Fig. 9). The red and green lines in the figure represent positive and negative correlations, respectively. In addition, the degree of connectivity of the nodes is a reflection of the significance of the genus in the network [51]. At the phylum level, both groups A and CK had seven phyla in their main flora composition, with an increase in the Bdellovibrionota and Nitrospirota phyla but a decrease in the Acidobacteriota and Firmicutes phyla in group A compared to group CK, thus further verifying that the added PFCS promoted the growth and reproduction of microflora associated with nitrogen metabolism. In group A, the four nodes with the highest connectivity showed bacteria belonging to the Bacteroidota, while in group CK the four nodes with the highest expression of Proteobacteria, these results were consistent with the main phyla of the microbial community. Except Christensenellaceae_R_7 and Ellin6067, Ferruginibacter and Saccharibacteria, the other genera in group A showed a significant positive correlation. In contrast, in the group CK, only four bacteria were positively correlated with each other and the other genus. It can be inferred that the addition of PFCS in group A helped to promote a positive correlation between the bacterial groups. However, the dominant bacterial flora was completely different between the two groups, indicating that adding PFCSs significantly affected the composition of the dominant bacterial genus.

Fig. 9.

Top 20 bacterial genera co-occurrence network of group CK (A) and group A (B)

RSM was used to optimize and determine the optimal parameters for PFCSs C/N ratio, temperature and HRT

The incorporation of surplus carbon derived from plant sources can result in the generation of excess activated sludge, which may subsequently exert a greater burden on subsequent residual sludge treatment. A response surface methodology (RSM) approach was employed to identify the optimal parameters for plant carbon source addition ratio, temperature, and HRT, with total nitrogen removal efficiency serving as the criterion for evaluation.

The TNR results were presented in Supplementary Table S4 with multiple regression fitting analysis, and the established quadratic regression model for: Y = 81.34 + 0.8606 A + 2.69 B − 0.1237 C − 1.26 AB − 0.38 AC − 0.8475 BC − 3.62 A² − 2.86 B² − 14.54 C². This model was analyzed for significance, and the results are shown in Table 2.

Table 2.

Analysis of variance for regression model

Source	Sum of Squares	df	Mean Square	F-value	p-value	Significant
Model	1106.80	9	122.98	488.90	< 0.0001	***
A-HRT	3.95	1	3.95	15.70	0.0074	**
B-C/N	38.65	1	38.65	153.64	< 0.0001	***
C-T	0.1225	1	0.1225	0.4871	0.5114
AB	3.16	1	3.16	12.55	0.0122	*
AC	0.5776	1	0.5776	2.30	0.1805
BC	2.87	1	2.87	11.42	0.0149	*
A²	43.71	1	43.71	173.79	< 0.0001	***
B²	27.24	1	27.24	108.31	< 0.0001	***
C²	704.96	1	704.96	2802.58	< 0.0001	***
Residual	1.51	6	0.2515
Lack of Fit	1.09	2	0.5436	5.15	0.0782
Pure Error	0.4221	4	0.1055
Cor Total	1108.31	15

* p < 0.05, ** p < 0.01, *** p < 0.001

The two-way interaction effects of HRT, C/N ratio and temperature on TN removal from wastewater were shown in Fig. 10. When the HRT was 10.14 h, the C/N ratio was 12.77, and the temperature was 25.13 ℃, the RSM predicted that the total nitrogen removal by the AAO process was 81.85%. Three parallel tests were conducted under this condition, and the actual test results showed that the average total nitrogen removal was 81.27%, with a deviation of only 0.58% from the model prediction, indicating that the model prediction was reliable.

Fig. 10.

Interactive effects of HRT and C/N ratio (A), C/N ratio and temperature (B), and HRT and temperature (C) on TN removal rate

It is evident from the above addition ratios that calculations demonstrate the potential of PFCS produced through microbial fermentation of agricultural waste (e.g. corn stover) to engender significant cost savings when compared to conventional carbon sources. Specifically, the total production cost of PFCS is $0.20/kg of bioavailable COD, including raw material procurement ($0.12/kg) and fermentation processing ($0.08/kg). In contrast, fossil-derived alternatives like methanol and sodium acetate cost $0.65/kg COD and $1.10/kg COD, respectively, based on 2023 market data(Alibaba Wholesale). For a full-scale WWTP with a capacity of 100,000 m³/day and a PFCS dosage of 5 mg/L, annual COD demand reaches 328.5 tons. Using PFCS instead of methanol reduces annual carbon source costs from $213,525 to $65,700, yielding $147,825 in savings (69% reduction).Beyond direct costs, PFCS mitigates environmental burdens. Lifecycle assessment (LCA) reveals that methanol production emits 2.5 kg CO₂/kg COD, whereas PFCS (from agricultural residue valorization) generates only 0.3 kg CO₂/kg COD, avoiding ~ 82 tons of CO₂ emissions annually. This aligns with circular economy principles by repurposing waste biomass and reducing reliance on fossil resources.Environmentally, PFCS exhibits an 88% lower carbon footprint (0.3 kg CO₂/kg COD) than methanol (2.5 kg CO₂/kg COD), achieving an annual emission reduction of 722.5 metric tons CO₂. This aligns with circular economy principles through valorization of 6.8 tons agricultural residue per ton PFCS produced, while maintaining total nitrogen (TN) removal efficiency > 82% at C/N ratios of 2.8–3.2, outperforming methanol-based systems (65–72% TN removal) under equivalent operational conditions.

We emphasize that these estimates are conservative, excluding potential revenue from sludge-derived biogas (enhanced by PFCS’s bioavailable carbon) or carbon credit incentives. Future studies will refine these models with operational data from pilot-scale trials.

Conclusion

The addition of PFCS increased the C/N ratio during AAO process in this study, enhanced ammonia oxidation efficiency in the wastewater treatment. The addition of PFCS not only significantly upregulated the copy number of genes associated with ammonia-oxidizing genes, but also contributed to changes in the structural diversity and composition of the microbial communities in group A, including significant increases in the relative abundance of the phyla Firmicutes and Nitrobacte, and the genera Nitrospira, Saccharimonadales, and Ferruginibacter, which were more effective in removing ammonia and nitrogen. Moreover, the addition of PFCS promoted synergistic effects between microbiota. In conclusion, the addition of inexpensive plant-based carbon sources is effective not only in improving denitrification in the AAO process but also in reducing the operating costs of wastewater treatment. The application of PFCS can reduce the quantity of solid waste generated from agricultural, forestry, and wetland plants. This is not only under the principles of sustainable development but also has a positive impact on the protection of the Earth’s ecosystem. To further validate the advantages and disadvantages of adding PFCS, future testing will be conducted in a range of wastewater treatment plant environments with extended operating periods.

Authors' contributions

HC: Writing-original draft, Conceptualization, Methodology, Investigation, Funding acquisition. JH: Writing-original draft, Formal analysis, Funding acquisition. DJ: Writing-review & editing, Software, Investigation, Formal analysis. XW: Writing-review & editing, Methodology, Formal analysis. XD: Writing-review & editing, Investigation, Formal analysis. JD: Writing-review & editing, Investigation, Formal analysis, Funding acquisition. SS: Writing-review & editing, Formal analysis. QX: Writing-review & editing, Formal analysis. CW: Investigation. DQ: Writing-review & editing, Supervision, Investigation, Validation, Funding acquisition.

Funding

The author(s) declare that financial support was received for the research, authorship, and/or publication of this article. This work was financed by Scientific Research Items Foundation of Hubei Educational Committee (Q20194303), Jingchu University of Technology Ph.D. Startup Fund (YY202446), Basic Public Welfare Research Program of Fujian Province of China (2024R1031001), National Natural Science Foundation of China (32102769), Foreign Cooperation Project of Fujian Academy of Agricultural Sciences of China (DWHZ2023-08), The Key Research and Development Program of Hunan Province of China (2023 WK2001).

Data availability

The data presented in the study are deposited in the repository NCBI https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1156371.

Declarations

Competing interests

The authors declare no competing interests.