Application of alkali-heated corncobs enhanced nitrogen removal and microbial diversity in constructed wetlands for treating low C/N ratio wastewater

Lack of carbon source is the main limiting factor in the denitrification of low C/N ratio wastewater in the constructed wetlands (CWs). Agricultural waste has been considered as a supplementary carbon source but research is still limited. To solve this problem, ferric carbon (Fe–C) + zeolite, Fe–C + gravel, and gravel were used as substrates to build CWs in this experiment, aiming to investigate the effects of different carbon sources (rice straw, corncobs, alkali-heated corncobs) on nitrogen removal performance and microbial community structure in CWs for low C/N wastewater. The results demonstrated that the microbial community and effluent nitrogen concentration of CWs were mainly influenced by the carbon source rather than the substrate. Alkali-heated corncobs significantly enhanced the removal of NO2−-N, NH4+-N, NO3—N, and TN. Carbon sources addition increased microbial diversity. Alkali-heated corncobs addition significantly increased the abundance of heterotrophic denitrifying bacteria (Proteobacteria and Bacteroidota). Furthermore, alkali-heated corncobs addition increased the copy number of nirS, nosZ, and nirK genes while greenhouse gas fluxes were lower than common corncobs. In summary, alkali-heated corncobs can be considered as an effective carbon source.


Introduction
The development of industry and agriculture has resulted in the discharge of large amounts of wastewater (Qin et al. 2023).China's rural wastewater volume has been close to 70 million tons per day, accounting for 30% of the nationwide wastewater (Feng et al. 2021).Compared with urban wastewater, rural wastewater sources are scattered.Furthermore, due to excessive application of nitrogen fertilizer, the C/N ratio of rural wastewater (about 2-5) is generally lower than that of urban wastewater (about 5-10) (Yuan et al. 2020).However, the economic level of rural areas is poorer than cities, inspection and maintenance of wastewater facilities in rural areas is not feasible (Si et al. 2018).So, there are still a large amount of nitrogen-rich wastewater discharged directly into rivers or lakes without treatment, leading to eutrophication (Jia et al. 2021a).Hence, unlike cities, the appropriate treatment of rural wastewater should be economical, simple and decentralized.
Constructed wetlands (CWs) are engineering techniques that artificially simulate the natural environment and have been proven to be a simple decentralized reactor with high nitrogen removal rates but without complex operations (Huang et al. 2022;Jia et al. 2021b).CWs are mostly composed of aquatic plants, substrates, water, and microorganisms (Wu et al. 2015), which remove nitrogenous pollutants from wastewater mainly through microbial nitrification-denitrification (Tan et al. 2021).Nitrification converts ammonium (NH 4 + -N) to nitrite (NO 2 1 3 and NO 2 − -N to N 2 by the participation of denitrifying bacteria (Luo et al. 2020).Denitrification plays a critical role in CWs, and denitrifying bacteria require electrons and nutrients from carbon sources (Ma et al. 2022;Fan et al. 2022).However, the C/N ratio of rural wastewater are gradually decreasing (Zhou et al. 2017).By now, the shortage of carbon source is the main limiting factor for denitrification in CWs.Wang et al. (2017) reported that the N removal rate of CWs was 70% when the carbon source was sufficient, but this value was only 40% when the carbon source was depleted.Researchers often mitigate this problem by adding supplementary carbon sources to CWs such as glucose, methanol, and acetate (Xu et al. 2018).Although the addition of these synthetic carbon sources can increase denitrification rates, but the CW operating costs also increased (Kaur et al. 2018).Therefore, it is necessary to develop a lowcost and easily accessible carbon source, especially for low C/N ratio wastewater treatment in rural.
Agricultural waste such as straw has been identified as alternatives to synthetic carbon sources (Zhou et al. 2022).China's annual straw production exceeds 1.0 × 10 9 t (Wang et al. 2022), with rice straw and corncobs accounting for 32.3% and 45.0%, respectively (Li et al. 2020).If these can be fully utilized, it will significantly reduce the cost of CWs operations.Tao et al. (2020) reported that corncobs can provide a sufficient carbon source for denitrification.Sun et al. (2019) reported that the addition of rice straw could increase the abundance of denitrifying bacteria.However, researchers have found that adding plant-based carbon sources directly to CWs also has significant drawbacks.Hang et al. (2016) reported that the cellulose and hemicellulose of plant are tightly linked to lignin by hydrogen bonds, making it difficult to be utilized by microorganism.Li et al. (2019) reported that the application of untreated plant carbon sources to CWs can cause excessive release of macromolecular organic matter, leading to deterioration of water quality.Therefore, some researchers have attempted to optimize the carbon sources by alkali-heated treatment, which can effectively promote the hydrolysis of lignin and cellulose with less relative mass loss of the carbon (Zheng et al. 2022;Sun et al. 2022).However, studies on denitrification and microbial community characterization in CWs with alkali-heated plant carbon sources are still limited.
In our previous works, Zhao et al. (2022) reported that the use of Fe-C + zeolite, Fe-C + gravel, and gravel as substrates for CWs was effective in improving CWs denitrification.Ferric-carbon (Fe-C) is made of zero-valent iron and activated carbon.Zeolite and gravel can adsorb NH 4 + by ion exchange (Jia et al. 2020).However, Fe-C micro-electrolysis and substrate adsorption has been difficult to purify rural wastewater with a continuously decreasing C/N ratio.Therefore, based upon our Zhao et al. (2022), we have designed the CWs by utilizing corncobs, rice straw and alkali-heated corncobs as the test carbon sources, aiming to (1) investigate the effect of different carbon sources on the nitrogen removal performance of CWs for low C/N wastewater; (2) analyze the microbial community structure and identify the bacteria governing the pollutant removal process; and (3) identify the optimal carbon source combined with nitrogen functional genes and greenhouse gases (GHGs) (N 2 O, CH 4 , and CO 2 ) emission fluxes.

Preparation of the carbon source
The corncobs and rice straw used in this experiment were purchased from a farmer in Beibei district of Chongqing municipality.The corncob was cut into about 2 cm 3 cube, and rice straw was cut into 2-3 cm pieces.Both were washed and then dried to a constant weight at 50 °C.A portion of the corncobs were soaked in 2% NaOH solution at 90 °C for 1 h, subsequently washed with deionized water to neutrality and dried at 50 °C to a constant weight.The following text abbreviates the alkali-heated corncobs as "preCorn".

Batch experiments
Carbon source materials (corncobs/rice straw/preCorn, 2.00 g) were wrapped in a 60 mesh nylon bag, hung in a 250 ml beaker and then immersed in 0.2 L of deionized water.Each treatment has three replicates.The experimental period was 11 days; chemical oxygen demand (COD) and total nitrogen (TN) in the soaking solution were analyzed every 24 h.The following equations were used to calculate the cumulative COD and TN release from carbon source (Sun et al. 2022): TR is the cumulative COD or TN release from carbon source (mg g −1 ).C T is the concentration of COD or TN in the soaking solution (mg L −1 ).V is the volume of soaking solution (0.2 L). m is the mass of carbon source (2 g).
Acorus calamus L. is a perennial cold-resistant herb that mostly survives in wetlands, and is commonly used in CW due to its easy accessibility, highly resistant and strong pollutant enrichment ability (Tao et al. 2020).After being cultured for 20 days (Table S1), four well-developed plants were selected and transplanted into each CW.The CWs were constructed and operated on May 12, 2022.Activated sludge extracted from a wastewater treatment plant of Chongqing was mixed with water to reach a mixture concentration of 1000 mg L −1 .The activated sludge mixture (15 L) was then inoculated into each CW on May 20.
Carbon source was placed in a 60-mesh nylon bag at 1% (w/w) of substrate fresh weight, and it was suspended vertically within the central PVC pipe.Two to three glass beads were added to each bag to prevent the sample from floating.The carbon source was added on May 26, which was designated as the first day of the experiment, for a total of 58 days.The hydraulic retention time of CWs was 48 h.The aeration time was 2 h per day (11:00-12:00 and 22:00-23:00) with a rate of 0.75 L min −1 .Aeration could increase the community richness and diversity of bacteria.The synthetic wastewater (C/N = 3) with 153.7 mg L −1 COD, 19.5 mg L −1 NH 4 + -N, and 28.8 mg L −1 NO 3 − -N was fed to the CW systems directly (Table S2).All analytical reagents were provided by Kelong Chemical Co., Ltd (Chengdu, China).

Water samples
Water samples were collected every 48 h, and the extraction time was from 9:00 am to 10:00 am.COD, NH 4 + -N, NO 3 − -N, NO 2 − -N, and TN were determined by standard methods (APHA 2005).Dissolved oxygen (DO) and pH were determined by a Mettler-multiparameter analyzer (Mettler Toledo SG98, Shanghai).

Quantitative PCR and microbial community assays
At the end of the experiment, substrates were extracted randomly from CWs at depths of 15 cm, 30 cm, and 45 cm, then mixed uniformly.The extracted of the biofilm on the substrate surface and DNA extraction as described in the previous study (Si et al. 2018).The primers used to amplify the nitrogen functional genes (amoA, nirS, nirK, and nosZ) were listed in Table S3.And the generated copies were quantified by a QuantiFluorTM-ST blue fluorescence quantitative system (Promega) with the TransGen AP221-02 method.High-throughput sequencing of bacterial communities was performed by Majorbio Bio-pharm Technology Co., Ltd (Shanghai, China).Uclust was used to cluster all these sequences into operational units (OTUs) with a 97% sequence identity.The Chao1, Shannon, and Simpson indices were calculated by Mothur (http:// www.mothur.org/).

Gas sampling
The static opaque chamber technique was chosen to determine the fluxes of CH 4 , N 2 O, and CO 2 .On the sampling date, gas samples were collected at 0, 10, 20, and 30 min after the chamber (120 cm height) was closed during 11:00-12:00 am using a 50-mL syringe.A fan was installed inside the chamber to ensure that the air in the chamber was well mixed.The N 2 O, CO 2 , and CH 4 concentrations were determined by a gas chromatograph (Agilent 7890A, USA) equipped with flame ionization detector (FID) and electron capture detector (ECD).The gas fluxes were calculated by the following equation (Zhou et al. 2022): F is the gas flux (μg m −2 h −1 for N 2 O, mg m −2 h −1 for CH 4 and CO 2 ).V is the chamber volume (m 3 ).A is the bottom area of the gas chamber (m 2 ), and dc/dt is the rate of gas concentration change.

Statistical analyses
All results are expressed as the means and standard errors (SE) that were obtained from three replications (mean ± SE, n = 3).The experimental data were analyzed using SPSS 22.0 (SPSS Inc., Chicago, USA) and the significance test was performed using the one-way ANOVA

Apparent release experiment
The dynamics of COD and TN released from the three carbon sources during the 11-day batch experiments are presented in Fig. S2.The COD and TN released from the residues increased from day 0 to day 3 and finally stabilized.
The COD increased rapidly at the beginning (Fig. S2), was probably due to soluble organic matter being solubilized and the release of carbon sources from biomass.The cumulative COD and TN released from rice straw were 145.2 mg g −1 and 2.3 mg g −1 , respectively, both were significantly higher than the others (P < 0.05).The cumulative COD and TN released from preCorn were 93.8 mg g −1 and 0.53 mg g −1 , respectively, which were 63.2% higher and 19.7% lower than those released from corncobs.Kirupa Sankar et al. ( 2018) reported that alkali heating can break the chemical bond between cellulose and stimulate the hydrolysis of plant residues.Zheng et al. (2021) reported that alkali heating could inhibit the release of NO 3 − -N from residues.Therefore, the COD/TN of preCorn soaking solution was 175.7, significantly higher than that of corncobs (86.8) and rice straw (62.8) (Fig. S2).The carbon source should have the advantage of high carbon and low nitrogen content.So, the release characteristics of preCorn were more suitable for practical needs.

pH and DO in CWs
The variation of DO and pH in different CWs are presented in Fig. S3.The average influent DO concentration was 7.5 mg L −1 .The decomposition of plant residues can consume large amounts of oxygen (Tang et al. 2022b).Among them, the average effluent DO concentration was 0.15 mg L −1 after rice straw addition, which was lower than others.Huang et al. (2019) reported that DO concentrations below 0.5 mg L −1 was essential for denitrification.In this experiment, the DO concentration after carbon sources addition ranged from 0.15 to 0.36 mg L −1 , which was below above standard.
The average influent pH was 7.4, and effluent pH ranged from 7.2 to 7.7.Statistical differences in pH changes were not always observed between influent and effluent (Fig. S3), which was consistent with Shen et al. (2019).The reason might be that although the decomposition of plant residues can release organic acids, the denitrification process can produce a certain amount of alkaline matter, which can inhibit the decrease of pH (Wu et al. 2022).In general, DO concentration and pH were not the main factors determining denitrification rate and microbial community in this experiment.

Effect of different carbon source additions on COD removal in CWs
The influent and effluent COD concentrations in different CWs are presented in Fig. 1.The effluent COD is mainly produced by the hydrolysis of carbon sources (Yang et al. 2018).Therefore, the effluent COD concentration was mainly influenced by the carbon source rather than the substrate.So, the treatment with corncobs was named A (A1, A2, A3), the treatment with preCorn was named B (B1, B2, B3), the treatment with the rice straw was named C (C1, C2, C3), and the treatment without carbon source was named CK (CK1, CK2, CK3).The effluent COD concentrations ranged from 6.5 to 75.6 mg L −1 during the experiment.After carbon sources addition, the effluent COD concentration peaked on the 10th day.This might be because the large release of COD from the carbon source (Fig. S2).Subsequently, with the maturation of biofilm on the substrate surface, the abundance of cellulose hydrolysis bacteria increased, which hydrolyzed carbonaceous organic matter suspended in the water and make it available to denitrifying bacteria (Shen et al. 2015).Thus, the effluent COD concentration gradually decreased from day 20 to day 58.Except for day 8 to day 12, the effluent COD concentrations were below the grade A (50 mg L −1 ) according to the Environmental Quality Standards for Surface Water of China (GB3838-2002).After carbon sources addition, the average removal rate of COD ranged from 81.3 to 84.7%, which was lower than CK (91.3%).Overall, untreated plant-based carbon sources addition limited COD removal.The reason may be the chemical bonds between hemicellulose and lignin would repress COD degradation.Considering the decomposition of bonds by alkali heat treatment, we speculated that COD releasing from preCorn was more biodegradable; thus, COD removal under treatment B was closer to CK (Fig. 1).+ -N concentration decreased from 19.9 to 1.0 mg L −1 .Jia et al. (2020) reported that CWs remove nitrogen primarily through nitrification and denitrification, with substrate adsorption accounting for less than 3%.So, the effluent nitrogen concentration was mainly influenced by the carbon source rather than the substrate.The average effluent NH 4 + -N concentration under treatment B (3.6 mg L −1 ) was significantly lower than the others (P < 0.05).Furthermore, the average NH 4 + -N concentrations under treatments A and C were 5.7 mg L −1 and 7.1 mg L −1 , which were lower than CK (8.0 mg L −1 ).Thus, the average NH 4 + -N removal rate under treatment B (93.7%) was significantly higher than others.The main pathway of NH 4 + -N removal is nitrification (Lai et al. 2020).Carbon source additions stimulated the removal of NH 4 + -N (Fig. 2).This might be because carbon source provided various organics for bacteria and increased the activity of hydroxylamine oxidoreductase in the CWs.The latter oxidizes hydroxylamine to nitrite, which is essential for accelerating biological nitrogen removal (Tao et al. 2020).On the other hand, hydrolysis of the carbon source can stimulate anaerobic ammonia oxidation in CWs, creating another important NH 4 + -N removal pathway (Tan et al. 2021).

− -N
The influent and effluent NO 3 − -N concentrations are presented in Fig. 3. Denitrification is considered to be the main pathway for NO 3 − -N removal (Luo et al. 2020).Denitrifying microorganisms are more likely to survive in environments with sufficient organic C (Lai et al. 2020).Therefore, from day 0 to 20, the effluent NO 3 − -N concentration was maintained at a low level when the carbon source was sufficient.Subsequently, NO 3 − -N concentration was increased instead under treatments C and CK after day 22.This may be because a large amount of NH 4 + -N was converted to NO 3 − -N through nitrification, but the carbon source was gradually consumed, resulting in a lower denitrification rate.During the experiment, the average effluent NO 3 − -N concentrations under treatments A and B were 0.41 mg L −1 and 0.26 mg L −1 , respectively.Both were significantly lower than C (4.6 mg L −1 ) and CK (5.8 mg L −1 ) (P < 0.05).Correspondingly, the average NO 3 − -N removal rates under treatments A and B were 97.7% and 98.6%, respectively, which were significantly higher than treatments C (85.5%) and CK (84.1%).This may be due to alkali heat treatment increased carbon release and prolonged the high carbon release cycle of the carbon source.Capua et al. 2022).The NO 2 − -N reduction was the ratelimiting step in nitrogen removal (Jia et al. 2020).From day 0 to day 20, the effluent NO 2 − -N concentration under CK decreased continuously but significantly higher than the others.When the carbon source is depleted, nitrite reductase will be inhibited by excessive competition for electrons acceptors with nitrate reductase, resulting in the accumulation of NO 2 − -N (Ge et al. 2012).Similar to NO 3 − -N, effluent NO 2 − -N concentrations under treatments C and CK were increased instead after day 25.Conversely, those under treatments A and B were maintained at low levels during the whole experimental period.The average effluent NO 2 − -N concentrations under treatments A, B, and C were 0.058, 0.043, and 0.25 mg L −1 , respectively, which were significantly lower than CK (0.70 mg L −1 ).NO 2 − -N is not beneficial for subsequent nitrification or direct discharging (Shen et al. 2015).In general, preCorn additions avoided NO 2 − -N accumulation.

TN
The  -2002).During the experiment, the effluent TN concentration in CWs decreased slightly.The average effluent TN concentration under treatment A, B, and C were 5.5, 1.9, and 11.0 mg L −1 , respectively, which were significantly lower than CK (15.9 mg L −1 ).The average TN removal rate under treatment B was 96.8%, which was significantly higher than treatments A (88.3%), C (77.4%), and CK (73.4%).It may be because preCorn addition stimulated the removal of NH 4 + -N (Fig. 2) and reduced the accumulation of NO 3 − -N and NO 2 − -N (Figs. 3 and 4).The results demonstrated that corncobs and preCorn additions were more effective for nitrogen removal.

Microbial community structures in CWs
By determining effluent COD and nitrogen concentrations, we found that the addition of common corncobs and preCorn achieved better pollutants removal rate than rice straw.To further investigate the microbial community structure of CWs with different carbon sources and to identify the bacteria governing the pollutant removal process, we performed high-throughput sequencing.The structure and abundance of microbial communities in CWs were analyzed (Fig. 6).Microorganisms rely on nutrients and attachment sites provided by the carbon sources (Sun et al. 2022).So, the bacterial community structure was mainly influenced by the carbon source rather than substrate.All coverages exceed 99%, suggesting that the results were representative.Sobs index indicates the measured operational taxonomic units (OTU) numbers.The number of OTU under treatment B (1517) was significantly higher than that under treatment A (1418) and CK (1320) (Table 2, P < 0.05).The richness index (Chao1) and diversity indices (Simpson, Shannon) reflect the richness and diversity of microbial communities (Zhao et al. 2018).Of these, the Shannon index emphasizes both the abundance and evenness of the species (Shannon 1948) and Simpson index reflects the proportion of dominant species in the sample (Simpson 1949).Compared with CK, treatment B significantly increased Chao1 (+ 12.9%) and Shannon indices (+ 9.4%), while decreasing Simpson index (− 43.9%) (Table 2, P < 0.05).Similar results were obtained by treatment A, but the differences were not always significant.Many experiments have shown that higher influent nitrogen concentrations lead to microbial lower diversity (Yu et al. 2022).preCorn additions enriched microbial abundance (Table 2), possibly related to the higher nitrogen removal efficiency, especially TN and NO 3 − -N (Fig. 5).
At the phylum level, Patescibacteria (26.1-33.9%),Actinobacteriota (19.2-20.9%),Chloroflexi (14.7-20.0%),Proteobacteria (11.8-17.0%),Bacteroidota (4.7-7.5%),Acidobacteriota (1.5-2.5%),Myxococcota (1.1-1.5%), and unclassified_ K__ norank_ d__ Bacteria (1.1-1.3%) were the top eight.The relative abundance of Patescibacteria under treatment B was significantly lower than CK, while there was no significant difference between treatment A and CK.Carbon source addition had no significant effect on the relative abundance of Actinobacteriota and Myxococcota.Compared with CK, the addition of carbon source reduced the relative abundance of Chloroflexi and Acidobacteriota, where it was significantly reduced by 26.5% and 39.4% under treatment A, respectively, but the decrease was not significant under treatment B. Conversely, the addition of carbon source significantly increased the relative abundance of Proteobacteria and Bacteroidota, which increased by 45.4% and 29.8% under treatment A, and by 54.5% and 63.8% under treatment B, compared with CK.The relative abundance of unclassified_ K__ norank_ d__ Bacteria was low, and there was no significant difference between treatments.Patescibacteria has synergistic effects on nitrogen fixation genes and can inhibit various nitrogen-degrading enzymes (Tang et al. 2022a); thus, it was negatively correlates with nitrogen removal (Fig. 7).Proteobacteria, Actinobacteriota, and Bacteroidetes can participate in denitrification and play an important role in nitrogen degradation (Zhang et al. 2022).Hence, the abundance of these three phyla was positively correlated with the nitrogen removal rate (Fig. 7).Plantbased carbon sources can enrich denitrifying microorganisms (mainly Proteobacteria and Bacteroidetes) (Fig. 6) by providing attachment sites and nutrients (Zhao et al. 2018).These improvements concomitantly resulted in decreased Patescibacteria abundance by reducing nitrogen (Fig. 6).In general, the improved nitrogen removal efficiency in CWs might be ascribed to the optimized structure of microbial community.COD removal rate was positively correlated with the relative abundance of Chloroflexi, Acidobacteriota, and Myxococcota (Fig. 7).These three phyla of bacteria have been demonstrated to have a strong ability to consume organic carbon, mostly survive in environments with sufficient DO and oligotrophic (Tang et al. 2022b;Zhao et al. 2018).The hydrolysis of plant residues would consume DO (Fig. S3) and thus reduce the abundance of these bacteria (Fig. 6).Conversely, alkali heat treatment can mitigate the oxygen deficiency by promoting hydrolysis of carbon sources and thereby maintain their growth.In general, alkali heat-treated carbon source could improve nitrogen removal without weakening COD removal.

Nitrogen functional genes in the CWs
The abundance of nitrogen functional gene was shown in Fig. S4.Both corncobs and preCorn stimulated the copies of nirS, nirK, and nosZ genes.Genes of nirK and nirS are encoding nitrite reductases, and nosZ is the only gene for nitrous oxide reductase (Hu et al. 2016).Carbon sources can supply food and energy for denitrifying bacteria (Li et al. 2019).Thus, the copy numbers of nirS, nirK, and nosZ genes under treatment B were 9.32 × 10 8 , 3.81 × 10 7 , and 6.95 × 10 7 copies g −1 (Fig. S4), respectively, which were significantly higher than CK.Moreover, the ratio of nosZ/ (nirS + nirK) under treatment B was 0.07, which was also higher than the others.The AOA-amoA and AOB-amoA genes reflect the activity of nitrifying bacteria (Zhao et al. 2022).The sum of AOA-amoA and AOB-amoA genes abundance under treatment A and B were 2.73 × 10 5 and 2.83 × 10 5 copies g −1 , which were lower than CK (3.13 × 10 5 copies g −1 ).The copy number of amoA gene decreased with increasing COD concentrations because the competition arising from organics oxidation (Ma et al. 2022).In this experiment, carbon sources addition reduced the copy number of amoA gene, but the difference was not significant (Fig. S4).In summary, carbon sources addition significantly increased the copy number of nirS, nirK, and nosZ genes without significantly decreasing that of amoA gene.This is still playing in favor of nitrogen removal.

Effect of different carbon source additions on GHG emissions in CWs
N 2 O is a by-product of nitrification and an intermediate product of denitrification (Luo et al. 2020).With the extension of the cycle, N 2 O flux decreased gradually, ranging from 48.4 to 2891.9 μg m −2 h −1 .The average N 2 O emissions under treatment A (1176.3 μg m −2 h −1 ) were significantly higher than those under treatment B (382.1 μg m −2 h −1 ) and CK (99.2 μg m −2 h −1 ).The higher N 2 O fluxes observed at CWs with carbon sources than CK could be attributed to the more NH 4 + -N was converted into NO 3 − -N (Fig. 2), which would produce more N 2 O (Fig. S5).The nitrous oxide reductase encoded by nosZ gene can reduce N 2 O to N 2 (Zhao et al. 2022).preCorn addition increased the nosZ/(nirS + nirK) ratio (Fig. S4), resulting in lower N 2 O emissions than corncobs (Fig. S5).
CH 4 can be produced by methanogenic in an anaerobic environment and will be consumed by methane-oxidizing bacteria with sufficient oxygen (Zhang et al. 2022).CO 2 is a major GHG, mainly produced by the respiration of microorganisms and plants (Zhou et al. 2022).During the experimental period, CH 4 and CO 2 emissions had a steadily increased trend (Fig. S5).The hydrolysis of carbon sources consumed a large amount of DO (Fig. S3), thus, carbon sources addition stimulated CH 4 emissions (Fig. S5).Carbon sources addition stimulated CO 2 emissions too (Fig. S5), which was consistent with previous studies (Chen et al. 2019;Yan et al. 2012).One explanation for this is that supplemental carbon positively affects microbial respiration by increasing nutrients.The lower CH 4 and CO 2 production observed under treatment B compared to treatment A, possibly because that alkali-heated corncobs were more easily decomposed by microbes and reduced DO consumption.However, the stimulation of GHG by supplementary carbon sources still needs further investigation.

Conclusion
This study investigated the differences in denitrification and microbial community structure of CWs with different carbon sources for treating low C/N ratio wastewater.Carbon source additions stimulated the denitrification of CWs.Among them, the average TN removal rate reached 96.8% after alkali-heated corncobs addition and had no significant effect on COD removal.The addition of alkali-heated corncobs significantly increased the microbial diversity and optimized the bacterial structure at the phylum level, especially enriching heterotrophic denitrifying bacteria (Proteobacteria and Bacteroidota).Furthermore, alkali-heated corncobs enriched nitrogen functional genes (nirS, nirK, and nosZ) while GHG fluxes were lower than common corncobs.The results from this experiment can provide the guidance for the practical operation of CWs using agricultural biomasses for treating low C/N ratio wastewater.
concentrations in different CWs are presented in Fig. 2.During this experiment, the effluent NH 4 The effluent NO 2 − -N concentration in different CWs was presented in Fig.4.NO 2 − -N is an important intermediate reduced product in nitrification-denitrification (Di

Fig. 1
Fig. 1 Dynamic changes of effluent COD concentration (I, II, III) and removal rate (IV) in different wetlands: (I) ferric-carbon (Fe-C) and zeolite as substrates; (II) Fe-C and gravel as substrates; (III) gravel influent and effluent TN concentrations in different CWs are presented in Fig. 5.The effluent TN concentration of the CWs with carbon source addition was lower than the class A standard (15 mg L −1 ) in the Discharge Standard of Pollutants for Municipal Wastewater Treatment Plant of China (GB18918

Fig. 5
Fig. 5 Dynamic changes of effluent TN concentration (I, II, III) and removal rate (IV) in different wetlands: (I) ferric-carbon (Fe-C) and zeolite as substrates; (II) Fe-C and gravel as substrates; (III) gravel

Table 1
Treatment for this experiment

Table 2
Alpha diversity index of different treatment