Removal of antibiotics and antibiotic resistance genes from domestic wastewater using mesocosm-scale constructed wetlands with different filter media

Abstract

This study aims to investigate the removal of antibiotics and antibiotic resistance genes (ARGs) in raw domestic wastewater by various mesocosm-scale constructed wetlands (CWs) with different filter media types. These included CW1 with gravel media, CW2 with biochar media and CW3 with zeolite media. The plant of choice in the wetland was Typha latifolia and the reactor was run for 12 cycles with each cycle running for 7 days after which new influent sample was fed. Antibiotic residues (ciprofloxacin and cefixime) concentration was analyzed by HPLC method. The removal of antibiotic resistant genes including qnrs, sul1, dfr1, blaTEM, blaCTXM, blaOXA and 16 S rDNA (indicator of total bacterial biomass) were analyzed by qPCR method at varying hydraulic retention time of 3 and 7 days. The conventional pollutants (nutrients and organic) were removed with a similar removal efficiency from the three CWs. The removal efficiencies of the biochar- and zeolite-layered wetlands had higher removal efficiencies for both antibiotics compared to wetland with gravel media. The CT means reflected the increased removal efficiency of genes by the CWs in the order of blaOXA > blaCTXM > sul1 > blaTEM > qnrS > dfr1 > 16 S rDNA. The reduction in concentrations of qnrS and blaCTXM genes differed significantly across the reactors depending upon filter media types while reduction in concentrations of qnrs and sul1 was significantly different under varying HRTs. Correlation analysis suggested the removal efficiency of ARGs to be significantly associated with total antibiotic concentration (p = 0.012) and total bacterial biomass (p < 0.001). The CW with zeolite as filter media and HRT of 7 days was opted as the best choice for contaminant removal. The findings from this study suggest that constructed wetlands are a promising treatment technology for removing emerging contaminants such as antibiotics and ARGs in domestic wastewater.

Introduction

The excessive use of antibiotics in both human and animal healthcare has resulted in the accumulation of these antibiotics, as well as antibiotic resistant genes, in domestic wastewater¹. Antibiotic residues that are not completely metabolized are excreted², thus, they are contributors to the contamination of water bodies and to the emergence of antibiotic resistant bacteria^3,4. Improving the comprehension of the fate of antibiotics in wastewater treatment processes is a must to solve this problem, specifically when considering water recycling schemes^5,6. A study shows that ARGs spread through different methods, one of them being horizontal gene transfer (through transformation, transduction, and conjugation)⁷, and their possible toxicities have attracted a lot of attention worldwide. Conventional biological wastewater treatment processes have been shown to be partially effective in the removal of some antibiotics, however many have still been detected at significant concentrations in secondary treated effluents⁸. Advanced treatment technologies, such as membrane filtration, activated carbon, and advanced oxidation processes, have demonstrated enhanced removal of antibiotics, but their efficacy in eliminating antibiotic resistant genes remains a concern⁹.

One promising solution is the use of constructed wetlands, a natural and eco-friendly technology that has shown great potential in removing a variety of contaminants from wastewater. Constructed wetland systems have already been used in the treatment of a wide range of wastewaters originated from domestic¹⁰, agricultural and industrial sources^11,12. These systems utilize natural processes involving vegetation, filter media, and microbial communities to degrade, transform, or remove contaminants from wastewater^13,14. Depending upon the design, operation and the type of wetland system employed, constructed wetlands can remove a variety of contaminants, including organic matter, nutrients¹⁵, heavy metals, and some organic micropollutants^16,17. Hybrid constructed wetlands with both horizontal and vertical flow of wastewater have been increasingly studied for the removal of nutrients, organic matter, pathogens, heavy metals, antibiotics, and antibiotic-resistant genes, owing to their enhanced treatment efficiency through the combined benefits of aerobic and anaerobic zones. Different classes of antibiotic resistant genes, including those conferring resistance to quinolones, sulfonamides, and beta-lactams, have been reported to be significantly reduced in constructed wetlands treating wastewater^18,19. However, these investigations were mostly carried out with varying hydraulic loading, wetland sizes, and vegetation types, making it difficult to isolate the specific role of the wetland media in emerging contaminant removal⁸. Although the effectiveness of constructed wetlands has been well acknowledged, many studies have tended to focus on the overall removal efficiency rather than the dynamics of ARGs in filter-specific media under operational conditions. In particular, there has been a lack of research that compares different wetland media in a single experimental study to assess the impact of the physicochemical properties of the filter media on antibiotic removal and ARGs. This is important for advancing from the current state of performance comparison to understanding the media-dependent removal characteristics and resistance mitigation strategies. The impact of media choice and hydraulic retention time (HRT) may offer some clues into the ARG persistence characteristics and the rational design of optimized wetlands for controlling emerging contaminants²⁰.

In this context, the current study assesses the removal of antibiotics and antibiotic resistance genes from domestic wastewater in three mesocosm-scale hybrid constructed wetlands with different filter media and hydraulic retention times. The targeted antibiotics were ciprofloxacin (CIP) and cefixime (CEF), which are commonly used in human healthcare and often found in wastewater because of the incomplete metabolism and improper disposal. The quantitative PCR (qPCR) analysis was used to measure the abundance of six ARGs, which were chosen according to their abundance in domestic wastewater²¹, together with 16 S rDNA as a proxy for total bacterial mass. The ARGs chosen included the qnrS gene, dfr1 gene, sul1 gene, and beta-lactamase genes (blaTEM, blaCTXM, blaOXA), which belong to different classes of antibiotic resistance. By combining filter media comparison with controlled hydraulic retention times and multi-level contaminant analysis, this research aims to shed light on the patterns of removal and help optimize constructed wetlands for reducing antibiotics and ARGs in wastewater (Fig. 1).

Fig. 1.

Schematic diagram of constructed wetland with interconnected treatment units and varying filter media types (gravel in CW1, biochar in CW2, and zeolite in CW3).

Materials and methodology

Designing of hybrid constructed wetland (HCW)

A hybrid subsurface flow CW was constructed with three rectangular operational units with dimensions 94 × 15 × 15 cm (length, width, and height) and a tank. All the units of the system as depicted in Fig. 1 were connected through polyvinylchloride pipes (length = 317 mm, inner diameter = 20 mm) and placed in a sequential manner. Three different CWs (CW1, CW2 and CW3) were employed, where each CW differed in the filter media type used in the B and C treatment units of the wetland systems. The media used in the first treatment unit (SS-VF) was composed of 2.5 cm of soil as the top layer, 5 cm of fine gravel (in CW1)/biochar (in CW2)/zeolite (in CW3) as the middle layer and 5 cm of coarse gravel (in CW1)/biochar (in CW2)/zeolite (in CW3) as the bottom layer. The second unit consisted of 2.5 cm of soil as the top layer, 5 cm of coarse gravel (in CW1)/biochar (in CW2)/zeolite (in CW3) as the middle layer and 5 cm of pebbles settled as the bottom layer. These two units were planted with Typha latifolia. The last unit consisted of sand bed (SB). The right end of each pod was slightly tilted to be in a declining position for facilitating the natural gravitational pull on the water. The distance between the first and the last pod was 1.5 m, and there was a 30 cm distance between each pod. Each pod had a water exit point that was 2.5 cm above the bottom of the pod. Before exposing to working conditions, the reactor was optimized for 2 weeks and fed with tap water to achieve plant acclimatization and stabilization of microbial growth in plant rhizosphere. During this time, the system was operated in a batch mode with a contact time of 3 days and no sampling was done.

Treatment of domestic wastewater and sample collection

Prior to the beginning of every new treatment phase, an initial characterization of the influent (raw wastewater from the sewage lines of the suburban community of Quaid-i-Azam University in Islamabad) was performed for a period of four months (September-December) every other week. This domestic wastewater originates from a total of 108 households residing in the colony with an average of 6 persons per household. The average wastewater production comes out to be 170 L person/day. A 12-liter sample of domestic wastewater was initially fed each time as influent into a tank (capacity 500 L) and flowed through the wetland at a rate of 0.5 L per hour. The tank was given a holding period of nearly two hours to allow for the sedimentation of any particle materials. This partially treated wastewater was then allowed to flow through the three functional (SS-VF, SS-HF, and Sand bed) units of the mesocosm in a subsequent manner. All the samples were then kept refrigerated and transported to the laboratory as soon as possible, where they were stored at 4 °C before analysis (within 24 h). Characterization of the effluent was done after its collection from the last processing unit using sterile 500 mL polypropylene bottles. The system was run in a semi-batch mode, with a newly collected influent sample introduced after each 7-day treatment cycle, to account for temporal fluctuation in wastewater composition. Instead of using a single homogenized wastewater source, this cycle-wise technique permitted evaluation of removal efficiency under diverse influent matrices.

The Hydraulic Retention Time (HRT) was defined as the average time wastewater remained within the porous media of the wetland system and was calculated using:

Where, = effective pore volume of the wetland media (L) and Q = volumetric flow rate (L day⁻¹).

The porosity (void ratio) of the filter media was multiplied by the total geometric volume of each treatment unit to determine the effective pore volume. The following porosity values were assumed based on values found in the literature: zeolite (0.40–0.50), charcoal (0.45–0.55), and gravel (0.35–0.40). To normalize system comparisons, an average porosity value of 0.40 was employed for HRT estimation.

Two HRTs in operation were assessed:

• 3 days, indicating treatment with a shorter contact time.

• 7 days, signifying treatment for a prolonged detention period.

By regulating the influent feeding length and permitting wastewater to stay in the system for the predetermined amount of time prior to effluent collection, these HRTs were accomplished.

All of the filter materials’ particle sizes were standardized prior to use in order to guarantee consistency and precise characterisation of the wetland media. The coarse gravel had a particle size range of 10 to 20 mm, while the fine gravel, which is utilized for the top layers, had a particle size range of 2 to 5 mm. The biochar’s particle size range was ground to between 1 and 4 mm. Before being packed into the reactors, all the materials had to be properly cleaned with distilled water to get rid of dust and weakly attached particles. The natural zeolite had a particle size range of 2 to 6 mm.

Biochar preparation

Biochar was prepared from the straw material of sugar cane collected from and was prepared through slow pyrolysis method. The reaction temperature was 400 °C, 600 °C, 800 °C. All of this reaction was done under oxygen limited conditions. Selection of biochar to be tested as filter media in constructed wetland was done due to number of benefits of biochar such as: Biochar can absorb in large number of exchangeable cations and has developed large surface oxygen functional groups and has greater pore structure. Rich and stable carbon, high porosity and large specific surface area (SSA) are the special properties that straw biochar had. It can exist in soil and water for a very long time and has strong adsorption and good stability.

Physicochemical analysis

Wastewater was characterized for the physicochemical parameters such as, pH, electrical conductivity (EC), total dissolved solids (TDS) and salts by their respective digital meters and conventional wastewater quality parameters. TDS in water samples were measured by standard method 1540 C. 5210 B standard method (5-day test) for estimating Biochemical oxygen demand (BOD) and a quick-test kit of range 25–1500 mg L − 1 (114541) by Merck was done for chemical oxygen demand (COD). TN (HJ 636–2012) was determined by a UV–vis spectrophotometer (Shimadzu Instrument Co. Ltd., UV-2450, Japan). TP was measured using ammonium molybdate spectrophotometric method (HJ 671–2013), where “HJ” refers to the national environmental protection standards issued by the Ministry of Ecology and Environment of China. For conducting this analysis all the reagents used were analytical grades and instruments were accurate and precise.

Quantification of antibiotics

In this study, cefixime and ciprofloxacin were screened for removal through the wetland reactors in the influent and effluent samples. Scion 6000 HPLC system equipped with UV detector 6410 was used. C18 column (4.6mmx150 mm) was used for running of the samples and data was collected using CompassCDS software. 0.45 μm membrane filters were used for filtering of the samples in an attempt to get rid of particulates. These filtered samples were then stored at 4 °C and analyzed within 24 h to avoid antibiotic degradation. Even though there was no need to perform extraction due to relatively high concentrations of antibiotics in raw domestic wastewater, filtration-based sample cleanup and timely analysis were ensured. For ciprofloxacin, the mobile phase consisted of 0.025 M phosphoric acid adjusted with triethylamine at pH 3 and acetonitrile (13:87). For cefixime, the mobile phase consisted of methanol and a phosphate buffer at pH 3.2 (solution of sodium dihydrophosphate 40 mM, with condensed phosphoric acid used for adjusting the pH). The column temperature and pressure were maintained at 35˚C and 10.57 Mpa respectively. Sample was injected at a volume of 10 µL and the flow rate of the mobile phase was 1.0 ml/min for detection of each antibiotic. The detection of ciprofloxacin and linezolid were monitored at a wavelength of 278 nm and 254 nm respectively. Calibration graphs were produced by preparing standard solutions of each antibiotic in the concentrations 10, 25, 50, 100 and 1000 mg/L. Calibration linearity (R² > 0.99), repeatability, and determining analytical detection limits before sample analysis were all part of method validation. Values were below the technique detection limit rather than showing a full absence of the target compound in situations where effluent concentrations were interpreted as total elimination. Standardized procedures were used to analyze each sample in order to guarantee analytical accuracy and repeatability.

Quantification of ARGs

DNA was extracted from the wastewater samples and under vacuum, sterile 0.22 μm mixed cellulose ester membrane filters (Millipore, USA) were used to filter 50 mL of effluent. Following the manufacturer’s instructions, the retained biomass on the membrane was aseptically divided into tiny pieces and then exposed to DNA extraction using the DNeasy PowerWater Kit (Qiagen, Germany). Every extraction was carried out twice for quality control. To keep an eye on any contamination, negative extraction controls were incorporated. The extracted DNA was analyzed by gel electrophoresis and NanoDrop 2000 Spectrophotometer for both quality and quantity respectively. Real-time quantitative PCR (qPCR) methods using SYBR Green Real Time qPCR Kit (Thermoscientific, US) were used to determine the abundance of the ARGs. The qPCR assays were run on an Applied Biosystems 7500 Fast Real-Time PCR System (ABI, USA). Temperature conditions for the quantification included initial denaturation at 95˚C for 10 min, followed by 40 cycles of 30 s at 95˚C and 1 min at 60˚C. Both positive and negative controls (Milli-Q water) were included in every run. A total of 40 cycles was applied to improve the chances of product formation from low initial template concentrations. Plasmid DNA was diluted in 10-fold series to generate calibration standard curves with the square of related coefficient (r²) of the standard curve > 0.99, and the amplification efficiency ranging from 95% to 110%. The samples were run in duplicates and the quantification of ARGs were expressed as absolute abundance (copies/mL) and relative abundance (target gene copies/16S rDNA gene copies). The primer sequences are listed in Table S1.

Statistical analysis

The removal efficiency (%) of the Hybrid CW for all the measured parameters was calculated using the Eq. 1:

Where, C_influent the concentration in influent (mg/L) and C_effluent the concentration in effluent (mg/L).

Removal efficiencies were interpreted in the context of analytical detection thresholds and cycle-wise influent variability to avoid overestimation of treatment performance.

Averages and standard deviations were calculated with Microsoft Excel, 2010 and statistical testing was done using SPSS version 23.0 (IBM, NY). Shapiro-Wilk test was done to assess the normal distribution of data. Statistical significance of difference in the gene removal through wetland systems was evaluated by Kruskal-Wallis test and the gene removal under different HRTs was evaluated by Mann-Whitney U test. Spearman correlation analysis was used to evaluate the statistical association between the removal efficiency of ARGs and other parameters.

Results

Over the course of four months (September–December), biweekly sampling was used to monitor the general wastewater quality parameters (pH, electrical conductivity (E.C.), total dissolved solids (TDS), and salts) as well as the conventional wastewater control parameters (total nitrogen (TN), total phosphorus (TP), biochemical oxygen demand (BOD), and chemical oxygen demand (COD). This resulted in multiple measurements per parameter for each constructed wetland.

All values are presented as mean ± standard deviation from triplicate measurements (n = 3 per cycle) for influent and effluent samples is shown in Table 1. In order to show the inherent variability in wastewater composition over time, the ranges of observed values are also given. This variability enables assessment of the system’s performance trends and temporal stability while reflecting variations in domestic wastewater inputs.

Table 1.

Physicochemical properties of influent and effluent wastewater during a four-month observation period (September–December) with biweekly sampling in mesocosm-scale constructed wetlands (CW1–CW3).

Wastewater parameter	Mean influent concentration (± SD, range)	Mean effluent concentration			Mean removal efficiency
		CW1 (± SD, range)	CW2 (± SD, range)	CW3 (± SD, range)	CW1 (%)	CW2 (%)	CW3 (%)
pH	8.32 ± 0.8 (7.5–9.2)	7.45 ± 0.3 (7.2–7.8)	7.74 ± 0.4 (7.4–7.9)	7.56 ± 0.2 (7.2–7.8)	9.57%	6.27%	8.38%
E.C (µS/cm)	578.75 ± 10.3 (560–595)	330 ± 4.5 (325–335)	319.5 ± 5.6 (296–331)	329.58 ± 4.2 (286–347)	41.79%	44.36%	41.98%
TDS (mg/L)	357.5 ± 17.38 (340–380)	259.08 ± 22.19 (236–272)	266.83 ± 16.77 (231–289)	263.83 ± 7.74 (235–288)	27.15%	25.09%	26.01%
Salts (mg/L)	225.08 ± 16.7 (205–245)	164.25 ± 19.37 (142–183)	143.83 ± 14.5 (116–173)	135.83 ± 6.68 (112–157)	26.63%	35.94%	39.23%
TN (mg/L)	34.75 ± 5.3 (28–42)	24.17 ± 4.3 (21–26)	21.08 ± 3.8 (18–23)	19.92 ± 4.6 (17–21)	30.23%	38.77%	42.40%
TP (mg/L)	4.33 ± 1.6 (2.5–6.2)	2.17 ± 1.1 (1.6–2.3)	1.33 ± 1.3 (1.1–1.5)	1 ± 1.2 (0.8–1.2)	51.85%	71.23%	78.45%
BOD (mg/L)	377.59 ± 47.36 (320–430)	317.17 ± 39.14 (304–323)	301.57 ± 42.59 (284–330)	285.48 ± 35.6 (245–308)	15.83%	20.03%	24.21%
COD (mg/L)	791.5 ± 31.4 (755–825)	588.75 ± 32.2 (559–612)	544.17 ± 32.5 (515–572)	579.75 ± 30.57 (536–595)	25.63%	31.26%	26.72%

For instance, effluent pH reductions across CW1–CW3 ranged from 6.27% to 9.57%, demonstrating how removal efficiencies varied over subsequent treatment cycles, while influent pH averaged 8.32 ± 0.8 (range: 7.5–9.2). To illustrate time-dependent trends in reactor performance, removal efficiencies for E.C., TDS, salts, TN, TP, BOD, and COD are also shown as percentage reductions. Across the majority of parameters, CW3 consistently attained the highest removal efficiencies, demonstrating the long-term impact of substrate type on treatment performance.

Occurrence and removal of antibiotics

The removal of ciprofloxacin (CIP) and cefixime (CEF) through three constructed wetlands (CWs) with different media types showed distinctly different trends in removal efficiencies, as Fig. 2 illustrates.

Fig. 2.

Temporal fluctuations in removal efficiency and residual antibiotic concentrations throughout 12 reactor cycles in three artificial wetland systems (CW1, CW2, and CW3). Remaining cefixime and ciprofloxacin concentrations (ng/L) are displayed in the lower and upper panels, respectively. For every wetland configuration, bars show the concentrations of influent and effluent, and lines show the associated % removal efficiency. The percentage removal efficiency is shown by secondary y-axes.

For CIP, the influent concentrations varied from 990 to 3500 ng/L. The removal efficiencies of CW1 fluctuated from 26.1% to 100%, with the lower rates in the initial cycles. CW2, however, was more stable in performance, with removal efficiencies ranging from 41.7% to 100%, and 100% removal being attained in several cycles. CW3 was better than the other wetland systems and was able to 100% removal in most cycles, particularly in the last cycles. As for CEF, the influent concentrations varied from 1200 to 2900 ng/L. The detection limits of the analytical methods were 5 ng/L for CIP and 10 ng/L for CEF, and “100% removal” refers to concentrations below these detection limits.

CT detection of genes by qPCR

The antibiotic resistance genes (ARGs) cumulative trend over successive 7-day reactor cycles was characterized by a consistent increase in mean CT values, thus reflecting a progressive gene abundance reduction. Indeed, lower CT values in influent samples correspond to an earlier detection as well as a higher abundance in the wastewater, while an increase in CT values in effluent samples reflects the efficiency of gene removal.

First, for the 16 S rDNA gene, the CT values of the influent ranged from 5.32 to 12.75, and CW3 at HRT 7 that achieved the highest removal efficiency (effluent CT = 22.56 at the final cycle). Likewise, the dfr1 gene presented its influent CT values from 13.43 to 27.95, and the most significant removal was by CW3 at HRT 7 (effluent CT = 34.90). The qnrS gene had the CT values of its influent fluctuating between 13.07 and 26.07, and the highest effluent CT (35.33) was produced by CW3 at HRT 7. The Sul1 gene CT values of the influent varied from 14.36 to 26.76, and the most pronounced removal was achieved by CW3 at HRT 7 (effluent CT = 35.86).

For blaCTXM, the influent CT values ranged from 15.49 to 32.40, with CW3 at HRT 7 resulting in the highest effluent CT (38.15). The TEM gene had influent CT values between 13.54 and 30.33, with CW3 at HRT 7 again showing the greatest removal efficiency (effluent CT = 38.15). Finally, for the blaOXA gene, influent CT values varied from 13.66 to 33.29, with CW3 at HRT 7 achieving the highest effluent CT (39.13) (Fig. 3).

Fig. 3.

Mean cycle threshold (CT) values of antibiotic resistant genes in influent and effluents of constructed wetlands packed with gravel, biochar, and zeolite media operated at hydraulic retention times (HRT) of 3 and 7 days over 12 treatment cycles. Bars represent mean ± standard deviation for each cycle.

In terms of relative abundance based on influent CT values, the genes were detected in the following order from most abundant to least abundant: 16 S rDNA < blaTEM < qnrS < dfr1 < sul1 < blaCTXM < blaOXA. Removal efficiency, as indicated by effluent CT values, followed the order: blaOXA > blaCTXM > sul1 > blaTEM > qnrS > dfr1 > 16 S rDNA, with CW3 at HRT 7 consistently demonstrating the highest gene removal across all genes.

Reduction in concentrations of antibiotic resistant genes with effect of filter media

The highest removal of 16 S rDNA was observed with biochar, with the concentrations of these bacteria in the effluent being the lowest at 1.05 × 10⁵ to 4.01 × 10⁵ copies/mL. However, zeolite had the second highest removal of 9.73 × 10⁴ copies/mL at an HRT of 7 days. The least effective medium was gravel. Biochar had the highest efficiency in removing blaTEM compared to zeolite and gravel. The concentrations of blaTEM were the lowest in the biochar medium compared to zeolite and gravel. Zeolite had the highest efficiency in removing blaCTX-M and blaOXA, reducing these gene concentrations to 1.15 × 10³ and 1.85 × 10³ copies/mL, respectively. Zeolite also had the highest efficiency in removing dfr1, sul1, and qnrS; the latter being nearly undetectable.

The removal efficiency varied across different media, with biochar achieving the highest percentage reductions for 16 S rDNA (92.4%) and blaTEM (89.7%), demonstrating its effectiveness for these target genes. Zeolite showed superior removal efficiency for all other genes, including blaCTXM (94.3%), blaOXA (96.2%), dfr1 (98.5%), sul1 (95.7%), and qnrS (99.1%), indicating its strong overall performance. A statistically significant difference in concentrations of qnrS, and blaCTXM genes, was found in the CWs systems (p = 0.040, and 0.036, respectively), which means that there are significant effects on removal capabilities of these genes. Meanwhile, removals of other genes such as 16 S rDNA, dfr1, sul1, blaTEM, and blaOXA, were not significantly different between the three wetland systems (p > 0.05) (Table 2), which shows they performed almost equally well.

Table 2.

Influent and mean effluent concentrations (copies/mL) of 16 S rDNA and selected antibiotic resistance genes (ARGs) in domestic wastewater treated by constructed wetlands with different filter media.

Gene	Influent concentration (copies/mL)	Mean effluent concentration (copies/mL)			p-value
		CW1	CW2	CW3
16 S rDNA	1.50 × 109	8.25 × 108	4.75 × 108	3.90 × 108	0.053
Dfr1	2.69 × 106	7.45 × 105	5.00 × 105	1.30 × 105	0.064
qnrS	7.60 × 106	2.97 × 106	1.80 × 106	1.26 × 106	0.040
Sul1	1.99 × 106	9.12 × 105	5.75 × 105	3.25 × 105	0.114
blaCTXM	4.15 × 105	9.86 × 104	1.32 × 104	4.76 × 103	0.036
blaTEM	2.29 × 106	8.55 × 105	2.14 × 105	2.92 × 105	0.095
blaOXA	4.15 × 103	2.16 × 103	1.82 × 103	3.75 × 102	0.131

Reduction in concentrations of antibiotic resistant genes with effect of hydraulic retention time

The constructed wetland effectively reduced ARG concentrations, with greater reductions observed at a longer retention time of 7 days compared to 3 days. Effluent concentrations decreased substantially for all genes, with 16 S rDNA reducing from 1.50 × 10⁹ to 3.20 × 10⁸ copies/mL, dfr1 from 2.69 × 10⁶ to 3.50 × 10⁵ copies/mL, qnrS from 7.60 × 10⁶ to 1.28 × 10⁶ copies/mL, sul1 from 1.99 × 10⁶ to 2.95 × 10⁵ copies/mL, blaCTXM from 4.15 × 10⁵ to 1.61 × 10⁴ copies/mL, blaTEM from 2.29 × 10⁶ to 2.08 × 10⁵ copies/mL, and blaOXA from 4.15 × 10³ to 5.82 × 10² copies/mL. A statistically significant effect of hydraulic retention time was found for qnrS and sul1 genes, which had significantly lower concentrations of effluents at HRT 7 days compared to HRT 3 days (p < 0.001), suggesting that longer retention times were associated with better removal efficiencies. In contrast, removal of 16 S rDNA, dfr1, blaCTXM, blaTEM, and blaOXA genes was similar for HRT 3 days compared to HRT 7 days (p > 0.05), suggesting that HRT was not as effective for removing these genes (Table 3).

Table 3.

Influent and mean effluent concentrations (copies/mL) of 16 S rDNA and selected antibiotic resistance genes (ARGs) in domestic wastewater treated by constructed wetlands under different hydraulic retention times (HRTs).

Gene	Influent concentration (copies/mL)	Mean effluent concentration (copies/mL)		P-value
		HRT 3 days	HRT 7 days
16 S rDNA	1.50 × 109	8.07 × 108	3.20 × 108	0.143
Dfr1	2.69 × 106	5.66 × 105	3.50 × 105	0.219
qnrS	7.60 × 106	2.73 × 106	1.28 × 106	< 0.001
Sul1	1.99 × 106	9.13 × 105	2.95 × 105	< 0.001
blaCTXM	4.15 × 105	6.16 × 104	1.61 × 104	0.266
blaTEM	2.29 × 106	6.99 × 105	2.08 × 105	0.266
blaOXA	4.15 × 103	2.32 × 103	5.82 × 102	0.315

Removal ratio of ARGs by the constructed wetlands

The 16 S rDNA reduction was seen across all media, with maximum removal at 7 days of HRT. Biochar exhibited the highest mean removal ratio of 1.96, followed by zeolite of 1.53, which emphasizes the importance of choice of media and extended HRT. In subsequent cycles, biochar at HRT 7 days reached a maximum removal ratio of 3.13. For dfr1, zeolite at HRT 7 days had a higher average removal ratio (2.63), with biochar removing 1.7. Except for 16 S rDNA and blaTEM, which had higher removal in CW2, all ARGs were better removed in CW3. At HRT 7 days, the maximum removal ratios were 3.32 for biochar and over 4.0 for zeolite.

The quinolone resistance gene qnrS also followed the same pattern, with a mean removal ratio of 2.44 for zeolite at HRT 7 days and maximum removal ratios of 3.98 and 4.36 in subsequent cycles; biochar removed 3.21. The sul1 gene had moderate but significant removal, with mean removal ratios of 2.18 for zeolite and 1.67 for biochar at HRT 7 days, and maximum values over 4.0 for zeolite in subsequent cycles. Statistical difference of removal ratio between the reactors was significant with a p value of 0.02 for blaCTX-M, 0.015 for sul1, 0.013 for qnrS, and < 0.001 for blaTEM, blaOXA, dfr1 and 16SrDNA.

At HRT 7, the blaCTXM ESBL marker showed peak mean removal ratios of 2.44 in zeolite and 2.1 in biochar, with later-cycle peaks getting close to 4.0. Similar to this, at HRT 7, blaTEM achieved a mean removal ratio of 2.07 in biochar and 1.61 in zeolite, peaking at 3.83 in mid-cycles. With a mean of 2.86 and peak values above 3.0, the blaOXA gene demonstrated better elimination in zeolite at HRT 7.

Statistical difference of removal ratio between the reactors was significant with a p value of 0.02 for blaCTX-M, 0.015 for sul1, 0.013 for qnrS, and < 0.001 for blaTEM, blaOXA, dfr1 and 16SrDNA (Fig. 4).

Fig. 4.

Log₁₀ removal ratios of antibiotic resistant genes in constructed wetlands packed with gravel, biochar, and zeolite media operated at hydraulic retention times (HRT) of 3 and 7 days over sequential reactor cycles. Data points represent mean values with error bars indicating standard deviation for each cycle.

Correlation between pollutant removal and gene removal

Spearman correlation analysis was performed to study the correlations between various wastewater contaminants removal amounts and antibiotic resistant gene removal. These contaminants included nutrient pollutants (TN, TP), organic pollutants (COD, BOD), emerging organic pollutants (∑ antibiotics) and total bacterial biomass (16 S rDNA). As shown in Table 4, Strong and significant positive correlations were observed between the removal efficiency of antibiotic-resistant genes and both ∑ antibiotics and total bacterial biomass (p < 0.05), whereas no significant correlation was found between the removal efficiency of antibiotic-resistant genes and COD, BOD, TN, or TP (R = 0.025–0.375, p > 0.05), suggesting that the removal of antibiotic-resistant genes is more directly influenced by the concentration of antibiotics and bacterial biomass in the wastewater rather than other parameters in the mesocosm-scale CWs.

Table 4.

Correlations between pollutant removal amounts and antibiotic resistant genes removal in the CWs by Spearman correlation analysis.

Pollutants Removal	Correlation Coefficient (r)	p - value
16 S rDNA	0.867	< 0.001**
∑ antibioticsa	0.696	0.012*
CODb	0.329	0.297
BODc	0.175	0.587
TNd	0.025	0.940
TPe	0.375	0.230

*Means that the correlation is significant at the 0.05 level (2-tailed).

**Means that the correlation is significant at the 0.01 level (2-tailed).

^aTotal antibiotics.

^bChemical oxygen demand.

^cBiological oxygen demand.

^dTotal nitrogen.

^eTotal phosphorous.

Discussion

The results from the present study showed variable removals of the contaminants including conventional wastewater quality parameters (COD, BOD, TN, and TP), two antibiotics (cefixime and ciprofloxacin), and a suite of antibiotic resistance genes representing different mechanisms of resistance among the three types of constructed wetlands. It was found that the CW3 exhibited the highest overall removal efficiency, particularly excelling in nutrient and organic pollutant reduction. CW2 also performed well, achieving notable reductions in conductivity and organic matter, while CW1 showed the lowest efficiency across most parameters, reflecting its comparatively limited treatment capacity. Studies such as²² Akinnawo, 2023 suggest that in systems with efficient microbial activity and optimized HRT, the removal efficiency of conventional wastewater contaminants tends to plateau regardless of the media type used. This might explain why comparable removal efficiencies for nutrient and organic contaminants was observed between the wetland systems.

For organic pollutants, CW1 had varying removal efficiencies of ciprofloxacin and cefixime residues, and in contrast CW2 and CW3 achieved a steady removal efficiency with complete removal of antibiotic residues. Similarly, the highest number of ARGs were removed by CW3 compared to other wetlands and that too at higher retention time, i.e., 7 days. The experimental design incorporated independent 7-day operational cycles with fresh influent dosing, which explains the observed variability in influent concentrations while enabling robust cycle-wise performance comparison. Wetlands, however, consistently demonstrated removal, suggesting strong performance; near-full removal probably represents detection limits rather than complete eradication. Long-term media aging was not evaluated because this was a mesocosm study. Stable efficiency throughout cycles indicate no short-term depletion, even if media like zeolite and charcoal may deteriorate over time. The removal of conventional wastewater contaminants by constructed wetlands (CWs) is generally lower than that achieved by conventional treatment systems such as activated sludge²³. However, by affecting microbial communities and contaminant transformation, filter media selection has a major impact on CW performance²⁴. Because of their physical and chemical characteristics, zeolite and biochar in this study probably improved microbial-mediated removal²⁵. Both media facilitate the breakdown of organic compounds and nutrients by offering a large surface area, adsorption sites, and a variety of microbial communities²⁶.

Complete removal of antibiotics has been reported in other hybrid CW systems, consistent with our findings. For example, Sakurai et al. 2021²⁷ removed ciprofloxacin to undetectable levels using a hybrid CW for blackwater, while Ayaz et al. 2015²⁸ achieved full removal of selected antibiotics from domestic wastewater. Differences in removal efficiency between media, such as zeolite versus volcanic material, have been attributed to variations in pH and pore size, which affect adsorption and microbial activity²⁹.

Although gravel sustains microorganisms and serves as a physical filter, its low surface areas and chemical affinities limit its ability to adsorb substances³⁰. In contrast, zeolite functions as an efficient pollutant remover through interactions involving ionic forces, H-bonding, and adsorption on microporous surfaces; biochar adsorbs antibiotics by hydrophobic forces, H-bonding, and π-π stacking³¹. It’s also possible that the zeolite’s catalytically active hydroxyl groups increased the elimination of antibiotics and ARG³².

The hydraulic retention time (HRT) also had a positive influence on ARG removal, as reflected by the lower gene abundance in the effluent at a retention time of 7 days compared to 3 days. However, qnrS and blaCTXM were significantly influenced by the type of medium used, whereas qnrS and sul1 were significantly influenced by retention time. Earlier studies have also reflected similar trends in ARG removal with an increase in HRT³³. Monsalves et al., 2022³⁴ also found that up to 36% higher removal efficiencies of ARGs (including qnr and tet genes) were observed when HRT was raised from approximately 10 h to 29 h. Although the total bacterial abundance was still detected, zeolite helped in the adsorption of extracellular DNA, thereby controlling the dissemination of resistant genes. Although the total bacterial gene markers, such as 16 S rRNA genes, and ARGs can be reduced in the CW systems, the extent of the reduction depends on the systems themselves. For instance, the removal of ARGs in the resulting waters in full-scale CWs ranged from 0.8 to 1.5 log units (Sabri et al., 2021). This means that the extent of the removal of ARGs, similar to the extent of the decline of the 16 S rDNA observed in the current study, may range from the observed efficiencies in the constructed wetland systems.

Although adsorption, biodegradation, and filtration are the processes believed to take part in the elimination of contaminants in constructed wetland systems, the results obtained in the current study allow for the interpretation of the processes at the substrate level^35,36. Considering the high affinity of fluoroquinolones for aluminosilicates, the high removal efficiency of the zeolite system implies that the ion exchange retention mechanisms may have been the predominant mechanisms in the elimination of antibiotics. This physicochemical immobilization process might not act as a sink but might actually extend the residence time for contaminants within these bioactive areas, thereby aiding subsequent biodegradation processes. The biochar system, on the contrary, displayed comparable antibiotic removal efficiency but with a lower rate of ARG clearance, which might suggest a shift towards biofilm-associated transformation processes. It is well documented that high surface roughness and redox groups on the surface of biochar facilitate high densities of microbial biofilm, which might facilitate degradation processes while also providing a conducive environment for ARG survival at low levels. The gravel system, which does not display strong properties for either sorption or catalytic activity, most likely displayed transient microbial processes and hydraulic filtration. These mechanistic processes might explain the variability observed for removal efficiency for various parameters.

It is important to note that the processes observed for ARG dynamics within this study might not be explained by a reduction in overall abundance. Longer durations of exposure to the substrates might lead to a reduction in opportunities for HGT due to a reduction in antibiotic selection pressure and increased competition for microbes within the biofilm. The change in the structure of the microbial ecology within the wetland is also indicated by the observation that the reduction of ARG is more apparent with longer hydraulic retention times. Additionally, while the overall bacterial population is still apparent, it has also been shown that substrates such as mineral zeolite enhance the adsorption of extracellular DNA, which might limit HGT.

The removal of ARG varied depending on the substrates used and the retention time due to gene-specific properties and ecological processes. For example, plasmid-encoded genes such as qnrS might be removed more rapidly with less selection pressure, while chromosomal-encoded genes might be removed more slowly. The removal of ARG is partially correlated to the removal of 16 S rDNA; however, it is not as simple as cell death for bacteria. Substrate effects might also include processes such as adsorption, competition, and redox processes on HGT; however, correlation does not necessarily mean causation. ARG removal is a function of a physical process of retention, microbial ecology, and selective pressure, with greater retention times and substrate properties conducive to adsorption favoring removal. Mechanistic validation would help to define gene-specific pathways. Deeper mechanistic validation (e.g., metagenomic analysis or mobile genetic element analysis) would be required to define gene-specific removal pathways and ecological resilience.

Overall, our results emphasize the importance of selecting optimal media and HRT to maximize contaminant and ARG removal. Zeolite consistently outperformed other media, highlighting the value of its adsorptive and catalytic properties, while biochar’s surface area and microbial support also contributed substantially. These findings suggest that integrating optimal media type with extended HRT can effectively mitigate environmental dissemination of antibiotic resistance in aquatic ecosystems. Further mechanistic studies, including metagenomic and mobile genetic element analysis, are needed to define gene-specific removal pathways.

Conclusion

The findings of this study demonstrated comparable removal efficiencies for two target antibiotics (ciprofloxacin and cefixime) and multiple ARGs (each representing a different major class of antibiotic), and other physicochemical contaminants in domestic wastewater using a hybrid constructed wetland (HCW) system with gravel, biochar, and zeolite as filter media. Among the media tested, zeolite-based media exhibited the highest removal rates, likely due to its microporous structure and high cation exchange capacity, enhancing sorption and degradation processes. Higher retention period in the wetland treatment system also enhanced the removal of the ARGs. The HCW achieved comparable or even superior removal efficiencies for antibiotics and ARGs compared to conventional wastewater treatment plants. These results highlight the potential of hybrid constructed wetlands as an effective and sustainable technology for mitigating antibiotic contaminants and resistance genes in domestic wastewater. The observed reduction of these pollutants was primarily attributed to a combination of sorption and biodegradation mechanisms. However, further investigations are required to explore the long-term stability, scalability, and detailed pathways of antibiotic and ARG removal across diverse wetland configurations and operational conditions.

Abbreviations

ARGs

Antibiotic resistance genes

CWs

Constructed wetlands

HRT

Hydraulic retention time

qnrS

Quinolone-resistant gene

dfr1

Trimethoprim-resistant gene

sul1

Sulfonamide-resistant gene

bla

Beta-lactamase resistant genes

Author contributions

Conceptualization, original draft writing, reviewing, and editing: AFA, TUR, AK, DF, AH. Formal analysis, investigations, reviewing, and editing: MA, AH, QA, FW, editing and funding resources by TB, MA, EMA, TMD, data validation, data curation, and supervision: SA, MA.

Funding

This study received funding from the Ongoing Research Funding Program (ORF-2026-197), King Saud University, Riyadh, Saudi Arabia.

Data availability

All data generated or analyzed during this study are included in this published article.

Declarations

Competing interests

The authors declare no competing interests.