Effective removal of iron, nutrients, micropollutants, and faecal bacteria in constructed wetlands cotreating mine water and sewage treatment plant effluent

Like many other countries, England and Wales face substantial challenges in restoring their streams, rivers, and lakes to good status in line with the objectives set in The Water Environment (Water Framework Directive (WFD)) (England and Wales) Regulations 2017. Currently, only 14% of the rivers in England have good ecological status (EAC 2022). To improve this condition, the Department for Environment, Food, and Rural Affairs (Defra) has set ambitious new targets under the Environment Act 2021 for water quality by 2038 that include halving the length of rivers polluted by harmful metals from abandoned mines and reducing phosphorus loadings from treated wastewater by 80% (Defra 2022b).

There is a perceived shortage of sustainable treatment technologies for reliable phosphorus removal at smaller scales that demand minimal operating and maintenance expertise and are suited to northern latitudes (Bunce et al. 2018). However, an intriguing win–win opportunity exists in the cotreatment of mine water with secondary-treated sewage treatment plant (STP) effluent in constructed wetlands (CWs), as synergies occur if phosphate from sewage is effectively removed via precipitation with iron from mine water to form ferric phosphate in the CW sediments (Johnson & Younger 2006; Younger & Henderson 2014; Wang et al. 2021). Johnson & Younger (2006) found that a pilot scale CW (25 × 25 × 0.45 m, L × W × H) cotreating mine water and STP in the UK removed ammonium-nitrogen and iron levels to below the proposed discharge limits, and the rate of phosphate removal was proportional to influent iron concentrations, suggesting that the presence of iron flocs in the mine water promotes efficient phosphate removal by adsorption and precipitation. Similarly, Wang et al. (2021) also found that a CW microcosm (0.89 × 0.2 × 0.2 m, L × W × H) cotreating acid mine drainage and domestic wastewater in China effectively removed phosphate and iron through precipitation. The pilot trial by Johnson and Younger informed the construction of the world's first full-scale free-water surface flow CW system for cotreatment of mine water and STP effluent, which has been operating since 2005 in the Lamesley area of Northeast England, to improve the water quality of the receiving River Team (Welsh 2005; CoalAuthority 2018).

CWs are nature-based, also known as passive treatment, systems with low operational/maintenance requirements (Wang et al. 2021). They tackle pollution by taking advantage of natural and freely available resources such as sunlight, plants, and microbes (Wang et al. 2021). Younger & Henderson (2014) investigated the performance of the Lamesley CWs a decade ago in terms of the removal of biological oxygen demand, ammoniacal nitrogen, suspended solids, phosphate, and iron (Fe). We were interested in understanding the performance of this CW system more comprehensively across the wider range of relevant water quality metrics that drive impacts on receiving rivers, including emerging issues like micropollutants and microorganisms. The public and political debate about the status of the rivers in England and Wales is increasingly concerned with their suitability for recreation and wildlife as citizen and governmental initiatives seek to protect rivers for bathing and fishing (WaterUK 2021; EAC 2022). ‘Rivers fit to swim in’ are nowadays a stated ambition of the Environmental Audit Committee (EAC 2022), government regulators, and authorities (DOHSC 2022). This means an increased focus on micropollutants and pathogens in rivers that may cause harm to wildlife and water users.

We, therefore, aimed to comprehensively assess how well combined treatment of abandoned coal mine water and secondary-treated wastewater in a CW protect the river environment from chemical and microbiological pollution. We hypothesized that the CW would (i) improve influent quality to meet all the treated effluent compliance limits, (ii) reduce chemical oxygen demand (COD), nutrients, heavy metals, faecal bacteria, and micropollutants beyond the mere dilution effect of blending secondary-treated wastewater with mine water, and consistently achieve the compliance limits under changeable weather conditions, (iii) create effluent that has no detrimental impact on the chemical, ecological, and bathing water status of the receiving river, and (iv) improve influent microbiome characteristics to produce treated effluent microbiomes resembling those of the receiving river.

We collected grab samples from seven locations around the CW area (Figure 1) comprising STP influent, STP effluent, mine water effluent, CW influent, CW effluent, and river upstream and downstream of the CW discharge. We investigated the performance of the five-cell wetland, because its outfall is located upstream of that of the four-cell wetland. We conducted the sampling in March, May, July, and August 2021 covering the spring period (March and May) and summer period (July and August) for every sample except the STP influent that was only obtained in May, July, and August. The STP influent characteristics have already been reported in a previous publication (Zan et al. 2023). Weather (rainfall) conditions for the sampling events are summarized in Table S2, in the SI. A scoping study with a reduced sampling schedule had been conducted in winter, January 2020 (before the outbreak of the COVID-19 pandemic and related laboratory lockdowns).

We analysed the water samples for temperature, pH, electrical conductivity, total dissolved solids (TDS), salinity, and dissolved oxygen (DO) in-situ using a precalibrated, handheld probe (Extech Instruments, Nashua, NH, USA) and a HQ40D Digital two-channel multimeter (HACH, Manchester, UK). We measured alkalinity using a digital titrator (HACH, Manchester, UK). COD, ammonium-nitrogen (⁠-N), nitrate-nitrogen (⁠-N), nitrite-nitrogen (⁠-N), total nitrogen (TN), phosphate-phosphorus (⁠-P), total phosphorus (TP), and fluoride (F⁻) were determined using HACH cuvette test kits LCI400, LCK339, LCK341, LCK238, LCK349, LCK349, and LCK323, respectively. Cuvette tests were performed following the manufacturer's instructions and evaluated in a HACH DR6000 Ultraviolet and Visible Spectrum Spectrophotometer. For quality assurance, we verified the cuvette tests with a blank solution (deionized (DI) water) and known concentration standards prepared from the respective nutrient salts to ensure that the result from the cuvette tests agreed with the standard concentration by ±5%. We also filtered the water samples through a cellulose acetate syringe filter (0.45 μm, 25 mm; VWR International, UK) to measure for anions using a Dionex High Pressure Ion Chromatography instrument (ThermoFisher, UK) and for dissolved organic carbon using a carbon analyser (Vario TOC cube, Elementar Analysen Systeme GmbH, Germany). Additionally, filtered water samples were acidified with 1% v/v concentrated nitric acid and analysed for metals using a Varian Vista-MPX Inductively Coupled Plasma-Optical Emission Spectrometer or Agilent Inductively Coupled Plasma Mass Spectrometry 7700 Series instrument, as appropriate for the metal concentration (Mayes et al. 2021). Certified 1,000 ppm standards (accuracy of ≤ ± 1.0%; VWR Chemicals, VWR International, UK) were diluted using 1% nitric acid solution for preparing calibration standards. Blanks and standards were run every seven samples to check analytical accuracy and precision.

We performed micropollutant analysis according to EPA method 1694 via Ultra Performance Liquid Chromatography-Mass Spectrometry (UPLC-MS/MS) (Waters, Elstree, UK). The instrument consists of a Waters Acquity UPLC™ system (Waters Corp., Milford, MA, USA) coupled to a Xevo TQ-S™ triple quadrupole mass spectrometer (Waters Micromass, Manchester, UK), equipped with electrospray ionization (ESI) interface (Waters, Watford, UK). The mass spectrometer was operated in both ESI modes using multiple reaction monitoring. Triplicate CW influent and effluent samples were analysed for each sampling event. We first submitted subsamples to the University of Bath for non-target screening via their liquid chromatography–high resolution mass spectrometry (LC-HRMS) system (Agilent, CA, USA). The LC-HRMS analyses were performed using an Agilent QTOF 6545 with Jetstream ESI spray source coupled to an Agilent 1260 Infinity II Quat pump HPLC with a 1260 autosampler, column oven compartment, and variable wavelength detector. We selected the most notable compounds for subsequent, quantitative UPLC-MS/MS analysis at Newcastle University. We quantified eight compounds comprising acetaminophen, diethyltoluamide (DEET), caffeine, carbamazepine, sulfapyridine, venlafaxine, sulpiride, and cetirizine. The detailed methodology including column details and instrument operating condition for both LC-HRMS and UPLC-MS/MS is provided in SI, Section 1.2 with Tables S3–S4.

We analysed faecal coliforms by membrane filtration according to Method 8074 (HACH, Manchester, UK). We also performed molecular microbiology methods using a combination of MinION nanopore sequencing of 16S rRNA gene amplicons and quantitative polymerase chain reaction (qPCR) methods, as previously described (Acharya et al. 2020a; Zan et al. 2022). We processed sequencing data using Matlab^© (Version R2019b, Mathworks, Portola Valley, CA, USA) for multivariate data analysis (cluster, principal component analysis (PCA), and analysis of similarities (ANOSIM)). We downloaded the taxonomic classification and quality of barcoded reads from the EPI2ME dashboard as a CSV file that contained information on run and read IDs and read accuracy, barcodes, and NCBI taxa IDs for classified reads. Then, we processed the CSV file with Matlab^© scripts published elsewhere (Zan et al. 2023). The detailed methodology including instrument details is provided in SI, Section 1.3 with Tables S5–S7 and Figure S4.

Data screening found suspected contamination of the mine water sampling tap by ingress of STP effluent from the blending tank in May and August, as explained in more detail in SI with Figures S5–S6. We, therefore, excluded these two outlier samples (mine water samples in May and August) from the final analysis. This exclusion did not affect the CW performance evaluation, as CW influent and effluent samples were available for all sampling events.

We used MS Excel© and two-tailed t-tests to evaluate the null hypothesis that there is no difference between the mean values of two sample groupings of interest, or one-tailed z-tests to investigate if the mean value of a parameter meets the desired standard. We used Primer7 software (primer-e, Auckland, New Zealand) to investigate the linkage between environmental parameters and microbial communities using the BEST (Bio-Env) procedure as described by Clarke et al. (2014).

Table 1 summarizes the conventional water quality parameters measured at different treatment stages as mean values and standard deviations of all the sampling events, while corresponding data for each event are provided in SI with Tables S8–S11.

Table 1

Conventional water quality parameters of the STP influent and effluent, mine water, and CW influent and effluent relative to the UK wastewater treatment work's compliance limits

.	STP influent .	STP effluent .	Mine water .	CW influent .	CW effluent .	Compliance limit .
-N (mg/L)	47.78 ± 27.06	3.07 ± 1.90	0.21 ± 0.26	0.91 ± 0.50	0.13 ± 0.16 (86.1%)	<40 and 3.5a,b
-N (mg/L)	0.19 ± 0.15	0.98 ± 1.43	0.004 ± 0.00	0.17 ± 0.18	0.06 ± 0.06 (62.5%)	N/A
-N (mg/L)	1.95 ± 1.38	24.97 ± 9.07	0.29 ± 0.24	6.81 ± 2.28	5.14 ± 2.30 (24.6%)	N/A
TN (mg/L)	80.73 ± 19.11	35.75 ± 9.26	1.40 ± 0.40	9.42 ± 4.1	6.43 ± 2.73 (31.6%)	N/A
-P (mg/L)	5.87 ± 2.71	2.70 ± 0.93	0.25 ± 0.35	0.86 ± 0.14	0.23 ± 0.07 (73.1%)	N/A
TP (mg/L)	6.89 ± 2.96	2.86 ± 0.96	0.23 ± 0.32	1.00 ± 0.20	0.26 ± 0.08 (74.2%)	N/A
Fluoride (mg/L)	1.22 ± 0.47	0.64 ± 0.18	0.60 ± 0.10	0.59 ± 0.15	0.59 ± 0.15	N/A
Alkalinity (mg/L CaCO3)	273.00 ± 47.84	51.00 ± 19.24	545.00 ± 7.07	408.25 ± 71.51	436.50 ± 84.08	N/A
Salinity (mg/L)	530.00 ± 29.82	396.75 ± 61.67	1,525 ± 35.36	1,205.00 ± 283.37	1,320.00 ± 154.92	N/A
pH	8.41 ± 0.13	7.20 ± 0.31	7.25 ± 0.23	7.12 ± 0.14	7.54 ± 0.19	N/A
Conductivity (μS/cm)	1,072.67 ± 64.27	804.00 ± 105.87	2,985 ± 49.5	2,335.00 ± 523.10	2,595.00 ± 238.12	N/A
TDS (mg/L)	749.00 ± 44.19	578.75 ± 84.55	2,100.00 ± 28.28	1,656.25 ± 374.37	1,815.00 ± 154.16	N/A
DO (% saturation)	22.10 ± 11.21	52.75 ± 15.81	62.79 ± 11.03	61.31 ± 8.30	62.17 ± 9.91	N/A
COD (mg/L)	636.83 ± 362.25	65.89 ± 23.69	11.88 ± 0.38	22.41 ± 4.71	16.38 ± 4.33 (26.0%)	<125c
DOC (mg/L)	22.77 ± 2.98	15.61 ± 2.01	3.44 ± 2.51	7.27 ± 3.12	6.72 ± 2.00	N/A
Temperature °C	17.13 ± 2.81	14.03 ± 5.95	11.60 ± 9.33	13.40 ± 5.99	13.73 ± 5.23	N/A
Faecal coliform (log10CFU/100 mL)	6.49 ± 0.35	4.98 ± 0.37	0 CFU/100 mL	3.87 ± 0.46	3.28 ± 0.27 (74.8%)	N/A
Heavy metals (μg/L)d
Fe	40.00 ± 10.00	45.00 ± 17.32	2,057.5 ± 689.43	880.00 ± 635.94	18.75 ± 8.54 (97.9%)	<2,000a,e,f
Mn	26.67 ± 15.28	43.75 ± 11.09	1,137.50 ± 38.89	756.25 ± 95.95	516.25 ± 296.18 (31.7%)	N/A
Pb	0.43 ± 0.15	0.00 ± 0.00	0.05 ± 0.07	0.07 ± 0.08	0.04 ± 0.04 (45.4%)	N/A
Zn	35.00 ± 8.66	57.50 ± 9.57	5.00 ± 7.07	27.50 ± 9.57	27.50 ± 28.72 (0.0%)	N/A
Cu	18.80 ± 12.65	10.00 ± 11.55	0.10 ± 0.15	1.66 ± 1.29	3.71 ± 1.59 (−)	N/A
As	0.88 ± 0.19	0.75 ± 0.29	0.21 ± 0.29	0.32 ± 0.23	0.17 ± 0.11 (48.4%)	N/A

.	STP influent .	STP effluent .	Mine water .	CW influent .	CW effluent .	Compliance limit .
-N (mg/L)	47.78 ± 27.06	3.07 ± 1.90	0.21 ± 0.26	0.91 ± 0.50	0.13 ± 0.16 (86.1%)	<40 and 3.5a,b
-N (mg/L)	0.19 ± 0.15	0.98 ± 1.43	0.004 ± 0.00	0.17 ± 0.18	0.06 ± 0.06 (62.5%)	N/A
-N (mg/L)	1.95 ± 1.38	24.97 ± 9.07	0.29 ± 0.24	6.81 ± 2.28	5.14 ± 2.30 (24.6%)	N/A
TN (mg/L)	80.73 ± 19.11	35.75 ± 9.26	1.40 ± 0.40	9.42 ± 4.1	6.43 ± 2.73 (31.6%)	N/A
-P (mg/L)	5.87 ± 2.71	2.70 ± 0.93	0.25 ± 0.35	0.86 ± 0.14	0.23 ± 0.07 (73.1%)	N/A
TP (mg/L)	6.89 ± 2.96	2.86 ± 0.96	0.23 ± 0.32	1.00 ± 0.20	0.26 ± 0.08 (74.2%)	N/A
Fluoride (mg/L)	1.22 ± 0.47	0.64 ± 0.18	0.60 ± 0.10	0.59 ± 0.15	0.59 ± 0.15	N/A
Alkalinity (mg/L CaCO3)	273.00 ± 47.84	51.00 ± 19.24	545.00 ± 7.07	408.25 ± 71.51	436.50 ± 84.08	N/A
Salinity (mg/L)	530.00 ± 29.82	396.75 ± 61.67	1,525 ± 35.36	1,205.00 ± 283.37	1,320.00 ± 154.92	N/A
pH	8.41 ± 0.13	7.20 ± 0.31	7.25 ± 0.23	7.12 ± 0.14	7.54 ± 0.19	N/A
Conductivity (μS/cm)	1,072.67 ± 64.27	804.00 ± 105.87	2,985 ± 49.5	2,335.00 ± 523.10	2,595.00 ± 238.12	N/A
TDS (mg/L)	749.00 ± 44.19	578.75 ± 84.55	2,100.00 ± 28.28	1,656.25 ± 374.37	1,815.00 ± 154.16	N/A
DO (% saturation)	22.10 ± 11.21	52.75 ± 15.81	62.79 ± 11.03	61.31 ± 8.30	62.17 ± 9.91	N/A
COD (mg/L)	636.83 ± 362.25	65.89 ± 23.69	11.88 ± 0.38	22.41 ± 4.71	16.38 ± 4.33 (26.0%)	<125c
DOC (mg/L)	22.77 ± 2.98	15.61 ± 2.01	3.44 ± 2.51	7.27 ± 3.12	6.72 ± 2.00	N/A
Temperature °C	17.13 ± 2.81	14.03 ± 5.95	11.60 ± 9.33	13.40 ± 5.99	13.73 ± 5.23	N/A
Faecal coliform (log10CFU/100 mL)	6.49 ± 0.35	4.98 ± 0.37	0 CFU/100 mL	3.87 ± 0.46	3.28 ± 0.27 (74.8%)	N/A
Heavy metals (μg/L)d
Fe	40.00 ± 10.00	45.00 ± 17.32	2,057.5 ± 689.43	880.00 ± 635.94	18.75 ± 8.54 (97.9%)	<2,000a,e,f
Mn	26.67 ± 15.28	43.75 ± 11.09	1,137.50 ± 38.89	756.25 ± 95.95	516.25 ± 296.18 (31.7%)	N/A
Pb	0.43 ± 0.15	0.00 ± 0.00	0.05 ± 0.07	0.07 ± 0.08	0.04 ± 0.04 (45.4%)	N/A
Zn	35.00 ± 8.66	57.50 ± 9.57	5.00 ± 7.07	27.50 ± 9.57	27.50 ± 28.72 (0.0%)	N/A
Cu	18.80 ± 12.65	10.00 ± 11.55	0.10 ± 0.15	1.66 ± 1.29	3.71 ± 1.59 (−)	N/A
As	0.88 ± 0.19	0.75 ± 0.29	0.21 ± 0.29	0.32 ± 0.23	0.17 ± 0.11 (48.4%)	N/A

Notes: Results were reported to two decimal places as Mean ± SD. STP effluent and CW influent and effluent were sampled in March, May, July, and August, the STP influent, in May, July, and August, and the mine water, in March and July. The numbers in parentheses represent the average percent removal of nutrients, COD, heavy metals, and faecal coliforms in the CWs.

^aThe Water Resources Act (1991): Consent to Discharge from the Environment Agengy (consent number 235/1891): site-specific consent for Birtley sewage treatment work and Lamesley CWs.

^bThe limit for STP effluent from (i) and CW effluent from (ii), respectively.

^cThe Environment Agency's compliance limits for treated wastewater discharge.

^dDissolved metal concentration (μg/L).

^eThe limit for CW effluent.

^fFor total iron (Fe) (μg/L).

View Large

The untreated sewage (STP influent) had a pH of 8 with high -N, COD, and faecal coliform levels. After primary/secondary settling and trickling filter treatment, the secondary-treated sewage was of neutral pH, and the COD and -N concentrations were below the compliance limit for the STP effluent of 125 and 40 mg/L, respectively (z-test, p-values < 0.01, Table 1). The mine water was not acidic (pH of 7) and had substantial alkalinity levels, presumably because of a contact with sandstone in overlaying geological strata, which helps establish a stable pH level, in line with previous studies reporting a pH of 7.1 with 755 mg/L CaCO₃ alkalinity for discharge from the Kibblesworth mine (Banks et al. 1997; Younger & Henderson 2014). Heavy metal levels in the mine water were higher for Fe and Mn, but lower for Zn, Cu, and As, as compared with the STP influent and effluent. We measured metals as dissolved concentration to avoid damaging analytical instruments, while the permissible limit for Fe was set for the total Fe concentration. But even the dissolved Fe in the mine water exceeded the permissible limit for total Fe substantially (z-test, p-value > 0.05, Table 1). Similarly, the Fe level previously reported for this mine water was also high (6,000 μg/L, for total Fe) (Younger & Henderson 2014). Further treatment is thus required to reduce the Fe level in the mine water.

The CW influent showed similar alkalinity, salinity, conductivity, and TDS characteristics to the mine water, and the pH remained at 7. High levels of Fe and Mn were also maintained. Since mine water had higher Fe, Mn, conductivity, TDS, salinity, and alkalinity than the STP effluent and contributed four of five parts of the blended water, only slight changes to these parameters were expected. Meanwhile, the STP effluent had much higher nutrient and faecal coliform, Cu, and Zn levels than the mine water. After the blending, an 80% reduction of these metrics will result simply from the dilution of the STP effluent with the mine water. All of this can be seen in Table 1 by comparing STP effluent, mine water, and CW influent characteristics.

Dilution by blending is no solution for pollution and the actual treatment achieved by the CW is revealed by comparing the effluent and influent characteristics. From Table 1, the COD level in the CW effluent was 26% lower than in the CW influent and complied with the compliance limit (125 mg/L). The -N concentration was significantly reduced by 86.1% from 0.91 ± 0.50 mg/L in the influent to 0.13 ± 0.16 mg/L in the effluent (t-test, p-value <0.05), which is well below the compliance limit specified for Lamesley CW discharge (3.5 mg/L) (z-test, p-value < 0.01). Nitrification of ammonium coupled with denitrification can remove TN in CWs (Vymazal 2007; Younger & Henderson 2014; Wang et al. 2021). Accordingly, mean values of all measured forms of nitrogen (⁠-N, -N, -N, and TN) were reduced in the CW effluent, relative to the CW influent. There was also a significantly lower TP concentration in the CW effluent than the influent (t-test, p-value < 0.01).

For the heavy metals, the dissolved Fe concentration was reduced by 97.9% from 880.00 ± 635.94 μg/L in the CW influent to 18.75 ± 8.54 μg/L in the effluent, which is more than a factor of 100 below the limit for the total Fe at 2,000 μg/L (z-test, p-value < 0.01). The simultaneous TP and Fe removal is the result of the sorption of phosphate by ferric hydroxide and precipitation as ferric phosphate (Dobbie et al. 2009; Younger & Henderson 2014). This CW is of neutral pH, and with the influent aeration in cascades, the Fe removal through oxidation and precipitation is enhanced (Wang et al. 2021). Ferrous iron (Fe²⁺) is rapidly converted to ferric iron (Fe³⁺) in oxidizing conditions and ferric iron can form ferric hydroxide at a pH of more than 3.5 (Jarvis et al. 2012). Moreover, with the mixture of wastewater and mine water, suspended solids in the wastewater provide nuclei and counterions (phosphate) for the generation of iron flocs, hence accelerating the precipitation of ferric hydroxide (ochre) and ferric phosphate (Johnson & Younger 2006). Certain microbes can also contribute towards the formation of such phosphate minerals by transforming organic phosphorus into phosphate (Gadd 2010). In this surface flow CW, the resulting sediment deposits did not cause any blockages, but after 15 years, the accumulated sediment needed to be removed to create free space for further treatment (NorthumbrianWater 2022).

In contrast to Fe, there was no significant difference in the Mn levels of the CW effluent and influent (t-test, p-value > 0.05). The persistent level of Mn is likely because the hydroxide solubility product of Mn is higher than for Fe, and very high pH (≈10) is required to immobilize Mn as hydroxide to get adsorbed onto the soil/sediment or get removed as Mn hydroxide precipitates (Jarvis et al. 2012). Furthermore, Mn removal is inhibited by high Fe concentrations in water (Neculita & Rosa 2019). Pb, Zn, Cu, and As were also inefficiently removed in the CWs, which could again be ascribed to complex geochemical factors. For example, Zn requires a pH of 8.2 for effective removal as its hydroxide (Jarvis et al. 2012). Cu concentrations were highly variable in both CW influents and CW effluents presumably due to the re-dissolution of the Cu precipitates and because they were also being released back by plants, but without a statistically significant difference in the mean values. These observations are important amendments to previous work, which only monitored Fe to assess heavy metal removal (Younger & Henderson 2014).

Tables S13–S14 in SI show how the concentrations and removal efficiencies (%) of the key pollutants by the CW varied between the four sampling events in March, May, July, and August. From the rainfall data in the 24 h before the sampling, May stood out as the sampling event with the wettest conditions and the highest STP effluent loading (Table S2 in SI). This wet weather seemed to affect nutrient removal. There was only 37% removal of -N in May, versus >90% removal in the other three sampling events. A slightly negative removal of -N also occurred in May, versus the positive removal in the other three sampling events. The performance of biological nitrogen removal will be influenced by rainfall and implications for HRT, as well as temperature (García et al. 2003; Shingare et al. 2019). Regarding temperature, winter sampling in our scoping study showed efficient removal of nutrients, faecal coliforms, and Fe in January (Table S15 in SI). A surprising lack of seasonal variation in biological nutrient removal by this CW was previously noted by Younger & Henderson (2014), who attributed it to the steady temperature of mine water all year round. The buffering of temperature variation is thus another benefit of cotreating mine water with sewage in CWs.

Faecal coliforms were reduced by 74.8% in the CW (Table 1), although without statistical significance owing to the high variability of concentrations. Additionally, in July we analysed for the abundance of Escherichia coli-producing extended-spectrum β-lactamases, which gives them resistance to commonly used antibiotics, including penicillins and cephalosporins (Rawat & Nair 2010) (Table S16 in SI). We found a substantially lower abundance of them in the CW effluent as compared with the influent (t-test, p-value < 0.01), indicating that the CW could remove the antibiotic-resistant E. coli.

There was >82% removal of the absolute abundance of genes from both Aliarcobacter and the faecal indicator bacteria by the CW in all sampling months. For Aliarcobacter (Figure 3(a)), the genus showed higher abundance in July and August relative to March and May. Aliarcobacter is normally found in untreated and treated sewage and the genus includes several pathogenic species (Chieffi et al. 2020). Some Aliarcobacter species remain viable in the water environment as they are aerotolerant (do not require oxygen for growth but can tolerate its presence) and can survive in low water temperatures (Chieffi et al. 2020). For genes from faecal indicator bacteria (Figure 3(b)), the genus Prevotella predominated in the 16S rRNA gene amplicon libraries of the CW influent samples in March, May, and July. Prevotella is also reported to be highly abundant in sewage (Fisher et al. 2015).

Several mechanisms could be responsible for the removal of these bacteria in the CW including sedimentation, filtration, adsorption, oxidation, solar disinfection, root exudation of biocides, and predation by protozoa (Wu et al. 2016). P. australis, the plant established in the CW, produces bactericidal substances such as phenolic compounds and secondary metabolites, e.g., tannins, terpenoids, alkaloids, and flavonoids that kill pathogenic/faecal indicator bacteria (Cowan 1999; Shingare et al. 2019; Dan et al. 2020). Pathogen removal in CWs can vary with seasons and weather conditions, being higher in the summer than the colder periods, which could be attributed to the increased temperature and UV radiation (Wu et al. 2016; Shingare et al. 2019). Moreover, CW showed higher pathogen removal in the dry/warm weather than in the wet weather (Alufasi et al. 2017; Shingare et al. 2019). Similarly, this study found higher removal in the summer (July and August) than in spring (March and May), and the lowest removal of genetic markers attributed to faecal bacteria was found in May (Figure 3(c) and 3(d)) when the weather was at the wettest. The high removal of rodA in August was different from the low removal of faecal coliform obtained from culturing in August (Table S13 in SI). But such low removal of faecal coliforms could be artefactual because faecal coliform counts in August were unusually low in the CW influent. Bacterial abundance estimates typically differ between culturing and qPCR methods and are generally higher by qPCR. This is because the culturing methods demonstrate the viability of the cells, although not all bacteria can be cultured, while in genomic methods the targeted genes could be from both viable and damaged cells or extracellular DNA (Figueroa-González & Pérez-Plasencia 2017; Acharya et al. 2020b). The rodA and HF183 results aligned with a previous study by Bunce et al. (2020) who found a similarly high abundance of these two genes (10⁶–10⁷ gene copies/100 mL) in small STP influents in the UK. In that study, the small STPs showed mean removal of around 98 and 95% for rodA and HF183, respectively. In summary, the CW could efficiently remove pathogens and faecal bacteria across all sampling events, but with reduced efficiency due to rainfall in May.

Table 2 shows the water quality of the River Team upstream and downstream of the CW discharge in relation to the WFD standards for rivers. The combined mine water and STP effluents contribute substantially to the flow in the River Team in dry weather conditions ((Welsh 2005) and Table S1 in SI), meaning that there is limited dilution of the discharge. From Table 2, there was no significant difference between -N concentration in the river upstream and downstream of the discharge (t-test, p-value > 0.05). The -N concentration was indicative of moderate status for the upstream, and of good status for the downstream, of the discharge meaning that the discharge may even have improved the river water quality in terms of ammonium concentrations.

Table 2

Conventional water quality parameters of the river upstream and downstream of the CW discharge relative to the Water Framework Directive's standards

.	River upstream .	River downstream .	Standarda .
-N (mg/L) 90th percentile	0.65 ± 0.24 0.96	0.33 ± 0.12 0.48	Quality 0.3 = high, 0.6 = good, 1.1 = moderate, 2.5 = poor
-N (mg/L)	0.17 ± 0.09	0.10 ± 0.03	N/A
-N (mg/L)	6.69 ± 1.96	5.52 ± 1.09	N/A
TN (mg/L)	9.09 ± 2.92	7.20 ± 1.70	N/A
-P (mg/L)	0.46 ± 0.19	0.31 ± 0.14	Qualityb 0.04 = high, 0.08 = good 0.19 = moderate, 1.03 = poor
TP (mg/L)	0.50 ± 0.19	0.33 ± 0.15	N/A
Fluoride (mg/L)	0.39 ± 0.15	0.53 ± 0.20	N/A
Alkalinity (mg/L CaCO3)	130.75 ± 34.64	312.00 ± 77.49	N/A
Salinity (mg/L)	414.00 ± 56.26	987.25 ± 168.81	N/A
pH 5th and 95th percentile	8.02 ± 0.30 7.54–8.02	7.64 ± 0.21 7.30–7.64	6–9
Conductivity (μS/cm)	852.25 ± 111.36	1,960.50 ± 305.28	N/A
TDS (mg/L)	595.25 ± 88.94	1,371.75 ± 206.80	N/A
DO (% saturation) 10th percentile	83.15 ± 6.08 75.36	72.59 ± 5.20 65.93	Quality 70 = high, 60 = good
			54 = moderate, 45 = poor
COD (mg/L)	25.53 ± 9.45	20.31 ± 3.24	N/A
DOC (mg/L)	9.24 ± 2.25	7.22 ± 1.09	N/A
Temperature °C 98th percentile	11.98 ± 5.02 22.30	12.15 ± 4.81 22.00	Quality 25 = high, 28 = good
			30 = moderate, 32 = poor
Faecal coliform (log10CFU/100 mL) 90th percentilec 95th percentilec	3.38 ± 0.48 10147.10 (8554.00)d 15265.47 (12868.79)d	3.31 ± 0.39 6508.83 (5486.95)d 9062.40 (7639.60)d	Qualitye (CFU/100 mL) 500 = excellentf 1,000 = goodf 900 = sufficientg
Heavy metals (μg/L)h
Fe	17.50 ± 5.00	25.00 ± 12.91	<1,000
Mn	140.00 ± 92.01 106.50±94.47i	435.00 ± 250.13 121.51±61.27i	<123 bioavailable
Pb	5.00 ± 5.77 0.18±0.19i	5.00 ± 4.08 0.19±0.19i	<1.2 bioavailable
Zn	35.00 ± 12.91 9.14±4.20i	33.75 ± 4.79 10.80±1.50i	<12.3 bioavailable
Cu	4.10 ± 1.04 0.16±0.04i	2.53 ± 1.16 0.08±0.04i	<1 bioavailable
As	0.34 ± 0.23	0.26 ± 0.20	<50

.	River upstream .	River downstream .	Standarda .
-N (mg/L) 90th percentile	0.65 ± 0.24 0.96	0.33 ± 0.12 0.48	Quality 0.3 = high, 0.6 = good, 1.1 = moderate, 2.5 = poor
-N (mg/L)	0.17 ± 0.09	0.10 ± 0.03	N/A
-N (mg/L)	6.69 ± 1.96	5.52 ± 1.09	N/A
TN (mg/L)	9.09 ± 2.92	7.20 ± 1.70	N/A
-P (mg/L)	0.46 ± 0.19	0.31 ± 0.14	Qualityb 0.04 = high, 0.08 = good 0.19 = moderate, 1.03 = poor
TP (mg/L)	0.50 ± 0.19	0.33 ± 0.15	N/A
Fluoride (mg/L)	0.39 ± 0.15	0.53 ± 0.20	N/A
Alkalinity (mg/L CaCO3)	130.75 ± 34.64	312.00 ± 77.49	N/A
Salinity (mg/L)	414.00 ± 56.26	987.25 ± 168.81	N/A
pH 5th and 95th percentile	8.02 ± 0.30 7.54–8.02	7.64 ± 0.21 7.30–7.64	6–9
Conductivity (μS/cm)	852.25 ± 111.36	1,960.50 ± 305.28	N/A
TDS (mg/L)	595.25 ± 88.94	1,371.75 ± 206.80	N/A
DO (% saturation) 10th percentile	83.15 ± 6.08 75.36	72.59 ± 5.20 65.93	Quality 70 = high, 60 = good
			54 = moderate, 45 = poor
COD (mg/L)	25.53 ± 9.45	20.31 ± 3.24	N/A
DOC (mg/L)	9.24 ± 2.25	7.22 ± 1.09	N/A
Temperature °C 98th percentile	11.98 ± 5.02 22.30	12.15 ± 4.81 22.00	Quality 25 = high, 28 = good
			30 = moderate, 32 = poor
Faecal coliform (log10CFU/100 mL) 90th percentilec 95th percentilec	3.38 ± 0.48 10147.10 (8554.00)d 15265.47 (12868.79)d	3.31 ± 0.39 6508.83 (5486.95)d 9062.40 (7639.60)d	Qualitye (CFU/100 mL) 500 = excellentf 1,000 = goodf 900 = sufficientg
Heavy metals (μg/L)h
Fe	17.50 ± 5.00	25.00 ± 12.91	<1,000
Mn	140.00 ± 92.01 106.50±94.47i	435.00 ± 250.13 121.51±61.27i	<123 bioavailable
Pb	5.00 ± 5.77 0.18±0.19i	5.00 ± 4.08 0.19±0.19i	<1.2 bioavailable
Zn	35.00 ± 12.91 9.14±4.20i	33.75 ± 4.79 10.80±1.50i	<12.3 bioavailable
Cu	4.10 ± 1.04 0.16±0.04i	2.53 ± 1.16 0.08±0.04i	<1 bioavailable
As	0.34 ± 0.23	0.26 ± 0.20	<50

Notes: Results were reported to two decimal places as Mean ± SD or percentile (in italics) of four sampling events. Values in percentile were provided for comparison with the standard as required by the directive.

The River Team is currently not a designated bathing river.

^aThe WFD Standards (England and Wales) 2015 for river.

^bBased on the standard for river upstream.

^cCFU/100 mL. Estimated numbers of E. coli (CFU/100 mL) are shown in parentheses after the faecal coliform numbers.

^dNot all but most faecal coliforms (about 75–93%) are E. coli (Hamilton et al. 2005). Hachich et al. (2012) recommended 84.3% for the conversion.

^eThe Bathing Water Regulations 2013 for E. coli in inland surface waters.

^fBased upon a 95th percentile evaluation.

^gBased upon a 90th percentile evaluation.

^hDissolved metal concentration (μg/L).

ⁱBioavailable concentration (μg/L) was calculated using the UKTAG tool. ‘Bioavailable’ means the fraction of the dissolved concentration of such metal is likely to result in toxic effects as determined using the UKTAG Metal Bioavailability Assessment Tool.

View Large

-P concentration in both upstream and downstream river samples indicated poor status with respect to this nutrient. Overall, there was no significant difference of TN and TP concentration in the river upstream and downstream samples (t-test, p-value > 0.05 for both TN and TP) implying no significant impact of the CW discharge on the nutrient status in the receiving river. High -P levels were already noted in the river upstream of the discharge, and therefore attributed to upstream sources, which include an STP at the East Tanfield. If the STP effluent from Birtley had been discharged directly into the river, it would have further augmented P levels in the river as there was higher -P concentration in the STP effluent (Table 1) than in the river upstream (Table 2). Cotreatment of STP effluent and mine water thus demonstrated a clear benefit of P removal that will benefit the River Team in terms of reduced P loading.

The pH of both the upstream and downstream samples was in the desired range. The DO as % saturation at the 10th percentile was indicative of the high status of the river in terms of its oxygenation. The water temperature in both upstream and downstream samples was also indicative of high status. There was significantly higher alkalinity, salinity, conductivity, and TDS in the river downstream relative to the river upstream (t-test, p-value < 0.01, for all). This is a consequence of the mine water characteristics and the poor removal of the main soluble ions in water like calcium, magnesium, chloride, and sulphate in the CW (Table S12 in SI).

In terms of heavy metals, the Fe, Pb, Cu, and As concentrations were below the standards (z-test, p-value < 0.01, for all) for both river upstream and downstream samples, while the mean values of Zn and Mn were only marginally below the limit without statistical significance (z-test, p-value > 0.05, for both). This indicates a potential risk of detrimentally affecting the receiving river because of the bioavailable levels of some metals (Mn from the mine water and Zn from the sewage) that are poorly removed in the CW.

There are currently no standards for bacteria in UK rivers except for two sites that are regulated as designated bathing waters (WFD 2017; Defra 2022a). However, this may change in the future as the public and the authorities become increasingly concerned about sewage impacts on recreation in rivers (DOHSC 2022; EAC 2022). The numbers of faecal coliform were high in both the river upstream and downstream samples, but lower in the downstream. They were converted to estimated numbers of E. coli to compare with the Bathing Water regulations. These estimated E. coli numbers exceeded the limit for sufficient bathing water status (900 CFU/100 mL as 90th percentile) by an order of magnitude.

Additionally, Table S19 in SI shows that the variables -P, pH, DO, and the temperature can best explain the dependency of the microbial community composition on the environmental conditions.

CWs designed for cotreatment of secondary-treated STP effluent and mine water in an average ratio of 1:4 had lower effluent than influent COD, nutrient, Fe, and faecal coliform levels and consistently achieved all compliance limits. However, levels of manganese, lead, copper, and zinc were not significantly reduced by the CW. The CW effectively removed the micropollutants acetaminophen, caffeine, and sulpiride, and to a lesser extent DEET, carbamazepine, sulfapyridine, venlafaxine, and cetirizine. The CW discharge did not detrimentally alter the nutrient status of the receiving river but could detrimentally affect the river due to the bioavailable concentration of manganese and zinc, which were near the respective guidance limits in the WFD. From the molecular microbiology data, the CW treatment resulted in the removal of putative human pathogen and faecal indicator bacteria, and consequently reduced the impacts of the STP effluent on the recreational value of the receiving river. CWs are a suitable nature-based treatment option to polish effluent from small STPs in rural areas by synergistic cotreatment with effluent from abandoned mines in a single system.

The study of J. Plaimart was sponsored by the Ministry of Higher Education, Science, Research, and Innovation of the Royal Thai Government. Northumbrian Water Limited and the Coal Authority are gratefully acknowledged for providing site access. Additional support for the molecular microbiology work was provided by EPSRC, Grant No. EP/P028527/1, BBSRC, Grant No. BB/T012471/1, and Royal Society, Grant No. ICA/R1/191241. For the purpose of open access, the author has applied a Creative Commons Attribution (CC BY) licence to any Author Accepted Manuscript version arising from this submission.

16S sequencing data generated in this project has been submitted to the NCBI Sequence Read Archive (SRA) with BioProject accession number PRJNA837409. Additional data created during this research are openly available (https://doi.org/10.25405/data.ncl.24937038). Please contact Newcastle Research Data Service at [email protected] for access instructions.

The authors declare there is no conflict.