Performance of pilot-scale horizontal subsurface flow constructed wetlands treating groundwater contaminated with phenols and petroleum derivatives

Ecological Engineering 95 (2016) 514–526

Contents lists available at ScienceDirect

Ecological Engineering journal homepage: www.elsevier.com/locate/ecoleng

Performance of pilot-scale horizontal subsurface flow constructed wetlands treating groundwater contaminated with phenols and petroleum derivatives Alexandros I. Stefanakis a,∗ , Eva Seeger a , Conrad Dorer b , Anja Sinke c , Martin Thullner a a

Department of Environmental Microbiology, UFZ – Helmholtz Centre for Environmental Research, Permoserstraße 15, 04318 Leipzig, Germany Department Isotope Biogeochemistry, UFZ – Helmholtz Centre for Environmental Research, Permoserstraße 15, 04318 Leipzig, Germany c BP International Limited, Sunbury on Thames, Middlesex TW16 7BP, UK b

a r t i c l e

i n f o

Article history: Received 30 October 2015 Received in revised form 21 June 2016 Accepted 26 June 2016 Keywords: Constructed wetlands Horizontal subsurface-flow Phenols MTBE Benzene Groundwater

a b s t r a c t Groundwater contaminated with various organic and inorganic pollutants is a major issue especially in historic industrial areas with chemical industries and refineries. In this study, Constructed Wetlands, as an ecological and environmentally sustainable alternative, were tested at pilot-scale for the removal of phenolic compounds and petroleum derivatives from contaminated groundwater in a pump-andtreat remediation research facility in Germany. Three horizontal subsurface-flow Constructed Wetlands (two planted, one unplanted) were fed with contaminated groundwater containing methyl tert-butyl ether (MTBE), benzene and ammonia. In two of the beds, a solution of phenol and m-cresol (15 and 2 mg/L or 314.5 and 45.5 mg/m2 /d, respectively) was injected to the groundwater inflow. Results showed a complete removal of the two phenolic compounds in the beds without any alteration in the MTBE and benzene removal rates. This indicated that Constructed Wetlands are versatile and can be used to effectively treat different pollutants simultaneously. Higher contaminant removal efficiency of planted systems confirms the positive role of plants presence and their ability to promote biodegradation. Spatial distribution analyses showed that the major portion of the removal took place in the first part of wetland length, which indicated that the systems could potentially deal with higher loads and can be used to optimize the system design. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Contaminated water at brownfields and industrial sites may pose a major environmental threat for ecosystems and public health, due to historic operational practices, incidents and leakages during storage and transport (Langwaldt and Puhakka, 2000; Seeger et al., 2011; Levchuk et al., 2014). Various toxic pollutants can end up in groundwater bodies, causing a significant ecological risk and posing a potential threat for public health, especially when these bodies would be used as drinking water sources or if they get in contact with surface waters. Among the various pollutants, fuel hydrocarbons like BTEX compounds (benzene, toluene, ethylbenzene, and xylenes), MTBE (methyl-tert-butyl-ether) and phenolic compounds are very com-

∗ Corresponding author. Present address: Constructed Wetland Competence Centre, Bauer Resources, P.O. Box 1186, Al Mina, Muscat, Oman. E-mail address: [email protected] (A.I. Stefanakis). http://dx.doi.org/10.1016/j.ecoleng.2016.06.105 0925-8574/© 2016 Elsevier B.V. All rights reserved.

mon in contaminated groundwater (Wu et al., 2006; Seeger et al., 2011; Van Afferden et al., 2011; Stefanakis et al., 2014). BTEX compounds and MTBE are highly soluble and mobile groundwater contaminants of great concern. Benzene, is considered as a human carcinogen (USEPA, 2015). Due to the related health risks, concentration limits have been regulated for both benzene (5 ␮g/L, respectively) in drinking water (USEPA, 2009). Phenols are organic contaminants present in wastewaters of different origin (Stefanakis et al., 2014), such as oil refineries and petrochemical industry (Abdelwahab et al., 2009), tanneries (Costa et al., 2008), olive mills (Herouvim et al., 2011), cork producing industry (Silva et al., 2015) and pulp and paper mills (Abira et al., 2005), while the usage of pesticides and disinfectants as well as natural sources like forest fires also contribute to phenol contamination (Stottmeister et al., 2010). Phenol presence in water and wastewater could be toxic to plants in case this water is (re)used for irrigation, while it may be toxic for bacteria too (Farré et al., 2001; Nair et al., 2008; Stefanakis et al., 2014). Phenol is one of the most toxic pollutants in wastewater, even at low concentrations

A.I. Stefanakis et al. / Ecological Engineering 95 (2016) 514–526

(Herouvim et al., 2011) and is a priority pollutant with a limit of 1 mg/L in wastewater (WHO, 2011) A pump-and-treat approach is often applied for remediation of impacted groundwater. The groundwater is pumped up and treated above ground using various physical and chemical methods such as membrane separation, adsorption onto porous media (e.g., activated carbon, zeolites), advanced oxidation processes (e.g., H2 O2 /O3 , H2 O2 /UV), air stripping and vapour extraction (Levchuk et al., 2014). However, these techniques may require significant expertise and complex equipment, may be accompanied by operational safety risks and finally also have high construction, operation and maintenance costs which makes them technically or financially infeasible. Constructed Wetlands (CWs) are considered today as an attractive sustainable alternative remediation technology with robust performance characteristics, reduced construction costs and minimum operation and maintenance costs (Kadlec and Wallace, 2009; Stefanakis et al., 2014). The technology of CWs is widely recognized as one of the “green” options for decentralized water/wastewater treatment. They have been effectively applied for the treatment of domestic and municipal wastewater, for wastewater produced from various industrial and agricultural installations, as also for sludge dewatering (Vymazal, 2009; Kadlec and Wallace, 2009; Stefanakis et al., 2009, 2011, 2014). The purification capacity of CWs relies on various naturally occurring physical, chemical and biological processes that take place within the system and degrade the various pollutants, as a result of the synergetic actions of the system components, i.e., substrate media, plant roots and microbial community (Stefanakis et al., 2014). Until today, the application of CWs for the treatment of groundwater contaminated with fuel hydrocarbons is limited. Studies have shown promising results for the removal of benzene and MTBE using horizontal subsurface-flow (HSSF) CWs (Seeger et al., 2011; Chen et al., 2012) or vertical flow (VF) CWs (Van Afferden et al., 2011; De Biase et al., 2013). Moreover, wastewaters with high phenol concentration (up to 500 mg/L) have been treated with CWs, e.g., domestic wastewater spiked with phenols (Tee et al., 2009), wastewater from olive mills (Del Bubba et al., 2004; Herouvim et al., 2011; Kapellakis et al., 2012), cork industry (Silva et al., 2015) and refineries (Knight et al., 1999) but respective applications for phenol contaminated groundwater have not been widely tested (Bedessem et al., 2007). As it is obvious, for these wastewater types inflow concentrations are usually higher than the concentration typically found in contaminated groundwater, while treatment often takes place in multistage CW systems. However, these studies provide good indications that phenolic compounds can be effectively treated in CW systems. The present study focuses on providing a better understanding of the role of the system parameters (e.g., plant presence, loading rate) for the overall phenolic compound removal efficiency, and investigating the response of the system when various contaminants like benzene, MTBE and phenolic compounds are

515

simultaneously present in the water. Specific objectives of the study are to evaluate: (i) the efficiency of HSSF CWs in the treatment of groundwater contaminated with phenolic compounds, benzene and MTBE, (ii) the role of system parameters such as plant presence for system performance, (iii) the possible interactions and/or treatment limitations due to the simultaneous presence of these different compounds, and (iv) the operational and environmental parameters that affect the performance. 2. Materials and methods 2.1. Site description The experimental facility is located near the town of Leuna, Saxony-Anhalt, Germany, next to former refinery facilities and an industrial area. There is an extensive contamination of local groundwater and soil due to the large-scale ammonia processing industry for the production of fertilizers and explosives and to petroleum refinement for more than 100 years. At this site a pump-and-treat approach is used to contain and remediate the groundwater. Part of this pumped water has been used over the past 10 years for a series of experimental studies with constructed wetlands (Jechalke et al., 2010; De Biase et al., 2013; Van Afferden et al., 2011; Seeger et al., 2011, 2013). 2.2. Project description The project consisted of two phases during which phenol and m-cresol were added to the groundwater injected into defined pilot-scale units: the preliminary phase (P1), which lasted about 3 months (14 August–24 October 2012) and included the setup of the experimental and operational procedures, the feasibility demonstration of the project approach and the testing of the applied evaluation methods, and the main phase (P2), which lasted about 8 months (8 April–27 November 2013) and covered the entire growth and active season of the plants. Table 1 presents the full timetable of the project and the important dates and respective actions taken during the project lifetime. During the preliminary phase, 11 field sampling campaigns were organized. The main phase included 14 sampling campaigns during the phenol/m-cresol injection period, one sampling campaign before the injections started and two campaigns after the end of the injections. Reed biomass was harvested and weighed during the first days of December 2013, almost one and a half month after the injections stopped. 2.3. Description of pilot-scale units Three pilot-scale HSSF CWs referred to as A–C were used. Each HSSF CW bed consisted of a steel basin (length × width × depth = 5.9 × 1.1 × 1.2 m, surface 6.5 m2 ; Fig. 1). All beds were filled with fine gravel (grain size 2–3.2 mm) up to

Table 1 Project timetable and experimental activities. Sampling took place simultaneously in all beds. Beds A (planted) and B (unplanted) received contaminated groundwater with injected phenol/m-cresol, bed C (planted) received contaminated groundwater without phenol/m-cresol. Experimental phase

Date

Activity

Preliminary (P1)

06 Aug–24 Oct 2012

Phenol/m-cresol injection period 11 field sampling campaigns

Main (P2)

08 Apr 2013

Field sampling (before injections period)

P2a (Q = 11 L/h)

17 Apr–5 Aug 2013

Phenol/m-cresol injection period 14 field sampling campaigns

P2b (Q = A, C: 11 L/h, B: 22 L/h)

5 Aug–23 Oct 2013 30 Oct, 27 Nov 2013 11 Dec 2013

2 field samplings (after injections period) Reed harvesting and weighing

516

A.I. Stefanakis et al. / Ecological Engineering 95 (2016) 514–526

Fig. 1. Cross section and dimensions of each pilot-scale CW and sampling points (i.e., blue dots) along the wetland length (inflow, 0.5 m, 1.9 m, 4.1 m, outflow) and at three different depths (0–5 cm, 30 cm, 80 cm). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.).

Table 2 Groundwater inflow characteristics. Values represent average and standard deviation (±␴) for the samples (n = 12) collected during the preliminary phase (P1; August–October 2012) and the main experimental phase (P2; April–October 2013). Parameter

Unit

Average ± ␴

Benzene MTBE Toluene Ethylbenzene m-p-Xylene o-Xylene NH4 + NO3 − NO2 − PO4 −3 Fe+2 SO4 −2 Cl− Ca+2 K+ Na+ Mg+2 Mn+2

mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L

10.2 ± 3.8 0.88 ± 0.32 0.002 ± 0.001 0.019 ± 0.017 0.009 ± 0.004 0.008 ± 0.003 27.1 ± 8.0 0.204 ± 0.164 < 0.010 1.80 ± 0.74 3.14 ± 0.71 76.0 ± 34.8 142.7 ± 29.8 204.1 ± 17.0 10.6 ± 1.3 143.1 ± 23.4 45.7 ± 3.2 1.2 ± 0.3

a height of 1 m. A small part of 20 cm at the inflow and outflow points was filled with coarse gravel (Fig. 1) to protect and avoid potential clogging of the inflow/outflow pipes with time. The presence of the coarse gravel pack in the beginning resulted in an evenly flow over the whole depth of the systems and equal concentrations of contaminants over the whole column. The last part of each CW length (4.70–5.70 m; Fig. 1) was designed as an open water compartment allowing for direct contact between the water surface and the atmosphere. The water level was set to 100 cm. Beds A and C were planted with common reeds (Phragmites australis), while bed B remained unplanted. All three beds were in operation for approximately 5 years before the start of this study, thus start-up period is avoided and the plant growth and biofilm development within the bed are considered to be optimum for the implementation of this study. Each HSSF CW was continuously supplied with contaminated groundwater pumped on-site from a near-by well at an inflow rate of 11 L/h, resulting in a nominal hydraulic residence time (HRT) of about 10 days and a hydraulic loading rate (HLR) of 0.040 m/d. Since 5 August 2013, the flow rate in the unplanted bed B was doubled (22 L/h) and the respective HLR was 0.081 m/d. Thus, main phase P2 is further divided into sub-phases P2a (0.040 m/d for all beds) and P2b (0.040 m/d for beds A and C, 0.081 m/d for bed B), as shown in Table 1. Groundwater characteristics are given in Table 2 for the entire experimental period. For all CWs, water inflow and outflow vol-

umes were quantified by flowmeters (15-min interval), allowing for contaminant loading rate to be calculated. 2.4. Phenol and m-cresol injections A solution of phenol and m-cresol was injected into the groundwater inflow of beds A and B. Bed C was operated as reference system receiving groundwater without any phenol injections. During the preliminary phase (P1), the nominal inflow concentrations were 5 mg/L and 1 mg/L for phenol and m-cresol, respectively. During the main experimental phase (P2), respective inflow concentrations were increased to 15 mg/L and 2 mg/L (with respective loads of 314.5 and 45.5 mg/m2 /d). The addition of the phenol/m-cresol solution into the groundwater inflow of beds A and B was realized via a T-valve using two peristaltic pumps with a nominal pumping rate of 120 mL/h. For each CW bed receiving the phenol/m-cresol solution, a storage tank was used to prepare the feeding solution. Therefore, for an inflow rate of 11 L/h, 45.8 g phenol and 8.9 mL m-cresol were dissolved in approximately 1 L tap water in the lab. This solution was diluted in each storage tank with 50 L tap water at the field site (concentration in the storage tank: 1365 mg/L phenol, 190 mg/L m-cresol). The pumping rate of each peristaltic pump was set to 60 mL/h. At a bi-weekly scheme the phenol/m-cresol solution in the storage tanks was replaced. Each time the tanks were emptied, cleaned and the new pollutant solution was mixed as described above. The residual phenol/m-cresol water from the storage tanks as well as the outflow water from CW beds A and B were further treated via an activated carbon filter and then via a slow-sand filter. After filtration, an external lab had to confirm that the final phenol/mcresol concentration was below the discharge limit. Only then, legal constrains allowed to add the outflow of these beds to the nearby discharge water stream of the entire wetland facility. 2.5. Sampling procedure and analyses Monthly samples of the inflow and outflow of each wetland were taken as part of the routine sampling activity of the wetland facility and were analyzed for benzene, MTBE, NH4 + -N and other water constituents (Seeger et al., 2011, 2013; De Biase et al., 2013). In addition, sampling campaigns were performed to monitor the concentrations of phenol and m-cresol: Influent and effluent water samples were collected from all beds every two weeks. Samples were taken from the inflow of beds A (INA ) and B (INB ) after the phenol/m-cresol injection, from the groundwater inflow (influent to bed C; INC ) and from the outflow of beds A–C (AOUT , BOUT and COUT , respectively). Once a month, samples were also taken from

A.I. Stefanakis et al. / Ecological Engineering 95 (2016) 514–526

517

Fig. 2. Variations of influent and effluent concentrations of phenol and m-cresol in CW beds A and B for the entire project life time.

various points within the wetland body to investigate the spatial distribution of phenol and m-cresol concentrations. For this, water was pumped using peristaltic pumps (type: REGLO digital, Ismatec) from nine points distributed along the wetland length (0.5 m, 1.9 m, 4.1 m; points a–c) and depth (0–5 cm, 30 cm, 80 cm; points 1–3), as shown in Fig. 1 (blue dots). Water samples were filled into autoclaved bottles. The samples used for the concentration analysis were acidified with HCl to pH 2 in order to inhibit microbial degradation processes. All samples were analyzed for the determination of phenol and mcresol concentration. Water samples were analyzed in duplicates for the concentration of organic compounds (benzene, MTBE, etc.) by a GC–MS system as described by Seeger et al. (2011).

2.5.1. Phenol and m-cresol concentration measurement by HPLC About 1 mL of each sample was transferred by a sterile 1 mL syringe into a 2 mL screw vial (WICOM, order number 41500) before the measurement. Phenol and m-cresol were quantified by means of high-pressure liquid chromatography (HPLC) (Shimadzu prominence line; liquid chromatograph: LC-20AB, degasser: DGU-20A3, auto sampler: SIL-20A, communications bus module: CBM-20A, diode array detector: SPD-M20A). 100 ␮L (non-extracted) sample were injected, separated on a C18 reverse-phase column (Merck, LiChroCART 125-4, LiChrophere 100, RP18e, 5 ␮m) with an oven temperature of 25 ◦ C and a flow of 1.0 mL/min. Phenol was detected at 269 nm and m-cresol at 271 nm. As eluents served 0.1% phosphoric acid (reagent A) and HPLC grade methanol (reagent B). The

518

A.I. Stefanakis et al. / Ecological Engineering 95 (2016) 514–526

elution program was as follows: 25% eluent A (5 min), linear increments of reagent A from 25% to 40% (from 5 to 18 min), linear increments of reagent A from 40% to 100% (from 18 to 20 min), afterwards reagent A was linearly decreased to 10% (from 20 to 22 min) and set back to the initial value of 25% (from 22 to 22.5 min) and kept until minute 30. Phenol had a retention time of about 7 min and mcresol of about 15 min. Equimolar mixtures of phenol and m-cresol served for calibration and were tested in the range of 1 ␮M until 300 ␮M (for environmental samples) and of 0.1 mM until 10 mM (for tank solutions) resulting in very good linearity. 2.6. Statistical tests In order to statistically investigate the significance level of the differences between the wetland units, paired t-tests were employed with a confidence interval for 95% probability. The paired t-test calculates the confidence interval of the mean of the differences of the parameters that were measured at the same sampling time during a time period. Therefore, results are not affected by the variability of a measurement versus time. 3. Results and discussion 3.1. Treatment efficiency The overall treatment performance of the pilot units is presented in terms of percent concentration reduction and mass removal, which are the most widely used expressions (Kadlec and Wallace, 2009; Stefanakis and Tsihrintzis, 2012). For the performance comparison, the approaches used are: average concentrations (inflow–outflow) and respective percent removal, and surface removal rate (SRR). The SRR is given by the difference between influent load (IL) and effluent load (EL) of each pollutant, and is expressed as the removed pollutant mass per surface area and time (g/m2 /d); it, thus, represents a useful parameter to assess system efficiency, which is also referred to in the literature

as areal load reduction (Kadlec and Wallace, 2009; Stefanakis and Tsihrintzis, 2012). 3.1.1. Phenol and m-cresol Variations of influent and effluent concentrations of the two injected phenolic compounds (phenol and m-cresol) in beds A and B are shown in Fig. 2 for the entire project life time, while mean concentration values and respective removal efficiencies are presented in Table 3. Data for bed C are not presented since this unit did not receive the artificial phenol/m-cresol solution and respective influent and effluent concentrations were always negligible (mean influent values 0.014 and 0.001 mg/L for phenol and mcresol, respectively, in the raw groundwater). The short preliminary phase P1 demonstrated the feasibility of the general project approach, confirmed the applied evaluation methods and gave the first results. Main conclusions are drawn from the results obtained during the main experimental phase P2 which covered the entire annual plant growth season. The first obvious result is the almost complete removal of both phenolic compounds in the HSSF CW beds. Planted bed A presented removal rates of up to 99% during phases P1 and P2b for both phenol and m-cresol. The efficiency of the unplanted bed B was always lower compared to bed A for all phases (89.1% and 94.0% m-cresol removal during phases P1 and P2b, respectively; Table 3). This is also a first indication of the potential positive effect of plant presence on the system performance. Fig. 2 also shows that during the entire experimental period of the study, the unplanted bed B always had higher effluent concentrations of both phenol and m-cresol compared to the planted bed A. From Fig. 2 it can be seen that for both phenolic compounds, a certain time period was needed in order to achieve the full response of the system to the phenol/m-cresol injections and the respective decline in effluent concentrations. The first samples (May 2nd) taken 15 days after the start of phenol/m-cresol injections (April 17th) showed that there still was a relatively high residual concentration of both phenol and m-cresol in the effluent of the two units

Table 3 Phenol and m-cresol influent and effluent concentrations and respective removal efficiencies of units A and B for the preliminary phase (P1; 14 Aug–24 Oct 2012), the main phase (P2; 8 Apr–23 Oct 2013) and the sub-phases P2a (8 Apr–5 Aug 2013) and P2b (5 Aug–23 Oct 2013). Phenol (mg/L) preliminary phase P1a

main phase P2a

total

total

INA

AOUT

%

INA

aver Sd min max

6.389 1.469 3.929 8.533

0.012 0.028 0.000 0.086

99.8 0.4 98.7 100.0

15.203 2.475 12.330 21.998

aver Sd min max

INB 6.639 1.616 4.058 9.028

BOUT 0.279 0.625 0.000 2.035

% 94.3 14.5 53.2 100.0

aver Sd min max

m-cresol (mg/L) INA AOUT 0.895 0.018 0.199 0.030 0.571 0.000 1.131 0.080

aver Sd min max

INB 0.962 0.234 0.576 1.318

BOUT 0.072 0.182 0.000 0.572

a b

P2ab AOUT

%

INA

0.661 2.384 0.000 8.940

95.3 17.2 35.7 100.0

15.321 2.996 12.330 21.998

INB 17.708 3.328 12.515 23.694

BOUT 1.401 2.964 0.000 11.424

% 90.8 20.2 22.3 100.0

% 97.7 3.8 90.6 100.0

INA 1.976 0.248 1.726 2.767

AOUT 0.195 0.467 0.000 1.561

% 89.1 28.5 9.6 100.0

INB 2.310 0.316 1.912 2.912

BOUT 0.342 0.504 0.000 1.575

P2bb %

INA

AOUT

%

1.148 3.149 0.000 8.940

91.8 22.7 35.7 100.0

15.046 1.820 12.725 16.913

0.011 0.018 0.000 0.043

99.9 0.1 99.7 100.0

INB 15.382 1.647 12.515 17.997

BOUT 2.174 3.788 0.044 11.424

% 85.1 25.8 22.3 99.8

INB 20.810 2.191 17.651 23.694

BOUT 0.370 0.730 0.000 1.839

% 98.4 3.2 91.9 100.0

% 90.5 22.9 23.9 100.0

INA 2.009 0.327 1.726 2.767

AOUT 0.335 0.594 0.000 1.561

% 83.7 29.2 23.9 100.0

INA 1.933 0.076 1.822 2.025

AOUT 0.007 0.011 0.000 0.026

% 99.6 0.6 98.7 100.0

% 83.9 24.0 24.2 100.0

INB 2.090 0.115 1.912 2.281

BOUT 0.488 0.600 0.000 1.575

% 76.4 28.8 24.2 100.0

INB 2.602 0.247 2.334 2.912

BOUT 0.149 0.281 0.000 0.700

% 94.0 11.1 72.3 100.0

For P1, P2, P2a and P2b sample numbers are 10, 14, 8 and 6, respectively. Respective HLR: P1, P2a = 0.040 m/d for all units, P2b = 0.040 m/d for units A and C, 0.081 m/d for unit B.

AOUT

A.I. Stefanakis et al. / Ecological Engineering 95 (2016) 514–526

519

Table 4 Phenol and m-cresol influent (IL) and effluent loads (EL) and respective surface load removal rates (SRR; mg/m2 /d) of units A and B for the preliminary phase (P1; 14 Aug–24 Oct 2012), the main phase (P2; 8 Apr–23 Oct 2013) and the sub-phases P2a (8 Apr–5 Aug 2013) and P2b (5 Aug–23 Oct 2013). Phenol (mg/m2 /d) preliminary phase P1a

main phase P2a

total

total

ILA

ELA

SRR

ILA

P2ab ELA

SRR

P2bb

ILA

ELA

SRR

ILA

ELA

SRR

aver sd min max

259.5 59.7 159.6 346.6

0.5 1.1 0.0 3.5

259.0 59.6 159.6 346.6

617.5 100.5 500.8 893.5

26.8 96.8 0.0 363.1

590.6 149.7 201.4 892.2

622.3 121.7 500.8 893.5

46.6 127.9 0.0 363.1

575.6 192.8 201.4 892.2

611.1 73.9 516.8 686.9

0.5 0.7 0.0 1.7

610.6 73.4 516.8 685.2

aver sd min max

ILB 269.6 65.6 164.8 366.7

ELB 11.3 25.4 0.0 82.7

ALR 258.3 79.9 94.0 361.3

ILB 1081.5 560.5 508.3 1924.7

ELB 63.4 122.4 0.0 464.0

ALR 1018.1 600.5 133.1 1898.4

ILB 624.8 66.9 508.3 731.0

ELB 88.3 153.8 1.8 464.0

ALR 536.4 185.3 133.1 729.2

ILB 1690.4 178.0 1433.8 1924.7

ELB 30.1 59.3 0.0 149.4

ALR 1660.3 153.1 1431.6 1898.4

aver sd min max

m-cresol (mg/m2 /d) ILA ELA 36.4 0.7 8.1 1.2 23.2 0.0 45.9 3.2

SRR 35.7 8.6 22.0 45.9

ILA 80.3 10.1 70.1 112.4

ELA 7.9 19.0 0.0 63.4

SRR 72.4 20.0 20.0 107.5

ILA 81.6 13.3 70.1 112.4

ELA 13.6 24.1 0.0 63.4

SRR 68.0 26.2 20.0 107.5

ILA 78.5 3.1 74.0 82.2

ELA 0.3 0.5 0.0 1.0

SRR 78.2 2.7 74.0 81.2

aver sd min max

ILB 39.1 9.5 23.4 53.5

SRR 36.2 14.9 2.5 53.5

ILB 139.1 66.2 77.7 236.5

ELB 16.5 23.1 0.0 64.0

SRR 122.6 74.6 20.4 236.5

ILB 84.9 4.7 77.7 92.6

ELB 19.8 24.4 0.0 64.0

SRR 65.1 25.6 20.4 92.6

ILB 211.4 20.1 189.6 236.5

ELB 12.1 22.8 0.0 56.9

SRR 199.3 34.7 148.3 236.5

a b

ELB 2.9 7.4 0.0 23.2

For P1, P2, P2a and P2b sample numbers are 10, 14, 8 and 6, respectively. Respective HLR: P1, P2a = 0.040 m/d for all units, P2b = 0.040 m/d for units A and C, 0.081 m/d for unit B.

(A and B) receiving the solution. After this first sampling event, phenol effluent concentrations rapidly dropped to values close to zero, especially in the planted bed A. For m-cresol, this decrease to low effluent values took place more gradually (Fig. 2b). Thus, it can be stated that the system needs about 20–30 days for phenol and 30–40 days for m-cresol to adjust to the presence of the new compounds. This also means that the system needs 2–4 times the residence time to adjust. This indicates a possible shift in the microbial community composition, which needs a certain time period to adjust to phenol and m-cresol. After this period, effluent concentrations remained low for both beds, with some variations occurring only in the unplanted bed B. Thus, slightly lower performance during phase P2a compared to phase P2b can be explained by this start-up period and the initially lower removal rates during the first month. Obtained results were further analyzed for influent (IL) and effluent (EL) loads and surface load removal rates (SRR; mg/m2 /d) for both phenol and m-cresol during the experimental periods (Table 4). During the preliminary phase P1, SRR values for beds A and B were similar (259 and 258.3 mg/m2 /d for phenol and 35.7 and 36.2 mg/m2 /d for m-cresol, respectively; Table 4) which indicated the potential of the system to remove higher loads. During the main phase P2a (with increased influent phenol concentrations and same HLR for both units), the planted bed A presented higher SRR (mean 575.6 and 68.0 mg/m2 /d for phenol and m-cresol, respectively) than the unplanted bed B (536.4 and 65.4 mg/m2 /d, respectively). SRR values for bed A were slightly increased during P2b to 610.6 mg/m2 /d (since there was no start-up period during this phase). The increase of the HLR in the unplanted bed B resulted in higher SRR for this unit during the phase P2b (1660.3 and 199.3 mg/m2 /d for phenol and m-cresol, respectively; Table 4). The fact that also with the increased HLR a nearly complete removal was achieved in the unplanted bed B implies that the system’s removal capacity was probably still not fully exceeded during this experiment.

Furthermore, results show that removal of phenol is favoured to m-cresol: removal rates of phenol were always higher than those of m-cresol in both beds which suggests that phenol is slightly easier to degrade than m-cresol. It is also interesting that these differences between phenol and m-cresol removal rates were higher in the unplanted bed B than in the planted bed A. As mentioned above, similar studies of phenol-contaminated groundwater treated with CWs do not exist in published literature yet, while published studies on phenol removal using CWs are relatively limited. Overall the CWs in this study seem to be more efficient than similar CW studies for phenol removal from various wastewater types. An efficiency of 77% phenol removal is reported for a pilot HSSF CW system treating paper mill wastewater, but for low influent concentrations (0.43–1.7 mg/L; Abira et al., 2005). Similar efficiency (72%) is mentioned for another pilot HSSF system (Rossmann et al., 2012) receiving aerated coffee processing wastewater containing 26.1 mg/L phenol with a higher HRT of 12 days compared to the present study. Generally, the role of the various phenol removal processes is not yet clearly defined. Current knowledge and experimental results imply that, in HSSF CW beds, biodegradation is the main mechanism for phenol transformation and removal (Imfeld et al., 2009; Stefanakis et al., 2014) and aerobic conditions seem to favour this process (Rossmann et al., 2012), although degradation may also proceed in the anaerobic zone (Stottmeister et al., 2010). 3.1.1.1. Spatial distribution. The spatial distribution of phenol and m-cresol concentration is presented in Fig. 3 as average of the spatial sampling campaigns during the shown periods. Generally, lower concentrations of the two phenolic compounds are found in all sampling points in the planted bed A compared to the unplanted bed B. In both beds, gradual removal takes place along the wetland length. This removal is more intense in the planted bed A, where the major portion of the influent concentra-

520

A.I. Stefanakis et al. / Ecological Engineering 95 (2016) 514–526

Fig. 3. Spatial distribution of phenol and m-cresol concentrations in the CW beds A and B along the wetland length (IN, 0.5 m, 1.9 m, 4.1 m, OUT) and for three different depths (0–5 cm, 30 cm, 80 cm) during the various operational periods. In bed A, the apparent increase in m-cresol concentrations from Periods P2a to P2b is caused by results from one of the P2b sampling campaigns, thus, increasing the otherwise lower concentration for this period.

tions is removed within the first parts of the wetland length. Higher concentrations are found in points closer to the inlet of both beds. In the planted bed A, highest concentrations of both phenol and m-cresol are found at the bottom parts of the bed than in the upper parts, while the top layer exhibits practically zero phenol concentration and very low m-cresol concentration. The same trend is also

observed in the unplanted bed B, although deeper parts contain higher concentration of both compounds than in bed A. Given that in the uppermost part of the top layer air is introduced from the atmosphere via diffusion, the top layer is dominated by aerobic conditions in both planted and unplanted beds and, thus, the majority of the microbial activity takes place there, which could explain the

A.I. Stefanakis et al. / Ecological Engineering 95 (2016) 514–526

521

Fig. 4. Variations of influent and effluent concentrations of MTBE and benzene in the three CW beds A–C for the entire project life time.

absence of the compounds. Furthermore, the plants in bed A promote the infiltration of oxygen and the root system is denser in the upper parts of the bed (Kadlec and Wallace, 2009; Stefanakis et al., 2014). The presence of plants may indirectly enhance the degradation of these compounds in the layers closer to the top of the bed, by emitting root exudates, which stimulate the growth of the microbial population, the community diversity and also the enzymatic activity (Pivetz, 2001; Stottmeister et al., 2003), and providing additional attachment area (besides the filter media grains) for the microbes along the root surface (Stottmeister et al., 2003; Gagnon et al., 2007; Stefanakis et al., 2014). Herouvim et al. (2011) also mentioned that the microbial activity was higher in planted than in unplanted CWs treating olive mill wastewater, which resulted in increased phenol removal through aerobic degradation.

However, as both, the planted bed A and the unplanted bed B, showed an almost complete removal of both phenolic compounds (Fig. 3), it can be assumed that anaerobic degradation of these compounds also occurs (Kumaran and Paruchuri, 1997; Stefanakis et al., 2014). In deeper parts of the planted bed with low root density (roots of P. australis usually don’t exceed depths of 50–60 cm), phenol and m-cresol removal probably takes place mainly via anaerobic pathways. In the unplanted bed, anaerobic conditions prevail in most of the wetland body, thus anaerobic degradation is assumed as the main removal mechanism. Moreover, it is also interesting that phenol and m-cresol removal efficiency in the unplanted bed B improved upon the increase of the applied load. The spatial removal of both compounds during period P2 altered and most of the influent concentration is removed in the

522

A.I. Stefanakis et al. / Ecological Engineering 95 (2016) 514–526

Fig. 5. Influent and effluent concentrations and load values of MTBE and benzene in the three CW beds A–C and respective removal rates.

upstream parts of the system (Fig. 3). This might be attributed to the fact that the microbial community needed a certain time period to adjust to the new compounds added in groundwater. During period P1 (start-up phase), a more uniform removal along wetland length

was observed. After this phase, removal rates in bed B increased (period 2) and more than half of the influent concentration was removed within the first half meter of the wetland length.

A.I. Stefanakis et al. / Ecological Engineering 95 (2016) 514–526

523

Fig. 6. Variations of influent and effluent values of air temperature, water loss, ammonium and sulphate in the three CW beds A–C for the entire project life time.

3.1.2. MTBE and benzene Among the BTEX compounds present in ground water, benzene and MTBE constitute the main organic contaminants (Table 2). Fig. 4 presents the concentration variation of the pilot systems throughout the entire experimental period (phases P1 and P2). During the winter pause (November 2012–March 2013), phenol injections were not carried out, but groundwater loadings and regular monitoring of the beds continued. Thus, Fig. 4 includes data for the entire monitoring period, i.e., phase P1, winter pause, phase P2. The effi-

ciencies of the systems are presented in Fig. 5 in terms of mean influent-effluent concentration, mean removal rates, influent loads and SRR values. Fig. 4 shows that there is a gradual increase in influent concentrations as the experiment proceeded from phase P1 to phase P2. Overall results showed that both planted beds (A and C) achieved always better removal rates compared to the unplanted bed B (52.8% and 82.2% for Bed A, 49.6% and 72.3% for bed C and 41.2% and 66.1% for bed B, for MTBE and benzene respectively; Fig. 5),

524

A.I. Stefanakis et al. / Ecological Engineering 95 (2016) 514–526

Table 5 Paired t-tests for the various comparisons between the wetland beds (* implies statistical significance at the 0.05 level). Comparison

INA ∼ INB INA ∼ AOUT INB ∼ BOUT INC ∼ COUT AOUT ∼ BOUT AOUT ∼ COUT INA ∼ ALRA INB ∼ ALRB INC ∼ ALRC ALRA ∼ ALRB ALRA ∼ ALRC AOUT ∼ BOUT [%] AOUT ∼ COUT [%]

Phenol

m-cresol

MTBE

mg/L

mg/m2 /d

mg/L

mg/m2 /d

0.038* 0.000* 0.000*

0.010* 0.000* 0.000*

0.014* 0.000* 0.000*

0.007* 0.000* 0.000*

0.006*

0.010*

0.019*

0.050*

0.006*

0.318 0.075*

0.143 0.019*

0.016*

0.017* 0.017*

as observed for the phenolic compounds too. Such dependency of benzene and MTBE removal rates on plant presence has been reported for similar CW systems by Seeger et al. (2011) and Chen et al. (2012). Moreover, both planted beds showed higher SRR values than the unplanted bed B, when operating under the same conditions (phases P1 and P2a). After the increase of the flow rate in unplanted bed B (phase P2b), SRR values increased for this unit although the percentage efficiency was still below the other two planted beds. MTBE removal in this study was higher compared to the study of Seeger et al. (2011) and similar to the study of Chen et al. (2012), where shallower systems (i.e., 50 cm depth) were used and double and similar, respectively, influent concentrations were applied. Moreover, higher removal efficiency was observed for benzene than for MTBE, which is also found in other HSSF CWs (Bedessem et al., 2007; Seeger et al., 2011), vertical flow soil filters (Van Afferden et al., 2011; De Biase et al., 2013) and aerated treatment ponds (Jechalke et al., 2010). The chemical structure, the relatively low growth of MTBE consuming microorganisms and the presence of BTEX compounds have been proposed as possible inhibitors of MTBE removal (Deeb et al., 2001; Salanitro et al., 2000; Schmidt et al., 2004; Jechalke et al., 2010). Moreover, it is reported that microbial degradation of benzene takes place at a higher rate than that of MTBE (Seeger et al., 2011), while volatilization (plant mediated or via air-water contact) was found to play a secondary role in MTBE and benzene removal at the study site (Reiche et al., 2010).

3.1.3. Ammonium and sulphate Ammonium was present in the groundwater inflow, at a concentration range of 25–30 mg/L. It reduced in all beds, with the two planted beds (A and C) showing a removal rate up to 40% (Fig. 6), which was higher than in the unplanted bed B (up to 10%). This efficiency is comparable to that reported for similar HSSF systems (Hench et al., 2003; Akratos and Tsihrintzis, 2007; Stefanakis et al., 2009; Seeger et al., 2013) and within the range of 40–50% removal performance reported by Vymazal (2007) for loadings between 250 and 630 g N/m2 /yr. It is known that microbial processes (nitrification, denitrification) dominate the transformation/removal of ammonium in HSSF CWs, while direct plant uptake is of secondary importance in the long-term (Stottmeister et al., 2003; Akratos and Tsihrintzis, 2007; Vymazal, 2007; Kadlec and Wallace, 2009; Stefanakis et al., 2009, 2014; Alvarez-Zaldívar et al., 2016). The simultaneous presence of nitrification/denitrification could explain the low concentration of oxidized nitrogen (nitrate < 0.4 mg/L, nitrite < 0.1 mg/L), as also reported elsewhere (Stefanakis et al., 2009; Akratos and Tsihrintzis, 2007; Faulwetter et al., 2009), although there is also the possibility that the competition for

mg/L 0.001* 0.002* 0.001* 0.130 0.314

0.182 0.398

Benzene mg/m2 /d 0.157 0.001* 0.112 0.001* 0.012* 0.315 0.000* 0.006* 0.000* 0.926 0.407

mg/L 0.002* 0.048* 0.001* 0.529 0.002*

mg/m2 /d 0.145 0.024* 0.001* 0.127 0.024* 0.529 0.026* 0.022* 0.001* 0.763 0.567

0.156 0.573

oxygen among the various degradation processes could limit ammonium transformation processes. Sulphate reduction was also observed in all beds. As Fig. 6 depicts, sulphate inflow concentration gradually decreased during the experimental period; mean influent values for the preliminary phase P1 and the main phase P2 were 100.5 and 27.4 mg/L, respectively. Removal rates varied from approximately 33% during phase P1 to 64–78% during phase P2, which is significantly higher than the mean reduction of 14% reported for HSSF systems by Kadlec and Wallace (2009). Sulphate removal in all beds indicates the presence of anaerobic conditions within the beds, which is also shown by the low oxygen concentrations measured (<0.5 mg/L). The high sulphate reduction coupled with the minimum ammonium removal (limited nitrification) implies that anaerobic conditions largely predominate at greater depth. The latter is confirmed by low redox values detected in all beds (< −400 mV) indicating the absence of oxygen and providing favourable conditions for the reduction of sulphate (Kadlec and Wallace, 2009). Additionally, based on oxygen transfer rates (0.3–10 g O2 /m2 /d) reported in literature for HSSF beds (Tyroller et al., 2010; Nivala et al., 2012), respective oxygen values for the pilot systems of this study would be high enough for pollutant breakdown. In this sense, measured low oxygen concentrations in the pilot beds imply high rates of oxygen consumption and associated microbial activity. However, it should also be mentioned that the beds used in the present study were deeper (1 m) than most of the systems reported in the literature (0.4–0.6 m), which also favours the creation of anaerobic conditions. The slight increase in pH from 7.2 (inflow) to 7.5–7.7 (outflow) in all beds supports the assumption of sulphate reduction in the CW systems.

3.2. Interactions of the added phenolic compounds with MTBE and BTEX The simultaneous fate of phenolic compounds and petroleum hydrocarbons in groundwater is a case rarely investigated in the literature. The experimental setup was selected in this way so that results could be drawn concerning the possible effects of the phenolic compounds addition on the removal of the other contaminants (MTBE, benzene, ammonia) and the overall treatment efficiency of the system. Thus, beds A and C were identical (same design and operation characteristics), but phenol injections took place only in bed A. Results obtained mainly from period P2 (which covered the entire plant growth season and almost completely the annual temperature range) are used to evaluate the possible alterations in system performance. Fig. 5 depicts higher removal rates and higher SRR for bed A for both, MTBE and benzene (41% and 58.6%, 20.2 and 334.6 mg/m2 /d),

A.I. Stefanakis et al. / Ecological Engineering 95 (2016) 514–526

compared to bed C (31% and 48.5%, 15.1 and 273.9 mg/m2 /d). At the same time, bed A managed to sufficiently remove the added phenolic compounds. The removal capacity of bed A reached 620 mg/m2 /d for phenol and 80 mg/m2 /d for m-cresol (Table 4), which exceeded the respective capacity for benzene and MTBE removal (respective SRR 335 mg/m2 /d and 20.2 mg/m2 /d). Benzene removal (average 41.1%) was lower compared to phenol and m-cresol (up to 99%), while both beds, A and C, showed similar efficiency in ammonium removal (40.7% and 38.3%, respectively; Fig. 6). Thus, the injected phenolic compounds were sufficiently removed in the planted bed A without causing any negative impact on the removal of MTBE and benzene. The differences in the treatment efficiency between the beds A and C were not statistically significant (Table 5), indicating that the system behaviour did not alter after the phenol injection in bed A. The overall performance of bed A outreached bed C, since it not only achieved higher removal rates for MTBE and benzene but also it removed almost completely the two injected phenolic compounds. Presumably, some natural variability in plant growth has contributed to this, given that the measured aboveground plant biomass was 15% higher in bed A than in C. Increased plant biomass implies a more extensive root system and, thus, potentially higher oxygen input, additional surface available for microbe attachment and biofilm development and increased microbial activity and exudates release by plant roots (Stottmeister et al., 2003; Imfeld et al., 2009; Seeger et al., 2011; Stefanakis et al., 2014; Haddaji et al., 2015). It is also interesting that the produced biomass was higher at the downstream end of both beds. This could be due to the root penetration to the open water compartment, where low pollutant concentrations are detected. Overall, the results demonstrate that the studied HSSF CW system is capable to simultaneously remove phenolic compounds and petroleum hydrocarbons at high rates.

3.3. Effect of plants and seasonal variations The role of plants was also investigated in the present study. Beds A and B were identical in terms of design, with the only difference being the absence of plants in bed B. Both beds received the same phenol/m-cresol injections. As Table 3 shows, in phase P2a the planted bed A achieved statistically significant higher removal rates for phenol and m-cresol than the unplanted bed B (91.8% and 85.1% for phenol and 83.7 and 76.4% for m-cresol, respectively). In phase P2b the flowrate in bed B was doubled which makes the comparison difficult; however, for both systems almost complete removal of the phenolic compounds was achieved. The same trend was observed in phase P2a for MTBE (35.3% and 18.4%, respectively in beds A and B) and benzene (46.8% and 14.7%, respectively). Ammonium removal was also higher in the planted bed (40.7%) than in the unplanted (10.0%). Improved performance in planted beds is generally accepted in the literature (Stottmeister et al., 2003; Akratos and Tsihrintzis, 2007; Stefanakis et al., 2009; Tee et al., 2009; Kurzbaum et al., 2010; Seeger et al., 2011; Chen et al., 2012; Seeger et al., 2013; Stefanakis et al., 2014). Plant root systems act as an attachment area for microbes and the development of biofilms which degrade a variety of pollutants (Stottmeister et al., 2003; Vymazal, 2007; Stefanakis et al., 2014). Plants also transfer oxygen into the rhizosphere through their roots which enhances aerobic microbial activities. Furthermore, they release organic compounds via their roots, thus, providing additional electron donors for anaerobic microbial processes (Stottmeister et al., 2003; Imfeld et al., 2009; Chen et al., 2012; Haddaji et al., 2015). Hence, plants promote an increased biodegradation rate of the various groundwater contaminants.

525

Air temperature variation is another factor that affects wetland performance. Higher temperatures stimulate microbial activity and the respective degradation of the various organic compounds (Kadlec and Wallace, 2009; Stefanakis et al., 2014). It is reported that temperature values within the range of 20–35 ◦ C favour optimum microbial mediated reactions, while the opposite happens for values below 15 ◦ C (Kadlec and Reddy, 2000). From May till October mean temperature values were above 10 ◦ C, while during summer months (June–August) temperature exceeded 20 ◦ C (Fig. 6). In all beds, increased removal rates for the two phenolic compounds were observed at elevated temperatures (Fig. 2), and the same is also observed for MTBE and benzene (Fig. 4) as well as for ammonium (Fig. 6). This is in agreement with previous CW studies showing higher benzene removal at higher temperatures (Tang et al., 2009; Chen et al., 2012; Seeger et al., 2011), and increased nitrification rates at higher temperatures (Vymazal, 2007; Stefanakis et al., 2014). High temperatures are also coupled with increased microbial activity due to optimum plant growth in summer months. This is shown by water losses via evapotranspiration which are higher during summer months (Fig. 6), indicating enhanced plant growth and root activity. Thus, increased removal rates in summer months might be a combined effect in planted beds. In turn, the absence of plants in bed B indicates a direct effect of air temperature variations on removal rates. 4. Summary and conclusions The present study is the first to investigate HSSF CWs for the simultaneous treatment of phenolic compounds and petroleum hydrocarbons. Overall results showed that these treatment systems are an effective remediation technology for groundwater contaminated with phenol, m-cresol, MTBE and benzene. The complete removal of phenol and m-cresol (620 mg/m2 /d and 80 mg/m2 /d respectively) in both planted and unplanted beds implies that the system possesses an even higher treatment capacity. Moreover, the presence of the phenolic compounds did not affect the removal of MTBE and benzene (20.2 and 334.6 mg/m2 /d, respectively). The presence of plants was found to improve the system contaminant removal performance. To this, the addition of the phenolic compounds did not cause any toxicity effects to plant health. The spatial distribution of phenols within the wetland body showed that, in the planted bed, a rapid concentration decrease takes place. In the unplanted bed, higher contaminant concentrations are detected at all sampling points within the bed. Generally, higher concentrations are found closer to the inlet and to the bottom parts of the bed. Moreover, similar concentrations between the last downstream sampling point of the wetland length and the effluent imply that the open water part is not a necessary design element. The role of the various removal processes of phenolic compounds are not yet clearly identified in HSSF CWs. However, the complete removal of the contaminants at the top layer of the bed where aerobic conditions prevail and create favourable conditions for microbial growth, implies that biodegradation is the main removal process. This also explains the initial lag time frame the microbial community needed for its adaptation to the injected phenolic compounds. Furthermore, the reduced yet high removal rates of the unplanted bed indicate that plant uptake is not significantly contributing to the observed contaminant removal. Acknowledgments This study was funded by BP International Ltd and by the Helmholtz Association via grant VG-NG-338 (GReaT MoDE) and program topic “CITE – Chemicals in the Environment.” The authors

526

A.I. Stefanakis et al. / Ecological Engineering 95 (2016) 514–526

would like to thank Peter Mosig (UFZ, Centre for Environmental Biotechnology) for technical support. References Abdelwahab, O., Amin, N.K., El-Ashtoukhy, E.-S.Z., 2009. Electrochemical removal of phenol from oil refinery wastewater. J. Hazard. Mater. 163, 711–716. Abira, M.A., Van Bruggen, J.J.A., Denny, P., 2005. Potential of a tropical subsurface constructed wetland to remove phenol from pre-treated pulp and papermill wastewater. Water Sci. Technol. 51 (9), 173–176. Akratos, C.S., Tsihrintzis, V.A., 2007. Effect of temperature, HRT, vegetation and porous media on removal efficiency of pilot-scale horizontal subsurface flow constructed wetlands. Ecol. Eng. 29, 173–191. Alvarez-Zaldívar, P., Centler, F., Maier, U., Thullner, M., Imfeld, G., 2016. Biogeochemical modelling of in situ biodegradation and stable isotope fractionation of intermediate chloroethenes in a horizontal subsurface flow wetland. Ecol. Eng. 90, 170–179. Bedessem, M.E., Ferro, A.M., Hiegel, T., 2007. Pilot-scale constructed wetlands for petroleum contaminated groundwater. Wat. Environ. Res. 79, 581–586. Chen, Z., Kuschk, P., Reiche, N., Borsdorf, H., Kästner, M., Köser, H., 2012. Comparative evaluation of pilot scale horizontal subsurface-flow constructed wetlands and plant root mats for treating groundwater contaminated with benzene and MTBE. J. Hazard. Mater. 209–210, 510–515. Costa, C.R., Botta, C.M.R., Espindola, E.L.G., Olivi, P., 2008. Electrochemical treatment of tannery wastewater using DSA® electrodes. J. Hazard. Mater. 153 (1–2), 616–627. De Biase, C., Carminati, A., Oswald, S.E., Thullner, M., 2013. Numerical modeling analysis of VOC removal processes in different aerobic vertical flow systems for groundwater remediation. J. Contam. Hydrol. 154, 53–69. Deeb, R.A., Hu, H.-Y., Hanson, J.R., Scow, K.M., Alvarez-Cohen, L., 2001. Substrate interactions in BTEX and MTBE mixtures by an MTBE-degrading isolate. Environ. Sci. Technol. 35 (2), 312–317. Del Bubba, M.D., Checchini, L., Pifferi, C., Zanieri, L., Lepri, L., 2004. Olive mill wastewater treatment by a pilot-scale subsurface horizontal flow (SSF-h) constructed wetland. Annal. Chim. 94, 875–887. Farré, M., Pasini, O., Alonso, M.C., Castillo, M., Barceló, D., 2001. Toxicity assessment of organic pollution in wastewaters using a bacterial biosensor. Anal. Chim. Acta 426, 155–165. Faulwetter, J.L., Gagnon, V., Sundberg, C., Chazarenc, F., Burr, M.D., Brisson, J., Camper, A.K., Stein, O.R., 2009. Microbial processes influencing performance of treatment wetlands: a review. Ecol. Eng. 35, 987–1004. Gagnon, V., Chazarenc, F., Comeau, Y., Brisson, J., 2007. Influence of macrophyte species on microbial density and activity in constructed wetlands. Water Sci Technol. 56 (3), 249–254. Haddaji, D., Bousselmi, L., Saadani, O., Nouairi, I., Ghrabi-Gammar, Z., 2015. Enzymatic degradation of azo dyes using three macrophyte species: Arundo donax, Typha angustifolia and Phragmites australis. Desalin. Water Treat. 53 (4), 1129–1138. Hench, K.R., Bissonnette, G.K., Sexstone, A.J., Coleman, J.G., Garbutt, K., Skousen, J.G., 2003. Fate of physical, chemical, and microbial contaminants in domestic wastewater following treatment by small constructed wetlands. Water Res. 37, 921–927. Herouvim, E., Akratos, C.S., Tekerlekopoulou, A., Vayenas, D.V., 2011. Treatment of olive mill wastewater in pilot-scale vertical flow constructed wetland. Ecol. Eng. 37, 931–939. Imfeld, G., Braeckevelt, M., Kuschk, P., Richnow, H.H., 2009. Monitoring and assessing processes of organic chemicals removal in constructed wetlands. Chemosphere 74, 349–362. Jechalke, S., Vogt, C., Reiche, N., Franchini, A.G., Borsdorf, H., Neu, T.R., Richnow, H.H., 2010. Aerated treatment pond technology with biofilm promoting mats for the bioremediation of benzene, MTBE and ammonium contaminated groundwater. Water Res. 44 (6), 1785–1796. Kadlec, R., Reddy, R., 2000. Temperature effects in treatment wetlands. Water Environ. Res. 73, 543–555. Kadlec, R.H., Wallace, S.D., 2009. Treatment Wetlands, 2nd ed. CRC Press, Boca Raton, FL. Kapellakis, I.E., Paranychianakis, N.V., Tsagkarakis, K.P., Angelakis, A.N., 2012. Treatment of olive mill wastewater with constructed wetlands. Water 4, 260–271. Knight, R.L., Kadlec, R.H., Ohlendorf, H.M., 1999. The use of treatment wetlands for petroleum industry effluents. Environ. Sci Technol. 33 (7), 973–980. Kumaran, P., Paruchuri, Y.L., 1997. Kinetics of phenol biotransformation. Water Res. 31 (1), 11–22. Kurzbaum, E., Zimmels, Y., Kirzhner, F., Armon, R., 2010. Removal of phenol in a constructed wetland system and the relative contribution of plant roots, microbial activity and porous bed. Water Sci. Technol. 62 (6), 1327–1334. Langwaldt, J.H., Puhakka, J.A., 2000. On-site biological remediation of contaminated groundwater: a review. Environ. Pollut. 107, 187–197. Levchuk, I., Bhatnagar, A., Sillanpää, M., 2014. Overview of technologies for removal of methyl tert-butyl ether (MTBE) from water. Sci. Total Environ. 476–477, 415–433.

Nair, C.I., Jayachandran, K., Shashidhar, S., 2008. Biodegradation of phenol. Afr. J. Biotechnol. 7 (25), 4951–4958. Nivala, J., Wallace, S., Headley, T., Kassa, K., Brix, H., 2012. Oxygen transfer and consumption in subsurface flow treatment wetlands. Ecol. Eng. 61 (B), 544–554. Pivetz, B., 2001. Ground water issue: Phytoremediation of contaminated soil and ground water at hazardous waste sites. EPA/540/S-01/500, February 2001. Includes bibliographical references (p. 17–22). Reiche, N., Lorenz, W., Borsdorf, H., 2010. Development and application of dynamic air chambers for measurement of volatilization fluxes of benzene and MTBE from constructed wetlands planted with common reed. Chemosphere 79, 162–168. Rossmann, M., de Matos, A.T., Abreu, E.C., Silva, F.F., Borges, A.C., 2012. Performance of constructed wetlands in the treatment of aerated coffee processing wastewater: removal of nutrients and phenolic compounds. Ecol. Eng. 49, 264–269. Salanitro, J.P., Johnson, P.C., Spinnler, G.E., Maner, P.M., Wisniewski, H.L., Bruce, C., 2000. Field-scale demonstration of enhanced MTBE bioremediation through aquifer bioaugmentation and oxygenation. Environ. Sci. Technol. 34, 4152–4162. Schmidt, T.C., Schirmer, M., Weiß, H., Haderlein, S.B., 2004. Microbial degradation of tert-butyl ether and tert-butyl alcohol in the subsurface. J. Contam. Hydrol. 70, 173–203. Seeger, E.M., Kuschk, P., Fazekas, H., Grathwohl, P., Kaestner, M., 2011. Bioremediation of benzene-, MTBE- and ammonia-contaminated groundwater with pilot-scale constructed wetlands. Environ. Pollut. 159, 3769–3776. Seeger, E.M., Maier, U., Grathwohl, P., Kuschk, P., Kaestner, M., 2013. Performance evaluation of different horizontal subsurface flow wetland types by characterization of flow behavior, mass removal and depth-dependent contaminant load. Water Res. 47, 769–780. Silva, W., Gomes, A., Simões, R., Pascoa, R., Albuquerque, A., Stefanakis, A.I., 2015. A lab-scale Constructed Wetland for wastewater treatment of the cork processing industry. In: 6th International Symposium on Wetland Pollutant Dynamics and Control, WETPOL 2015, York, UK, 13–18 September. Stefanakis, A.I., Tsihrintzis, V.A., 2012. Effects of loading, resting period, temperature, porous media, vegetation and aeration on performance of pilot-scale vertical flow constructed wetlands. Chem. Eng. 181–182, 416–430. Stefanakis, A.I., Akratos, C.S., Gikas, G.D., Tsihrintzis, V.A., 2009. Effluent quality improvement of two pilot-scale, horizontal subsurface flow constructed wetlands using natural zeolite (clinoptilolite). Microporous Mesoporous Mater. 124 (1–3), 131–143. Stefanakis, A.I., Akratos, C.S., Tsihrintzis, V.A., 2011. Effect of wastewater step-feeding on removal efficiency of pilot-scale horizontal subsurface flow constructed wetlands. Ecol. Eng. 37 (3), 431–443. Stefanakis, A.I., Akratos, C.S., Tsihrintzis, V.A., 2014. Vertical Flow Constructed Wetlands: Eco-Engineering Systems for Wastewater and Sludge Treatment. Elsevier Publishing Inc., Amsterdam. Stottmeister, U., Wiebner, A., Kuschk, P., Kappelmeyer, U., Kaestner, M., Bederski, O., Mueller, R., Moormann, H., 2003. Effects of plants and microorganisms in constructed wetlands for wastewater treatment. Biotechnol. Adv. 22, 93–117. Stottmeister, U., Kuschk, P., Wiessner, A., 2010. Full-scale bioremediation and long-term monitoring of a phenolic wastewater disposal lake. Pure Appl. Chem. 82, 161–173. Tang, X., Eke, P.E., Scholz, M., Huang, S., 2009. Processes impacting on benzene removal in vertical-flow constructed wetlands. Bioresour. Technol. 100, 227–234. Tee, H.C., Seng, C.E., Noor, A. Md., Lim, P.E., 2009. Performance comparison of constructed wetlands with gravel- and rice husk-based media for phenol and nitrogen removal. Sci. Total Environ. 407, 3563–3571. Tyroller, L., Rousseau, D.P.L., Sant, S., García, J., 2010. Application of the gas tracer method for measuring oxygen transfer rates in subsurface flow constructed wetlands. Water Res. 44, 4217–4225. USEPA, 2009. List of Drinking Water Contaminants and MCLs, From https://www. epa.gov/ground-water-and-drinking-water/table-regulated-drinking-watercontaminants. USEPA, 2015. Integrated Risk Information System (IRIS), From http://www.epa. gov/iris. Van Afferden, M., Rahman, K.Z., Mosig, P., De Biase, C., Thullner, M., Oswald, S.E., Müller, R.A., 2011. Remediation of groundwater contaminated with MTBE and benzene: the potential of vertical-flow soil filter systems. Water Res. 45, 5063–5074. Vymazal, J., 2007. Removal of nutrients in various types of constructed wetlands. Sci. Total Environ. 38, 48–65. Vymazal, J., 2009. The use constructed wetlands with horizontal sub-surface flow for various types of wastewater. Ecol. Eng. 35, 1–17. WHO, 2011. Guidelines for Drinking-Water Quality, fourth edition. WHO, Switzerland. Wu, Y., Lerner, D.N., Banwart, S.A., Thornton, S.F., Pickup, R.W., 2006. Persistence of fermentative process to phenolic toxicity in groundwater. J. Environ. Qual. 35, 2021–2025.