Fate and distribution of arsenic in a process-designed pilot-scale constructed wetland treatment system

Fate and distribution of arsenic in a process-designed pilot-scale constructed wetland treatment system

Ecological Engineering 68 (2014) 251–259 Contents lists available at ScienceDirect Ecological Engineering journal homepage: www.elsevier.com/locate/...

1MB Sizes 1 Downloads 26 Views

Ecological Engineering 68 (2014) 251–259

Contents lists available at ScienceDirect

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

Fate and distribution of arsenic in a process-designed pilot-scale constructed wetland treatment system Jeffrey P. Schwindaman a , James W. Castle a,∗ , John H. Rodgers Jr. b a b

Department of Environmental Engineering and Earth Sciences, Clemson University, Clemson, SC 29634, USA School of Agricultural, Forest, and Environmental Sciences, Clemson University, Clemson, SC 29634, USA

a r t i c l e

i n f o

Article history: Received 10 August 2013 Received in revised form 29 January 2014 Accepted 29 March 2014 Keywords: Constructed wetland Arsenic Fate and distribution Mass balance Treatment

a b s t r a c t The fate and distribution of arsenic in simulated groundwater was determined in a pilot-scale constructed wetland treatment system (CWTS) designed to promote specific biogeochemical processes for arsenic removal. Two CWTS series were designed to promote co-precipitation and sorption of arsenic with iron oxyhydroxides under oxidizing conditions, and two series were designed to promote precipitation of arsenic with sulfide and co-precipitation of arsenic with iron sulfide under reducing conditions. Measured conditions in the CWTS were within ranges favorable for the targeted processes. Arsenic removal was significantly greater (˛ = 0.05) in an oxidizing series amended with zero-valent iron (ZVI) than in the other series, with removal extents, efficiencies, and rate coefficients ranging from 6 to 79 ␮g L−1 , 40 to 95%, and 0.13 to 0.77 d−1 , respectively. The majority of inflow arsenic retained in the first reactor of each series partitioned to the sediment (88–99%), while the remainder partitioned to Typha latifolia. A greater percentage of inflow arsenic was retained in the sediment of the first reactor of the two oxidizing series (20 and 13%) than in the first reactor of the two reducing series (6 and 7%). Addition of ZVI enhanced arsenic removal from the aqueous phase in both oxidizing series and reducing series and increased the percentage of inflow arsenic partitioned to sediment. A vertical concentration gradient developed over time in the sediment, with 74–85% of sediment-bound arsenic accumulated in the upper 6 cm and the remaining percentage below 6 cm. Results from this study demonstrate that a CWTS can decrease the concentration of arsenic in simulated groundwater to below the World Health Organization (WHO) drinking water quality guideline of 10 ␮g L−1 primarily by transferring arsenic from the aqueous phase to the sediment. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Constructed wetland treatment systems (CWTSs) can offer a low-cost, low-maintenance approach for treating arseniccontaminated water. CWTSs can be designed to target specific biogeochemical processes (i.e. sorption, precipitation, coprecipitation, and volatilization) to remove constituents of concern (COCs) from the aqueous phase or transform COCs into less bioavailable forms (Dorman et al., 2009; Pham et al., 2011; Rodgers and Castle, 2008). Arsenic commonly exists in natural waters as arsenite [As (III)] or arsenate [As (V)] (Francesconi and Kuehneit, 2002). Arsenite is the predominant species of arsenic in groundwater and is more

∗ Corresponding author at: 340 Brackett Hall, Clemson, SC 29634, USA. Tel.: +1 864 656 5015. E-mail address: [email protected] (J.W. Castle). http://dx.doi.org/10.1016/j.ecoleng.2014.03.049 0925-8574/© 2014 Elsevier B.V. All rights reserved.

toxic to humans than arsenate (Sharma and Sohn, 2009). Under reducing conditions and near-neutral pH, arsenite most often occurs as H3 AsO3. Under oxidizing conditions and near-neutral pH, arsenate often occurs as H2 AsO4 − and HAsO4 2− (Henke and Hutchison, 2009). The form of arsenate as an oxyanion allows for electrostatic interaction with constituents in soil and sediment, such as iron oxyhydroxides. Arsenic removed from the aqueous phase in a CWTS can be retained in sediment, plants, or lost due to volatilization (Rahman et al., 2011). Arsenic can be transferred to the sediment of a CWTS via precipitation, co-precipitation, and sorption (Lizama et al., 2011). Precipitation and co-precipitation can remove arsenic from the aqueous phase by direct formation of insoluble arsenic complexes or by incorporation of trace amounts of arsenic into newly formed insoluble compounds (Henke and Hutchison, 2009). Under reducing conditions, dissimilatory sulfate reduction results in production of hydrogen sulfide, which can react with dissolved arsenic to precipitate insoluble As–S complexes (Cohen, 2006; Kirk et al.,

252

J.P. Schwindaman et al. / Ecological Engineering 68 (2014) 251–259

Table 1 Targeted biogeochemical conditions to promote processes for arsenic removal from arsenic-contaminated water within wetland cells of the pilot-scale constructed wetland treatment system. Removal process Co-precipitation and sorption with Fe–oxyhydroxides Precipitation of As–S and co-precipitation of As with Fe–S a b c d

Sediment redox potential (mV) a

Oxidizing Eh >−50 Reducingd Eh −50 to −250

Dissolved oxygen (mg L−1 )

pH (S.U.)

≥2 ≤2

4–9b 5–8c

Buddhawong et al. (2005). Cheng et al. (2009). Lizama et al. (2011). Rahman et al. (2011).

2010; Lizama et al., 2011) or co-precipitate arsenic with insoluble iron-sulfide minerals (Ford et al., 2006; Henke, 2009; Paul et al., 2009). Oxidation of reduced iron can result in formation of insoluble iron oxyhydroxides (e.g. ferrihydrite, goethite, etc.) into which trace amounts of arsenic can be incorporated (Henke and Hutchison, 2009). Arsenic also can be removed from the aqueous phase in a CWTS by sorption to iron oxyhydroxides or to organic detritus (Lizama et al., 2011; Sundberg et al., 2006). Plants can retain arsenic in a CWTS via sorption to roots and submerged shoots, as well as translocation to emergent shoots and tips (An et al., 2011; Blute et al., 2004; Sundberg-Jones and Hassan, 2007). Many wetland plants, including Typha latifolia (broadleaf cattail), translocate oxygen from the atmosphere to the rhizosphere via radial oxygen loss from roots (Doyle and Otte, 1997; Li et al., 2011). This process modifies redox chemistry around the roots and promotes precipitation of iron oxyhydroxides. Ferric iron plaque, enriched in arsenic, has been observed on roots of T. latifolia in wetland sediments and is possibly the primary mechanism by which arsenic is sequestered by wetland plants (Blute et al., 2004). This hypothesis is supported by studies that found approximately an order of magnitude greater mass of arsenic associated with wetland plant roots compared with shoots (Buddhawong et al., 2005; Dushenko et al., 1995). Previous studies evaluated performance of CWTSs designed to remove arsenic in terms of removal extents, removal efficiencies, and first-order removal rate coefficients (Dorman et al., 2009; Eggert et al., 2008; Spacil et al., 2011), but did not consider fate and partitioning of arsenic in CWTSs. Therefore, the objectives of this study were to (1) measure biogeochemical conditions in pilot-scale CWTS reactors designed to promote oxidation processes and reactors designed to promote reduction processes for removal of arsenic from arsenic-contaminated water; (2) compare arsenic removal from the aqueous phase between oxidizing reactors and reducing reactors; and (3) determine the fate and distribution of arsenic in the CWTS. 2. Materials and methods 2.1. Pilot-scale constructed wetland system and measurement of conditions Pilot-scale CWTSs were designed and constructed to produce ranges of conditions (sediment oxidation–reduction potential, dissolved oxygen concentration, and pH) that promote specific biogeochemical processes of arsenic removal from the aqueous phase (Table 1). Two CWTS reactor series were designed to promote co-precipitation and sorption of arsenic with iron oxyhydroxides under oxidizing conditions (“oxidizing series”), and two series were designed to promote precipitation of arsenic with sulfide and co-precipitation of arsenic with iron sulfide under reducing conditions (“reducing series”) (Fig. 1). All four CWTS series were built and operated in a climate-controlled greenhouse in Clemson, SC. Each series consisted of four treatment reactors, with each reactor contained in a 265-L (70-gal) Rubbermaid® utility tank

(92-cm long by 74-cm wide by 61-cm deep). Each reactor in both the reducing and oxidizing series, with the exception of the third reactor in each of the two oxidizing series, contained: (1) 30-cm of river sand from 18-Mile Creek in Clemson, SC, (2) water to a depth of 25 cm, and (3) approximately 20 T. latifolia plants harvested from an aquaculture pond in Clemson, SC. The third reactor in each of the two oxidizing series, designed to promote increased atmospheric oxygen diffusion, contained a 50-cm thickness of approximately 3-cm diameter granitic gravel, water to a depth of 5 cm, and was unvegetated. Water levels in the reactors remained constant during the experiment. In the two reducing series, the upper 5 cm of sediment in each reactor was amended with pelletized gypsum (1% v/v) as a source of sulfate for dissimilatory sulfate reduction and with hay and shredded hardwood mulch (5% v/v) to provide a nutrient source for sulfate-reducing bacteria (SRB). One oxidizing series and one reducing series were amended with zero-valent iron filings (ZVI; Peerless Metal Powders & Abrasive, Detroit, MI) by distributing 20 g per reactor by hand into the water column approximately every 14 days. Simulated arsenic-contaminated groundwater was formulated by mixing chemical constituents with municipal water in a 5678-L (1500-gal) detention basin. Use of simulated water provides greater experimental control over water characteristics and eliminates the economic burden of acquiring, shipping, and storing large volumes of actual groundwater. The simulated water, based on composition of actual arsenic-contaminated groundwater in the Lakshmipur District of Bangladesh (Kinniburgh and Smedley, 2001), was formulated to contain 0.30 mg L−1 Astotal (0.18 mg L−1 as As III and 0.12 mg L−1 as As V), 55 mg L−1 Ca, 50 mg L−1 Mg, 180 mg L−1 Na, 13 mg L−1 K, 220 mg L−1 HCO3 , 220 mg L−1 Cl, 18 mg L−1 SO4 , 15 mg L−1 Si, 1.2 mg L−1 P, and 3.0 mg L−1 Fe. As III was added as As2 O3 and As V as As2 O5 . Concentrations of chemical constituents in groundwater of the Lakshmipur District are within the range in groundwater of other districts in central and southern Bangladesh and are therefore considered representative of arsenic-contaminated groundwater from the shallow Holocene aquifer, which is the source of most drinking water in Bangladesh. pH of the simulated water was 7.1, and conductivity was 890 (␮S cm−1 ). Conductivity of the simulated water was measured during the experiment to confirm consistency of total cation and anion concentrations and not as a parameter influencing the targeted biogeochemical processes in wetland cells. Water was transferred from the detention basin into the first reactor of each series by a piston pump (FMI® QG400) at a flow rate of 128 mL min−1 , resulting in a nominal hydraulic retention time (HRT) of 24 h per reactor or 96 h per series. Reactors were connected by PVC pipe fittings located at 4 cm below the top of each reactor. Reactors were arranged with decreasing elevation from the first to fourth reactor in each series to induce gravity flow. Dissolved oxygen (DO) concentration, pH, and sediment oxidation–reduction (redox) potential were measured bi-weekly in each reactor. DO concentration and pH were measured using YSI® (model 55) and Orion® (model A325) field instruments, respectively. To measure sediment redox potential, one platinum-tipped

J.P. Schwindaman et al. / Ecological Engineering 68 (2014) 251–259

253

Fig. 1. Schematic diagram of pilot-scale constructed wetlands for treatment of arsenic-contaminated water. Series 1 and 2 were designed to promote oxidizing conditions, and series 3 and 4 were designed to promote reducing conditions. Series 1 and 4 were amended biweekly with 20 g ZVI per reactor. The third reactor in each of the two oxidizing series was unplanted and contained a 50-cm thickness of granitic gravel.

electrode was installed approximately 2 cm into the sediment in the inflow area and one in the outflow area of each reactor. Based on results of previous studies (e.g. Chague-Goff, 2005; Knox et al., 2006, 2010; Ye et al., 2001), 2 cm is considered the optimal depth to install the electrodes for measuring conditions representative of the biogeochemical treatment processes. Electrodes remained in situ for the duration of the experiment. Sediment redox potential was measured using a GDT-11 multi-meter connected to in situ electrodes and an Accumet® calomel reference electrode (Faulkner et al., 1989). Statistical differences were determined by ANOVA and post hoc t-tests with ˛ = 0.05 using Microsoft EXCEL. Tests were parametric. Normality of data was assessed using the Anderson–Darling Test. Data sets with p-values >0.05 were considered to be normally distributed and deemed suitable for parametric statistical tests.

plasma mass-spectrometry (ICP-MS) (Thermo Scientific X Series) according to EPA Method 200.8 (USEPA, 1994a). Arsenic removal extent, removal efficiency, and removal rate coefficient were determined for each CWTS series. Removal extent is defined as the concentration of arsenic in the outflow (␮g L−1 ). Removal efficiency was calculated using Eq. (1): removal efficiency (%) =

[C]0 − [C] × 100 [C]0

(1)

where [C]0 is inflow concentration (␮g L−1 ) and [C] is outflow concentration (␮g L−1 ). First-order rate kinetics are often used to model removal of COCs in a CWTS (Eggert et al., 2008; Horner et al., 2012; Lizama et al., 2011; Wong et al., 2006). Removal rate coefficient (k, day−1 ) was calculated using Eq. (2): k=

− ln([C]/[C]0 ) HRT

(2)

2.2. Comparison of arsenic removal between oxidizing and reducing reactors

where HRT (day) is the nominal hydraulic retention time (i.e. the time between sampling series inflow and outflow).

Arsenic concentrations were measured in samples from inflow to each series and outflow from each reactor. Arsenic removal was determined from the measured concentrations and compared between oxidizing and reducing reactors. Flow rate from each piston pump was measured prior to each sampling using a graduated cylinder and stopwatch. Flow rates remained consistent throughout the experiment. Aqueous samples were collected bi-weekly from 8/31/12 to 3/15/13 during 14 sampling periods from each treatment reactor sequentially according to the HRT (24 h per reactor). For example, samples of inflow water entering the first reactor were collected at t = 0 h, samples of outflow water from the first reactor were collected at t = 24 h, from the second reactor at t = 48 h, from the third reactor at t = 72 h, and from the fourth reactor at t = 96 h. In this way, a single volume of water was theoretically sampled as it passed through a series. Samples were collected in acid-washed 50-mL polypropylene centrifuge tubes from sampling ports between each reactor. Twenty-five milliliters of sample were centrifuged for 20 min at 8000 rpm. 1 mL of supernatant was transferred gravimetrically to an acid-washed 15-mL polypropylene centrifuge tube and brought to a volume of 10 mL with 2% trace metal grade nitric acid solution. Samples were analyzed for arsenic using inductively-coupled

2.3. Fate and distribution of arsenic Sediment cores and plant samples were collected on day 1 of the experiment (8/17/12, prior to introduction of arsenic on the same day), day 81 (11/6/12), day 141 (1/5/13), and day 188 (2/21/13). Using a 1.91-cm diameter coring device, one 30-cm long sediment core representing the entire sediment thickness was collected from the front, one from the middle, and one from the rear of the first reactor in each series. Each core was sectioned into 6-cm intervals, and the three samples from each interval were combined and homogenized. This procedure resulted in 5 sediment samples from each series, for a total of 20 samples. One T. latifolia plant was collected from the front and one from the back of the first reactor in each series. The number of T. latifolia plants in the first reactor of each series was recorded at the time of sampling. Plants were rinsed with deionized water and sectioned with stainless steel shears into roots, submerged shoots, emergent shoots, and tips. Shears were acid-washed and rinsed with deionized water prior to sectioning and rinsed with deionized water between collections of plant samples. Approximately 10 g of each plant sample and 30 g of each sediment sample (wet weight) were dried for 8 h at 100 ◦ C. The dry bulk density of each sediment sample was determined

254

J.P. Schwindaman et al. / Ecological Engineering 68 (2014) 251–259

Table 2 Measured sediment redox potential (mV) for each sampling date. Values outside targeted range (Table 1) are indicated by (*). Series

Reactor

1 1 1 1 2 2 2 2 3 3 3 3 4 4 4 4

1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4

8/31 2012

9/9 2012

9/26 2012

10/14 2012

33 −5 255 66 6 35 423 35 −142 *8.5 −171 *−24 −120 110 *21 *−21

*−52 −5 379 −2 −23 6 407 *−59 −189 *−35 −192 −189 −119 −99 −54 *−48

69 −48 206 44 −5 36 152 *−56 −181 *93 −191 −212 −60 −84 *4 −62

23 *−94 209 *−77 2 *−62 47 *−62 −194 *76 −144 −200 −64 −171 *−7 −131

10/28 2012 −6 *−101 191 4 −7 −82 27 −47 −155 *−44 −189 −216 −56 −86 *7 −147

11/13 2012 18 *−134 379 *−122 4 *−51 323 *−76 −193 *−3 *6 −108 −120 −150 *53 −93

gravimetrically. Sediment and plant samples were stored at 4 ◦ C prior to chemical analysis. 0.5 g each of dried sediment and plant samples were placed in Teflon® MARS microwave digestion tubes (CEM Corporation) with 10 mL trace-metal grade nitric acid (67%) (Fisher Scientific) and digested using EPA method 3051 (USEPA, 1994b). The digestate was transferred into a 50-mL centrifuge tube, and the volume brought to 25 mL with deionized water. Samples were centrifuged, diluted, and analyzed for arsenic using the same procedure that was used for aqueous samples. Translocation factor (TF), which is the ratio of arsenic concentration in above-ground plant tissues (i.e. shoots and tips) to arsenic concentration in plant roots, was calculated using Eq. (3) (Li et al., 2011; Sundberg-Jones and Hassan, 2007): TF =

[As]above ground [As]roots

(3)

where [As]above ground is arsenic concentration in above-ground plant tissues (sum of concentrations in shoots and tips; mg kg−1 plant dry weight) and [As]roots is arsenic concentration in the roots (mg kg−1 plant dry weight). Aqueous bioconcentration factor (BCF), which is the ratio of arsenic concentration in plant tissue (i.e. roots, shoots, and tips) to arsenic concentration in the aqueous phase, was calculated using Eq. (4) (Sundberg-Jones and Hassan, 2007): BCF =

[As]plant [As]water

(4)

where [As]plant is arsenic concentration in plant tissue (sum of arsenic concentrations in roots, shoots, and tips; mg kg−1 plant dry weight) and [As]water is arsenic concentration in the aqueous phase (mg L−1 ). One grab sample from the upper 2 cm of sediment was collected from the front, one from the middle, and one from the back of the first reactor of each series on day 14 of the experiment (8/31/12), day 94 (11/19/12), day 140 (1/4/13), and day 188 (2/21/13). Samples were scooped with an acid-washed metal spatula into a 50-mL centrifuge tube and sealed underwater. The three samples from each reactor were composited and analyzed for acid volatile sulfide (AVS) using the modified diffusion method (Leonard et al., 1996). AVS is defined as the sulfide extracted from sediment by 1-N HCl (Di Toro et al., 1992; Leonard et al., 1996) and interpreted as the reactive fraction of sulfide available to bind metals (Di Toro et al., 1992; Keon et al., 2001; Wilkin and Ford, 2002). Sulfide was measured using an ion-selective electrode to determine AVS concentration. Previous studies (Dorman et al., 2009; Horner et al., 2012; Rahman et al., 2011; Rousseau et al., 2004; Spacil et al., 2011; Wong

11/29 2012 *−134 *−124 323 52.5 49 −42 60 −47 −80 *29 *48 −68 *−42 −88 *35 −161

12/16 2012 *−150 −20 427 −42 132 33 210 −37 −151 *107 *17 −50 −53 −88 *28 −233

1/4 2013

1/18 2013

2/1 2013

2/15 2013

3/1 2013

3/15 2013

*−91 −10 421 *−80 105 53 154 *−108 −203 *55 *−47 −137 *−44 −82 *−10 −200

*−211 −44 341 *−114 69 21 174 −32 −185 *83 −62 −118 −51 −125 *5 −167

*−119 *−71 497 *−77 126 −49 116 −42 −168 *88 *−35 −60 *−38 −75 *9 −200

46 *−61 427 *−138 209 13 224 −32 −185 *62 *19 −56 *−35 −74 *−31 −189

*−215 −8 384 *−96 −25 6 245 −45 −173 *50 *−1 *−30 *−12 −173 −65 −241

*−225 −42 359 45 *−68 24 372 *−59 −180 *51 *15 −86 *−21 −109 −63 −229

et al., 2006) have shown that the greatest percentage removal of inflow mass of a COC most often occurs in the first reactor of a CWTS series; therefore, the first reactor of each series was chosen as the system for a chemical mass balance Eq. (5). Min = Mout + Msed + Mplant + Munacc

(5)

where Min is mass of arsenic that entered the system as inflow (mg) during time t (duration of the experiment over which the mass balance is applicable), Mout is mass of arsenic that exited the system as outflow (mg) during time t, Msed is mass of arsenic retained in sediment (mg) after time t, Mplant is mass of arsenic retained in T. latifolia (mg) after time t, and Munacc is mass of arsenic unaccountable (i.e. a loss or gain from mass balance calculation) (mg). The Msed term was calculated from arsenic concentrations from sediment samples according to Eq. (6): Msed =

n 

(Csed,i sed,i Vsed,i )

(6)

i=1

where Csed,i is arsenic concentration in a sediment sample from depth interval i (mg arsenic g−1 sediment dry weight) after time t, sed,i is dry bulk density of a sediment sample from interval i (g cm−3 ), and Vsed,i is volume of sediment within interval i, based on dimensions of the reactor and thickness of the interval i (cm3 ). Mplant was calculated from the arsenic concentration in T. latifolia according to Eq. (7): Mplant =

n 

(mplant,i Cplant,i nplant )

(7)

i=1

where mplant ,i is dry mass of plant tissue (g), Cplant ,i is arsenic concentration in plant tissue (mg arsenic g−1 plant dry weight) after time t, and nplant is number of plants per reactor. Substituting Eqs. (6) and (7) into Eq. (5) and expressing mass in and mass out of the system as a mass loading rate (Qin Cin and Qout Cout , respectively) results in Eq. (8): Qin Cin t = (Qout Cout t) +

+

 n 

 n 



(mplant,i Cplant,i nplant )

i=1



(Csed,i sed,i Vsed,i )

+ Munacc

(8)

i=1

where Qin is volumetric flow rate of water entering the system as inflow (L d−1 ), Qout is volumetric flow rate of water leaving the system as outflow (L d−1 ), Cin is inflow arsenic concentration (mg L−1 ),

J.P. Schwindaman et al. / Ecological Engineering 68 (2014) 251–259

136.4 110.5 151.9 86.0 38.1 68.3 42.6 123.4 83.4 68.7 117.1 34.3 120.4 68.2 0.03 0.23 0.36 0.07 0.11 0.05 0.11 0.03 0.01 0.02 b 0.00 0.02 0.04 b 0.00 128.1 103.9 149.4 98.2 79.2 113.5 85.7 165.1 117.8 146.0 148.5 77.8 151.0 223.0 0.19 0.39 0.02 0.08 0.16 0.17 0.14 0.06 0.06 0.06 b 0.00 b 0.00 0.02 0.07 65.1 50.0 113.0 61.6 49.2 60.5 63.9 136.6 97.3 123.1 161.8 79.0 130.6 126.8 a

b

Mean inflow concentration. No decrease in concentration from inflow to outflow.

0.37 0.55 0.19 0.30 0.63 0.77 0.71 0.45 0.31 0.45 0.18 0.14 0.13 0.28 137.7 241.2 183.7 106.0 108.1 127.4 117.4 180.2 121.8 155.6 132.6 79.1 161.4 172.5 8/31/12 9/9/12 9/26/12 10/14/12 10/28/12 11/13/12 11/29/12 12/16/12 1/4/13 1/18/13 2/1/13 2/15/13 3/1/13 3/15/13

32.2 25.2 58.9 27.2 7.6 5.5 6.9 29.2 33.3 24.1 60.0 45.4 78.7 56.5

Rate coefficient (d−1 ) Removal extent (␮g L−1 ) Removal extent (␮g L−1 )

Rate coefficient (d−1 )

Removal extent (␮g L−1 )

Rate coefficient (d−1 )

Series 3 Series 2 Inflow concentration (␮g L−1 )

Series 1 a

Arsenic concentration decreased from the inflow to outflow of all pilot-scale CWTS series. Arsenic removal was significantly greater (˛ = 0.05) in the oxidizing series amended with ZVI than in any other series, with removal extent, efficiency, and rate coefficient ranging from 6 to 79 ␮g L−1 , 40 to 95%, and 0.13 to 0.77 d−1 respectively (Table 3). In the oxidizing series not amended with ZVI, removal extent, efficiency, and rate coefficient were 49–162 ␮g L−1 , 0–79%, and 0.00–0.39 d−1 , respectively. Outflow concentration of arsenic was less than the WHO drinking water quality guideline of 10 ␮g L−1 in the oxidizing series amended with ZVI for 3 of the 14 sampling periods. In the reducing series amended with ZVI, removal extent and removal efficiency were 34–152 ␮g L−1 and 0–69%, respectively, compared to 78–223 ␮g L−1 and 0–76%, respectively, in the reducing series not amended with ZVI. The addition of ZVI significantly improved the removal extent and efficiency of oxidizing series (p = 1.7 × 10−6 and 1.2 × 10−7 , respectively) and reducing series (p = 3.9 × 10−3 and 1.7 × 10−2 , respectively). Enhanced arsenic removal in the series amended with ZVI is attributed to ZVI providing a source of iron for co-precipitation and sorption of arsenic with iron oxyhydroxides in the oxidizing series and a source of iron for co-precipitation of arsenic with iron sulfide in the reducing series.

Sampling period

3.2. Arsenic concentration and removal

Table 3 Inflow concentrations, outflow concentrations (removal extents), and rate coefficients for each treatment series.

Sediment redox potential, DO concentration, and pH were within the targeted ranges (Table 1) for the majority of measurements in all 4 series (Table 2, Supplementary Tables S1 and S2). In some reactors (e.g. reactor 4 in series 4, Table 2), sediment redox potential decreased during the experiment, which may have been due to the accumulation of organic detritus in the vicinity of the redox probes. In reducing series 3 and 4 mean sediment redox potential was lower (−77 and −79 mV, respectively) and mean DO concentration less (2.0 and 1.8 mg L−1 , respectively) compared to oxidizing series 1 and 2 (45 and 51 mV, respectively, and 6.6 and 7.4 mg L−1 ). Lower sediment redox potential and lower DO concentration in series 3 and 4 compared to series 1 and 2 are attributed to the consumption of oxygen by aerobic microbes during biodegradation of organic matter in the reducing series. Mean sediment redox potential (298 and 201 mV, respectively) and mean DO concentration (9.0 and 11.6 mg L−1 , respectively) were significantly greater (˛ = 0.05) in the third reactors of oxidizing series 1 and 2 than in the other reactors of either series. The greater redox potentials and higher DO concentrations in the third reactors are attributed to lower organic matter content in the substrate (unplanted granitic gravel) compared to the other reactors (river sand planted with T. latifolia) and to greater atmospheric oxygen diffusion due to shallower water depth (5 cm) compared to the other reactors (25 cm). In oxidizing series 1 and 2 pH values in the planted reactors (range of 6.9–8.4 and 6.9–8.7, respectively; Supplementary Table S2) were less than in the unplanted third reactors (7.1–9.6 and 7.7–9.9, respectively). This difference in pH is interpreted as caused by organic acids in the planted reactors. New growth of T. latifolia was observed in all planted reactors over the course of the 188-day experiment.

Series 4

3.1. Conditions in constructed wetland system

Removal extent (␮g L−1 )

b

3. Results and discussion

0.00 0.20 0.12 0.08 0.29 0.16 0.23 0.09 0.11 0.22 0.03 0.20 0.11 0.23

Rate coefficient (d−1 )

Cout is outflow arsenic concentration (mg L−1 ), and t is time (days) over which the mass balance is applicable.

255

256

J.P. Schwindaman et al. / Ecological Engineering 68 (2014) 251–259

Table 4 Distribution of arsenic in the first treatment reactor of each series at the conclusion of the experiment (188 days). Values are mg arsenic (and percentage of total inflow arsenic). Series

Inflow

Sediment

Plants

Outflow

Unaccountable

1 2 3 4

4128 (100.0%) 4294 (100.0%) 4766 (100.0%) 4949 (100.0%)

809 (19.6%) 541 (12.6%) 270 (5.7%) 325 (6.6%)

23.2 (0.6%) 9.8 (0.2%) 3.1 (0.1%) 45.5 (0.9%)

2904 (70.3%) 3662 (85.3%) 4695 (98.4%) 3794 (76.7%)

392 (9.5%) 80.7 (1.9%) −202 (−4.2%) 784 (15.8%)

3.3. Fate and distribution of arsenic Over the duration of the experiment (188 days), the total inflow mass of arsenic (Min , Eq. (5)) ranged from 4128 to 4949 mg among the four series, with the percentage of total inflow mass of arsenic removed by each series ranging from 16 to 76%. In the first reactor of each series, 2–30% of inflow arsenic was retained in sediment (Msed , Eq. (5)), retained in plants (Mplant , Eq. (5)), or was unaccountable (Munacc , Eq. (5)), and the remainder was passed from outflow of the first reactor into the second reactor (Mout , Eq. (5)) (Table 4). The majority of inflow arsenic retained in the first reactor of each series (Msed + Mplant , Eq. (5)) partitioned to the sediment (88–99%), with the remainder partitioned to T. latifolia. Mass of arsenic in sediment increased over the duration of the experiment from a range of 52–89 mg among the four series at the beginning of the experiment to a range of 270–809 mg among the four series at the end of the experiment. Initial arsenic concentrations in sediment, determined prior to introduction of arsenic into the system, were within a narrow range (0.2–0.7 mg arsenic kg−1 sediment dry weight) (Fig. 2). Over time, however, a concentration depth profile developed with a significant decrease (˛ = 0.05) in arsenic concentration with depth (Fig. 2). At the conclusion of the experiment, the upper 6 cm of sediment (upper sediment core interval) accounted for 74–85% of total sediment-bound arsenic, with the remainder at 6–30 cm depth. The vertical arsenic concentration gradient is attributed to the majority of arsenic removal having occurred upon initial contact with sediments at the sediment–water interface, with decreasing removal as water moved downward from the sediment–water interface. Conditions in the sediment can also play a role in the vertical distribution of arsenic. For example, in a previous study of laboratory-scale constructed wetlands, Buddhawong et al. (2005) found sediment redox potential at the top of the reactor bed to be approximately 250 mV greater than at the bottom. A decrease with depth in sediment redox potential of this magnitude in an oxidizing reactor could mobilize arsenic through the reductive dissolution of iron oxyhydroxides, and in a reducing reactor could result in redox conditions unfavorable for dissimilatory sulfate reduction (<−250 mV). Mass of arsenic retained in sediment (Msed , Eq. (5)) was greater in oxidizing series (809 and 541 mg in series 1 and 2, respectively) than in reducing series (270 and 326 mg in series 3 and 4, respectively). A greater percentage of inflow arsenic was removed in the first reactor of the oxidizing series amended with ZVI and the first reactor of the reducing series amended with ZVI (30 and 23%, respectively) than in the first reactor of the unamended oxidizing and unamended reducing series (15 and 2%, respectively). The first reactor of the oxidizing series amended with ZVI (series 1) removed the greatest percentage of inflow arsenic (30%) compared to the first reactor of other series (2–23%). In both oxidizing series, sediment redox potential and DO concentration were favorable for precipitation of iron oxyhydroxides (Table 2, Supplementary Table S1), which can remove arsenic from the aqueous phase via co-precipitation and sorption (Cheng et al., 2009), but not for dissimilatory sulfate reduction. In the current pilot-scale study, the

removal rate coefficient for series 1 (0.13–0.77 d−1 ) was within the range of removal rate coefficients from previous bench-scale batch reactor studies with arsenic and ZVI under oxidizing conditions (Bang et al., 2005; Su and Puls, 2001), in which pseudo-firstorder rate coefficients (k = −d[As]/dt) ranged from 0.09 to 0.84 d−1 depending on the speciation of arsenic, pH, and type of ZVI. The concentration of arsenic in the upper 6 cm of sediment was greater in the series amended with ZVI (23.1 mg As kg−1 sediment) than in the unamended series (12.8 mg As kg−1 sediment). ZVI provided a source of iron for precipitation of iron oxyhydroxides under oxidizing conditions and therefore improved arsenic removal in series 1 compared to the unamended series. In both reducing series, sediment redox potential and DO concentration were favorable for dissimilatory sulfate reduction necessary for precipitation of arsenic sulfide and co-precipitation of arsenic with iron sulfide (Table 2, Supplementary Table S1), but not for the precipitation of iron oxyhydroxides. Throughout the experiment, AVS concentration was greater by approximately two orders of magnitude in reducing series 3 and 4 (100–643 mg L−1 ) than in oxidizing series 1 and 2 (0–8 mg L−1 ). Greater AVS concentrations in reducing series compared to oxidizing series are attributed to dissimilatory sulfate reduction in the reducing series. AVS concentrations in the reducing series amended with ZVI (range from 101 to 643 mg L−1 ) were greater than in the unamended reducing series (100–234 mg L−1 ), which is attributed to the ZVI serving as an electron donor for dissimilatory sulfate reduction (Karri et al., 2005). Between 1 and 12% of inflow arsenic retained in the first reactor of each series (Msed + Mplant , Eq. (5)) partitioned to T. latifolia. Concentrations of arsenic in T. latifolia ranged from 6.0–15.5 mg As kg−1 plant (dry weight) among the four series initially, increased after 81 days to 65.1–200.2 mg As kg−1 plant, and decreased by the conclusion of the experiment (188 days) to 6.4–62.4 mg As kg−1 plant (Fig. 3). At the conclusion of the experiment, the roots and submerged shoots accounted for 87–97% of total plant-bound arsenic, with the remainder of plant-bound arsenic in the emergent shoots and tips. Throughout the experiment, arsenic concentrations were greater in roots than in the above-ground plant tissues (TF ≤ 1, Eq. (3)), indicating little translocation of arsenic from roots to shoots (Table 5). Sequestration of arsenic in the roots and submerged

Table 5 Arsenic bioconcentration factors (BCF, Eq. (4)) and translocation factors (TF, Eq. (3)) for Typha latifolia in the first treatment reactor of each series (dry weight). Series

8/31/12

11/6/12

1/5/13

2/21/13

BCF (L kg−1 ) 1 2 3 4

84.7 77.0 49.7 53.4

1616 1593 542 1668

886 871 915 1530

320 172 53.7 520

TF 1 2 3 4

0.9 0.7 0.9 1.0

0.3 0.4 0.5 8.6

1.3 2.0 0.2 6.4

0.7 0.1 0.5 0.5

J.P. Schwindaman et al. / Ecological Engineering 68 (2014) 251–259

257

Fig. 2. Vertical distribution of arsenic in sediment in the first reactor of each series during the 8/17/12 (A), 11/6/12 (B), 1/5/13 (C), and 2/21/13 (D) sampling period. The initial concentrations of arsenic in sediment were within a narrow range among intervals (A), compared to later sampling dates (B–D), in which arsenic concentrations were significantly greater (˛ = 0.05) from 0 to 6 cm than from 6 to 30 cm.

Fig. 3. Distribution of arsenic in Typha latifolia in the first reactor of each series during the 8/17/12 (A), 11/6/12 (B), 1/5/13 (C), and 2/21/13 (D) sampling period. 8/17/12 is day 0 of the experiment, before arsenic was introduced, showing background concentration (scale expanded to show differences in concentrations between plant tissues).

258

J.P. Schwindaman et al. / Ecological Engineering 68 (2014) 251–259

shoots of T. latifolia is preferable to translocation to emergent shoots and tips in terms of mitigating risk of exposure to herbivores that could feed on the exposed portion of wetland plants. In all plant samples, arsenic concentration was greater in the plant tissue than in the water (BCF > 1 L kg−1 , Eq. (4)), indicating partitioning of arsenic from water to T. latifolia. BCF decreased in all series between 11/6/12 and 2/21/13 (Table 5). Plant shoots were green with new shoots emerging in each reactor during the 11/16/12 sampling period, whereas plant shoots were brown and dormant during the 2/21/13 sampling period. These observations are consistent with greater arsenic uptake by T. latifolia during periods of active growth. Arsenate [As (V)] shares the same tetrahedral coordination as phosphate and can compete with phosphate for plant uptake (Carbonell et al., 1998; Li et al., 2011). Arsenite [As (III)], also tetrahedrally coordinated, is similar in size to silicic acid (H2 SiO3 ) and is available for plant uptake as a substitute for silicon (Zhao et al., 2009). Mass of arsenic associated with T. latifolia (Mplant , Eq. (5)) was greater in series amended with ZVI (23 and 46 mg in series 1 and 4, respectively), than in series not amended with ZVI (10 and 3 mg in series 2 and 3, respectively). Both arsenate and arsenite have an affinity for iron oxyhydroxides, which have been observed as plaques associated with T. latifolia roots (Blute et al., 2004). We hypothesized that amendment of ZVI to series 1 and 4 would increase the concentration of arsenic associated with T. latifolia roots by promotion of iron oxyhydroxide plaque formation. Interestingly, the arsenic concentration associated with the roots was not consistently greater in series 1 and 4 (range from 5.5 to 155.0 and 3.2 to 40.4 mg As kg−1 root, respectively) than in unamended series 2 and 3 (9.2–138.8 and 3.2–93.8 mg As kg−1 root, respectively). The minimal retention of arsenic in plants is consistent with other mass balance studies (Singhakant et al., 2009; Ye et al., 2003). This finding could lead to the conclusion that plants play only a minor role in the CWTS; however, the wetland plants have several benefits beyond direct arsenic uptake. Detritus from decaying plant material provides organic matter for dissimilatory sulfate reduction, binding sites for SRB, and sorption sites for arsenic. In addition, plants likely prevent resuspension of sediment by acting as a wind break. These effects may help explain why previous studies (Rahman et al., 2011; Singhakant et al., 2009) have observed greater arsenic removal in planted wetlands than unplanted wetlands. Plants also provide a source of detritus for accretion of wetland sediment. There is sometimes concern over arsenic accumulation in sediment (Christophoridis et al., 2009; Liber et al., 2011); however, sediment added internally to a wetland (accumulation of detritus and deposition of suspended solids) can result in sediment accretion ranging from mm year−1 to cm year−1 (Kadlec and Wallace, 2009), thus diluting concentration of COCs, such as arsenic, in the sediment. Between 2 and 16% of inflow arsenic was unaccountable (Table 4). It is possible for inorganic arsenic to undergo biologically mediated reduction to methylated arsines which can escape the system via volatilization (Frankenberger and Arshad, 2002). Previous studies have attributed a portion of unaccountable arsenic to loss from volatilization (Rahman et al., 2011; Ye et al., 2003); however, volatile arsines rarely occur in natural environments except under extremely reducing conditions (<−500 mV) (Sharma and Sohn, 2009). In the current study, the lowest sediment redox potential was −282 mV; therefore it is unlikely that arsenic was lost due to volatilization. Interestingly, the highest percentages of unaccountable arsenic were in series 1 (10%) and series 4 (15%), both amended with ZVI. ZVI has a high affinity for arsenic with a sorption capacity ranging from 732 to 1771 mg As kg−1 ZVI depending on the speciation of arsenic (Su and Puls, 2001), and was

heterogeneously distributed in the first reactor of series 1 and series 4. It is possible that unaccountable arsenic in series 1 and 4 was sorbed on the surface of ZVI. Results from the current pilot-scale investigation represent a step forward in developing a low-cost, long-term solution to arsenic contamination of groundwater in developing countries such as Bangladesh. Treatment performance goals of successful demonstration and full-scale CWTSs are achieved by designing the systems based on parameters including hydraulic loading rates, hydraulic retention times, and wetland surface area, which can be determined using removal rate coefficients from pilot-scale studies such as the current study (e.g. Knight et al., 1999; Mooney and Murray-Gulde, 2008; Rousseau et al., 2004). In addition, results from our pilot-scale study demonstrate that incorporation of oxidizing conditions and amendment with iron can benefit design of CWTSs for treating arsenic-contaminated groundwater. ZVI in the form of iron filings is a recommended low-cost source of iron (e.g. Bang et al., 2005) for co-precipitation and sorption of arsenic using CWTSs. Accretion of the sediment in a CWTS by accumulation of plant detritus provides continued capacity of a CWTS to sequester constituents of concern. Based on previous studies, full-size CWTSs analogous to the pilot-scale system of the current study will function for at least 40–50 years (Kadlec and Wallace, 2009). 4. Conclusions A pilot-scale CWTS was designed and built to produce biogeochemical conditions that promoted targeted processes for the removal of arsenic from arsenic-contaminated water. Arsenic removal data demonstrate that a CWTS could be used to decrease arsenic concentration in the aqueous phase to below the WHO water quality guideline of 10 ␮g L−1 . The majority of arsenic removed from the aqueous phase during the 188-day experiment partitioned to sediment (88–99%), while the remainder partitioned to T. latifolia. The percentage of inflow arsenic retained in sediment of the first reactor of the two oxidizing series (20 and 13% for series 1 and 2, respectively) was greater than the percentage of inflow arsenic retained in sediment of the first reactor of the two reducing series (6 and 7% for series 3 and 4, respectively). The addition of ZVI to oxidizing series and to reducing series enhanced arsenic removal from the aqueous phase (mean removal efficiency of 72 and 42%, respectively) compared to unamended series (27 and 20%, respectively) and increased the mass of arsenic retained in sediment. By the end of the 188-day experiment, a vertical concentration gradient had developed in the sediment, with 74–85% of sediment-bound arsenic accumulated in the upper 6 cm and the remainder below 6 cm. The majority of arsenic retained in T. latifolia was associated with the roots and submerged shoots (87–97%), with little translocation to the emergent shoots and tips. The mass of arsenic associated with T. latifolia was greater in series amended with ZVI (23 and 46 mg for series 1 and 4, respectively), than in series not amended with ZVI (10 and 3 mg for series 2 and 3, respectively). Results of this pilot-scale study demonstrate that a CWTS can decrease the concentration of arsenic in arsenic-contaminated water primarily by transferring arsenic from the aqueous phase to the sediment. Results, including removal rate coefficients calculated in this study, can be used to design an on-site demonstration or full-scale CWTS as the next step toward using CWTS technology for treating arsenic-contaminated groundwater. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ecoleng. 2014.03.049.

J.P. Schwindaman et al. / Ecological Engineering 68 (2014) 251–259

References An, J., Kim, J., Kim, K., Park, J., Lee, J., Jang, M., 2011. Natural attenuation of arsenic in the wetland system around abandoned mining area. Environ. Geochem. Health 33, 71–80. Bang, S., Korfiatis, G.P., Meng, X., 2005. Removal of arsenic from water by zero-valent iron. J. Hazard Mater. 121 (1–3), 61–67. Blute, N.K., Brabander, D.J., Sutton, S.R., Newville, M.G., Rivers, M.L., 2004. Arsenic sequestration by ferric iron plaque on cattail roots. Environ. Sci. Technol. 38, 6074–6077. Buddhawong, S., Kuschk, P., Mattusch, J., Wiessner, A., Stottmeister, U., 2005. Removal of arsenic and zinc using different laboratory model wetland systems. Eng. Life Sci. 5 (3), 247–252. Carbonell, A.A., Aarabi, M.A., DeLaune, R.D., Gambrell, R.P., Patrick Jr., W.H., 1998. Arsenic in wetland vegetation: availability, phytotoxicity, uptake and effects on plant growth and nutrition. Sci. Total Environ. 217, 189–199. Chague-Goff, C., 2005. Assessing the removal efficiency of Zn, Cu, Fe, and Pb in a treatment wetland using selective sequential extraction: a case study. Water Air Soil Pollut. 160, 161–179. Cheng, H., Hu, Y., Luo, J., Xu, B., Zhao, J., 2009. Geochemical processes controlling fate and transport of arsenic in acid mine drainage (AMD) and natural systems. J. Hazard. Mater. 165, 13–26. Christophoridis, C., Dedepsidis, D., Fytianos, K., 2009. Occurrence and distribution of selected heavy metals in the surface sediments of Thermaikos Gulf, N. Greece. Assessment using pollution indicators. J. Hazard. Mater. 168, 1082–1091. Cohen, R.R.H., 2006. Use of microbes for cost reduction of metal removal from metals and mining industry waste streams. J. Clean. Prod. 14, 1146–1157. Di Toro, D.M., Mahony, J.D., Hansen, D.J., Scott, K.J., Carlson, A.R., Ankley, G.T., 1992. Acid volatile sulfide predicts the acute toxicity of cadmium and nickel in sediments. Environ. Sci. Technol. 26, 96–101. Dorman, L., Castle, J.W., Rodgers Jr., J.H., 2009. Performance of a pilot-scale constructed wetland system for treating simulated ash basin water. Chemosphere 75 (7), 939–947. Doyle, M.O., Otte, M.L., 1997. Organism-induced accumulation of iron, zinc and arsenic in wetland soils. Environ. Pollut. 96 (1), 1–11. Dushenko, W.T., Bright, D.A., Reimer, K.J., 1995. Arsenic bioaccumulation and toxicity in aquatic macrophytes exposed to gold-mine effluent: relationships with environmental partitioning, metal uptake and nutrients. Aquat. Bot. 50, 141–158. Eggert, D.A., Rodgers Jr., J.H., Huddleston, G.M., Hensman, C.E., 2008. Performance of pilot-scale constructed wetland treatment systems for flue gas desulfurization waters. Environ. Geosci. 15 (3), 115–130. Faulkner, S.P., Patrick Jr., W.H., Gambrell, R.P., 1989. Field techniques for measuring wetland soil parameters. Soil Sci. Soc. Am. J. 53, 883–890. Ford, R.G., Wilkin, R.T., Hernandez, G., 2006. Arsenic cycling within the water column of a small lake receiving contaminated ground-water discharge. Chem. Geol. 228, 137–155. Frankenberger, W.T., Arshad, M., 2002. Volatilization of arsenic. In: Frankenberger Jr., W.T. (Ed.), Environmental Chemistry of Arsenic. Marcel Dekker, Inc., New York (Chapter 16). Francesconi, K.A., Kuehneit, D., 2002. Arsenic compounds in the environment. In: Frankenberger Jr., W.T. (Ed.), Environmental Chemistry of Arsenic. Marcel Dekker, Inc., New York (Chapter 3). Henke, K.R., 2009. Arsenic in natural environments. In: Henke, K. (Ed.), Arsenic Environmental Chemistry, Health Threats, and Waste Treatment. Wiley, West Sussex, UK (Chapter 3). Henke, K.R., Hutchison, A., 2009. Arsenic chemistry. In: Henke, K. (Ed.), Arsenic Environmental Chemistry, Health Threats and Waste Treatment. Wiley, West Sussex, UK (Chapter 2). Horner, J.E., Castle, J.W., Rodgers Jr., J.H., Murray Gulde, C., Myers, J.E., 2012. Design and performance of pilot-scale constructed wetland treatment systems for treating oilfield produced water from Sub-Saharan Africa. Water Air Soil Pollut. 223, 1945–1957. Kadlec, R.H., Wallace, S.D., 2009. Treatment Wetlands, 2nd ed. CRC Press, Boca Raton, FL. Karri, S., Sierra-Alvarez, R., Field, J.A., 2005. Zero valent iron as an electron-donor for methanogenesis and sulfate reduction in anaerobic sludge. Biotechnol. Bioeng. 92 (7), 810–819. Keon, N.E., Swartz, C.H., Brabander, D.J., Harvey, C., Hemond, H.F., 2001. Validation of an arsenic sequential extraction method for evaluating mobility in sediments. Environ. Sci. Technol. 35, 2778–2784. Kinniburgh, D.G., Smedley, P.L. (Eds.), 2001. Arsenic Contamination of Groundwater in Bangladesh. British Geological Survey Technical Report WC/00/19, Volume 2, Final Report. British Geological Survey, Keyworth, http://www.bgs.ac.uk/downloads/start.cfm?id=2222 (accessed 17.01.14.). Kirk, M.F., Roden, E.E., Crossey, L.J., Brealey, A.J., Spilde, M.N., 2010. Experimental analysis of arsenic precipitation during microbial sulfate and iron reduction in model aquifer sediment reactors. Geochim. Cosmochim. Acta 74, 2538–2555.

259

Knight, R.L., Kadlec, R.H., Ohlendorf, H.M., 1999. The use of treatment wetlands for petroleum industry effluents. Environ. Sci. Technol. 33, 973–980. Knox, A.S., Nelson, E.A., Halverson, N.V., Gladden, J.B., 2010. Long-term performance of a constructed wetland for metal removal. Soil Sediment Contam. 19 (6), 667–685. Knox, A.S., Paller, M.H., Nelson, E.A., Specht, W., Halverson, N.V., Gladden, J.B., 2006. Metal distribution and stability in constructed wetland sediment. J. Environ. Qual. 35, 1948–1959. Leonard, E.N., Cotter, A.M., Ankley, G.T., 1996. Modified diffusion method for analysis of acid volatile sulfides and simultaneously extracted metals in freshwater sediment. Environ. Toxicol. Chem. 15 (9), 1479–1481. Li, H., Ye, Z.H., Wei, Z.J., Wong, M.H., 2011. Root porosity and radial oxygen loss related to arsenic tolerance and uptake in wetland plants. Environ. Pollut. 159 (1), 30–37. Liber, K., Doig, L.E., White-Sobey, S.L., 2011. Toxicity of uranium, molybdenum, nickel, and arsenic to Hyalella Azteca and Chironomus dilutus in water-only and spiked-sediment toxicity tests. Ecotox. Environ. Saf. 74, 1171–1179. Lizama, A.K., Fletcher, T.D., Sun, G., 2011. Removal processes for arsenic in constructed wetlands. Chemosphere 84 (8), 1032–1043. Mooney, F.D., Murray-Gulde, C., 2008. Constructed treatment wetlands for flue gas desulfurization waters: full-scale design, construction issues, and performance. Environ. Geosci. 15, 131–141. Paul, C.J., Ford, R.G., Wilkin, R.T., 2009. Assessing the selectivity of extractant solutions for recovering labile arsenic associated with iron (hydr)oxides and sulfides in sediments. Geoderma 152, 137–144. Pham, M.P.T., Castle, J.W., Rodgers Jr., J.H., 2011. Biogeochemical process approach to design and construction of a pilot-scale wetland treatment system for an oilfield produced water. Environ. Geosci. 18, 157–168. Rahman, K.Z., Wiessner, A., Kuschk, P., van Afferden, M., Mattusch, J., Müller, R.A., 2011. Fate and distribution of arsenic in laboratory-scale subsurface horizontalflow constructed wetlands treating an artificial wastewater. Ecol. Eng. 37 (8), 1214–1224. Rodgers Jr., J.H., Castle, J.W., 2008. Constructed wetland systems for efficient and effective treatment of contaminated waters for reuse. Environ. Geosci. 15 (1), 1–8. Rousseau, D.P.L., Vanrolleghem, P.A., De Pauw, N., 2004. Model-based design of horizontal subsurface flow constructed treatment wetlands: a review. Water Res. 38, 1484–1493. Sharma, V.K., Sohn, M., 2009. Aquatic arsenic: toxicity, speciation, transformations, and remediation. Environ. Int. 35, 743–759. Singhakant, C., Koottatep, T., Satayavivad, J., 2009. Enhanced arsenic removals through plant interactions in subsurface-flow constructed wetlands. J. Environ. Sci. Health A 44 (2), 163–169. Spacil, M.M., Rodgers Jr., J.H., Castle, J.W., Chao, W.Y., 2011. Performance of a pilotscale constructed wetland treatment system for selenium, arsenic, and lowmolecular-weight organics in simulated fresh produced water. Environ. Geosci. 18 (3), 145–156. Su, C., Puls, R.W., 2001. Arsenate and arsenite removal by zerovalent iron: kinetics, redox transformation, and implications for in situ groundwater remediation. Environ. Sci. Technol. 35, 1487–1492. Sundberg, S.E., Hassan, S.M., Rodgers Jr., J.H., 2006. Enrichment of elements in detritus from a constructed wetland and consequent toxicity to Hyalella azteca. Ecotox. Environ. Saf. 64, 264–272. Sundberg-Jones, S.E., Hassan, S.M., 2007. Macrophyte sorption and bioconcentration of elements in a pilot constructed wetland for flue gas desulfurization wastewater treatment. Water Air Soil Poll. 183, 187–200. USEPA, 1994a. Method 200.8: Determination of Trace Elements in Waters and Wastes by Inductively Coupled Plasma-Mass Spectrometry. United States Environmental Protection Agency, Office of Research and Development, Cincinnati, OH. USEPA, 1994b. Method 3051: Microwave Assisted Acid Digestion of Sediments, Sludges, Soils, and Oils. United States Environmental Protection Agency, Office of Research and Development, Cincinnati, OH. Wilkin, R.T., Ford, R.G., 2002. Use of hydrochloric acid for determining solidphase arsenic partitioning in sulfidic sediments. Environ. Sci. Technol. 36 (22), 4921–4927. Wong, T.H.F., Fletcher, T.D., Duncan, H.P., Jenkins, G.A., 2006. Modelling urban stormwater treatment – a unified approach. Ecol. Eng. 27, 58–70. Ye, Z.H., Lin, Z.-Q., Whiting, S.N., De Souza, M.P., Terry, N., 2003. Possible use of constructed wetland to remove selenocyanate, arsenic, and boron from electric utility wastewater. Chemosphere 52 (9), 1571–1579. Ye, Z.H., Whiting, S.N., Lin, Z.-Q., Lytle, C.M., Qian, J.H., Terry, N., 2001. Removal and distribution of iron, manganese, cobalt, and nickel within a Pennsylvania constructed wetland treating coal combustion by-product leachate. J. Environ. Qual. 30, 1464–1473. Zhao, F.J., Ma, J.F., Meharg, A.A., McGrath, S.P., 2009. Arsenic uptake and metabolism in plants. New Phytol. 181, 777–794.