In situ denitrification and DNRA rates in groundwater beneath an integrated constructed wetland

Water Research 111 (2017) 254e264

Contents lists available at ScienceDirect

Water Research journal homepage: www.elsevier.com/locate/watres

In situ denitrification and DNRA rates in groundwater beneath an integrated constructed wetland M.M.R. Jahangir a, b, c, *, O. Fenton b, C. Müller d, e, R. Harrington f, P. Johnston a, K.G. Richards b a

Department of Civil, Structural & Environmental Engineering, Trinity College Dublin, Ireland Teagasc Environment Research Centre, Johnstown Castle, Co. Wexford, Ireland Department of Soil Science, Bangladesh Agricultural University, Mymenisngh, 2202, Bangladesh d School of Biology and Environmental Science, University College Dublin, Belfield, Ireland e Department of Plant Ecology (IFZ), Justus-Liebig University Giessen, Germany f Vesi Environmental Ltd., Co. Cork, Ireland b c

a r t i c l e i n f o

a b s t r a c t

Article history: Received 30 June 2016 Received in revised form 26 December 2016 Accepted 6 January 2017 Available online 7 January 2017

Evaluation of the environmental benefits of constructed wetlands (CWs) requires an understanding of their impacts on the groundwater quality under the wetlands. Empirical mass-balance (nitrogen in/nitrogen out) approaches for estimating nitrogen (N) removal in CWs do not characterise the final fate of N; þ where nitrate (NO
3 -N) could be reduced to either ammonium (NH4 -N) or N2 with the potential for significant production of N2O. Herein, in situ denitrification and DNRA (dissimilatory nitrate reduction to ammonium) rates were measured in groundwater beneath cells of an earthen lined integrated constructed wetland (ICW, used to remove the nutrients from municipal wastewater) using the 15N-enriched NO
3 -N push-pull method. Experiments were conducted utilising replicated (n ¼ 3) shallow (1 m depth) and deep (4 m depth) piezometers installed along two control planes. These control planes allowed for the assessment of groundwater underlying high (Cell 2, septic tank waste) and low (Cell 3) load cells of the ICW. Background piezometers were also installed off-site. Results showed that denitrification (N2ON þ N2-N) and DNRA were major NO
3 -N consumption processes accounting together for 54e79% of the total biochemical consumption of the applied NO
3 -N. Of which 14e16% and 40e63% were consumed by denitrification and DNRA, respectively. Both processes differed significantly across ICW cells indicating that N transformation depends on nutrient loading rates and were significantly higher in shallow compared to the deep groundwater. In such a reduced environment (low dissolved oxygen and low redox potential), higher DNRA over the denitrification rate can be attributed to the high C concentration and high TC/NO
3 -N ratio. Low pH (6.5e7.1) in this system might have limited denitrification to some extent to an incomplete state, evidenced by a high N2O-N/(N2O-NþN2-N) ratio (0.35 ± 0.17, SE). A relatively higher N2O-N/(N2O-NþN2-N) ratio and higher DNRA rate over denitrification, suggest that the end products of N transformations are reactive. This N2O can be consumed to N2 and/or emitted to the atmosphere. The DNRA rate and accumulation of NHþ 4 -N indicated that the ICW created a suitable þ groundwater biogeochemical environment that enhanced NO
3 -N reduction to NH4 -N. This study -N attenuation to reactive forms of N in the groundwater showed that CWs significantly influence NO
3 beneath them and that solely focusing on within wetland NO
3 -N attenuation can underestimate the environmental benefits of wetlands. © 2017 Elsevier Ltd. All rights reserved.

Keywords: Push-pull 15 N enrichment Denitrification N2O emissions DNRA Integrated constructed wetland

1. Introduction

* Corresponding author. Teagasc Environment Research Center, Johnstown Castle, Co. Wexford, Ireland. E-mail address: [email protected] (M.M.R. Jahangir). http://dx.doi.org/10.1016/j.watres.2017.01.015 0043-1354/© 2017 Elsevier Ltd. All rights reserved.

Constructed wetlands (CWs) are an emerging technology used globally to remove pollutants from wastewaters and landfill leachate (Gill et al., 2014; Harrington et al., 2007; Søvik et al., 2006; Tanner and Sukias, 2011). Integrated constructed wetlands (ICW)

M.M.R. Jahangir et al. / Water Research 111 (2017) 254e264

comprise series of inter-connected ponds, what we call here cells, which are located at the down hydraulic gradient from each other. Recently such systems have been used for sediment capture from agricultural and urban runoff (Ockenden et al., 2014; Scholz and Lee, 2005) as well as stormwater (Carleton et al., 2000; Davies and Bavor, 2000). Such systems facilitate a number of physical, chemical and biological processes that relate to species of nitrogen (N) and phosphorus (P). In such systems N removal efficiencies of through-flowing water when treating wastewater are low (e.g. 40e50% of total N added; Vymazal, 2007) and highly variable across CW types (Jahangir et al., 2016). Indeed the conventional influent e effluent balance approach for removal efficiencies does not explain whether the N concentrations have been reduced by dilution and dispersion, or removed by biological assimilation, or transformed to other forms. It must be noted that the environmental benefits of CWs can be diminished if the removed N species þ are reactive (e.g. nitrate- NO
3 -N, ammonium- NH4 -N and nitrous oxide- N2O). As these products of N cycling processes can cause air and water quality concern e.g. surface and groundwater contamination and direct/indirect N2O emissions to the atmosphere (Jahangir et al., 2016) they must be accounted for. In addition, the fate of the removed N from such a system is not yet understood. The maximum allowable concentration (MAC) of NHþ 4 -N for surface and groundwater are 0.14 and 0.23 mg L
1, respectively (Craig et al., 2010). Studies have shown that groundwater underlying CW can have higher NHþ 4 -N concentrations (0.01e12.6 mg
1 NHþ 4 -N L ) than in the surface discharges from the same CW (Dzakpasu et al., 2014). Dzakpasu et al. (2012) measured from 4.3 to
1 27.8 mg NHþ in water samples collected from lysimeters 4 -N L installed below the compacted soil bed of ICW cells. Ammonium is a biologically available form of N that is a significant aquatic contaminant and is more harmful than NO
3 -N when discharged to surface waters (Burgin et al., 2013). However, the sources and fate of NHþ 4 -N in groundwater below the CWs have not been investigated. Therefore a wider understanding of N cycling processes and the end products of different processes in groundwater beneath CW would help elucidate how such losses of N and associated gases negatively impact water and air quality. To our knowledge, no investigations have assessed the impacts of CWs on N cycling processes in the underlying groundwater of CW cells. In fact only a few processes ultimately remove total N from the wastewaters in the form of benign N (N2), whereas the vast majority of processes convert N to its more reactive forms (Vymazal, 2007). Hence, understanding N transformation processes beneath CWs is crucial to evaluate their potential for sustainable environmental benefits. Denitrification is a multistep biological process which uses NO
3N as an electron acceptor and carbon (C) as an electron donor and
causes respiratory reduction of NO
3 -N to NO2 -N to N2O to N2 via NO. One advantage of denitrification in CWs is that it acts as a ‘permanent’ sink if the N is lost permanently as N2. However, when the process is not fully reduced to N2, the intermediate, N2O, can be emitted. Nitrous oxide is one of the major greenhouse gases (GHGs) in contributing to global warming and atmospheric and ground level ozone depletion (Beaulieu et al., 2011; IPCC, 2013). If the denitrification process dominates within the system over other processes, the risk of pollution swapping (the increase of one pollutant e.g. N2O as a result of decrease of another e.g. NO
3 -N) will be increased. Nitrous oxide produced in CW soils and in underlying groundwater can be emitted to the atmosphere directly by ebullition, degassing and transpiration by vascular plants. But alternatively it can be transported via groundwater flow and emitted elsewhere (Riya et al., 2010). Past studies have shown that GHGs in CW soils were correlated significantly to the surface emissions (Beaulieu et al., 2011; Riya et al., 2010). An estimation of N2O/ (N2OþN2) along with the total denitrification rate is therefore

255

essential to understand N2O production and consumption in such systems. Dissimilatory nitrate reduction to ammonium (DNRA) is a biological process which can produce NHþ 4 -N in CW soils and in the underlying groundwater. Unlike denitrification, DNRA conserves N in the ecosystem as NHþ 4 -N until it is assimilated by microbes/ plants or oxidised to NO
3 -N (Giblin et al., 2013). In recent years, N cycling studies have increasingly investigated DNRA in various ecosystems to explore its importance in N cycling (Rütting et al., 2011) but controls on DNRA are relatively unknown (Burgin et al., 2013; Giblin et al., 2013; van den Berg et al., 2015; Kraft et al., 2015). The conditions that favour the occurrence of either denitrification or DNRA are still in debate (Rütting et al., 2011; van den Berg et al., 2015; Yoon et al., 2015) and DNRA is probably the least studied N transformation process in wetlands (Vymazal, 2007). Ammonium that is taken up by plants can be recycled as organic N, which again can be transformed to NHþ 4 -N by ammonification. In a 15N isotopic experiment, O'Luanaigh et al. (2010) þ showed that NHþ 4 -N transformations occurred changing NH4 -N to þ organic N and back to NHþ -N within the CW. The NH -N produced 4 4 in soils and subsoils can be fixed in soil clay but after saturation of the exchange site, it can be transported to surface waters via groundwater. This is evident in many studies where drainage water from heavy clay subsoils shows high NHþ 4 -N concentrations in discharge water along open ditches or subsurface tile networks (e.g. Necpalova et al., 2012; Ibrahim et al., 2013). During transport the
oxidation rate of NHþ 4 -N to NO3 -N is subject to the physicochemical conditions along the flow path which may be limited in such subsurface system (Vymazal, 2007). Occurrence of NHþ 4 -N, as the dominant form of N in groundwater below clay loam textured soils, has been reported by past researchers (e.g. Kartal et al., 2007; Mian et al., 2009; Necpalova et al., 2012). The main question arises as to whether the NHþ 4 -N found in the underlying groundwater has been produced in situ or whether it has diffused out of the sediment at the bottom of the artificial wetland cells. Hence, accumulation of biologically available N by DNRA has major implications for our understanding of how CW ecosystems will respond to increases in the N loads they receive from wastewaters. Recent isotopic methodological developments using in situ or ex situ (intact cores in laboratory incubation studies) techniques are suitable to further understand N cycle processes (Lee et al., 2009; O'Luanaigh et al., 2010). The push-pull method has been used previously to measure in situ denitrification rates in estuarine/river sediments (Huygens et al., 2013), in groundwater below arable and grassland (Jahangir et al., 2013a) and in riparian zones (Addy et al., 2002; Harrison et al., 2011; Kellogg et al., 2005). To our knowledge this method has not been reported yet in groundwater below CWs. Therefore the objectives of this study are to quantify: a) denitrification and DNRA rates and N2O/(N2OþN2) ratios in shallow (beneath the ICW bed) and deep (4 m below the ICW bed) groundwater beneath two cells of the ICW receiving contrasting nutrient loads and b) compare relative contributions of denitrification and DNRA on total biological NO
3 -N reduction. 2. Materials and methods 2.1. Experimental site An ICW is a kind of CW consisting of a series of inter-connected wetland cells located down gradient from each other. The surface flow ICW site (0.32 ha in size, a 5-cell ICW (Cells 1, 4 and 5 not used in the present study) is located in Dunhill village, Waterford, southeastern Ireland (Fig. 1). Groundwater background piezometers (1A eshallow, 1B edeep, Fig. 1) examine groundwater not affected by the ICW. Two virtual control planes (Fig. 1) of

256

M.M.R. Jahangir et al. / Water Research 111 (2017) 254e264

Fig. 1. Experimental site in Ireland showing groundwater flow direction, the locations of groundwater (GW) background piezometers, ICW cell layout (Cell 1 is inactive and not used in this study, Cell 2 ¼ high load and Cell 3 ¼ low load) and the replicated piezometers along control plane 1 (2A, 2B, 3A, 3B, 4A, 4B) and control plane 2 (5A, 5B, 6A, 6B, 7A, 7B).

piezometers (Groundwater Control Plane 1 ¼ 2A-B, 3A-B, 4A-B and Groundwater Control Plane 2 ¼ 5A-B, 6A-B, 7A-B) were installed (0.2 m from the edge of each cell) to examine groundwater below the cells. As groundwater flow direction (Fig. 1) is towards the river moving from Cell 1 to Cell 3, these control planes investigated (at shallow and deep levels) groundwater emanating from up-gradient cells. Therefore Control Plane 1 examines groundwater underlying Cell 2 (high load), whereas control plane 2 examines groundwater underlying Cell 3 (low load). The site is being used to treat municipal waste water from a village of 500 (person equivalent) PE wastewaters and stormwater for a period of 15 years. After construction, the ICW was underlain by 1.5 m of subsoil, with the upper 0.5 m compacted using local soil material (sand, silt and clay were 26, 47 and 27%, respectively) to reduce the saturated hydraulic conductivity (ks) to 1  10
8 m s
1. The ks of shallow and deep groundwater zones measured by falling head tests in July 2014 were 2.0  10
8 and 3.5  10
8 m s
1, respectively. Three replicated multilevel piezometers (0.025 m inner diameter; 1 m screened interval at base) were installed on the bank of the ICW cells (0.2 m from edge of ICW Cell) at two depths: 1 m below ground level (bgl), representative of shallow groundwater within the subsoil beneath the ICW bed, and 4 m bgl, representative of deeper groundwater below the ICW (Fig. 1). Two additional piezometers were located up hydraulic gradient from the ICW Cells being investigated installed at similar depths as those adjacent to the ICW Cells (Fig. 1). These multilevel piezometers were used for groundwater quality monitoring below the ICW Cells. Mean annual rainfall during January 2013 to December 2014 inclusive was 1050 mm. Mean annual groundwater temperature was 11  C. The estimated mean vertical flow rate of water from CW cells to groundwater (velocity, V ¼ saturated hydraulic conductivity, ks  hydraulic gradient, i) was 74 mm/year. 2.2. In situ push-pull method The in situ push-pull experiment (adopted from Addy et al.,

2002) was conducted from March to May 2014. The pushepull experiment consisted of collection of groundwater from a piezometer, amendment with 50-at% 15N-enriched NO
3 -N as KNO3 (purity 99%) and a conservative tracer (bromide- Br
) injection into the same piezometer from which it was collected (‘‘push’’) and incubated for 6 h and then pumping it back (‘‘pull’’). The conservative tracer was used to estimate the loss of NO
3 -N by dilution and dispersion. The incubation period for 6 h was determined based on a push-pull pre-test, conducted to assess the maximum drifting time of groundwater solution for microbial denitrification and DNRA for which at least 50% of recovery of the Br
was achieved. For this, the push-pull pre-test was
1 conducted repeatedly with 20 mg NO
groundwater so3 -N L lution coupled with 20 mg Br
L
1 (conservative tracer). Six litre groundwater solutions were injected into each piezometer at a slow rate of 12 L h
1 to minimise the change in hydraulic gradient surrounding the piezometer. As the Br
recovery (pulled concentration/initial pushed concentration  100) was substantially lower (10e58%) than its targeted rate, two more push-pull pre-tests were conducted with groundwater solution of similar
concentrations of NO
3 -N and Br and incubated for 12 h and 6 h. In between each push-pull pre-test, groundwater was sampled for Br
analysis to ensure that it returned to the background concentration. For the 15NO
3 -N push-pull test, 6 L groundwater was collected from each piezometer in a plastic container (carboy) using a peristaltic pump (Model 410, Solinst Canada Ltd.) with a Teflon outlet tube at a rate of 12 L h
1. Groundwater was amended immediately with 20 mg N L
1 isotopically

1 enriched (50-at% 15N) NO
as KBr 3 -N as KNO3 and 20 mg Br L and injected into the respective piezometer at the same rate it was pumped. The 6 L solution fills approximately 18 kg of soils (total porosity 46% and bulk density 1350 kg m
3; estimated using the equations and constants described by Saxton et al., 1986). In brief, the total amount of aquifer materials covered by the solution was calculated using Eqn. (1) below:

M.M.R. Jahangir et al. / Water Research 111 (2017) 254e264

 Mt ¼






Vt
Vg *Bd Porosity of aquifer

29N2 ¼

29R 1 þ 29R þ 30R

(2)

30N2 ¼

30R 1 þ 29R þ 30R

(3)

(1)

where Mt is the total mass of aquifer materials (kg), Vt is the total volume of solution (m3), Vg is the volume of gravel pack (m3) and Bd is the bulk density (kg m
3). To determine ambient concentrations in the dosing solution, the groundwater solution was sampled during the injection phase for dissolved oxygen (DO), Br
and dissolved gases (N2O and N2) and hydrochemistry (dissolved organic C- DOC, dissolved organic NDON, total N- TN, total C- TC, redox potential- Eh, and sulphate
SO2
4 ). The amended solution (enriched N and Br ) was incubated in situ for 6 h (time between the end of injection and start of pumping back). Groundwater was pumped back from the piezometer (twice the injected volume) at a slow rate (12 L h
1) using a peristaltic pump with a Teflon outlet. Samples for dissolved gases and hydrochemistry were taken at every 400 ml interval. For dissolved N2O and N2, a 160 ml sample was taken in a glass serum bottle and sealed with a butyl rubber septum and an aluminium crimp cap. The sealed serum bottles were then submerged under water in a polystyrene box and stored at 4  C in a cold room until analysis on the following day. The Br
recoveries were estimated as C/Co; where C was the tracer's concentrations in the pulled groundwater following incubation and Co was the tracer's concentrations in the original pushed groundwater (Freeze and Cherry, 1979).

257

28N2 ¼ 1 - (29N2þ30N2)

(4)

Similarly, ratios of 45R (45/44) and 46R (46/44) were used for calculation of molecular fraction of 45, 46 and 44 N2O-N. The atomic fraction, i.e. 15N in N2 or 15N in N2O was then calculated using Eqns. (5) and (6) below:

15N
N2 ¼

29N2 þ 2*30N2 *moles 2

15N
N2 O ¼

45N2 O þ 2*46N2 O *moles 2

(5)

(6)

The total mass of N2OeN or N2 was then transformed to the mass of 15N2OeN or 15N2 by multiplying it by the respective 15N sample enrichment proportion (ratio of pulled atom % of the dissolved N2O-N and N2 to pushed NO
3 eN atom %, both corrected for ambient atom %). Production of gases attributable to denitrification was calculated from the differences between the ambient concentrations and the concentrations obtained after 6-h incubation. Gas production rates for 15N2OeN and 15N2-N were expressed as mg N kg
1 soil d
1 as below:

 Total mass of 15 N O
N and 15 N
N per volume of water pulled  2 2 *24 Rates mg Nkg
1 d
1 ¼ Dry mass of soil per volume of water*incubation period

NO
3 -N and

15

(7)

15

NHþ 4 -N

2.3. Dissolved gas extraction and denitrification rate calculation

2.4. Analysis of

In the laboratory, dissolved N2O and N2 in ambient, pushed and pulled samples were extracted using the phase equilibration He headspace extraction technique (Lemon, 1981; Davidson and Firestone, 1988). For a detailed description of the He headspace technique see Jahangir et al. (2012). In brief, exactly 140 ml water of the 160 ml glass serum bottle was removed using a plastic syringe (BD Plastic Ltd.) with simultaneous injection of 140 ml of high purity He through a Teflon inlet tube connected to a cylinder (He:water ratio 7:1; v/v). The serum bottles were shaken for 6 min placed end to end on a Gyrotory shaker (Model G-10, New Brunswick Scientific Co., USA) and left for a standing period of 2 h. Headspace samples were then taken for the analysis of N2O and N2 concentrations and the 15N enrichment of N2O and N2 in preevacuated 12 ml exetainers (Labco Inc. Wycomb, UK). Concentrations and 15N enrichment of N2O and N2 were determined on a dual-inlet isotope ratio mass spectrometer (Stable Isotope Facility, UC Davis, Davis, CA) as described by Mosier and Schimel (1993). Details of calculation of 15N2O and 15N2 concentrations and denitrification rates (mg N kg
1 d
1) were previously explained in Jahangir et al. (2013a). The total masses of N2OeN and N2-N gases (mg/moles) were calculated using equations and constants provided by Tiedje (1982) and Mosier and Klemedtsson (1994). Briefly, for N2-N, ratios of 29R (29/28) and 30R (30/28) were used for calculation of molecular fraction of 29, 30 and 28 N2 using Eqns. (2)e(4) below:

þ The diffusion method for 15N enrichment of NO
3 -N and NH4 -N in groundwater samples was adapted from Brooks et al. (1989). Samples were analysed first for TN on a TOC analyser (TOC-V cph/ cpn, Shimadzu Corporation, Kyoto, Japan). Groundwater total oxidised N, nitrite and NHþ 4 -N were analysed with an Aquakem 600 discrete analyser (Aquakem 600A, Vantaa, Finland). As groundwater TN varies with different samples, the required volume of sample was taken in a 250 ml kilner jar to keep TN content between 20 and 100 mg. Total volume of the sample was made up to 50 ml by adding deionised water. Potassium chloride salt was added to the sample in kilner jar required to make a 2M KCl extract and shaken gently to dissolve the salt (Fig. 2). Five millimetre diameter disks of filter paper (GF/D) were cut with a paper punch and washed in 0.5M K2SO4. A small ball of ‘Bluetack’ adhesive was stuck on the underside of the kilner jar lid. Hypodermic needles (BD Microlance™ 3, Becton Dickinson & Co. Lt.) were bent to 90 and the base of the needle was stuck into the Bluetack ensuring the needle was suspended under the centre of the lid (Fig. 2). Next 10 microliters of 2.5 M KHSO4 was pipetted onto the filter paper disks (acid trap). These disks were stored by resting them across the top of a 100 ml beaker for about 10 min. Two filter discs were placed onto each needle (per kilner jar) by piercing them one by one using forceps. For NHþ 4 -N, an approximate 0.7 g scoop of MgO heavy powder (BDH Chemicals) was added to the suspension, the lid was immediately screwed on and the jar swirled gently for 10 s and stored in an incubator at 55  C for three days with occasional swirling (once a

258

M.M.R. Jahangir et al. / Water Research 111 (2017) 254e264

only by physical processes and not taken up by biological processes. 2.5. Hydrochemistry and conservative tracer (Br
) analysis

Fig. 2. Suspension of acid trap into the Kilner Jar used for analysing 15 NHþ 4 -N from the groundwater samples.

NO
3 -N and

15

day). During swirling care was taken to prevent the alkaline suspension touching the acid trap which can neutralise it. After 3 days, disks were removed and placed (2 disks from each kilner jar) in tin capsules using forceps, dried at 55  C for an hour and placed on microplate. For NO
3 -N, an approximate 0.4 g Devarda's alloy (Fisher Chemicals) was added to the same suspension in the kilner jar followed by another 0.25 g MgO heavy powder, the lid bearing fresh filter disks immediately screwed on and the jar swirled gently for
10 s. A similar procedure to that for NHþ 4 -N was followed for NO3 -N in transferring from disks to tin capsules and microplate. The samples in the microplate were sent to the Isotope Facilities UC Davis for determination of concentrations and 15N enrichment of
NHþ 4 -N and NO3 -N on a dual-inlet isotope ratio mass spectrometer
(UC Davis, Davis, CA). The total mass of NHþ 4 -N or NO3 -N was then 15 15 þ
transformed to the mass of NH4 -N or NO3 -N by multiplying it by the respective 15N sample enrichment proportion (ratio of
þ
pulled at.% of the NHþ 4 -N or NO3 -N to pushed at.% of NH4 -N or NO3 N, both were corrected for ambient at.%). Nitrate consumption or
1
1 NHþ d ) were calculated 4 -N production (DNRA) rates (mg N kg using Eqn. (8) below:

For DO, samples were collected in a 12 ml exetainer (Labco Ltd., Wycombe, UK) after slowly overflowing of approximately 10 ml excess water, which was closed immediately using a double septum (butyl rubber þ Teflon) stopper. Samples were submerged underwater in a polystyrene box, stored at 4  C and analysed the following day. Dissolved oxygen was measured by membrane inlet mass spectrometry (Kana et al., 1994). Groundwater pH, electrical conductivity and redox potential (Eh) were measured using a multiparameter probe (Troll 19500, In Situ Inc., USA). For hydrochemistry, samples were taken in a 50 ml plastic tube (BD Plastic Ltd.) and stored in a cold room at 4  C until analysis within one
week. Groundwater was analysed for NO
3 -N and Br using DX-120 ion chromatography (Metrohm UK Ltd.). Groundwater table depth was measured using a standard electrical dip metre. Dissolved organic carbon was analysed using total organic carbon analyser (TOC-V cph/cpn, Shimadzu Corporation, Kyoto, Japan). Groundwater SO2
concentration was measured using a turbimetric 4 method (Askew and Smith, 2005).

2.6. Statistical analysis Data were analysed for the analysis of variance (ANOVA) using a general linear model in SAS (SAS Institute Inc, 2009). As most of the variables were log normally distributed, log transformations were used. For the concentration of denitrification end products (N2O and N2), N2O mole fractions, denitrification rate (N2ON þ N2-N), DNRA rate, total NO
3 consumption, denitrification/
total NO
3 consumption ratio and DNRA/total NO3 consumption ratio, the effects of groundwater depth and Cell were examined along with their interactions. Where significant differences between factors were found, the Tukey Kramer HSD all pairs multiple comparison test was used to distinguish specific differences. The influence of hydrochemical variables on denitrification and DNRA were determined by stepwise multiple regression. Differences of mean Br
recovery between shallow and deep groundwater and between denitrification and DNRA were tested with a paired sample t-test.

 Total mass of 15 NH þ
N or NO

N produced or consumed per volume of water pulled  3 4 Rates mg Nkg
1 d
1 ¼ *24 Dry mass of soil per volume of water*incubation period

(8)

3. Results Mass of soils was calculated based on the amount of soils occupied by 1 L water (total porosity 46% and bulk density 1350 kg m
3). Total biological NO
3 -N consumption was calculated following a mass balance as shown by Eqn. (9) below:
Total biochemical NO
3 -N consumption ¼ total NO3 -N consumption
- physical NO3 -N removal (9)

Physical removal indicated reduction of concentrations due to dilution and advective dispersion. This was estimated from the reduction in tracer (Br
) concentrations, which can be reduced

3.1. Groundwater hydrogeochemical properties prior to the experiment The background ambient NHþ 4 -N, DON and TN concentrations in both shallow and deep groundwater were higher (Table 1) relative to off site piezometers' (i.e. 1A-1B, Fig. 1). The NO
2 -N concentrations in all piezometers beneath the CW were negligible. Equally, TC and DOC concentrations were high in groundwater below the CW. Groundwater in shallow and deep piezometers presented conditions of a highly reduced environment with mean Eh ranging from
29 to
36 m V and mean DO concentrations ranging from

M.M.R. Jahangir et al. / Water Research 111 (2017) 254e264

259

Table 1 Groundwater hydrochemical properties (mean ± SE, n ¼ 3), prior to the experiment, in upgradient (control, not replicated) and beside the ICW Cells; Eh - redox potential; DOdissolved oxygen. C conc. (mg C L
1)

NHþ 4 -N

NO
3 -N

DON

TN

DOC

TC

0.24 0.17 5.70 8.80 3.20 3.00

1.13 0.44 0.43 0.27 0.15 0.15

0.69 0.38 2.00 1.40 1.40 0.40

2.02 0.96 7.70 10.4 4.50 3.30

4.76 3.55 33.1 10.9 27.4 7.74

17.7 22.1 90.5 72.5 84.3 62.4

± ± ± ±

2.5 2.3 1.4 1.8

± ± ± ±

0.05 0.16 0.02 0.03

± ± ± ±

0.4 0.5 0.3 0.1

± ± ± ±

0.46 to 0.53 mg L
1. The Eh and DO values showed high spatial variability in both shallow and deep groundwater. The pH in groundwater ranged from 6.5 to 7.1. The C:N ratio of the ambient groundwater (1A-1B) was between 12 and 18. 3.2. Recovery of conservative tracer The initial Br
concentration, injected into individual piezometers, was 20 mg L
1 after which decreased to 15 and 12 mg L
1 in shallow and deep groundwater in 6 h, respectively (Fig. 3). Recovery of the conservative tracer was significantly lower in deep groundwater than that in shallow groundwater (p < 0.05). The mean three highest recoveries of tracer, termed the core plume, ranged from 58 to 73 and 43e59% in shallow and deep groundwater, respectively. The highest values obtained in the first 3 L of the pull, were used to measure denitrification and DNRA rates to minimise the uncertainty caused by dilution and dispersion (Addy et al., 2002; Harrison et al., 2011). 3.3. Denitrification rates and N2O-N/(N2O-NþN2-N) ratios Mean N2O and N2 concentrations in comparison with initial concentrations present in the injected solution and in the pulled water samples after 6 h incubation are presented in Fig. 4. Differences in the denitrification (N2O-N þ N2-N) rates between the shallow and deep groundwater were marginally significant (p ¼ 0.09). Considering multiple comparisons between the cells, total denitrification rates were significantly higher in Cell 2 (ca. 1063 ± 241 and 341 ± 128 m^mg N kg
1 d
1, mean ± SE, in shallow and deep groundwater, respectively) than in Cell 3 (ca. 123 ± 38 and 218 ± 192 mg N kg
1 d
1, mean ± SE, in shallow and deep groundwater, respectively) (Table 2). Total denitrification rates showed higher spatial variability in deep (coefficient of variation, CV 65e153%) than that in shallow groundwater (CV 39e53%). Mean groundwater N2O production rates were similar between cells (p > 0.05) with mean values below Cell 2 and Cell 3 of 99 ± 28 and 86 ± 74 mg N kg
1 d
1, respectively. However, mean N2O production rates were significantly lower (p < 0.05) in shallow (ca. 42 ± 29 mg N kg
1 d
1) than the deep groundwater (ca. 143 ± 17 mg N kg
1 d
1). Mean N2 production rates, were significantly higher (p < 0.01) beneath Cell 2 (603 ± 389 mg N kg
1 d
1) than Cell 3 (84 ± 26 mg N kg
1 d
1) and were also higher in shallow (551 ± 441 mg N kg
1 d
1) compared to deep (136 ± 26 mg N kg
1 d
1) groundwater. Deep groundwater showed higher spatial variability (CV 132e172%) in N2 production rates than the shallow equivalent (CV 40e50%). Based on the N2O and N2 rates denitrification product ratios (N2ON/(N2O-NþN2-N) were significantly higher (p < 0.01) in the deep groundwater than in the shallow groundwater. Considering multiple comparisons between Cells, the N2O-N/(N2O-NþN2-N) ratios beneath Cell 2 (ca. 0.06 ± 0.04 and 0.34 ± 0.32 in shallow and deep groundwater, respectively) were significantly lower

3.0 2.7 1.6 1.7

± ± ± ±

2.8 6.6 2.1 4.9

25 Br- conc. (mg L-1)

Off-site: shallow Off-site: deep Shallow: Cell 2 Shallow: Cell 3 Deep: Cell 2 Deep: Cell 3

N conc. (mg N L
1)

± ± ± ±

6.8 2.3 4.1 3.1

Eh (mV)

DO (mg L
1)

pH

80.0 28.6
30.5
24.4
36.6
28.8

3.73 3.40 0.52 0.53 0.42 0.47

7.5 7.2 6.9 7.1 6.7 7.0

± ± ± ±

5 6 5 5

± ± ± ±

0.1 0.1 0.1 0.1

± ± ± ±

0.0 0.0 0.1 0.1

Cell 2

Shallow

20

Cell 3

15 10 5 0 initial 25

-1 Br conc. (mg L )

Depth/ICW Cells

20

Deep

0.31

0.63

0.94

1.26 Cell 2 Cell 3

15 10 5 0 initial 0.31 0.63 0.94 1.26 [Cumulative pulled volume/ Total pushed volume]

Fig. 3. Push Pull Br
concentrations (mg L
1) in initially injected groundwater solutions plotted against cumulative pull volume/total push volume in shallow and deep groundwater from the 6-h in situ NO
3 -N pushepull test; n ¼ 3.

than beneath Cell 3 (0.09 ± 0.02 and 0.90 ± 0.09 in shallow and deep groundwater, respectively). Multiple linear regression analysis following a backward method revealed the following model for total denitrification: log denitrification ¼ 28.8 þ 0.18 DOC
0.50 NO
3
N
0.18 2 SO2
4
3.32 pH; (Adjusted R ¼ 0.54; p < 0.01; n ¼ 9)

(10)

3.4. Dissimilatory nitrate reduction to ammonium (DNRA) rate Mean NHþ 4 -N concentration compared with initial concentration present in the injected solution and in the pulled water samples after 6 h incubation has been presented in Fig. 5. The DNRA rate, estimated from the conversion of the applied 15NO
3 -N to 15NHþ 4 -N, was higher (p < 0.05) in shallow than in the deep groundwater (Table 2). Among the two cells, the DNRA rate was significantly higher (p < 0.05) beneath Cell 2 (ca. 1063 ± 241 and 341 ± 128 mg kg
1 N d
1 in shallow and deep groundwater, respectively) than Cell 3 (ca. 123 ± 38 and 218 ± 192 mg N kg
1 d
1

260

M.M.R. Jahangir et al. / Water Research 111 (2017) 254e264

Fig. 4. Push Pull N2O-N and N2-N concentrations (mg N L
1) in initially injected groundwater solutions plotted against cumulative pull volume/total push volume in shallow and deep groundwater from the 6-h in situ NO
3 -N pushepull test; n ¼ 3.

Table 2
Mean (±SE) of total denitrification (N2O-N þ N2-N) rate, N2O-N/(N2O-N þ N2-N) ratio, NO
2 -N production rate, DNRA rate, and total NO 3-N consumption rate at two different groundwater depths below two constructed wetland Cells (n ¼ 3). Depth

Cell

Denitrification rate (mg N kg
1 d
1)

N2O-N/(N2O-N þ N2-N) ratio

NO
2 -N production rate (mg N kg
1 d
1)

DNRA rate (mg N kg
1 d
1)

Total NO
3 -N consumption rate (mg N kg
1 d
1)

Shallow

Cell 2 Cell 3

1063 ± 241 123 ± 38

0.06 ± 0.04 0.09 ± 0.02

83.9 ± 49.9 24.6 ± 6.7

2277 ± 898 1448 ± 375

4593 ± 938 2473 ± 863

Deep

Cell 2 Cell 3

341 ± 128 218 ± 192

0.34 ± 0.32 0.90 ± 0.09

5.7 ± 2.8 2.8 ± 1.2

804 ± 440 738 ± 201

1824 ± 273 1732 ± 328

in shallow and deep groundwater, respectively). There was no significant correlation between cell and depth observed (p > 0.05). Similar to the other hydrochemical properties, the DNRA rate exhibited high spatial variability in shallow and deep groundwater, with CVs ranging from 63 to 70% and from 72 to 149%, respectively. Similar to the in situ DNRA rate, ambient NHþ 4N concentrations, measured before the onset of the experiment, were higher in the shallow groundwater than in the deep groundwater (Table 1). Unlike the in situ DNRA rate, ambient NHþ 4N concentrations beneath Cell 2 were similar (p > 0.05). DNRA was
found to be significantly related to TC/NO
3 -N, pH, NO3 -N and DO (Eqn. (11)). log DNRA ¼
6.5 þ 0.07 TC/NO
3
N þ 1.27 2 pH þ 0.05NO
3
N þ 0.19DO; (Adjusted R ¼ 0.84; p < 0.001; n ¼ 9) (11)

3.5. Total NO
3 -N reduction rate Through the coupling of an in situ 15NO
3 -N push-pull experiment and use of a conservative tracer, it was possible to calculate 15 the total 15NO
NO
3 -N consumption, incorporating 3 -N attenuation by both physical (dilution/dispersion) and biochemical processes. Irrespective of cell, reduction in NO
3 -N concentrations by dilution/dispersion/diffusion was higher in the deep groundwater compared to the shallow groundwater. In contrast, biological NO
3N consumption was higher in shallow groundwater than deep groundwater. Nitrate consumption by biochemical processes was significantly higher (p < 0.05) in shallow than in deep groundwater (Table 2). Significantly higher biochemical NO
3 -N consumption was observed in the groundwater beneath Cell 2 (ca. 4593 ± 938 and 1824 ± 273 mg N kg
1 d
1 in shallow and deep groundwater, respectively) than beneath Cell 3 (ca. 2473 ± 863 and

M.M.R. Jahangir et al. / Water Research 111 (2017) 254e264

261


1
Fig. 5. Push Pull NHþ 4 -N and NO2 -N concentrations (mg N L ) in initially injected groundwater solutions plotted against cumulative pull volume/total push volume in shallow and deep groundwater from the 6-h in situ NO
3 -N pushepull test; n ¼ 3.

1732 ± 328 mg N kg
1 d
1 in shallow and deep groundwater, respectively). No significant interaction effect of Cell and depth were observed (p > 0.05). 3.6. Contribution of denitrification and DNRA to total NO
3 -N consumption

Relative proportion of biochem nitrate removal processes

Mean NO
3 -N consumption by denitrification over the total biochemical NO
consumption (denitrification rate/total 3 -N biochemical NO
3 -N consumption rate) during the incubation period was similar in shallow and deep groundwater (p > 0.05) (Fig. 6). Considering multiple comparisons, NO
3 -N consumption by denitrification over the total biochemical NO
3 -N consumption differed significantly between Cell 2 (ca. 0.23 ± 0.01 and 0.19 ± 0.09 in shallow and deep groundwater, respectively) and Cell 3 (ca. 0.08 ± 0.05 and 0.10 ± 0.04 in shallow and deep groundwater,

1.00 0.80

Other biochem processes DNRA

0.60

Denitrification

0.40 0.20

respectively) (p < 0.05). Mean NO
3 -N consumption by DNRA was higher than denitrification (p < 0.001). Nitrate consumption by DNRA over the total biochemical NO
3 -N consumption differed significantly between the two depths of groundwater, being higher in shallow than in deep groundwater (p < 0.05) (Fig. 6). It was significantly lower beneath Cell 2 (ca. 0.50 ± 0.13 and 0.39 ± 0.18 in shallow and deep groundwater, respectively) than beneath Cell 3 (ca. 0.63 ± 0.10 and 0.41 ± 0.06 in shallow and deep groundwater, respectively). The contribution of other processes (e.g. immobilization and fixation by soil matrix) was much lower when compared to denitrification and DNRA, accounting for 1078 ± 468 and 728 ± 144 mg N kg
1 d
1 in shallow and deep groundwater, respectively. Nitrate consumption by other biochemical processes over the total biochemical consumption (rate of other biochemical processes/total biochemical NO
3 -N consumption rate) was significantly lower in shallow than deep groundwater (p < 0.05). Equally, it was lower in Cell 2 (ca. 0.27 ± 0.21 and 0.42 ± 0.19 in shallow and deep groundwater, respectively) than in Cell 3 (ca. 0.28 ± 0.12 and 0.49 ± 0.12 in shallow and deep groundwater, respectively) (Fig. 6). 4. Discussion Investigations into the environmental benefits of CWs focus primarily on the within wetland nutrient attenuation process (Jahangir et al., 2016). Uniquely the current study focuses on the effect of CW on the underlying groundwater quality and the effect that recharge from the CW has on groundwater NO
3 -N transformations.

0.00 Cell 2

Cell 3

Shallow

Cell 2

Cell 3

Deep

Fig. 6. Relative contributions of denitrification, DNRA and other biochemical processes 15 to the total biochemical NO
NO
3 -N consumption during the in situ 3 -N push pull test at two different groundwater depths below two constructed wetland cells.

4.1. Denitrification rates The groundwater denitrification rates observed in this study are comparable to denitrification rates measured in previous studies within CW microcosms (Xue et al., 1999), and sea sediments (Song et al., 2013). Availability of substrate C (energy source for

262

M.M.R. Jahangir et al. / Water Research 111 (2017) 254e264

denitrifiers, electron donors) and NO
3 -N (electron acceptor) coupled with anoxic conditions are important factors controlling denitrification. The groundwater underlying the CW had low DO and redox potential with high C content (Table 1). The groundwater denitrification rate was significantly correlated with DOC (r ¼ 0.68; p < 0.01), TC (r ¼ 0.47; p < 0.01), TN (r ¼ 0.49; p < 0.1), DON (r ¼ 0.69; p < 0.01), SO2
4 (r ¼
0.54; p < 0.05) and temperature (r ¼ 0.52; p < 0.05). There was a significant multiple regression between denitrification rate and DOC, -NO3-N, -SO2
and epH 4 (Eqn. (10)). The negative relationship between denitrification rate and groundwater NO
3 -N concentration suggests that there are other factors influencing denitrification. Lower denitrification rates can also be attributed to lower denitrifier number (Barrett et al., 2013). Other physico- chemical conditions can inhibit denitrification such as S2
/H2S, produced through SO2
4 reduction, as indicated by the negative relationship between groundwater SO2
4 and denitrification (Dalsgaard and Bak, 1994, Eqn. (10)). Denitrification rates were higher in the shallow groundwater due to higher C substrate availability. Previously, Janse and Van Puijenbroek (1998) and Needelman et al. (2007) related wetland denitrification to nutrients and organic matter availability. The effect of pH on denitrification has been highlighted previously; Bakken et al. (2012) showed that N2O reduction to N2 decreased with pH. The variation in denitrification rate between the different wetland cells is likely to result from the different ambient groundwater C and N concentrations and aquifer ks rates. The effect of nutrient substrate control on denitrification has been widely reported (Grebliunas and Perry, 2016; McCarty et al., 2007; Sirivedhin and Gray, 2006). Aquifer ks has been previously negatively correlated with groundwater denitrification rate (Fenton et al., 2009, 2011; Jahangir et al., 2013b). The lower N2O-N/(N2O-NþN2-N) ratios in shallow groundwater and beneath Cell 2 indicated their higher potential for complete denitrification than observed in the deep groundwater and beneath Cell 3. Incomplete denitrification in groundwater beneath the CW can enhance N2O emissions and thus diminish environmental benefits by causing pollution swapping (Jahangir et al., 2016; IPCC, € m et al., 2007). However, ultimate exchange of N2O 2014; Stro produced in deep groundwater with the atmosphere can be low because further reduction to N2 may occur during transport. Riya et al. (2010) showed indirect emissions of N2O accounted for 2.69% of the total N2O emissions in CWs. 4.2. Dissimilatory nitrate reduction to ammonium (DNRA) rates The high DNRA rate observed in this system has important implications for the effect of CW on the underlying groundwater hydrogeochemistry. Comparisons of the estimated DNRA rate with other studies are very limited due to lack of in situ data. However, these results correspond (within the same order of magnitude) to the DNRA rates measured in freshwater sediments (Scott et al., 2008) and estuaries (Gardner et al., 2006). Both shallow and deep groundwater underlying this CW system had highly reduced conditions, redox potential ranging from
29 to
36 mV and DO from 0.46 to 0.53 mg L
1. The microbes that carry out DNRA favour more reduced environments (Takaya, 2002), higher C/N ratios and pH (Kraft et al., 2015; Yoon et al., 2015, 2015). The DNRA rate decreased with the depth of groundwater, and correlated with DOC concentrations (r ¼ 0.37; p < 0.05), TC concentrations (r ¼ 0.64; p < 0.05), TN (r ¼ 0.53; p < 0.01); DON (r ¼ 0.64; p < 0.01), ambient NHþ 4 -N (r ¼ 0.46; p < 0.05), pH (r ¼ 0.43; p < 0.05), SO2
4 (r ¼ 0.49; p < 0.01) and temperature (r ¼ 0.74; p < 0.05). The differences in the groundwater DNRA rate below Cell 2 and Cell 3 could be attributed to the higher C concentrations beneath Cell 2 than in Cell 3. Even though

the NO
3 -N injection rate was similar beneath all cells, the higher dilution/dispersion in Cell 3 may decrease NO
3 -N concentrations that might have resulted in limited NO
3 -N availability for DNRA and thus a decrease in DNRA rate beneath this Cell. The positive relationship with DNRA and ambient SO2
suggests that 4 enhancement of DNRA by sulphate reducing bacteria can occur in such a system (Gardner et al., 2006; Giblin et al., 2013; Rysgaard et al., 1996). Ammonium produced by DNRA can remain in that form only until it has contact with an aerobic environment (Tesoriero et al., 2000) after which it is oxidised to NO
3 -N (Thayalakumaran et al., 2008). However, the fate of NHþ 4 -N produced in groundwater beneath the CW needs to be investigated further to ascertain its potential environmental impact. 4.3. Relative contribution of denitrification and DNRA to biochemical NO
3 -N removal Total NO
3 -N consumption by biochemical processes decreased with depth, being higher in shallow than that of deep groundwater. This can be attributed to the lower C substrates and the associated C/N ratios. In the deep groundwater availability of NO
3 -N decreased quickly to microbes due to the quicker dilution and dispersion, being with higher saturated hydraulic conductivity and lower tracer recovery than the shallow ground water. Past research showed that saturated hydraulic conductivity corresponds to lower denitrification rate (Fenton et al., 2011). Beneath the CW coupled denitrification and DNRA accounted for NO
3 -N consumption of 79 and 54% in shallow and deep groundwater, respectively. These are within the range of NO3-N consumption of 6e99% reported by Washbourne et al. (2011) in a lake ecotype. DNRA was the dominant consumption process accounting for 63 and 40% of the total biochemical N consumption in shallow and deep groundwater, respectively. Recent studies have suggested that DNRA rather than denitrification can be a more important process in wetland sediments (Burgin and Hamilton, 2008) and coastal ecosystems (Giblin et al., 2013). Burgin and Hamilton (2007) reported that DNRA rates were higher than denitrification is wetland and lake ecotypes. The differences between denitrification and DNRA may be due to the availability of organic matter, because DNRA is favoured when high C:N ratios prevail (especially high TC/NO
3 -N ratio) and denitrification is favoured when carbon supplies are limiting in comparison with N (Korom, 2002; Kelso et al., 1997). The TC/NO
3N has been revealed as one of the important predictors of DNRA (Eqn. (11)). Generally DNRA is favoured in a system when NO
3 -N becomes limiting in comparison with C (Burgin and Hamilton, 2007). Yoon et al. (2015) argued that denitrification occurs under electron donor-limiting conditions (that is, low C/N ratios), whereas respiratory ammonification is preferred under N oxyanion-limiting conditions (that is, high C/N ratios). It is again supported by Fazzolari et al. (1998) who demonstrated that high þ glucose-to-NO
3 ratios increased NH4 -N and lowered N2O production. In arable soil, Schmidt et al. (2011) observed a strong correlation between C/N ratio and respiratory ammonification activity. Mean TC/NO
3 -N ratio of the samples analysed for denitrification and DNRA rates was 13.4 ± 1.9 (SE). The environmental factors that affected denitrification and DNRA were common for both processes (i.e. DOC, NO
3 -N, pH and temperature) other than þ the TC/NO
3 -N and ambient NH4 -N which was positively correlated with DNRA but did not affect denitrification. This indicates that DNRA was either enhanced by the abundance of DNRA performing bacteria which were stimulated by the addition of NO
3 -N. Higher DNRA to total biochemical NO
3 -N reduction may also be attributed to the presence of macrophytes. Certain macrophytes may greatly increase the proportion of DNRA to denitrification,

M.M.R. Jahangir et al. / Water Research 111 (2017) 254e264

possibly due to increased C availability from root exudates and elevated O2 levels (Nijburg and Laanbroek, 1997). Aerenchymatous plants release O2 into the root zone (Nijburg et al., 1997), and this in turn tends to enhance DNRA over denitrification as DNRA is less inhibited by O2 presence than denitrification, especially at high C:N ratios (Fazzolari et al., 1998). Our findings are in agreement with previous research pertaining to comparative studies on denitrification and DNRA e.g. Pett-Ridge et al. (2006) showed that DNRA is less sensitive to O2 level than denitrification. Multiple regression revealed that, within such a DO level (<1.0 mg L
1), predicted DNRA rate increased with DO. Other biochemical processes contributed to 28 and 46% of total biochemical N transformations in shallow and deep groundwater, respectively. Such processes may include microbial assimilation, plant uptake and fixation in soil clay minerals (Sahrawat, 2004; Clement et al., 2002). Plant N uptake may occur to a limited extent within the shallow groundwater zone but would be unlikely in the deeper groundwater. Aquifer media NHþ 4 -N fixation might be higher in the deep groundwater due to less saturation of the exchange sites, resulting from lower ambient NHþ 4 -N concentrations than the shallow groundwater. Microbial assimilation would be expected to be lower in the deep groundwater as microbial abundance has been reported to decrease with depth (Barrett et al., 2013). The recharge from the CW to the underlying groundwater directly influenced groundwater NO
3 -N transformations and consumption. Traditionally CWs have been used to treat waste waters from a range of municipal and agricultural sources. The current study found that CWs can also contribute to further attenuation of NO
3 -N pollution in the regional groundwater bodies through enhanced denitrification and DNRA. However, DNRA resulted in elevated groundwater NHþ 4 -N occurrence in the groundwater below the CWs but this NHþ 4 -N could be attenuated through further N transformations such as nitrification, denitrification or anammox. While NHþ 4 -N is an important water pollutant, the environmental fate and transport of the NHþ 4 -N generated in the groundwater is uncertain. 5. Conclusions On the basis of this study, ICWs influence NO
3 -N attenuation in the groundwater beneath them and solely focusing on within wetland NO
3 -N attenuation can underestimate the environmental benefits of wetlands. In this study, Dissimilatory NO
3 -N reduction
to NHþ 4 -N and denitrification were major NO3 -N consumption processes in the groundwater beneath the integrated constructed wetland and whose rates reduced with depth. The elevated groundwater NHþ 4 -N that was observed correlated with the DNRA þ rate indicating that NO
3 -N reduction to NH4 -N in the groundwater is an important process beneath CWs. Acknowledgements The research was funded by Irish Research Council and Department of Agriculture, Food and Marine in Association with The University of Dublin, Trinity College. The authors are very thankful to Paul Carroll at Waterford County Council, Ireland and Vesi International Ltd., Cork for facilitating access to the ICW site for conducting the experiment. We would like to extend our gratitude to Denis Brennan, Cathal Somers and John Murphy at Teagasc Environment Research Centre for their help in field and lab work. This study was also associated with the German Science Foundation research unit DASIM (FOR2337) “Denitrification in Agricultural Soils: Integrated control and Modelling at various scales”.

263

References Addy, K., Kellogg, D.Q., Gold, A.J., Groffman, P.M., Ferendo, G., Sawyer, C., 2002. In situ push-pull method to determine groundwater denitrification in riparian zones. J. Environ. Qual. 31, 1017e1024. Askew, F.E., Smith, R.K., 2005. Inorganic non-metallic constituents, 4500-SO2
4 ; Sulphate; Method 4500-SO2
4 . turbimetric method. In: Eaton, et al. (Eds.), Standard Methods for the Examination of Waters and Waste Water, twenty-first ed. American Public Health Association, NW Washington, pp. 4e188. ISBN 087553-047-8 (2001e3710). Bakken, L.R., Bergaust, L., Liu, B., Frostegård, A., 2012. Regulation of denitrification at the cellular level: a clue to the understanding of N2O emissions from soils. Philos. Trans. R. Soc. B 367, 1226e1234. Barrett, M., Jahangir, M.M.R., Lee, C., Smith, C.J., Bhreathnach, N., Collins, G., Richards, K.G., O'Flaherty, V., 2013. Abundance of denitrification genes under different peizometer depths in four Irish agricultural groundwater sites. Environ. Sci. Pollut. Res. 20, 6646e6657. Beaulieu, J.J., Tank, J.L., Hamilton, S.K., Wollheim, W.M., Hall, R.O., Mulholland, P.J., 2011. Nitrous oxide emission from denitrification in stream and river networks. Proc. Natl. Acad. Sci. 108, 214e219. Brooks, P.D., Stark, J.M., McInteer, B.B., Preston, T., 1989. Diffusion method to prepare soil extracts for automated N-15 analysis. Soil Sci. Soc. Amer. J. 53, 1707e1711. Burgin, A.J., Hamilton, S.K., 2007. Have we overemphasized the role of denitrification in aquatic ecosystems? A review of nitrate removal pathways. Front. Ecol. Environ. 5 (2), 89e96. 2
Burgin, A.J., Hamilton, S.K., 2008. NO
3 driven SO4 production in freshwater ecosystems: implications for N and S cycling. Ecosystem 11 (6), 908e922. Burgin, A.J., Hamilton, S.K., Gardner, W.S., McCarthy, M.J., 2013. Nitrate reduction, denitrification, and dissimilatory nitrate reduction to ammonium in wetland sediments. In: DeLaune, R.D., Reddy, K.R., Richardson, C.J., Megonigal, J.P. (Eds.), Methods in Biogeochemistry of Wetlands, vol. 10. SSSA Book Series, Madison, USA, pp. 307e325. Carleton, J.N., Grizzard, T.J., Godrej, A.N., Post, H.E., Lampe, L., Kenel, P.P., 2000. Performance of a constructed wetlands in treating urban stormwater runoff. Water Environ. Res. 72 (3), 295e304. Clement, J.C., Pinay, G., Marmonier, P., 2002. Seasonal dynamics of denitrification along topohydrosequesences in three different riparian wetlands. J. Environ. Qual. 31, 1025e1037. Craig, M., Mannix, A., Daly, D., 2010. Groundwater quality. In: McGarrigle, M., Lucey, J., O Cinneide, M. (Eds.), Water Quality in Ireland, 2007-2009. Environmental Protection Agency, Ireland, pp. 15e40. Dalsgaard, T., Bak, F., 1994. Nitrate reduction in a sulfate-reducing bacterium, desulfovibrio desulfuricans, isolated from rice paddy soil: sulfide inhibition, kinetics, and regulation. Appl. Environ. Microbiol. 60, 291e297. Davidson, E.A., Firestone, R.K., 1988. Measurement of nitrous oxide dissolved in soil solution. Soil Sci. Soc. Am. J. 52, 1201e1203. Davies, C.M., Bavor, H.J., 2000. The fate of stormwater-associated bacteria in constructed wetland and water pollution control pond systems. J. Appl. Microbiol. 89, 349e360. Dzakpasu, M., Scholz, M., Harrington, R., Jordan, N.J., McCarthy, V., 2012. Characterising infiltration and contaminant migration beneath earthen-lined integrated constructed wetlands. Ecol. Eng. 41, 41e51. Dzakpasu, M., Scholz, M., Harrington, R., McCarthy, V., Jordan, S.N., 2014. Groundwater quality impacts from a full-scale integrated constructed wetland. Groundw. Monit. Rem. 34 (3), 51e64. Fazzolari, E., Nicolardot, B., Germon, J.C., 1998. Simultaneous effects of increasing levels of glucose and oxygen partial pressures on denitrification and dissimilatory nitrate reduction to ammonium in repacked soil cores. Eur. J. Soil Biol. 34, 47e52. Fenton, O., Healy, M.G., Henry, T., Khalil, M.I., Grant, J., Baily, A., Richards, K.G., 2011. Exploring the relationship between groundwater geochemical factors and denitrification potentials on a dairy farm in southeast Ireland. Ecol. Eng. 37, 1304e1313. Fenton, O., Richards, K.G., Kirwan, L., Khalil, M.I., Healy, M.G., 2009. Factors affecting nitrate distribution in shallow groundwater under a beef farm in South Eastern Ireland. J. Environ. Manag. 90, 3135e3146. Freeze, R.A., Cherry, J.A., 1979. Ground Water. Prentice Hall, New Jersey. Gardner, W.S., McCarthy, M.J., An, S., Sobolev, D., 2006. Nitrogen fixation and dissimilatory nitrate reduction to ammonium (DNRA) support nitrogen dynamics in Texas estuaries. Limnol. Oceanogr. 51, 558e568. Giblin, A.E., Tobias, C.R., Song, B., Weston, N., Banta, G.T., Rivera-Monroy, V.H., 2013. The importance of dissimilatory nitrate to ammonium (DNRA) in the nitrogen cycle of coastal ecosystems. Oceanography 26 (3), 124e131. Gill, L.W., Ring, P., Higgins, N., Johnston, P.M., 2014. Accumulation of heavy metals in a constructed wetland treating road runoff. Ecol. Eng. 70, 133e139. Grebliunas, B.D., Perry, W.L., 2016. The role of C: N:P stoichiometry in affecting denitrification in sediments from agricultural surface and tile-water wetlands. Springer Plus 5, 359. http://dx.doi.org/10.1186/s40064-016-1820-6. Harrington, R., Carroll, P., Carty, A.H., Keohane, J., Ryder, C., 2007. Integrated constructed wetlands: concept, design, site evaluation and performance. Int. J. Water 3 (3), 243e255. Harrison, M.D., Groffman, P.M., Mayer, P.M., Kaushal, S.S., Newcomer, T.A., 2011. Denitrification in alluvial wetlands in an urban landscape. J. Environ. Qual. 40, 634e646.

264

M.M.R. Jahangir et al. / Water Research 111 (2017) 254e264

Huygens, D., Trimmer, M., Rütting, T., Müller, C., Heppell, C.M., Lansdown, K., Boeckx, P., 2013. Biogeochemical N cycling in wetland ecosystems: 15N isotope techniques, Methods in biogeochemistry of wetlands. In: Reddy, K.R., Megonigal, J.P., Delaune, R.D. (Eds.), Soil Sci. Soc. Amer., vol. 30, pp. 553e591. Ibrahim, T.G., Fenton, O., Richards, K.G., Fealy, R.M., Healy, M.G., 2013. Spatial and temporal variations of nutrient loads in overland flow and subsurface drainage from a marginal land site in south-east Ireland. Biol. Environ. Proc. R. Ir. Aca. 113B (2), 1e18. IPCC, 2014. IPCC 2013 Supplement to the 2006 IPCC guidelines for national greenhouse gas inventories: 675 wetlands. In: Hirashi, T., Krug, T., Tanabe, K., Srivastava, N., Baasansuren, J., 676 Fukuda, M., Troxler, T.G. (Eds.), Switzerland, IPCC Task Force on National Greenhouse Gas 677 Inventories, 354. IPCC, 2013. Working group I contribution to the IPCC fifth assessment report climate change 2013: the physical science basis summary for policymakers. Intergov. Panel Clim. Change 36. Jahangir, M.M.R., Fenton, O., Gill, L., Mueller, C., Johnston, P., Richards, K.G., 2016. Carbon and nitrogen dynamics and greenhouse gas emissions in constructed wetlands treating wastewater: a review. Hydrol. Earth Syst. Sci. 20, 109e123. Jahangir, M.M.R., Johnston, P., Addy, K., Khalil, M.I., Groffman, P.M., Richards, K.G., 2013a. Quantification of in situ denitrification rates in groundwater below an arable and a grassland system. Water Air Soil Pollut. 24, 1693. Jahangir, M.M.R., Johnston, P., Barrett, M., Khalil, M.I., Groffman, P., Boeckx, P., Fenton, O., Murphy, J., Richards, K.G., 2013b. Denitrification and indirect N2O emissions in groundwater: hydrologic and biogeochemical influences. J. Contam. Hydrol. 152, 70e81. Jahangir, M.M.R., Johnston, P., Grant, J., Somers, C., Khalil, M.I., Richards, K.G., 2012. Evaluation of headspace equilibration methods for measuring greenhouse gases in groundwater. J. Environ. Manag. 111, 208e212. Janse, J.H., Van Puijenbroek, P.J.T.M., 1998. Effects of eutrophication in drainage ditches. Environ. Pollut. 102, 547e552. Kana, T.M., Darkangelo, C., Hunt, M.D., Oldham, J.B., Bennett, G.E., Cornwell, J.C., 1994. Membrane inlet mass spectrometer for rapid high precision determination N2, O2 and Ar in environmental water samples. Anal. Chem. 66, 4166e4170. Kartal, B., Kuypers, M.M.M., Lavik, G., Schalk, J., Op den Camp, H.J.M., Jetten, M.S.M., Strous, M., 2007. Anammox bacteria disguised as denitrifiers: nitrate reduction to dinitrogen gas via nitrite and ammonium. Environ. Microbiol. 9 (3), 635e642. Kellogg, D.Q., Gold, A.J., Groffman, P.M., Addy, K., Stolt, M.H., Blazejewski, G., 2005. In situ groundwater denitrification in stratified, permeable soils underlying riparian wetlands. J. Environ. Qual. 34, 524e533. Kelso, B.H.L., Smith, R.V., Laughlin, R.J., Lennox, S.D., 1997. Dissmilatory nitrate reduction in anaerobic sediments leading to river nitrate accumulation. Appl. Environ. Microbiol. 63 (12), 4679e4685. Korom, S.F., 2002. Natural denitrification in the saturated zone: a review. Water Resour. Res. 28 (6), 1657e1668. Kraft, B., Tegetmeyer, H.E., Sharma, R., Klotz, M.G., Ferdelman, G., Hettich, R.L., Geelhoed, J.S., Strous, M., 2015. The environmental controls that govern the end product of bacterial nitrate respiration. Science 345 (6197), 676e679. Lee, C., Fletcher, T.D., Sun, G., 2009. Nitrogen removal in constructed wetland systems. Eng. Life Sci. 9 (1), 11e22. Lemon, E., 1981. Nitrous oxide in freshwaters of the Great Lakes basins. Limnol. Ocean. 26, 867e879. McCarty, G.W., Mookherji, S., Angier, J.T., 2007. Characterization of denitrification activity in zones of groundwater exfiltration within a riparian wetland ecosystem. Biol. Fertil. Soils 43, 691e698. Mian, I.A., Riaz, M., Begum, S., Bhatti, A., Cresser, M., 2009. Ammonium: a mobile cation in N -impacted soils? Soil quality¼environment quality. In: Creamer, R., Holden, N., Griffiths, B., Schmidt, O., Richards, K., et al. (Eds.), Conference Program and Book of Abstracts of the Joint SSSI/BSSS Annual Meeting 2009. SSSI/ BSSS, Johnstown Castle, Wexford, Ireland, p. 97. Mosier, A.R., Klemedtsson, L., 1994. Measuring denitrification in the field. In: Weaver, et al. (Eds.), Methods of Soil Analysis. Part 2, pp. 1047e1065. SSSA Book Ser 5. Mosier, A.R., Schimel, D.S., 1993. Nitrification and denitrification. In: Knowles, R., Blackburn, T.H. (Eds.), Nitrogen Isotope Techniques. Academic Press, Orlando, pp. 181e208. Necpalova, M., Fenton, O., Casey, I., Humphreys, J., 2012. N leaching to groundwater from dairy production involving grazing over the winter on a clay-loam soil. Sci. Total Environ. 432, 159e172. Needelman, B., Kleinman, P., Strock, J., 2007. Improved management of agricultural drainage ditches for water quality protection: an overview. J. Soil. Water Conserv. 62, 171e178. Nijburg, J.W., Coolen, M.J.L., Gerards, S., Klein Gunnewiek, P.J.A., Laanbroek, H., 1997. Effects of nitrate availability and the presence of glyceria maxima on the composition and activity of the dissimilatory nitrate-reducing bacterial community. Appl. Environ. Microbiol. 63 (3), 931e937.

Nijburg, J.W., Laanbroek, H.J., 1997. The fate of 15N-Nitrate in Healthy and Declining Phragmites australis stands. Microb. Ecol. 34, 254e262. O'Luanaigh, N.D., Goodhue, R., Gill, L.W., 2010. Nutrient removal from on-site domestic wastewater in horizontal subsurface flow reed beds in Ireland. Ecol. Eng. 36, 1266e1276. Ockenden, M.C., Deasy, C., Quinton, J.N., Surridge, B., Stoate, C., 2014. Keeping agricultural soil out of rivers: evidence of sediment and nutrient accumulation within field wetlands in the UK. J. Environ. Manag. 135, 54e62. Pett-Ridge, J., Silver, W.L., Firestone, M.K., 2006. Redox fluctuations frame microbial community impacts on N-cycling rates in humid tropical forest soil. Biogeochemistry 81, 95e110. Riya, S., Zhou, S., Nakashima, Y., Terada, A., Hosomi, M., 2010. Direct and indirect greenhouse gas emissions from vertical flow constructed wetland planted with forage rice. Kagaku Kogaku Ronbunshu 36, 229e236. Rütting, T., Boeckx, P., Müller, C., Klemedtsson, L., 2011. Assessment of the importance of dissimilatory nitrate reduction to ammonium for the terrestrial nitrogen cycle. Biogeoscience 8, 1169e1196. Rysgaard, S., Risgaard-Petersen, N., Sloth, N.P., 1996. Nitrification, denitrification, and nitrate ammonification in sediments of two coastal lagoons in southern France. Hydrobiologia 329, 133e141. Sahrawat, K.L., 2004. Ammonium production in submerged soils and sediments: the role of reducible iron. Commun. Soil Sci. Plant Anal. 35, 399e411. SAS Institute Inc, 2009. SAS/STAT 9.2 User's Guide, second ed. SAS Institute Inc., Cary, NC. Saxton, K.E., Rawls, W.J., Romberger, J.S., Papendick, J.I., 1986. Estimating generalized soil-water characteristics from texture. Soil Sci. Soc. Am. J. 50 (4), 1031e1036. Schmidt, C.S., Richardson, D.J., Baggs, E.M., 2011. Constraining the conditions conducive to dissimilatory nitrate reduction to ammonium in temperate arable soils. Soil Biol. Biochem. 43, 1607e1611. Scholz, M., Lee, B.H., 2005. Constructed wetlands: a review. Int. J. Environ. Stud. 62, 421e447. Scott, J.T., McCarthy, M.J., Gardner, W.S., Doyle, R.D., 2008. Denitrification, dissimilatory nitrate reduction to ammonium, and nitrogen fixation along a nitrate concentration gradient in a created freshwater wetland. Biogeochemistry 87, 99e111. Sirivedhin, T., Gray, K.A., 2006. Factors affecting denitrification rates in experimental wetlands: field and laboratory studies. Ecol. Eng. 26, 167e181. Song, G.D., Liu, S.M., Marchant, H., Kuypers, M.M.M., Lavik, G., 2013. Anammox, denitrification and dissimilatory nitrate reduction to ammonium in the East Sea sediment. Biogeoscience 10, 6851e6864. Søvik, A.K., Augustin, J., Heikkinen, K., Huttunen, J.T., Necki, J.M., Karjalainen, S.M., Klove, B., Liikanen, A., Mander, U., Puustinen, M., Teiter, S., Wachniew, P., 2006. Emission of the greenhouse gases nitrous oxide and methane from constructed wetlands in Europe. J. Environ. Qual. 35, 2360e2373. €m, L., Lamppa, A., Christensen, T.R., 2007. Greenhouse gas emissions from a Stro constructed wetland in southern Sweden. Wetl. Ecol. Manag. 15, 43e50. Takaya, N., 2002. Dissimilatory nitrate reduction metabolisms and their control in fungi. J. Biosci. Bioeng. 94, 506e510. Tanner, C.C., Sukias, J.P.S., 2011. Multi-year nutrient removal performance of three constructed wetlands intercepting drainage flows from intensively grazed pastures. J. Environ. Qual. 40, 620e633. Tesoriero, A.J., Liebscher, H., Cox, S.E., 2000. Mechanism and rate of denitrification in an agricultural watershed: electron and mass balance along groundwater flow paths. Water Resour. Res. 36 (6), 1545e1559. Thayalakumaran, T., Bristow, K.L., Charlesworth, P.B., Fass, T., 2008. Geochemical conditions in groundwater systems: implications for the attenuation of agricultural nitrate. Agril. Water Manag. 95, 103e115. Tiedje, J.M., 1982. Denitrification. In: Page, et al. (Eds.), Methods of Soil Analysis. Part 2. Agronomy Monograph No. 9, second ed. ASA and SSSA, Madison, pp. 1011e1025. van den Berg, E.M., van Dongen, U., Abbas, B., van Loosdrecht, M.C.M., 2015. Enrichment of DNRA bacteria in a continuous culture. ISME J. 9, 2153e2161. Vymazal, J., 2007. Removal of nutrients in various types of constructed wetlands. Sci. Total Environ. 380, 48e65. Washbourne, I.J., Crenshaw, C.L., Baker, M.A., 2011. Dissimilatory nitrate reduction pathways in an oligotrophic aquatic ecosystem: spatial and temporal trends. Aquat. Microb. Ecol. 65, 55e64. Xue, Y., Kovacic, D.A., David, M.B., Gentry, L.E., Mulvaney, R.L., Lindau, C.W., 1999. In situ measurements of denitrification in constructed wetlands. J. Environ. Qual. 28, 263e269. € ffler, F.E., 2015. Denitrification Yoon, S., Cruz-García, C., Sanford, R., Ritalahti, K.M., Lo versus respiratory ammonification: environmental controls of two competing

dissimilatory NO3 /NO2 reduction pathways in Shewanella loihica strain PV-4. ISME J. 9, 1093e1104.