Nutrient removal and microbial communities ...

Science of the Total Environment 484 (2014) 167–175

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Nutrient removal and microbial communities' development in a young unplanted constructed wetland using Bauxsol™ pellets to treat wastewater Laure M. Despland a,b,c,⁎, Malcolm W. Clark a,d, Tony Vancov b, Michel Aragno c a

School of Environment, Science & Engineering, Southern Cross University, PO Box 157, Lismore, NSW 2480, Australia Department of Primary Industries NSW, 1243 Bruxner Highway, Wollongbar, NSW 2477, Australia c Laboratory of Microbiology, Institute of Biology, University of Neuchâtel, PO Box 158, 2009 Neuchâtel, Switzerland d Marine Ecology Research Centre, Southern Cross University, PO Box 157, Lismore, NSW 2480, Australia b

H I G H L I G H T S • • • • •

Bauxsol™ pellets efficiently treat wastewater in an unplanted constructed wetland. Combined bio-geochemical processes remove N 95% PO34 − and ~26% nitrogen. Pellets dominate PO34 − removal and remain under-saturated. Distinct bacterial communities between pellets and soil Co-existence of ammonia-oxidising bacteria and archaea, anammox and denitrifiers

a r t i c l e

i n f o

Article history: Received 27 November 2013 Received in revised form 5 March 2014 Accepted 9 March 2014 Available online xxxx Editor: Simon Pollard Keywords: Bauxsol™ pellets Constructed wetland Municipal wastewater treatment Microbial communities' development Phosphate and nitrogen removal

a b s t r a c t Municipal wastewater was treated over a six month period in an unplanted constructed wetland with a lower soil layer and an upper Bauxsol™ pellet layer. The interactions between Bauxsol™ pellets, soil, effluent and microbial communities demonstrated a positive influence on contaminant removal. Bauxsol™ treated effluent showed N 95% phosphate removal and ~26% nitrogen removal during the trial. Substantial quantities of nitrate, trace-metals and Colwell P were bound to the pellets, whereas only ammonium was bound to the soil. The structure of microbial communities analysed by denaturing gradient gel electrophoresis (DGGE) showed distinct bacterial communities attached to Bauxsol™ pellets and soil owing to differences in geochemistry and microenvironmental conditions. Polymerase chain reaction (PCR) amplification of specific marker genes (i.e. bacterial and archaeal amoA genes, nosZ gene, and hzo gene) was used to evaluate the presence of microbial communities associated with nitrogen transformation. Data revealed the co-existence of aerobic ammonia-oxidising bacteria, anaerobic ammonia-oxidising bacteria (anammox) and denitrifiers attached to Bauxsol™ pellets and ammoniaoxidising bacteria and archaea attached to soil. This study successfully demonstrates that Bauxsol™ pellets are a suited alternative media for constructed wetland to treat wastewater effectively removing phosphate and serving as biomass support particles for bacterial communities associated with nitrogen-cycling. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Constructed wetland (CW) systems are commonly used as a tertiary or polishing treatment step to further physically, chemically and biologically remove contaminants (Kadlec and Wallace, 2009). The substrate and plant root zones provide considerable reactive surface area ⁎ Corresponding author at: School of Environment, Science & Engineering, Southern Cross University, PO Box 157, Lismore, NSW 2480, Australia. Tel.: + 612 6620 3650; fax: +612 6621 2669. E-mail addresses: [email protected], [email protected] (L.M. Despland).

http://dx.doi.org/10.1016/j.scitotenv.2014.03.030 0048-9697/© 2014 Elsevier B.V. All rights reserved.

for adsorption (i.e. ligand exchange and surface precipitation) of complexing ions such as phosphates and trace-metals. Sorption capacity depends, among other things, on the initial concentration of ions, contact time, effluent pH, substrate binding capacity, and biomass uptake capacity (Akhurst et al., 2006; Kadlec and Wallace, 2009; Rhue and Harris, 1999). Some of these limitations may be overcome by using high binding capacity substrates that maintain adequate hydraulic conductivity (Gray et al., 2000; Wood and McAtamney, 1996; Zhang et al., 2007). Moreover, the substrate materials and plant roots act as biomass supports on which microorganisms may grow and develop as biofilms (Kadlec and Wallace, 2009; Madigan et al., 2012).

168

L.M. Despland et al. / Science of the Total Environment 484 (2014) 167–175

The formation of a redox gradient in CWs (bottom part anoxic, upper part oxic), the presence of oxic/micro-oxic sheaths around plant roots, and the presence of oxic/anoxic micro-zones within porous substrates and associated biofilms allow aerobic and anaerobic microbial activities to occur in close proximity (Gobat et al., 2004; Kadlec and Wallace, 2009; Krasnits et al., 2009; Madigan et al., 2012; Stottmeister et al., 2003). These microbial activities mostly transform nitrogen in CWs via a combination of nitrification, denitrification and anammox processes, and several studies have reported the co-existence of aerobic and anaerobic bacteria involved in the nitrogen cycle in CWs (Dong and Sun, 2007; Iasur-Kruh et al., 2010; Krasnits et al., 2009; Shipin et al., 2005; Truu et al., 2005). The presence of microbial communities associated with transforming nitrogen may be determined and monitored using polymerase chain reaction (PCR) amplification of specific marker genes such as bacterial and archaeal amoA genes (nitrification), nosZ gene (denitrification) and hzo gene (anammox). We have previously reported on using this technique to reveal the co-existence of ammonia-oxidising bacteria, denitrifiers and anammox organisms in a submerged up-flow wastewater filter system (Despland et al., 2012). Denaturing gradient gel electrophoresis (DGGE) analysis of 16S rDNA genes amplified by PCR may also be used in wastewater treatment studies to observe, compare and contrast changes in microbial community structure, composition and diversity. Several groups have used PCR-DGGE analysis in CW studies to correlate changes in microbial community to specific physicochemical characteristics (e.g. pH, electrical conductivity, porosity, and nutrient content) of the substrate material (Calheiros et al., 2009; Truu et al., 2005), and spatial (depth) effects to increasing carbon, nitrogen and oxygen supply in the upper layer (Iasur-Kruh et al., 2010; Truu et al., 2005). Bauxsol™ (a circum-neutral seawater-neutralised bauxite refinery residue) is a complex mix of minerals predominantly composed of iron and aluminium oxy-hydroxides. Bauxsol™ has been extensively used for water and soil remediation owing to its physicochemical characteristics such as high surface to volume ratio (up to 100 m2/g); high binding capacity (metals: N 1500 mEq/kg; phosphorus: N2% by mass); moderate acid neutralising capacity; insolubility and high nondispersion capacity (Clark et al., 2009; Collins et al., 2014; Hanahan et al., 2004; McConchie et al., 1999). Bauxsol™ application studies on trace-metal and phosphate removal from water, acid mine drainage, acid sulfate soil, and sewage effluent show that it is cost effective and works across a broad pH range (Akhurst et al., 2006; Clark et al., 2006, 2008; Collins et al., 2014; Despland et al., 2011). However, issues associated with the low hydraulic conductivity of Bauxsol™ powders in flowing water can be overcome by utilising porous cement-bound pellets (Despland et al., 2010a). These pellets also show excellent trace-metal and nutrient binding in wastewater treatment and they act as biomass support particles (BSP) for environmental bacterial communities including groups linked to nitrogen cycle (Despland et al., 2011, 2012, 2010b). Although our previous findings have demonstrated the efficiency of Bauxsol™ pellets to treat wastewater, no work has been conducted on how microbial communities and treatment effectiveness develop over time in a model constructed wetland (i.e. Bauxsol™ pellets in contact with underlying soil). Consequently, this paper investigates and reports on the temporal and spatial phosphate and nitrogen removals, investigates the development, partitioning and structural differences of attached bacterial communities between the soil and Bauxsol™ layers, and assesses the presence/absence of microbial communities involved in the nitrogen cycle. 2. Materials and methods 2.1. Experimental design & sampling Unplanted horizontal flow wetlands were designed as slightly U-shaped, rectangular canals made of PVC (300 cm long × 50 cm

wide × 25 cm deep). The experimental canal was filled up with a 5 cm deep lower clay soil layer (i.e. soil coming from the wetland located at the South Lismore Sewage Treatment Plant (STP) NSW, Australia) and a 5 cm deep upper Bauxsol™ pellet layer (5–10 mm diameter; Despland et al., 2010a), whereas the control canal was filled with a 5 cm deep lower clay soil layer and a 5 cm deep upper basalt gravel layer (5–10 mm diameter; Fig. 1); a soil layer was introduced to simulate a wetland ecosystem, and both canals were unplanted to simplify the system and facilitate data interpretation. Secondary treated effluent (i.e. before alum dosing) from the South Lismore STP was first pumped into a 200 L drum and then delivered to both experimental and control canals by a filter/solenoid system at 16 mL/min (i.e. the same inlet water for both canals). A slope of 0.25° allowed the effluent to be gravity fed with a hydraulic retention time (HRT) of ~5 days (a typical time to remove contaminants in CW; Kadlec and Wallace, 2009; Reed et al., 1995). Outlets were drilled just above the upper Bauxsol™/gravel layer at the distal end of both canals, allowing water to discharge continuously. To ensure an even distribution and effluent flow (the hydraulic conductivity was much greater in the upper Bauxsol™/gravel layer) and utilisation of the full treatment profile, canals were fitted with an alternating T- and U-shaped baffles to increase tortuosity (Fig. 1, Figs. S1 and S2). Piezometers, with a side opening covered by a fine mesh to prevent solid infiltration, were installed mid-canal (1.5 m) and in close proximity (2.8 m) to the distal end of each canal to sample water from the lower and upper layers (Fig. 1; Fig. S1). Double layers of shade cloth placed above and around the canals prevented algal growth (without inducing complete darkness; Fig. 1); regular cleaning and system maintenance prevented particulate clogging at the inlet. Liquid samples were taken after equilibration (i.e. water flowing just at the upper layer–air interface) and then at regular intervals (T = 1, 2, 4, 6, 8, 12, 16, 20, and 24 weeks) from the inlet, from the experimental and control canal outlets, as well as from the piezometers in both lower and upper layers (Fig. S1). Piezometers were first emptied and allowed to re-fill prior to sampling to ensure sample freshness. Solid samples were taken as small sediment cores of 2 cm diameter extracted at 0.6 m (±0.15), 1.3 m (±0.15), 1.8 m (±0.15) and 2.5 m (± 0.15) from the proximal end of the canals. The sediment core consisted of three layers that were analysed separately: the upper layer (Bauxsol™ pellet in the experimental canal, gravel in the control canal), the interface layer (i.e. between the upper and lower layers, which represents a mix of Bauxsol™ pellets and soil in the experimental canal, and a mix of gravel and soil in the control canal), and the lower layer (soil in both canals) (Fig. S2). Duplicate cores at each distance were combined, using one sample from the canal edge and one from the mid-line of the canal (Fig. S2). Core voids were re-filled with fresh materials washed with Milli-Q water to avoid any pH/alkalinity spikes. Samples were first taken at 0.15 m above the distance (+ 0.15 m) to avoid any disturbance, if any, from the refilled core voids (Fig. S2). 2.2. Liquid and solid physicochemical analyses All inlet, outlet, and piezometer waters were analysed for: pH (APHA 4500 H+-b), electrical conductivity (EC) (APHA 2510-b), dissolved oxygen (DO) concentrations (APHA 2810-b), total nitrogen (TN) (APHA − 4500 N-c), ammonium (NH+ 4 ) (APHA 4500 NH3-h), nitrite (NO2 ) − − − (APHA 4500 NO2 -c), nitrate (NO3 ) (APHA 4500 NO3 -f), total phosphorus (TP) (APHA 4500 P-h), orthophosphate (PO3− 4 ) (APHA 4500 P-g) and redox potential (American Public Health Association, 1998). Inlet and outlet waters were analysed for biochemical oxygen demand (BOD) (APHA 5210-b), calcium (Ca) (APHA 3120 ICPOES), and temperature (APHA 2550-b) (American Public Health Association, 1998). Bauxsol™ pellets and soil of the experimental canal (proximal (0.6 m) and distal (2.5 m) ends) and gravel and soil of the control canal (proximal (0.6 m) and distal (2.5 m) ends) were analysed for: TP and Colwell phos− phorus (Rayment and Higginson, 1992), NH+ 4 and NO3 using a 1:10 KCl


169

Fig. 1. Canal setup: empty canal with baffles and piezometer system (top left); lower soil layer (top right); experimental canal with upper Bauxsol™ pellet layer (bottom left); control canal with upper gravel layer (bottom right).

extract (APHA 4500). Additional pre- and post-experiment solids from both the proximal and distal ends of each layer of each canal were also analysed for TN, total carbon (TC), and total organic carbon (TOC) using the LECO CNS 2000 Analyser, and inorganic carbon (IC; calculation of: TC − TOC), C/N ratio (calculation of: TOC / TN), and trace-metals including silver (Ag), arsenic (As), lead (Pb), cadmium (Cd), chromium (Cr), copper (Cu), manganese (Mn), nickel (Ni), selenium (Se), zinc (Zn), mercury (Hg), iron (Fe) and aluminium (Al) using a Nitric/HCl digest (APHA 3120 ICPMS) (American Public Health Association, 1998).

amplified fragments corresponding to the 16S rDNA gene's V3 region was undertaken on extracts containing N 20 ng/μL (i.e. all inlets, available liquids and solids from the 4th, 5th and 6th months, and soil preexperiment). DNA extracted from Bauxsol™ pellets and soil in the experimental canal and gravel and soil from the control canal at 5 and 6 months, as well as DNA extracted from inlets at 0, 5, and 6 months, were subjected to functional PCR amplification targeting specific genes involved in nitrogen transformation (bacterial and archaeal amoA genes, nosZ gene, and hzo gene). Details of the PCR/DGGE setting and primer sets used are described in Despland et al. (2012).

2.3. DNA extraction, PCR/DGGE on 16S rDNA genes, and functional PCR amplification of genes involved in the nitrogen cycle

2.4. Statistical analyses

DNA extraction on liquids and solids was performed following the method described in Despland et al. (2012). DGGE separation of PCR

DGGE data sets (i.e. profiles) were statistically analysed using the ePRIMER (version 6) software package (Clarke and Gorley, 2006) to

10.0 9.5 9.0

pH

8.5 8.0 7.5 7.0 6.5 6.0 0-1 mth

1-2 mths

2-3 mths

3-4 mths

4-5 mths

5-6 mths

Time Fig. 2. pH of water samples taken from the inlet (–∎–), the control canal outlet (–○–), the experimental canal outlet (–✳–), the experimental canal at mid-canal (1.5 m) in the Bauxsol™ layer (histogram vertical line) and in the soil layer (histogram horizontal line), at the distal end of the canal (2.8 m) in the Bauxsol™ layer (histogram vertical dash) and in the soil layer (histogram horizontal dash) between 0 and 1, 1 and 2, 2 and 3, 3 and 4, 4 and 5, and 5 and 6 months. ±Standard error.

170


100

PO43- removal [%]

90 80 70 60 50 40 30 0-1 mth

1-2 mths

2-3 mths

3-4 mths

4-5 mths

5-6 mths

Time Fig. 3. Percentage removal of orthophosphate in the experimental canal outlet (–✳–) and in the control canal outlet (–○–) between 0 and 1, 1 and 2, 2 and 3, 3 and 4, 4 and 5, and 5 and 6 months.

compare bacterial communities between the experimental and control canals, and within canal layers. Sample richness (i.e. average total number of bands) and Shannon diversity index (H′) were calculated using the same software. Non-metric multi-dimensional scaling (MDS) analysis on fourth root-transformed data and Bray–Curtis similarity matrix was used to identify the similarity between selected solid samples according to sampling time and treatment (i.e. experimental and control canals).

3. Results 3.1. Geochemistry 3.1.1. pH, EC, redox potential, dissolved oxygen and temperature Water pH, electrical conductivity and redox potential readings in the inlet and the control canal outlet were stable during the entire trial (6.8 ± 0.1, 683 ± 29 μS, and +189 ± 2 mV in the inlet; 7.4 ± 0.1, 744 ± 43 μS, and +189 ± 2 mV in the control canal outlet; Fig. 2). No differences were observed between the gravel and soil layers or between the mid-canal (1.5 m) and the distal end of the canal (2.8 m). In contrast, the experimental canal outlet showed increased pH readings over time (0–1 month 8.3 ± 0.4; 5–6 months 9.6 ± 0.1). Up until 4 months, water pH at the distal end of this canal (2.8 m) was at least one pH unit lower and thereafter a shift was observed (Fig. 2). An electrical conductivity spike (up to 14,180 μS) was registered in the experimental canal outlet over the first 4 months. In the experimental canal outlet the redox potential was around + 209 ± 9 mV over the first 4 months that then declined to around +175 ± 3 mV.

On average dissolved oxygen concentration was stable during the trial in all waters (inlet 2.2 ± 0.3 mg/L, experimental canal outlet 4.8 ± 0.4 mg/L, and control canal outlet 3.5 ± 0.2 mg/L). Data also show that the average temperature in the inlet, experimental canal outlet and control canal outlet increased with time (0–1 month 16.9 ± 0.7 °C; 5–6 months 27.1 ± 1.9 °C) coinciding with a change in season to summer. 3.1.2. Phosphorus and trace-metals Total phosphorus (TP) concentration in the inlet was primarily orthophosphate (86%) and fluctuated from 4.0 mg/L to 6.9 mg/L. In the experimental canal, phosphate removal averaged 97.6 ± 0.9% over the entire trial (Fig. 3). TP and Colwell P were predominantly bound to Bauxsol™ pellets, and the uptake increased over time without signs of saturation from 0 mg/kg to 400 ± 69 mg/kg for TP and from 22 ± 13 mg/kg to 310 ± 36 mg/kg for Colwell P (Table 1A and B). The soil in the experimental canal showed only moderate binding (up to 58 ± 6 mg/kg for TP and 16 ± 1 mg/kg of Colwell P; Table 1A and B). In contrast, the control canal had a much lower phosphate removal (average of 52.4 ± 11.3%; Fig. 3) and had a high TP loss (up to 1 ± 0.25 g/kg or 0.1%). However, the soil in the control canal had a much higher TP loading (up to 495 ± 67 mg/kg) than the soil in the experimental canal (Table 1A). Bauxsol™ pellets also showed high trace-metal binding, where As, Pb, Cd, Cr, Cu, Ni, Se, Zn, Mn, and Al were effectively bound to the pellets in the experimental canal (Table S1). In contrast, the control canal revealed that gravel lost all trace-metals except Cr and some Al. Moreover, soil in the control canal did not bind trace-metals (except Mn and Zn) unlike the soil in the experimental canal (Table S1).

Table 1 Average bound total phosphorus (A) and Colwell phosphorus (B) on solids from the experimental canal (Bauxsol™ pellets and soil) and from the control canal (gravel and soil) between 0 and 1, 1 and 2, 2 and 3, 3 and 4, 4 and 5, and 5 and 6 months. All data are in mg/kg. a ±Standard error. 0–1 month

1–2 months

2–3 months

3–4 months

4–5 months

5–6 months

A) Bound total phosphorus Experimental Bauxsol™ Soil Control Gravel Soil

Canal

−9.7 7.9 −1088 287

± ± ± ±

32.4 11.5 247 111

88 ± 14 ± −884 ± 257 ±

19 23 184 76

203 20 −865 193

± ± ± ±

126 44 302 16

342 58 −464 282

± ± ± ±

13 6 99 104

343 56 −547 407

± ± ± ±

12 4 182 21

400 56 −765 496

± ± ± ±

69 4 36 67

B) Bound Colwell phosphorus Experimental Bauxsol™ Soil Control Gravel Soil

22 9.2 0.8 4.1

± ± ± ±

13 2.9 1.4 1.6

89 16 5.9 3.9

28 1 5.3 1.3

167 14 14 5.2

± ± ± ±

23 2 1 2.6

202 10 14 9.6

± ± ± ±

12 2 2 1.8

244 13 15 13

± ± ± ±

30 4 1 2

310 14 16 16

± ± ± ±

36 3 2 1

a

Material/layer

± ± ± ±

Negative values represent a loss in mass relative to an earlier condition, i.e., solids are a source rather than a sink.


B

35 30

NH4+ (mg/L)

TN (mg/L)

A

25 20 15 10 0-1 mth

1-2 mths

2-3 mths

3-4 mths

4-5 mths

171

18 16 14 12 10 8 6 4 2 0 0-1 mth

5-6 mths

1-2 mths

2-3 mths

Time

D

7 6 5 4 3 2 1 0

4-5 mths

5-6 mths

4-5 mths

5-6 mths

18

NO3- (mg/L)

NO2- (mg/L)

C

3-4 mths

Time

16 14 12 10 8

0-1 mth

1-2 mths

2-3 mths

3-4 mths

4-5 mths

0-1 mth

5-6 mths

1-2 mths

2-3 mths

Time

3-4 mths

Time

Fig. 4. Nitrogen concentration (mg/L) in the inlet (–∎–), the experimental canal outlet (–✳–), and the control canal outlet (–○–) between 0 and 1, 1 and 2, 2 and 3, 3 and 4, 4 and 5, and 5 − − and 6 months. A) Total nitrogen, B) ammonium (NH+ 4 ), C) nitrite (NO2 ), and D) nitrate (NO3 ). ±Standard error.

3.1.3. Nitrogen Average total nitrogen (TN) concentration was 25.0 ± 2.3 mg/L for the inlet (a decrease was recorded after 3 months, linked to an ammonium decrease), while TN was 18.4 ± 1.2 mg/L in the experimental canal outlet and 17.9 ± 1.3 mg/L in the control canal outlet; this was consistent between the upper and lower layers of both canals (Fig. 4A). Consequently, the total nitrogen reduction rates were ~ 26% and ~ 28% in the experimental and control canal outlets, respectively. Despite ammonium (NH+ 4 ) concentration variability in the inlet (2.5 to 16.9 mg/L NH+ 4 -N), almost all ammonium was removed after 2 months by both canals (b 1 mg/L NH+ 4 -N in the experimental outlet and b0.1 mg/L NH+ 4 -N in the control outlet; Fig. 4B). Moreover, both canal soils became loaded with ammonium (up to 72 mg/kg NH+ 4 -N in the experimental canal, and up to 87 mg/kg NH+ 4 -N in the control canal; Table 2A). Nitrite (NO− 2 ) concentrations were 1.0 ± 0.2 mg/L NO− 2 -N in the inlet during the first month and around 0.4 ± 0.1 mg/L NO− 2 -N thereafter (Fig. 4C). In the experimental canal outlet, nitrite concentrations were above 1.5 mg/L NO− 2 -N until 4 months (spike to − 9.6 mg/L NO− 2 -N at week 6), and around 0.3 ± 0.1 mg/L NO2 -N thereafter. In contrast, the control canal outlet had nitrite concentrations − b0.2 mg/L NO− 2 -N after 1 month (Fig. 4C). Nitrate (NO3 ) concentrations -N in the inlet (Fig. 4D). Up until averaged 12.6 ± 0.9 mg/L NO− 3 2 months, the nitrate concentration in the experimental canal outlet

and control canal outlet was at 10.3 ± 0.8 mg/L NO− 3 -N and at 11.3 ± 1.2 mg/L NO− 3 -N (i.e. lower than the inlet), respectively. However, after 3 months nitrate increased and averaged 14.6 ± 1.2 mg/L NO− 3 -N and 16.1 ± 1.4 mg/L NO− 3 -N (i.e. higher than the inlet; Fig. 4D). High nitrate concentrations were found on Bauxsol™ pellets (113 ± 30 mg/kg; Table 2B). 3.1.4. C/N ratio and biochemical oxygen demand The C/N ratio (i.e. TOC/TN) of 1.9 and 20.6 was found on preexperiment Bauxsol™ pellets and soil, respectively, and it fell to 17.3 ± 3.1 on Bauxsol™ pellets and 13.4 ± 0.1 on soil in the experimental canal post-experiment. In contrast, the control canal displayed a lower C/N ratio post-experiment in the gravel layer (1.3 ± 0; C/N ratio of 0.5 on pre-experiment gravel) and a slightly higher C/N ratio in the soil layer (14.5 ± 0.8). Biochemical oxygen demand (BOD) fluctuated from 5 to 60 mg/L in the inlet and was b 5 mg/L in both canal outlets during the entire trial, consequently BOD reduction was 84 ± 5%. 3.2. Microbiology 3.2.1. Attached bacterial community DGGE analysis of the 16S rDNA amplified products from experimental canal solids showed a well-adapted bacterial community attached to

Table 2 Average bound ammonium (A) and nitrate (B) on solids from the experimental canal (Bauxsol™ pellets and soil) and from the control canal (gravel and soil) between 0 and 1, 1 and 2, 2 and 3, 3 and 4, 4 and 5, and 5 and 6 months. All data are in mg/kg. ±Standard error. Canal A) Bound ammonia Experimental Control

B) Bound nitrate Experimental Control

Material/layer

0–1 month

1–2 months

2–3 months

3–4 months

4–5 months

5–6 months

Bauxsol™ Soil Gravel Soil

8.6 35 15 31

± ± ± ±

1.4 20 7 19

15 55 27 48

± ± ± ±

8 20 5 20

26 52 41 52

± ± ± ±

4 12 10 21

14 52 38 62

± ± ± ±

7 12 13 10

8.8 52 25 62

± ± ± ±

1.5 11 0 10

16 72 33 87

± ± ± ±

6 9 7 14

Bauxsol™ Soil Gravel Soil

137 1.6 6.7 4.9

± ± ± ±

97 1.7 5.6 3.4

85 0.03 3.2 1.6

± ± ± ±

7 0.24 0.7 1.0

57 ± 0.8 ± 3.3 ± 2.8 ±

18 0.4 1.4 1.7

77 0.9 2.0 2.8

± ± ± ±

38 0.3 0.1 1.7

91 1.0 2.6 0.9

± ± ± ±

25 0.4 0.5 0.2

127 1.1 4.0 1.1

± ± ± ±

61 0.3 0.9 0.5

172


Fig. 5. DGGE analysis (triplicate) of 16S rDNA gene products amplified from total DNA extracted from solids samples. A) On the 4th month in the experimental canal (Bauxsol™ and soil layers); B) on the 6th month in the experimental canal (Bauxsol™ and soil layers); C) on the 4th month in the control canal (gravel and soil layers); D) on the 6th month in the control canal (gravel and soil layers). St = DNA marker standard from wastewater isolates.

Bauxsol™ pellets (Fig. 5). DGGE profiles of Bauxsol™ and soil layers were similar at 4 months (Figs. 5A and 6), but changed to two distinct profiles at 6 months (Figs. 5B and 6). Furthermore, increased richness (i.e., number of bands) over time on the Bauxsol™ pellet profiles was observed (from 8 bands at 4 months to 14 bands at 6 months) and stronger differences in profile richness along the length of the canal were registered (Bauxsol™ layer: b1.5 m 12 bands, N 1.5 m 17 bands; soil layer: b 1.5 m 9 bands, N 1.5 m 16 bands). Temporal differences within Bauxsol™ pellet profiles (Figs. 5A and B, and 6) were also apparent. In addition, a higher Shannon diversity index at the Bauxsol™/soil interface of the experimental canal was noted (Bauxsol™ 2.02 ± 0.10; interface 2.21 ± 0.16; soil 1.90 ± 0.25; Fig. 6).

In contrast, an increased similarity between the two layers in the control canal was observed at 4 and 6 months (Figs. 5C and D, and 6) and the Shannon diversity index was comparable between the gravel, the interface, and the soil (gravel 2.08 ± 0.33; interface 2.05 ± 0.18; soil 2.10 ± 0.31; Fig. 6). Additionally, the bacterial community developing on gravel in the control canal was different to the bacterial community on Bauxsol™ pellets in the experimental canal (Figs. 5 and 6).

3.2.2. Functional genes linked to nitrogen transformation Aerobic ammonia-oxidising bacteria (AOB) carrying the amoA gene were present in the inlet and on pre-experiment soil, on Bauxsol™ pellets and soil of the experimental canal, and on gravel of the control canal (Table 3). In contrast, ammonia-oxidising archaea (AOA) carrying the amoA gene were only found on pre-experiment soil, on soil of the experimental canal, and on gravel and soil of the control canal (Table 3). Denitrifiers carrying the nosZ gene and anaerobic ammoniaoxidising bacteria (anammox) carrying the hzo gene were both detected in the inlet and on Bauxsol™ pellets (experimental canal) and gravel (control canal; Table 3).

Table 3 Presence (+) or absence (−) of bands on gel electrophoresis of functional PCR amplification products targeting bacterial nitritation (amoA AOB gene), archaeal nitritation (amoA AOA gene), bacterial denitrification (nosZ gene), and anaerobic ammonia-oxidising (anammox) (hzo gene). Samples: inlet; pre-experiment soil; Bauxsol™ pellets and soil (experimental canal), and gravel and soil (control canal) towards the end of the experiment (5 and 6 months, combined). Samples

Fig. 6. Non-metric multi-dimensional scaling (MDS) of attached bacterial communities generated by the analysis of solid DGGE 16S rDNA gene products. Fourth-root transform and Bray–Curtis matrix were used. Experimental canal: at 4 months (▼) in Bauxsol™ and soil layers; at 6 months (●) in Bauxsol™, interface and soil layers. Control canal: at 4 months (Δ) in gravel and soil layers; at 6 months (□) in gravel, interface and soil layers.

Target gene

Location

Type

amoA AOB

amoA AOA

nosZ

hzo

Inlet Pre-experiment Experimental canal

Water Soil Bauxsol™ Soil Gravel Soil

+ + + + + −

− + − + + +

+ − + − + −

+ − + − + −

Control canal


4. Discussion Bauxsol™ pellets developed for this study are highly suitable for the treatment of municipal wastewater in a simple (i.e. unplanted) constructed wetland system. They have an excellent and wellinterconnected macro- and micro-porosity (Fig. S3) that provides a high surface area for geochemical reactions to occur and multiple micro-environments for the growth of a wide range of microorganisms. 4.1. Geochemical reactions 4.1.1. pH, EC and DO spikes The temporal pH increase in the experimental canal outlet is attributed to the release of unreacted calcium oxide from the cement binder used in the production of the pellets (Despland et al., 2010a) and this has been observed in previous studies (Despland et al., 2011, 2010b). However, the soil layer in this canal acted to buffer the extent of the increase. This buffering capacity initially lowered the water pH at the distal end of the canal (2.8 m) during the first 4 months (Fig. 2), thereby delaying the appearance of an outlet pH spike. At similar bed volumes in the Despland et al. (2011) study, the pH was some 2 pH units higher. The lower pH recorded mid-canal (1.5 m) after 4 months (Fig. 2) is an indication that the pellets were (at least in the first half of the canal) free of unreacted calcium oxide. Rinsing the pellets with freshwater as a pre-treatment and/or correcting the pH post-treatment would moderate the pH spike. Similarly, the observed EC spike correlated with the calcium spike (1737 ± 135 mg/L) in the experimental canal outlet and it is also attributed to the release of unreacted calcium oxide (i.e. a rise of calcium strongly affect the EC). Increased DO readings between inlet and both canal outlets were most probably caused by a constant diffusion of oxygen at the air–upper layer interface, but may also be attributed to diatom and green algal growth. 4.1.2. Phosphorus removal and trace-metal binding The outstanding TP removal in the experimental canal (N 95%, Fig. 3; under the Australian guidelines for advanced wastewater treatment; National Water Quality Management Strategy, 1997) is explained by the high phosphate binding capacity of Bauxsol™ (Akhurst et al., 2006; Clark et al., 2006; Despland et al., 2011; Hanahan et al., 2004). Other studies using different materials in constructed wetland systems have reported phosphate removal rates N 90% [e.g. laterite (Wood and McAtamney, 1996), maerl (Gray et al., 2000), and steel-slag (Zhang et al., 2007)]. However, as shown by a previous study (Despland et al., 2011) and the current study, Bauxsol™ pellets present other advantages such as high trace-metal binding (Table S1) through a combination of surface adsorption and new mineral precipitation (Clark et al., 2009, 2008; Collins et al., 2014; Hanahan et al., 2004). Similar to the findings of Akhurst et al. (2006), Hanahan et al. (2004), and Clark et al. (2009, 2006) for Bauxsol™ powders, phosphate removal by Bauxsol™ pellets in this current study was most likely a combination of ligand exchange, chemisorption and surface precipitation, with abundant Ca2 + and Mg2 + present in the Bauxsol™ pellet matrix forming MgHPO4 and CaHPO4. Moreover, the data show that phosphate saturation of Bauxsol™ pellets has not occurred in the experimental canal. Hence, the underlying soil in the experimental canal was unlikely to be P saturated, suggesting a prolonged canal life expectancy. Estimations based on the extrapolating bed volume from a previously reported mesocosm column experiment (Despland et al., 2011) suggest that an unplanted Bauxsol™ pellet constructed wetland should remove N50% of the TP for at least five years. In addition, the high concentration of Colwell P found on the Bauxsol™ pellets towards the end of the experiment (Table 1B) suggests that a loaded Bauxsol™ material may be a potential fertiliser; hence saturated pellets may have further reuse options and provide the possibility of P mining of waste effluents.

173

Differences in control and experimental soil TP binding capacity are readily explained by the lower pH (7.4 ± 0.2) found in the control soil (i.e. dominant phosphate species as H2PO− 4 ), which facilitates electrostatic and chemical attraction (Brady and Weil, 2008), and by higher phosphorus concentration reaching the soil layer (Rhue and Harris, 1999). Despite the higher phosphate removal by the soil layer of the control canal, overall control canal phosphorus removal quickly declined as saturation occurred; gravel provided limited to no phosphorus binding shown by high TP losses (Table 1A). These losses were most likely chemical weathering of basalt surface minerals and subsequent weak binding (i.e. ion exchange forming outer-sphere complex) which is highly reversible (Rhue and Harris, 1999). Consequently, in the control canal the soil is the only active P removing component, thus the canal longevity severely declines as the soil layer rapidly saturates with P. 4.1.3. Nitrogen removal Geochemical processes may explain part of the nitrogen reduction in both canal outlets (Fig. 4). High ammonium removal rates (Fig. 4B) may in part be due to geochemical binding as showed by the loading of ammonium found on the soils in both canals (Table 2A). Efficient binding of NH+ 4 to soil particles is due to the high affinity of the soil organomineral complexes (Brady and Weil, 2008; Gobat et al., 2004). Furthermore, the higher removal rate of ammonium found in the control canal outlet may be attributed to struvite (MgNH4PO4) precipitation (Babic-Ivancic et al., 2002; Despland et al., 2011). Any struvite precipitation in the experimental canal will be substantially weaker as high calcium concentrations (i.e. Bauxsol™ pellet curing process) interfere with struvite formation (Babic-Ivancic et al., 2002) and phosphate was most likely bound as simple MgHPO4 and CaHPO4. However, Clark et al. (2008) found for digested sewage sludge decants that struvite precipitation could account for a 50% reduction in NH+ 4 when using Bauxsol™, but only accounted for 15% of the P removal. Nitrate removal in the experimental canal before 3 months (Fig. 4D; ~26%) may partially be explained by strong nitrate binding on Bauxsol™ pellets (Table 2B). It may be hypothesised that the binding was via a ligand exchange mechanism but to date Bauxsol™ has not been reported for its nitrate binding capacity. Clark (unpublished data) suggests an established equilibrium between nitrate and Bauxsol™ regardless of the nitrate concentration (1–30 mg/L). Despite the strong nitrate binding in the experimental canal, nitrate concentrations at the outlet remained relatively unchanged possibly through the activity of Nitro-bacteria that oxidise nitrite into nitrate (i.e., nitratation; second step of the nitrification process). 4.2. Geochemistry influencing microbiology and vice-versa 4.2.1. Attached bacterial community DGGE analysis of the 16S rDNA amplified products from experimental canal solids shows a well-adapted bacterial community attached to the Bauxsol™ pellets (Fig. 5A and B) and at 6 months a distinct bacterial community attached to the soil (Figs. 5B and 6). This distinct growth and observed divergence between bacterial communities attached to Bauxsol™ pellets and gravel (Figs. 5 and 6) suggest that material geochemistry directly influences the micro-environmental conditions and generates different ecological niches (Figs. 2, 3 and 4; Tables 1 and 2). Several studies (Calheiros et al., 2009; Truu et al., 2005) show that specific material physicochemical characteristics and subsequent effluent composition changes have significant impacts on bacterial community diversity and structure. Bauxsol™ pellet micro-porosity (Fig. S3) suggests that the pellets may provide steeper redox and nutrient diffusion gradients, which then impact on the bacterial community structure. Other studies (Iasur-Kruh et al., 2010; Krasnits et al., 2009; Ragusa et al., 2004; Tietz et al., 2007; Truu et al., 2005) have shown that CW depth impacted microbial community diversity because of different nutrient (e.g. carbon, nitrogen, and oxygen) supply rates.

174


The higher Shannon diversity index at the interface in the experimental canal suggests an ecotonal diversity (Fig. 6), which is typically seen when two adjacent ecological systems (i.e. Bauxsol™ pellet layer and soil layer) possessing specific physicochemical and microbial characteristics interact (Gobat et al., 2004). Although the bacterial community was attached to the gravel and soil in the control canal (Fig. 5C and D), no observable ecotonal diversity was noted at the interface (Fig. 6), most probably because of greater geochemical similarity between the materials. Moreover, the gravel had a much lower influence than Bauxsol™ pellets on the characteristics of microbial communities, and therefore the soil in the control canal governed the communities' development on gravel (Figs. 5C and D, and 6). Temporal differences within Bauxsol™ pellet profiles (Figs. 5A and B, and 6) suggest species' specificity and may show a shift in the bacterial community's ecological strategy from a r-strategy (i.e. rapid colonisation of new environment by species with a high rate of reproduction) to a K-strategy (i.e. specific and competitive species well adapted to the environment) (Gobat et al., 2004; Madigan et al., 2012). A K-strategy shift may further explain profile richness differences found along the experimental canal length at 6 months (i.e. higher at the distal end). Ragusa et al. (2004) showed that microbial biomass takes up to 100 days to stabilize in constructed wetlands, with recently established wetlands, like this experiment, requiring some time to build stable (climax) bacterial communities. 4.2.2. Evidence of biological activities and presence of microorganisms linked to nitrogen transformation Biological activities (i.e. active biofilm) were found in the experimental canal as suggested by the post-experiment C/N ratio (17.3 ± 3.1 and 13.4 ± 0.1, in the Bauxsol™ and soil layers, respectively); C/N ratios between 6 and 25 are indicative of a nitrogen-rich soil litter that is readily accessed by decomposers (Brady and Weil, 2008; Gobat et al., 2004). In addition, the Bauxsol™ pellet C/N ratio was almost double the C/N ratio found in a previous study using Bauxsol™ pellets in a column system (Despland et al., 2012), implying some influence from the experimental soil layer. Similarly, relatively good biological activity in the control canal was apparent despite the low C/N ratio found in the gravel layer (1.3 ± 0), presumably because of the underlying active soil layer (14.5 ± 0.8; indication of a nitrogen-rich soil readily accessed by decomposers). Further evidence of biological activities was substantiated by water quality improvements after 6 months. BOD reductions in both canal outlets (84 ± 5%) are within Australian guidelines (National Water Quality Management Strategy, 1997) and are much greater than other studies in CWs; Krasnits et al. (2009) 70%, Iasur-Kruh et al. (2010) 50%, and Calheiros et al. (2009) 35%. Attached microbial communities associated with nitrogen transformation were confirmed by functional PCR-amplification analysis (Table 3) and suggest that these microorganisms are responsible for some of the observed ammonium, nitrite and nitrate concentration changes over time. However, additional quantitative PCR (qPCR) and reverse transcription PCR (RT-PCR) analyses are required to confirm this because particular gene presence does not necessarily signify intense biological activity. Likewise, failure to detect positive PCR products does not preclude their presence, given that negative results may be a consequence of PCR bias or low gene copy numbers. PCR-amplification of the bacterial amoA gene indicates that AOB enter the canals from the inlet waters and pre-experiment soil (Table 3). Spatial diversity of AOB within the experimental canal most probably occurred in response to variable environmental conditions (e.g. alkaline pH, changes in ammonium concentrations; Figs. 2 and 4B). These findings are supported by many studies on AOB population distributions in constructed wetlands (Gorra et al., 2007; Ibekwe et al., 2003; Truu et al., 2005). Hence, it is likely that Bauxsol™ pellets provide more favourable niches than gravel, and greater AOB ecological guild stability from higher porosity (Fig. S3), and surface area to volume

ratio. This observation agrees with previous studies on AOB adaptation to different substrates (Gorra et al., 2007; Truu et al., 2005). In contrast, the irregular AOB presence in the control canal suggests competition for NH+ 4 with heterotrophic fast growing microorganisms (Truu et al., 2005). Similarly, PCR-amplification of the archaeal amoA gene suggests that AOA came from the pre-experiment soil. The young age of the system (i.e. 6 months) and the very slow growth rate of these microorganisms (Erguder et al., 2009) possibly account for the low AOA detection rates in the experimental canal (Table 3). More mature CWs report large AOA population numbers (Park et al., 2006; Truu et al., 2005), and given enough time, CW using Bauxsol™ pellets could establish a substantial AOA population. Amplification of the hzo gene indicates that anammox entered the canals from the inlet waters (Table 3). Alternatively, the data could also suggest that anammox bacteria may have come from the underlying soil where it is possible that the initial soil transfer disturbed the soil anammox populations, therefore requiring some time to recover (i.e. no detection in the disturbed pre-experiment soil). The low concentrations detected here may also be due to the fact that anammox organisms have reportedly very slow growth rates (0.003/h growth rate, or a 10.6 day doubling time; Jetten et al., 2001). Hence, a more mature Bauxsol™ pellet CW may develop an important anammox population, similar to that illustrated by Shipin et al. (2005) and Dong and Sun (2007). Similar to anammox, PCR-amplification of the nosZ gene suggests that denitrifier populations were introduced by the inlet wastewater (Table 3). DO concentrations of approximately 3.6 mg/L in both canals suggest constant oxygen diffusion across the air and upper layer, and consequently, it is likely that denitrifiers present and active, being facultative aerobes, utilised oxygen as an electron acceptor before nitrate. The potential activities by strict aerobic microorganisms (i.e. AOB), facultative anaerobic microorganisms (i.e. denitrifiers) and strict anaerobic microorganisms (i.e. anammox bacteria) attached to Bauxsol™ pellets in the experimental canal imply the presence of anoxic micro-zones towards pellet centres or biofilm facilitating anaerobic processes. Comparably, the possible activities of the AOB, AOA, denitrifiers and anammox bacteria present in the control canal necessitate anoxic micro-zones of the biofilm growing on gravel. The presence of oxic and anoxic zones in and around solid particles is well documented (Gobat et al., 2004; Madigan et al., 2012), with several studies reporting the co-existence of aerobic and anaerobic bacteria in CWs (Dong and Sun, 2007; Iasur-Kruh et al., 2010; Krasnits et al., 2009; Shipin et al., 2005; Truu et al., 2005). Consequently, the presence of such diverse bacterial groups within Bauxsol™ pellets is not particularly surprising. However, unlike other substrates, the high phosphate and trace-metal binding capacity confirms the utility of pelletised Bauxsol™ as an ideal and effective material for constructed wetlands. 5. Conclusion This experiment demonstrates and highlights that, after some adaptation time, the bacterial communities established in the Bauxsol™ pellet layer and in the soil layer were distinct from each other, due in part to differences in geochemistry and micro-environmental conditions. The co-existence of attached aerobic ammonia-oxidising bacteria, anammox and denitrifiers on Bauxsol™ pellets and aerobic ammoniaoxidising bacteria and archaea attached to the soil, shows the potential for a complex biological nitrogen cycle. However, further qPCR and RTPCR analyses are required to confirm microbial activities. Coupled with nitrogen cycling, the strong geochemical processes that generate N95% phosphate removal, and nitrate, ammonium, and trace-metal binding make Bauxsol™ pellets an ideally suited alternative to conventional constructed wetland materials. Moreover, the high P-removal capacity particularly Colwell P bound to Bauxsol™ pellets indicates that saturated pellets would be an excellent fertiliser, thereby increasing their reuse value, reducing waste production, and increasing sustainability. Further


research should focus on a larger scale study with planted Bauxsol™ pellet constructed wetland system to add rhizosphere and mycorrhiza effects and probably induce a much longer life expectancy of the constructed wetland. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.scitotenv.2014.03.030. References Akhurst DJ, Jones GB, Clark M, McConchie D. Phosphate removal from aqueous solutions using neutralised bauxite refinery residues (Bauxsol™). Environ Chem 2006;3: 65–74. American Public Health Association. Standard methods for the examination of water and wastewater. Washington DC: American Public Health Association, American Water Works Association and Water Environment Federation; 1998. Babic-Ivancic V, Kontrec J, Kralj D, Brecevic L. Precipitation diagrams of struvite and dissolution kinetics of different struvite morphologies. Croat Chem Acta 2002;75:89–106. Brady NC, Weil RR. The nature and properties of soils. Upper Saddle River, N.J: Prentice Hall; 2008. Calheiros CSC, Duque AF, Moura A, Henriques IS, Correia A, Rangel A, et al. Substrate effect on bacterial communities from constructed wetlands planted with Typha latifolia treating industrial wastewater. Ecol Eng 2009;35:744–53. Clark M, Gnanam F, McConchie D, Fergusson L. The treatment of wastewaters with high phosphorous and high ammonia loadings using modified bauxite refinery residues. Chin J Geochem 2006;25:141. Clark MW, McConchie DM, Johnston M. Bauxite refinery residues, potential for waste utilisation. In: Lavelle JR, editor. Waste management: research, technology and developments. New York: Nova Science; 2008. Clark M, Berry J, McConchie D. The long term stability of metal-laden Bauxsol reagent under different geochemical conditions. Geochem Explor Environ Anal 2009;9:101–12. Clarke KR, Gorley RN. PRIMER v6: user manual/tutorial. Plymouth: PRIMER-E; 2006. Collins RN, Clark MW, Payne TE. Solid phases responsible for MnII, CrIII, C II, Ni, CuII and Zn immobilisation by a modified bauxite refinery residue (red mud) at pH 7.5. Chem Eng J 2014;236:419–29. Despland LM, Clark MW, Aragno M, Vancov T. Minimising alkalinity and pH spikes from Portland cement-bound Bauxsol (seawater-neutralized red mud) pellets for pH circum-neutral waters. Environ Sci Technol 2010a;44:2119–25. Despland LM, Vancov T, Clark MW, Aragno M. Bauxsol™ pellets: bacterial communities and phosphate removal in wastewater treatment. Proceedings of the IWA World Water Congress & Exhibition, Montreal, Canada; 2010b. Despland LM, Clark MW, Vancov T, Erler D, Aragno M. Nutrient and trace-metal removal by Bauxsol pellets in wastewater treatment. Environ Sci Technol 2011;45:5746–53. Despland LM, Vancov T, Aragno M, Clark MW. Diversity of microbial communities in an attached-growth system using Bauxsol™ pellets for wastewater treatment. Sci Total Environ 2012;433:383–9. Dong Z, Sun T. A potential new process for improving nitrogen removal in constructed wetlands: promoting coexistence of partial-nitrification and ANAMMOX. Ecol Eng 2007;31:69–78. Erguder TH, Boon N, Wittebolle L, Marzorati M, Verstraete W. Environmental factors shaping the ecological niches of ammonia-oxidizing archaea. FEMS Microbiol Rev 2009;33:855–69. Gobat JM, Aragno M, Matthey W. The living soil: fundamentals of soil science and soil biology. Enfield, New Hampshire, USA: Science Publishers; 2004. Gorra R, Coci M, Ambrosoli R, Laanbroek HJ. Effects of substratum on the diversity and stability of ammonia-oxidizing communities in a constructed wetland used for wastewater treatment. J Appl Microbiol 2007;103:1442–52.

175

Gray S, Kinross J, Read P, Marland A. The nutrient assimilative capacity of maerl as a substrate in constructed wetland systems for waste treatment. Water Res 2000;34: 2183–90. Hanahan C, McConchie D, Pohl J, Creelman R, Clark M, Stocksiek C. Chemistry of seawater neutralization of bauxite refinery residues (red mud). Environ Eng Sci 2004;21: 125–38. Iasur-Kruh L, Hadar Y, Milstein D, Gasith A, Minz D. Microbial population and activity in wetland microcosms constructed for improving treated municipal wastewater. Microb Ecol 2010;59:700–9. Ibekwe AM, Grieve CM, Lyon SR. Characterization of microbial communities and composition in constructed dairy wetland wastewater effluent. Appl Environ Microbiol 2003;69:5060–9. Jetten MSM, Wagner M, Fuerst J, van Loosdrecht M, Kuenen G, Strous M. Microbiology and application of the anaerobic ammonium oxidation (‘anammox’) process. Curr Opin Biotechnol 2001;12:283–8. Kadlec RH, Wallace S. Treatment wetlands. Boca Raton, FL: CRC Press; 2009. Krasnits E, Friedler E, Sabbah I, Beliavski M, Tarre S, Green M. Spatial distribution of major microbial groups in a well established constructed wetland treating municipal wastewater. Ecol Eng 2009;35:1085–9. Madigan MT, Martinko JM, Stahl DA, Clark DP. Brock biology of microorganisms. Harlow: Pearson Education; 2012. McConchie D, Clark M, Hanahan C, Fawkes RIn: Gaballah I, Hager J, Solozabal R, editors. The use of seawater-neutralised bauxite refinery residues (red mud) in environmental remediation programs. Proceedings of the Rewas '99 Global Symposium on Recycling, Waste Treatment, and Clean TechnologySan Sebastion, Spain: The Minerals, Metals, and Materials Society; 1999. p. 391–400. National Water Quality Management Strategy. Australian guidelines for sewerage systems: effluent management. Agriculture and Resource Management Council of Australia and New Zealand, Australian and New Zealand Environment and Conservation Council; 1997. Park HD, Wells GF, Bae H, Criddle CS, Francis CA. Occurrence of ammonia-oxidizing archaea in wastewater treatment plant bioreactors. Appl Environ Microbiol 2006; 72:5643–7. Ragusa SR, McNevin D, Qasem S, Mitchell C. Indicators of biofilm development and activity in constructed wetlands microcosms. Water Res 2004;38: 2865–73. Rayment GE, Higginson FRAustralian laboratory handbook of soil and water chemical methods, vol. 3. Port Melbourne: Inkata Press; 1992. Reed SC, Middlebrooks EJ, Crites RW, Reed SC, Middlebrooks EJ, Crites RW. Natural systems for waste management and treatment. New York: McGraw-Hill; 1995. Rhue RD, Harris WG. Phosphorus sorption desorption reactions in soils and sediments. In: Reddy KR, OConnor GA, Schelske CL, editors. Phosphorus biogeochemistry in subtropical ecosystems. Florida: Lewis Publishers; 1999. p. 187–206. Shipin O, Koottatep T, Khanh NTT, Polprasert C. Integrated natural treatment systems for developing communities: low-tech N-removal through the fluctuating microbial pathways. Water Sci Technol 2005;51:299–306. Stottmeister U, Wiessner A, Kuschk P, Kappelmeyer U, Kastner M, Bederski O, et al. Effects of plants and microorganisms in constructed wetlands for wastewater treatment. Biotechnol Adv 2003;22:93–117. Tietz A, Kirschner A, Langergraber G, Sleytr K, Haberl R. Characterisation of microbial biocoenosis in vertical subsurface flow constructed wetlands. Sci Total Environ 2007; 380:163–72. Truu J, Nurk K, Juhanson J, Mander U. Variation of microbiological parameters within planted soil filter for domestic wastewater treatment. J Environ Sci Health Part A Toxic/Hazard Subst Environ Eng 2005;40:1191–200. Wood RB, McAtamney CF. Constructed wetlands for waste water treatment: the use of laterite in the bed medium in phosphorus and heavy metal removal. Hydrobiologia 1996;340:323–31. Zhang XL, Zhang S, He F, Cheng SP, Liang W, Wu ZB. Differentiate performance of eight filter media in vertical flow constructed wetland: removal of organic matter, nitrogen and phosphorus. Fresenius Environ Bull 2007;16:1468–73.