e c o l o g i c a l e n g i n e e r i n g 2 9 ( 2 0 0 7 ) 8795

available at www.sciencedirect.com

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

What is the role of Phragmites australis in experimental constructed wetland lters treating urban runoff?
Byoung-Hwa Lee, Miklas Scholz
Institute for Infrastructure and Environment, School of Engineering and Electronics, College of Science and Engineering, The University of Edinburgh, William Rankine Building, The Kings Buildings, Edinburgh EH9 3JL, Scotland, UK

a r t i c l e
Article history:

i n f o

a b s t r a c t
The aim of this research was to assess the role of the macrophyte Phragmites australis (Cav.) Trin. ex Steud. in experimental temporarily ooded vertical-ow wetland lters treating urban runoff. For 2 years, hydrated nickel and copper nitrate were added to sieved road runoff to simulate contaminated primary treated urban runoff. During the rst year, 5-day biochemical oxygen demand (BOD) removal efciencies were lower in planted lters than in unplanted lters. However, the BOD removal performances of all lters were virtually similar irrespective of the planting regime during the second year. The nutrient removal

Received 19 April 2006 Received in revised form 4 August 2006 Accepted 4 August 2006

Keywords: Phragmites australis Urban runoff Wetland lter Biochemical oxygen demand Suspended solids Nitrogen Copper Nickel pH Biomass harvesting

performance of planted lters was more efcient and stable throughout the season particularly after the lters matured compared to that of unplanted lters. A substantial amount of nitrogen was also removed by harvesting P. australis, though metals were not, when compared to those retained in the lters. Furthermore, higher concentrations of nickel in the efuent were recorded in the planted lters, despite their ability to take up the heavy metals. P. australis provided undesirable conditions for precipitation of Ni by lowering the pH in the processes. Finally, after applying shock loadings of Cu, higher Cu outow concentrations were recorded for planted in comparison to unplanted lters. 2006 Elsevier B.V. All rights reserved.

1.
1.1.

Introduction
Macrophytes

and the effect of different plant species on the treatment wetland has been controversial (Scholz, 2006).

1.2.
Vegetation is the principal component of a wetland system. The common plants in wetlands are common reed (Phragmites spp.), cattail (Typha spp.), rush (Juncus spp.), and bulrush (Scirpus spp.). However, the most common plant species worldwide is Phagmites australis (Cav.) Trin. ex Steud. (IWA, 2000; Scholz, 2006). Although macrophytes are widely used within treatment wetlands in Europe and Northern America (Cooper et al., 1996; Kadlec and Knight, 1996), the role of macrophytes

Pollution control with macrophytes

Several previous studies reported a considerable contribution of macrophytes to pollutant removal. Karathanasis et al. (2003) reported that the 5-day at 20 C N-allylthiourea biochemical oxygen demand (BOD) removal efciency was lower in unplanted systems (63%) than in planted systems (7075%). The removal efciency of total suspended solids was also signicantly lower in unplanted systems (46%) than in planted

Corresponding author. Tel.: +44 131 6506780; fax: +44 131 6506554. E-mail address: m.scholz@ed.ac.uk (M. Scholz). 0925-8574/$ see front matter 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.ecoleng.2006.08.001

88

e c o l o g i c a l e n g i n e e r i n g 2 9 ( 2 0 0 7 ) 8795

systems (8890%) for subsurface-ow (SSF) wetlands. Gray et al. (2000) also described that the planted system removed more chemical oxygen demand from sewage than the unplanted system (75% removal compared to 48%) in the surface-ow (SF) wetlands. Furthermore, wetland plants require nutrients for growth and reproduction. Kadlec et al. (2005) reported that nitrogen (648%) was retained in plants in gravel-bed SSF wetlands. The uptake capacity of macrophytes is roughly between 30 and 150 kg P ha1 year1 and between 200 and 2500 kg N ha1 year1 (IWA, 2000). Greenway (2005) reported that constructed wetlands can be designed to maximize the removal of nutrients by enhancing macrophyte diversity through the incorporation of lagoons, shallow-water wetlands and subsurface-ow wetlands into the treatment train. This paper addresses the role of constructed wetlands in nutrient removal in Queenslands wetlands (Australia), and presents a good and relevant reference list.

1.3.

Controversial role of macrophytes

Some researchers did not detect any signicant difference between planted and unplanted systems. Balizon et al. (2002) reported that differences observed between wetland systems comprising of duckweed, reed and algae were insignicant. Scholz and Xu (2002) and Scholz (2006) also suggested that the BOD removal performance was virtually similar irrespectively of the planting regimes of constructed wetlands. Macrophytes can assimilate pollutants in their tissue and provide a surface and an environment for microorganisms to grow (Vymazal, 2002). Moreover, macrophytes create good conditions for the sedimentation of suspended solids (SS) and prevent erosion by reducing the velocity of the water in the wetland. The growth of roots within aggregates helps to decompose organic matter and prevents clogging by creating channels for the water to pass through the intermittently loaded vertical-ow system. The macrophytes transport approximately 90% of the oxygen available in the rhizosphere. This stimulates both aerobic decomposition of organic matter and the growth of nitrifying bacteria (Reddy et al., 1989; Brix, 1997; Scholz, 2006). However, when compared to microorganisms, macrophytes only play a secondary role in the degradation of organic matters in wetland systems (Stottmeister et al., 2003). Organic matter accumulates in wetlands over time through the annual turnover of macrophytes in terms of leaves and shoots. Organic matter binds heavy metals directly, and provides a carbon and energy source for microbial metabolism. Thus, macrophytes can be an essential part in the long-term functioning of wetlands (Gladden et al., 2002; Batty, 2003; Scholz, 2006). Previous researches have shown that macrophytes contribute insignicantly to heavy metal removal and that they are not important sinks for metals. Mays and Edwards (2001) reported that iron (Fe) and manganese (Mn) contents in the plants were only 1 and 2%, respectively, of the annual Fe and Mn loading in the wetlands. Similar ndings were reported for Fe removal (0.07% of the annual loading) by plants (Mitsch and Wise, 1998).

Contradictory results indicate a signicant role of macrophytes in heavy metal removal. Batty (2003) reported that P. australis took up almost 100% of Fe supplied at a concentration of 1 mg L1 . Fe removal efciency at a Woolley Colliery (West Yorkshire, UK) was also shown to improve from 70 to 95% if the system was planted (Younger et al., 2002). Cheng et al. (2002) found that more than 30% of the copper (Cu) and Mn loading rates accumulated in Cyperus alternifolius in vertical-ow experimental systems. P. australis also showed a very high adsorption afnity for copper, cadmium, nickel, lead and zinc from aqueous solutions (Southichak et al., 2006). Macrophytes also have a negative impact on wetland management. In fall, macrophytes loose their leaves, so that the BOD concentration will increase due to the increase of nutrients in the litter zone (Cooper et al., 1996; Scholz and Xu, 2002; Scholz, 2006). Furthermore, the overall storage area for water will be reduced due to an accumulation of macrophyte debris. It follows that the retention time of the wastewater in the constructed wetland will decrease if the inow and outow rates remain constant (Scholz et al., 2002; Scholz, 2006). It has become generally accepted knowledge that the physical function (i.e. erosion control and provision of surface area for microorganism) of macrophytes signicantly improves the performance of wetlands. In comparison, the metabolism of macrophytes, such as uptake and oxygen release, affects the treatment process to different extents depending on the design of constructed wetlands (Brix, 1997). While macrophytes signicantly affect the removal of pollutants in horizontal SSF systems with long hydraulic retention times, their role is minor in pollutant removal by SF systems that have comparatively short retention times (Karathanasis et al., 2003; Stottmeister et al., 2003). However, the contribution of macrophytes to the removal of pollutants for temporarily ooded vertical-ow systems has not been investigated as intensively as, for example, for horizontal-ow systems. Furthermore, the development of macrophytes in constructed wetlands treating urban runoff under severe environmental conditions, as well as their subsequent effect on the pollutant removal efciency, was not examined in detail.

1.4.

Research aim and objectives

The aim of this study was to assess the role of P. australis in experimental wetland lters treating urban runoff. The objectives were to analyze the pollutant removal performances for various lter designs and their effects on the pollutant removal efciencies under varying environmental conditions in a cold climate.

2.
2.1.

Materials and methods

Study site

Twelve wetland lters treating pretreated gully pot liquor were located outdoors at The Kings Buildings Campus (The University of Edinburgh, Scotland) to assess the ltration performance in a cold climate (09/09/0221/09/04). Replicates were not used, because previous experience with similar lters (e.g.

e c o l o g i c a l e n g i n e e r i n g 2 9 ( 2 0 0 7 ) 8795

89

Fig. 1 (a) Packing order (expressed as relative heights) of aggregates and (b) operational conditions (black represents application) of temporarily ooded vertical-ow wetland lters, of which Filters 4, 6, 8 and 1012 were planted with Phragmites australis (Cav.) Trin. ex Steud.

All lters simulate constructed treatment wetlands. However, different wetland lters are similar to various other natural treatment processes. For example, Filters 1 and 2 (controls) were similar to wastewater stabilization ponds or gully pots (extended storage) without a signicant amount of aggregates. In comparison, Filters 3, 5, 7 and 9 were similar to gravel and slow sand lters and Filters 4, 6, 8 and 10 were typical reed bed lters (Fig. 1). The reed bed lters contained gravel and native P. australis (Cav.) Trin. ex Steud., all of similar total biomass weight during planting and from the same local source. However, Filters 5, 6, 9 and 10 also contained adsorption media. Additional natural adsorption media (Filtralite and Frogmat) were used. Filtralite pellets, which contain 3% calcium oxide (CaO) and have diameters between 1.5 and 2.5 mm, are associated with enhanced metal and nutrient reduction (Brix et al., 2001; Scholz and Xu, 2002). Furthermore, Frogmat (a natural product based on raw barley straw) has a high adsorption area and is therefore likely to be associated with a high heavy metal reduction potential. The use of other lter media with high adsorption capacities such as activated carbon (e.g. Scholz and Martin, 1998; Scholz et al., 2002) and oxide-coated sand (e.g. Sansalone, 1999) has been discussed elsewhere. Filters 11 and 12 were more complex in their design and operation. The top water layer of both lters was aerated (with air supplied by air pumps) to enhance oxidation (minimizing zones of reducing conditions) and nitrication (Green et al., 1998; Cheng et al., 2002). Filter 12 received approximately 153% of Filter 11s mean annual inow volume and load. A higher hydraulic load should result in greater stress on P. australis and biomass.

Scholz and Xu, 2002) has shown that ltration performances between similar lters are obviously similar and that there are frequently no signicantly different treatment performances between such statistical experimental setups. Moreover, the additional sampling costs are usually not justied and data variations are similar between small and large wetland systems (Scholz, 2006). Gully pot liquor is concentrated surface runoff, which is detained in the wet gully pot until it overows into the sewer due to incoming surface runoff from new rainfall events. This concentrated road runoff was collected from randomly selected gully pots on the campus, the nearby housing estates and two major roads. After mixing both the sediment and the water phase within the gully pot, water was collected by manual abstraction with a 2 L beaker.

2.3.

Environmental conditions and operation

2.2.

Filter design and media composition

Round drainage pipes were used to construct the experimental lters. All 12 vertical-ow wetland lters were designed with the following dimensions: height = 83 cm and diameter = 10 cm. The outlet of each constructed wetland lter comprised a valve at the bottom of each lter. Different packing order arrangements of lter media and plant roots were used in the wetland lters (Fig. 1). The P. australis plants were collected from a local grower and supplier. Four established plants with healthy rhizomes were planted densely in each lter. It was not practically possible to measure biomass without disturbing the plants. Aboveground biomass was therefore estimated (Table 1).

The lters were designed to operate in batch ow mode to reduce pumping and computer control costs. The inow waters of Filter 2 and Filters 712 were dosed with hydrated copper nitrate (Cu(NO3 )2 3H2 O) and hydrated nickel nitrate (Ni(NO3 )2 6H2 O) to simulate contaminated urban water runoff. All lters were periodically inundated (100%) with pretreated inow gully pot liquor and partially drained (50%) or entirely drained (0%) to encourage air penetration through the aggregates (Cooper et al., 1996; Scholz and Xu, 2002). Raw gully pot liquor was sieved (pore size of 2.5 mm) to simulate preliminary and primary treatment efuent. This procedure is in line with common practice in the wastewater industry (Cooper et al., 1996; Tchobanoglous et al., 2003). Since 22 September 2003, the pH value of the inow has been articially raised by the addition of sodium hydroxide (NaOH) to the sieved gully pot liquor. In consequence, the mean inow pH has increased from 6.7 to 8.1. Table 2 summarizes typical inow water quality characteristics for the contaminated urban water. The empty bed volumes varied between approximately 3.5 and 5.5 L depending on the lter composition and lter maturity (i.e. accumulation of debris). Considering that the lters were operated only two (Filters 111) or three (Filter 12) times per week in batch ow mode, the lters received between 1.0 and 2.4 L urban water per day. The corresponding loads can be easily calculated with Table 2, and loads and removal rates for selected variables and lters are shown in Tables 3, 5 and 6.

90

e c o l o g i c a l e n g i n e e r i n g 2 9 ( 2 0 0 7 ) 8795

2.4.

Metal nitrates

which is the standard carrier gas in most national analytical laboratories (APHA, 1995).

Cu and nickel (Ni) were selected as additional heavy metals for investigation, because they are commonly occurring contaminants from road vehicles and are not easily bio-available (Cooper et al., 1996; Kadlec and Knight, 1996; Scholz and Xu, 2002). It follows that these metals are likely to accumulate within the sediment and debris of wetland lters. As the buildup continues, metal toxicity increases, as does the risk of severe pollution due to leaching (Scholz et al., 2002). Some heavy metals do accumulate easily in constructed wetlands, but may be released if environmental conditions change; e.g. road gritting (containing salt) in winter. Such transformation processes are not well understood (Norrstrom and Jacks, 1998). Hydrated copper nitrate (Cu(NO3 )2 3H2 O) and hydrated nickel nitrate (Ni(NO3 )2 6H2 O) were added to the inow water of Filter 2 and Filters 712. This gave total concentrations of dissolved Cu and Ni of approximately 1 mg L1 for each metal. It is comparable to gures reported for urban water heavily contaminated with heavy metals and mine wastewater (Cooper et al., 1996; Kadlec and Knight, 1996; Mungur et al., 1997; Scholz and Xu, 2002). Concerning the dosed inow water, the background concentration for nitrate-nitrogen (including nitrite-nitrogen) was only approximately 0.497 mg L1 . Therefore, the contribution of the introduced nitrate-nitrogen amounted to 65% (or approximately 0.917 mg L1 ) of the overall nitrate-nitrogen (including nitrite-nitrogen) load. On 24 September 2004, 1 g Cu as copper nitrate (Cu(NO3 )2 ) was added to Filters 1, 3, 4, 5, 6, and 12 to assess the toxic shock effect and the long-term trend of Cu retention performance of wetland lters. The potential breakthrough of Cu in heavily contaminated wetland lters was also examined. This added copper nitrate contained 1.95 g nitrate, where considering the annual amount of Cu load per lter (e.g. 153 and 127 mg for Filter 8 during the rst and second year, respectively), 1 g of Cu per lter is the equivalent of a 7-year load.

2.6.

BOD, nutrient and other determinations

The BOD was determined in all water samples with the OxiTop IS 12-6 system, a manometric measurement device, supplied by the Wissenschaftlich-Technische Werkstatten (WTW), Weilheim, Germany. Nitrication was suppressed by adding 0.05 mL of 5 g L1 N-allylthiourea (WTW Chemical Solution No. NTH 600) solution per 50 mL of sample water. Nutrients (nitrate-nitrogen, ammonia-nitrogen, and orthophosphate-phosphorus) and all other variables (suspended solids, total solids, turbidity, dissolved oxygen, pH, redox potential, conductivity and temperature) were determined according to standard methods (Allen, 1974; APHA, 1995). Oxidized aqueous nitrogen was determined as the sum of nitrate-nitrogen and nitrite-nitrogen. Nitrate-nitrogen was reduced to nitrite-nitrogen by cadmium and determined as an azo dye at 540 nm using a Perstorp Analytical EnviroFlow 3000 ow injection analyzer following diazotization with sulfanilamide and subsequent coupling with N-1-naphthylethylendiamine dihydrocloride (Allen, 1974). Ammonia-nitrogen and ortho-phosphate-phosphorus interactions were determined by automated colorimetry from their reactions with hypochlorite and salicylate ions in solution in the presence of sodium nitrosopentacyanoferrate (nitroprusside) and with acidic molybdate to form a blue phosphomolybdenum complex, respectively. The colored complexes formed were measured spectrophotometrically at 655 and 882 nm, respectively, using a Bran and Luebbe autoanalyzer (Allen, 1974). Nutrients were also determined for the dried biomass of P. australis (see Section 2.5). Finally, Tubifex tubifex (Sludgeworm) worms were counted by visual examination (i.e. by eye) in the entire outow of each lter and mean numbers were calculated per liter.

3.
3.1.

Results and discussion

Development of P. australis biomass

2.5.

Metal determinations

Metal concentrations were determined in the raw gully pot liquor, sieved gully pot liquor (partially used as actual inow water for some lters, see above), contaminated (added metal nitrates) sieved gully pot liquor (partially used as actual inow water for some lters) and the outow waters from the experimental rig. Metals were also determined for the dried (i.e. at 105 C overnight) biomass of P. australis. Random plant samples were taken from points at least 83 cm above the bottom of the lters. It follows that the stems were cut approximately 20 cm above the top of the layer of debris. Ashed plant samples were digested under reux in aqua regia for 2 h, cooled and ltered (Whatman No. 5412) and subsequently made up to 100 mL with de-ionized water ready for analysis. A Varian Spectr AA 400 Atomic Absorption Spectrometer with a GTA-96 graphite furnace tube atomizer was used for the standard analysis of nickel and copper. Notched GTA partition tubes (coated) were applied. The carrier gas was argon,

Aboveground P. australis was harvested three times during the operation periods, each time at the end of fall. Dry weights of P.

Table 1 Dry weight of harvested aboveground Phragmites australis (Cav.) Trin. ex Steud. for all planted lters, and the corresponding total nitrogen concentrations for 2004, where the plants were most mature Filter numbers Dry weight (g) 2002
4 6 8 10 11 12 8.0 4.7 1.0 1.1 1.5 6.0

Total nitrogen (mg) 2004

88.7 79.4 81.0 139.3 130.6 120.0

2003
20.6 23.9 28.5 21.7 19.5 23.5

2004
23.8 32.2 29.6 48.4 40.2 33.3

e c o l o g i c a l e n g i n e e r i n g 2 9 ( 2 0 0 7 ) 8795

91

Table 2 Typical stormwater inow characteristics for the rst (09/22/0209/21/03) and second (09/22/0309/21/04) year of the experiment Variable Unit First year
Nickel (dissolved) Copper (dissolved) BODa Nitrate-nitrogen Ammonia-nitrogen ortho-Phosphate-phosphorus Suspended solids Dissolved oxygen pH Redox potential Conductivity Water temperature
a

Mean Second year

1.02 1.02 89.2 1.38 1.45 0.10 853.9 3.07 8.07 44.4 2227.2 12.0

Standard deviation First year

0.143 0.036 49.29 1.008 2.058 0.149 377.75 2.493 0.411 112.72 11182.22 5.78

Second year
0.042 0.018 55.50 1.220 1.934 0.136 1420.85 1.490 1.082 93.44 4620.82 5.65

mg L1 mg L1 mg L1 mg L1 mg L1 mg L1 mg L1 mg L1 mV S C

1.06 1.03 61.1 1.45 1.65 0.06 335.7 4.70 6.69 142.5 5139.1 10.7

Five-day at 20 C N-allylthiourea biochemical oxygen demand.

Table 3 Comparison of 5-day at 20 C N-allylthiourea biochemical oxygen demand (BOD) reduction rates of planted (Filters 4, 6, 8 and 1012) and unplanted (Filters 1 to 3, 5, 7 and 9) lters Filter number Loading (g m2 day1 )
3 4 5 6 7 8 9 10 6.03 5.64 6.38 6.82 6.28 5.85 5.91 6.32

First year Removal (g m2 day1 )

5.08 4.07 5.40 4.80 5.41 4.49 5.02 4.28

Second year Reduction rate (%)

84.2 72.2 84.6 70.4 86.1 76.8 84.9 67.7

Loading (g m2 day1 )
5.20 4.74 5.90 7.11 5.79 5.44 5.45 5.82

Removal (g m2 day1 )
5.02 4.61 5.59 6.73 5.63 5.25 5.29 5.57

Reduction rate (%)

96.5 97.3 94.7 94.7 97.2 96.5 97.1 95.7

australis were determined for all planted lters (Table 1). These data show that the dry weights for all lters increased consistently. The amount of nutrients removed through harvesting is therefore likely to have increased also. With respect to the effect of pH on the growth of P. australis, a high pH (Table 2) did not harm the growth of this macrophyte, if one is to consider its biomass, harvested in 2004. Furthermore, the lters loaded with heavy metalcontaminated water produced more P. australis biomass (Table 1), indicating that the development of this macrophyte was not inhibited when exposed to high concentrations of metals, which is in agreement with previous research (McCabe et al., 2001; Ye et al., 2003). Considering the systematic and stratied experimental setup, there seems to be no logical explanation for the obvious differences in biomass of P. australis (except for the possibility of random variation). The gully pot liquor used contained high concentrations of salt (Table 2), which is harmful to some wetland plants. However, according to previous ndings (Cooper et al., 1996), the maximum salinity tolerance value for P. australis is about 45 g L1 . Considering that the measured conductivity inow values are 5139 and 11,182 S for the rst and second year of operation (Table 2), respectively, the presence of salts should not stunt the growth of P. australis signicantly.

3.2.

P. australis effect on BOD removal

As presented in Table 3, the reduction rates of unplanted lters (Filters 3, 5, 7, and 9) during the rst year of operation were consistently higher than those of the corresponding planted lters (Filters 4, 6, 8, and 10). This suggests that unplanted lters are more efcient in BOD removal for relatively new wetland lters. This was also supported by the Wilcoxon signed-rank statistic. During the rst year of operation, all the p-values were less than 0.05 under the hypothesis that

Table 4 Comparison of efuent 5-day at 20 C N-allylthiourea biochemical oxygen demand (BOD) concentrations for planted (Filters 4, 6, 8 and 1012) and unplanted (Filters 1 to 3, 5, 7 and 9) lters Filter number Wilcoxon p-values First year
3 and 4 5 and 6 7 and 8 9 and 10 0.01 0.01 0.00 0.00

Second year
0.51 0.66 0.36 0.19

92

e c o l o g i c a l e n g i n e e r i n g 2 9 ( 2 0 0 7 ) 8795

Table 5 Comparison of nitrogen reduction rates in Filters 7 and 8 Filter number Loading (g m2 day1 )
Ammonia-nitrogen 7 8 Nitrate-nitrogen 7 8 0.111 0.105

First year Removal (g m2 day1 )

0.052 0.082

Second year Reduction rate (%)

46.9 78.7

Loading (g m2 day1 )
0.085 0.081

Removal (g m2 day1 )
0.070 0.077 0.039 0.084

Reduction rate (%)

81.7 95.8 41.7 95.5

0.110 0.106

0.029 0.099

26.6 93.7

0.093 0.088

efuent BOD values of planted lters are similar or lower compared to the BOD values of the unplanted wetland lters. This indicates that efuent BOD values of planted lters are higher than those for unplanted lters (Table 4). These ndings are supported by the initial presence of relatively high numbers of large red T. tubifex worms in planted lters. T. tubifex is known as an indicator for organically polluted water including high BOD and low DO concentrations (Kadlec and Knight, 1996; Scholz, 2006). Efuent water from planted lters had a mean of 1.0 (S.D. 0.74) T. tubifex per liter. In contrast, unplanted lters contained a mean of 0.1 (S.D. 0.86) T. tubifex per liter. This provides further evidence that P. australis adversely affected the efuent water quality in the present study. However, T. tubifex were virtually absent in all lters since spring 2003 (i.e. after the setting-up period). Another possible reason for the relatively low efuent BOD concentration in the unplanted lters is the lack of vegetation cover, which may result in extra aeration and oxidation of the organic load (Thomas et al., 1995). Furthermore, plant debris decayed within the wetland lters, and hence subsequently increased the organic loading in planted lters during the winter season. However, p-values 0.05 during the second year of operation indicate that P. australis did not signicantly affect the removal performance of organic matter as reported elsewhere (Lim et al., 2001). This result suggests that the water treatment potential of the developed (i.e. matured) biolm substrate offsets the negative impact of P. australis on the treatment during the second year of operation (Scholz and Xu, 2002; Scholz et al., 2002; Scholz, 2006). Additionally, the mature root system enhances the capacity of transporting oxygen to the substrate and provides a large surface area for microorganisms to grow on (Cooper et al., 1996; Scholz, 2006).

and ltration by the root network. Therefore, few differences in SS removal performances between wetland lters with and without P. australis indicate that the contribution of this macrophyte to the physical SS removal processes was not considerable in the temporarily ooded vertical-ow wetland lters (see also Tanner et al., 1995; Thomas et al., 1995). Clogging problems were not observed during the operation periods in planted lters. In contrast, unplanted lters (Filters 5, 7 and 9) were clogged on 12 December 2003. This suggests that P. australis provides good ltration conditions by preventing the lters from clogging (see also Brix, 1997).

3.4.

P. australis effect on nitrogen removal

As presented in Table 5, ammonia-nitrogen reduction efciencies for planted lters (Filter 8) were always higher than for comparable unplanted lters (Filter 7). Furthermore, nitratenitrogen reduction efciencies for planted lters were much higher than those for unplanted lters. In particular, the nitrate-nitrogen concentrations increased through the treatment process in unplanted lters during the second year of operation. In contrast, Vymazal (2002) reported that the removal of total nitrogen is lower for vegetated beds compared to non-vegetated beds. However, he had only assessed horizontal-ow systems for wastewater treatment.

3.3.

P. australis effect on suspended solids removal

The p-values of the Wilcoxon signed-rank test for suspended solids of the efuent between planted and unplanted lters were 0.05, indicating that the SS removal performances of most lters were virtually similar. These results clearly show that the SS removal performance is not affected by P. australis, contradicting the results of Karathanasis et al. (2003), who reported that the vegetated systems exhibit nearly twice as high removal efciencies when compared to an unplanted system. Brix (1997) described that the higher SS removal performances in planted systems are attributed to larger surface areas, reduced water velocities and reinforced settling

Fig. 2 Mass loading and removal rates of ammonia-nitrogen (NH4 -N) in the unplanted Filter 7 and planted (Phragmites australis (Cav.) Trin. ex Steud.) Filter 8. The fraction of the total squared error that is explained by each equation is dened as R2 .

e c o l o g i c a l e n g i n e e r i n g 2 9 ( 2 0 0 7 ) 8795

93

Table 6 Mass loading and removal rates of heavy metal in Filters 7 and 8
Filter number

First year Loading (g m2 day1 ) Removal (g m2 day1 )

0.071 0.062 0.074 0.072

Second year Reduction rate (%)

89.4 81.5 96.2 97.3

Loading (g m2 day1 )
0.068 0.064 0.067 0.063

Removal (g m2 day1 )
0.062 0.057 0.064 0.061

Reduction rate (%)

92.1 89.0 95.4 95.5

Nickel 7 8 Copper 7 8

0.080 0.076 0.077 0.074

Fig. 2 shows a linear relationship between the loading and removal rates of ammonia-nitrogen in both lters. In comparison to Filter 7, higher removal rates and more consistent removal performances were observed in Filter 8. Plant uptake and oxygen transfer improved the reduction of ammonianitrogen. The total nitrogen removal by harvesting P. australis was estimated. These values were calculated from the concentration of total nitrogen and the corresponding weight of harvested P. australis in November 2004. The total quantities of accumulated nitrogen in the stems and leaves of P. australis are shown in Table 1. Data for earlier years are likely to be unreliable, because they are inuenced by the setup period. Considering the total removed amount of ammonia-nitrogen (221 mg) and nitrate-nitrogen (241 mg) in Filter 8 during the second year of operation, a substantial amount (17.5%) of nitrogen could be removed by regular harvesting in fall. For a more detailed discussion on biomass and nutrient mass balance within vertical-ow constructed wetlands, the work published by Tanner (1996) and Meuleman et al. (2003) should be consulted. For example, ndings indicate that early (i.e. in October rather than December) harvesting of nutrients in The Netherlands has resulted in higher removal rates (Meuleman et al., 2003).

Despite the ability of P. australis to take up metals, efuent concentrations of Ni for the planted lter (Filter 8) were higher than those for the unplanted lter (Filter 7). Higher concentrations of Ni in the efuent of planted lters are most likely attributed to their lower pH values, compared to those of unplanted lters. Fig. 3 illustrates the changes of the inuent and efuent pH for planted and unplanted lters. The efuent pH values (6.726.93) for the planted lters were slightly lower than pH values (7.197.31) for the unplanted lters. The conversion of ammonium to nitrite results in the formation of hydrogen ions (Eq. (1)) and, subsequently, the hydrogen ions are neutralized by bicarbonate ions during the nitrication process (Eq. (2)). Both a decrease in bicarbonate alkalinity and an increase in the carbon dioxide lowers the pH. Theoretically, 7.2 mg of alkalinity (as CaCO3 ) are required to neutralize the hydrogen ions produced by the oxidation of 1 mg of ammonia-nitrogen to nitrite-nitrogen. Thus, a low pH can signicantly reduce the rate of nitrication. Below a pH of 7.2, the rate falls rapidly, approaching zero at a pH of 6 (Tchobanoglous et al., 2003; Rich, 2005). NH4 + + 1.5O2 2H+ + H2 O + NO2 H+ + HCO3 CO2 + H2 O (1) (2)

3.5.

P. australis effect on heavy metal removal

Mean loading and removal rates for Ni and Cu are presented in Table 6. Reduction rates of Cu were high (>95%) regardless of the planting regime during the 2 years of operation. In contrast, the Ni reduction rates of the unplanted Filter 7 were higher than those of the planted Filter 8. This is also supported by the Wilcoxon signed-rank analysis result (p = 0.00) from comparisons of efuent Ni concentrations between Filters 7 and 8. Metal loads removed by harvesting were negligible compared to those retained in the lters. For example, 0.3 and 0.1% of Ni and Cu loads, respectively, accumulated in the aboveground tissues of P. australis during the second year. Therefore, it can be concluded that most of the heavy metal loads were accumulated in the sediment rather than taken up by P. australis in the lters. This conrms earlier ndings by Scholz and Xu (2002) and Du Laing et al. (2006). Furthermore, Peverly et al. (1995) reported that roots accumulate elevated concentrations of metals including copper. However, roots were not assessed in this study because of their detrimental effect on the small experimental lters.

A lower pH level in a planted lter is most likely due to a more active nitrication process compared to an unplanted lter. In addition, plants utilize nitrogen and can contribute to the lowering of pH through respiration and litter decomposition processes (Collins et al., 2004). A similar relationship

Fig. 3 Changes of inuent and efuent pH in the unplanted Filter 7 and planted (Phragmites australis (Cav.) Trin. ex Steud.) Filter 8.

94

e c o l o g i c a l e n g i n e e r i n g 2 9 ( 2 0 0 7 ) 8795

removal process as plant uptake through the root system greatly facilitated the nitrogen removal. Furthermore, a substantial amount of total nitrogen (e.g. between 80 and 140 mg per lter in 2004) was removed by harvesting P. australis. The amount of metals removed by harvesting was negligible (<1% on average) when compared to those retained in the lters. Furthermore, higher efuent Ni concentrations were recorded in the planted lters, despite their ability to take up heavy metals. P. australis provided undesirable conditions for precipitation of Ni by lowering the pH in the processes.

Acknowledgements
Fig. 4 Variation of copper (Cu) concentrations in the unplanted Filter 3 and planted (Phragmites australis (Cav.) Trin. ex Steud.) Filter 4. The authors wish to acknowledge support provided by Ms. M. Wagner, Ms. S. Guilbert and Mr. F. McGowan. Mr. A.J. Sadowski proofread this paper.

references
between pH change and nitrogen removal was observed by Sun et al. (2003). Therefore, it can be inferred that P. australis is likely to provide undesirable conditions for precipitation of Ni by lowering the pH. Collins et al. (2004) reported that the water in planted wetland lters had lower pH values, but higher Fe and Mn concentrations than water in unplanted lters. The effect of P. australis on the removal of metals becomes remarkable in the long-term due to accumulation of metals in the litter zone (e.g. Du Laing et al., 2006; Scholz, 2006). After 2 years of operation, selected lters (see below) were loaded with toxic shock loads of Cu to assess whether the wetland lters could retain heavy metals in the long-term, and also to determine the fate of Cu in the lter. The breakthroughs of Cu were observed in both Filters 3 and 4 (Fig. 4). However, higher peaks of Cu for Filter 4 (planted lter) in comparison to Filter 3 (unplanted lter) were also recorded. Lower Cu removal performance of the planted lters is attributed to the relatively low pH values in the planted lters, as discussed above. A relatively low redox potential related to the peaks of Cu in the outow of Filter 3. This is most likely attributed to ferric iron, which is formed at approximately +120 mV. The correlation coefcients between Cu and the redox potential of the outows were 0.31. Findings indicate that metal retention in the wetland lters is more susceptible to the change of environmental conditions such as pH and redox potential in the long term. Furthermore, the presence of P. australis has a great impact on metal retention by lowering the pH in aged wetland lters.

4.

Conclusions

P. australis (Cav.) Trin. ex Steud. had a negative impact on the 5-day at 20 C N-allylthiourea biochemical oxygen demand removal processes, although they provided good ltration conditions by preventing the lters from becoming clogged. The nitrogen removal performance of planted lters was more efcient and stable throughout the year (particularly after the lters have matured) compared to unplanted lters. P. australis was found to contribute signicantly to the nitrogen

Allen, S.E., 1974. Chemical Analysis of Ecological Materials. Blackwell, Oxford, UK. American Public Health Association (APHA), 1995. Standard Methods for the Examination of Water and Wastewater, 19th ed. APHA-AWWWA-WEF, Washington, DC, USA. Balizon, M.E., Dolmus, R., Quintana, J., Navarro, Y., Donze, M., 2002. Comparison of conventional and macrophyte-based systems for the treatment of domestic wastewater. Water Sci. Technol. 45, 111116. Batty, L.C., 2003. Wetland plantsmore than just a pretty face? Land Contam. Reclam. 112, 173180. Brix, H., 1997. Do macrophytes play a role in constructed treatment wetlands? Water Sci. Technol. 35, 1117. Brix, H., Arias, C.A., del Bubba, M., 2001. Media selection for sustainable phosphorus removal in subsurface ow constructed wetlands. Water Sci. Technol. 44, 4754. Cheng, S., Grosse, W., Karrenbrock, F., Thoennessen, M., 2002. Efciency of constructed wetlands in decontamination of water polluted by heavy metal. Ecol. Eng. 18, 317325. Collins, B., McArthur, J.V., Sharitz, R.R., 2004. Plant effects on microbial assemblages and remediation of acidic coal pile runoff in mesocosm treatment wetlands. Ecol. Eng. 23, 107115. Cooper, P.F., Job, G.D., Green, M.B., Shutes, R.B.E., 1996. Reed Beds and Constructed Wetlands for Wastewater Treatment. WRc plc., Swindon, UK. Du Laing, G., Van Ryckegem, G., Tack, F.M.G., Verloo, M.G., 2006. Metal accumulation in intertidal litter through decomposing leaf blades, sheaths and stems of Phragmites australis. Chemosphere 63, 18151823. Gladden, J.B., Specht, W.L., Nelson, E.A., 2002. Comparison of Constructed Wetland Mesocosms Designed for Treatment of Copper-contaminated Wastewater. Westinghouse Savannah River Company (Contract no. DE-AC09-96SR18500 with the U.S. Department of Energy). Gray, S., Kinross, J., Read, P., Marland, A., 2000. The nutrient assimilative capacity of maerl as a substrate in constructed wetland system for waste treatment. Water Res. 34, 21832190. Green, M., Friedler, E., Safrai, I., 1998. Enhancing nitrication in vertical ow constructed wetland utilizing a passive air pump. Water Res. 32, 35133520. Greenway, M., 2005. The role of constructed wetlands in secondary efuent treatment and water reuse in subtropical Australia. Ecol. Eng. 25, 501509.

e c o l o g i c a l e n g i n e e r i n g 2 9 ( 2 0 0 7 ) 8795

95

IWA Specialist Group on Use of Macrophytes in Water Pollution Control, 2000. Constructed Wetlands for Pollution Control. Scientic and Technical Report Number 8. IWA Publishing, London, UK. Kadlec, R.H., Knight, R.L., 1996. Treatment Wetlands. CRC Press, Boca Raton, FL, USA. Kadlec, R.H., Tanner, C.C., Hally, V.M., Gibbs, M.M., 2005. Nitrogen spiraling in subsurface-ow constructed wetlands: implications for treatment response. Ecol. Eng. 25, 365381. Karathanasis, A.D., Potter, C.L., Coyne, M.S., 2003. Vegetation effects on fecal bacteria, BOD, and suspended solid removal in constructed wetlands treating domestic wastewater. Ecol. Eng. 20, 157169. Lim, P.E., Wong, T.F., Lim, D.V., 2001. Oxygen demand, N and copper removal by free-water-surface and subsurface-ow constructed wetlands under tropical conditions. Env. Int. 26, 425431. Mays, P.A., Edwards, G.S., 2001. Comparison of heavy metal accumulation in a natural wetland and constructed wetlands receiving acid mine drainage. Ecol. Eng. 16, 487500. McCabe, O.M., Daldwin, J.L., Otte, M.L., 2001. Metal tolerance in wetland plants? Minerva Biotechnol. 13, 141149. Meuleman, A.F.M., van Logtestijn, R., Rijs, G.B.J., Verhoeven, J.T.A., 2003. Water and mass budgets of a vertical-ow constructed wetland for wastewater treatment. Ecol. Eng. 20, 3144. Mitsch, W.J., Wise, K.M., 1998. Water quality, fate of metals, and predictive model validation of a constructed wetland treating acid mine drainage. Water Res. 32, 18881900. Mungur, A.S., Shutes, R.B.E., Revitt, D.M., House, M.A., 1997. An assessment of metal removal by laboratory scale wetland. Water Sci. Technol. 35, 125133. Norrstrom, A.C., Jacks, G., 1998. Concentration and fractionation of heavy metals in roadside soils receiving de-icing salts. Sci. Total Env. 218, 161174. Peverly, J.H., Surface, J.M., Wang, T.G., 1995. Growth and trace metal absorption by Phragmites australis in wetlands constructed for landll leachate treatment. Ecol. Eng. 5, 2135. Reddy, K.R., Patrick, W.H., Lindau, C.W., 1989. Nitricationdenitrication at the plant root-sediment interface in wetlands. Limnol. Oceanogr. 34, 10041013. Rich, L.G., 2005. Nitrication in aerated lagoons and with intermittent sand lters. Technical Note 6. Available at http://www.ces.clemson.edu/ees/rich/technotes. Sansalone, J.J., 1999. Adsorptive inltration of metals in urban drainagemedia characteristics. Sci. Total Env. 235, 179188.

Scholz, M., 2006. Wetland Systems to Control Urban Runoff. Elsevier, Amsterdam, The Netherlands. Scholz, M., Hohn, P., Minall, R., 2002. Mature experimental constructed wetlands treating urban water receiving high metal loads. Biotechnol. Prog. 18, 12571264. Scholz, M., Martin, R.J., 1998. Control of bio-regenerated granular activated carbon by spreadsheet modelling. J. Chem. Tech. Biotechnol. 71, 253261. Scholz, M., Xu, J., 2002. Comparison of constructed reed beds with different lter media and macrophytes treating urban stream water contaminated with lead and copper. Ecol. Eng. 18, 385390. Southichak, B., Nakano, K., Nomura, M., Chiba, N., Nishimura, O., 2006. Phragmites australis: a novel biosorbent for the removal of heavy metals from aqueous solution. Water Res. 40, 22952302. Stottmeister, U., Wiesner, A., Kuschk, P., Kappelmeyer, M., Kaster, M., 2003. Effects of plants and microorganisms in constructed wetlands for wastewater treatment. Biotech. Adv. 22, 93117. Sun, G., Gray, K.R., Biddlestone, A.J., Allen, S.J., Cooper, D.J., 2003. Effect of efuent recirculation on the performance of a reed bed system treating agricultural wastewater. Proc. Biochem. 39, 351357. Tanner, C.C., 1996. Plants for constructed wetland treatment systemsa comparison of the growth and nutrient uptake of eight emergent species. Ecol. Eng. 7, 5983. Tanner, C.C., Clayton, J.S., Upsdell, M.P., 1995. Effect of loading rate and planting on treatment of dairy farm wastewaters in constructed wetlands: removal of oxygen demand, suspended solids and fecal coliforms. Water Res. 29, 1726. Tchobanoglous, G., Burton, F.L., Stensel, H.D., 2003. Wastewater Engineering: Treatment and Reuse, 4th ed. (revised). Metcalf & Eddy Inc., McGraw-Hill Companies Inc., NY, USA. Thomas, P.R., Glover, P., Kalaroopan, T., 1995. An evaluation of pollutant removal from secondary treated sewage efuent using a constructed wetland system. Water Sci. Technol. 32, 8793. Vymazal, J., 2002. The use of sub-surface constructed wetlands for wastewater treatment in the Czech Republic: 10 years experience. Ecol. Eng. 18, 633646. Ye, Z.H., Baker, A.J.M., Wong, M.H., Willis, A.J., 2003. Copper tolerance, uptake and accumulation by Phragmites australis. Chemosphere 50, 795800. Younger, P.L., Banwart, S.A., Hedin, R.S., 2002. Mine Water: Hydrology, Pollution, Remediation. Kluwer Academic Publications, London, UK.