Nutrient processing capacity of a constructed wetland in Western Ireland

The magazine publisher is the copyright holder of this article and it is reproduced with permission. Further reproduction of this article in violation of the copyright is prohibited. To contact the publisher: http://www.agronomy.org/

ABSTRACT
In Ireland, constructed wetland systems are increasingly being used to perform tertiary treatment on municipal waste effluent from small towns and villages located in areas whose receiving waters are deemed sensitive. The bedrock formation in the west of Ireland is primarily karst limestone and where the overburden-soil cover is very shallow, such waters are highly sensitive to pollution sources, as little or no natural attenuation and/or treatment will occur. Constructed wetland technology has been seen to offer a relatively low-cost alternative to the more conventional tertiary treatment technologies, particularly when dealing with low population numbers in small rural communities. This paper examines the waste treatment performance, in terms of nutrient (P and N) reduction, of a recently constructed surface-flow wetland system at Williamstown, County Galway, Ireland. Performance evaluation is based on more than two years of water quality and hydrological monitoring data. The N and P mass balances for the wetland indicate that the average percentage reduction over the two-year study period is 51% for total N and 13% for total P. The primary treatment process in the wetland system for suspended solids (between 84 and 90% reduction), biological oxygen demand (BOD) (on average, 49% reduction), N, and P is the physical settlement of the particulates. However, the formation of algal bloom during the growing season reduces the efficiency of the total P removal.
    Abbreviations: BOD, biological oxygen demand; FWS, free-water surface; PE, population equivalent; TSS, total suspended solids.
    OVER THE PAST two decades, much research attention has been focused on quantifying the effectiveness and improving the design of constructed wetlands in treating wastewater at either secondary or tertiary (polishing) stages. It is now recognized that constructed wetlands can provide an effective and economic way of treating liquid effluents. The application of constructed wetland technology is being used worldwide to cleanse effluents from a variety of sources, for example, municipal waste effluent from cities and towns, sewage from single households and small rural communities, agricultural runoff, dairy washings, mine drainage, urban and motorway storm runoff, and landfill leachate.
    In Ireland, application of wetland technology is in its infancy relative to North America and other European countries. The last five years have seen a significant introduction of constructed wetland technology as a sewage treatment system to single households and small rural communities and as a tertiary system to larger villages and towns. There is also a growing use of constructed wetlands in farmyard waste management and in the treatment of urban stormwater runoff, in particular treatment of motorway runoff. As a treatment option it has widespread appeal due to its relative low running cost and its visual and environmental acceptance by the general public.
    In Ireland, as in quite a number of other countries, water authorities are reluctant to permit the use of constructed wetlands as stand-alone secondary treatment systems. In many cases they are being recommended only as an add-on polishing system to either conventional secondary treatment systems at the town and village scale or to septic tank systems at the household and small community scale, particularly in areas of sensitive receiving waters. The primary reasons for this reluctance is the scarcity of reliable long-term performance data relevant to the particular country and climate and their general poor winter treatment performances.
    There is a considerable body of literature on the role of constructed wetlands in tertiary treatment. Much of this is in relation to N and P reduction capacity.

NITROGEN REMOVAL BY THE WETLAND
    Both free-water surface (FWS) and subsurface (SF) constructed wetlands are very effective in N removal, with up to 70 to 80% removal being reported (Knight et al., 1993). Removal efficiencies are very dependent on residence time and temperature. Research performed by Bachand and Horne (2000) show that water temperature and organic carbon availability affect denitrification rates. Longer retention times also result in enhanced settlement of particulate organic N within the wetland. Nitrogen removal in constructed wetlands is accomplished primarily by physical settlement, denitrification, and plant and microbial uptake. Plant uptake does not represent permanent removal unless plants are routinely harvested.
    In a previous study (Cooper, 1996), a one-cell FWS constructed wetland was used as a combined storm and tertiary reed bed, treating effluent from an English village with a population equivalent of 382. The area of the wetland was 380 m[sup2], giving an areal loading of approximately 1 m[sup2] per population equivalent (PE). Operating under a normal hydraulic loading of around 15 cm d[sup-1], the wetland performed well, giving effluent ammonia and total oxidized N concentrations of 0.37 and 9.4 mg L[sup-1], respectively. However, during storm events, where the hydraulic loading increased to a maximum of around 181 cm d[sup-1], an upward trend in the effluent ammonia and a slight increase in total oxidized N was recorded. This suggests that decreasing retention time may adversely affect the efficacy of treatment.
    Studies indicate that, dependent on the organic loading regime, all types of wetlands have a good nutrient-processing efficacy under European climatic conditions (Coombes, 1990; Luederitz et al., 2001). In a study of four wetlands (two FWS and two vertical flow wetlands), Leuderitz et al. (2001) found that, under an organic load of 3.24 g BOD m[sup-2] d[sup-1], a FWS wetland in Loburg, Germany had a 95 and 69% reduction in BOD and total N, respectively. Similarly, Coombes (1990) found that under an organic load varying from 1.14 to 14 g BOD m[sup-2] d[sup-1], up to a 93% BOD reduction was recorded in a subsurface wetland in Kirmington, England. However, in the study, a surface flow bypass in some wetlands gave rise to a variation in the performance data. This may have affected the nitrification-denitrification processes within the wetland.

PHOSPHORUS IMMOBILIZATION BY THE WETLAND
    Unlike N, there is no gaseous loss in P removal from a system. The only P removal mechanism is through short-term or long-term storage. Uptake by biota, including bacteria, algae, and duckweed (Lemna spp.), as well as macrophytes, provides an initial removal mechanism (Kadlec, 1997). However, this is only a short-term P storage as 35 to 75% of P stored is eventually released back into the water upon dieback of algae and microbes (Richardson and Craft, 1993; White et al., 2000). The only long-term P storage in the wetland is via peat accumulation. The efficiency of long-term peat storage is a function of the loading rate and also depends on the amount of native iron, calcium, aluminum, and organic matter in the substrate (Wood and Hensman, 1989; Shatwell and Cordery, 1999).
    Lake and reservoir sediments have been shown to act as P sinks (Froelich, 1988; Sundby et al., 1992; Richardson and Craft, 1993; White et al., 2000). At P loadings of less than 5 g P m[sup-2] yr[sup-1], wetland sediment can absorb greater than 90% of the total incoming P (Faulkner and Richardson, 1989). Phosphorus-containing particles settle to the substrate and are rapidly covered by a continuous accumulation of settled sediment. Provided appropriate measures are taken to stabilize the bed sediment (i.e., flow-through velocities are controlled), continuous accumulation of sediment will leave some P too deep within the substrate to be reintroduced back into the water column (Shatwell and Cordery, 1999).
    However, a small portion of the P in the substrate may be reintroduced back into the water column by a process known as desorption (Meyer, 1979; Froelich, 1988; Sundby et al., 1992). Froelich (1988) used the term "phosphate buffering" to describe this process and defined it as "fast and slow, reversible and irreversible phosphate 'fixation' onto or off of soil or fluvial particles, whether the mechanism is via disequilibrium with a discrete mineral phase or with phosphate adsorbed on surfaces." Conceptually, the ability of a substrate to bind P can be defined by the concentration gradient between the water column and upper layers of the substrate. Based on the assumption that the upper layer of the substrate strives to be in equilibrium with the overlying water, an "equilibrium phosphorus concentration" can be defined. This is obtained experimentally from an adsorption isotherm plot that graphs the equilibrium concentration to which ortho-P will settle when added to a sediment sample (from the wetland). The P concentration of the water column relative to this value gives an indication of the ongoing ability of the substrate to bind (and subsequently) immobilize P.
    The objective of this study is to quantify the nutrient reduction efficiency of a surface-flow constructed wetland under Irish climatic conditions. Using a nutrient mass balance (based on concentration and flow data), the role of sedimentation and biological processes in nutrient removal will be quantified. Also, the effect of ambient water temperature on these processes will be analyzed.

MATERIALS AND METHODS

SITE DESCRIPTION
    The site of the wetland is located on the outskirts of the town of Williamstown in North County Galway, Ireland. The bedrock formation of the region is primarily karst limestone overlain by a thin layer of glacial deposits. Catchment drainage is via a series of small streams that disappear under ground via a swallow hole located approximately 500 m east of the site. This underlying aquifer system is classified as a Regionally Important Karst Aquifer of extreme vulnerability (Daly, 1985). As a consequence of the sensitivity of the receiving water body and the lack of suitable streamflow for initial dilution, the waste effluent requires the provision of tertiary treatment in addition to the conventional (20/30) secondary treatment. Constructed wetland technology was seen to offer a relatively low-cost, effective alternative to the more conventional tertiary wastewater treatment technologies, particularly in this case where the PE is not very high.
    In the summer of 1998, construction of a secondary treatment package plant (aeration chamber and clarifier) and a free-water surface constructed wetland was completed, consisting of two reed bed cells and one retention pond cell connected in series. The three cells of the wetland are constructed as shallow lagoons enclosed by boulder clay embankments and lined with a high-density polyethylene (HDPE) liner. Details of the wetland system are presented in Table 1. Two wetland emergent plant species were used: common reed [Phragmites australis (Cav.) Trin. ex Steud.] and common cattail (Typha latifolia L.). Floating vegetation planted in the pond system was a variety of water lily (Nuphar spp.) and duckweed.
    The design mean inflow rate to the treatment works was 79 m[sup3] d[sup-1], which represents a PE of 330. The observed mean inflow rate (Table 2) for the 1998-1999 study period was 55 m[sup3] d[sup-1]. Analysis of the rainfall-inflow hydrographs showed that approximately 28% of this discharge was storm runoff from a number of roofs and yards that had not been separated. Therefore, the actual PE discharging to the treatment system was 147 (based on flow volumes). The secondary treatment plant was designed to produce an effluent quality better than 20 mg L[sup-1] BOD, 30 mg L[sup-1] suspended solids (SS), and 10 mg L[sup-1] ammonia (NH[sub3]-N). The measured effluent quality leaving the package treatment plant for the study period had a mean BOD of 18 mg L[sup-1], total suspended solids (TSS) of 52 mg L[sup-1], and NH[sub3]-N of 2.2 mg L[sup-1].

HYDROLOGICAL MONITORING
    The flow of water within the wetland system was controlled by V-notch weirs, which were located at the outlet of the first, second, and third cells. An electromagnetic flow meter located at the inlet of the first cell and druck pressure transducers (0.06% accuracy) located in the V-notch chambers at the outlet of each cell allowed continuous measurement of the flow rate through the system. A rating relationship was derived for each of the V-notch weirs from on-site volumetric flow measurements. The accuracy of the electromagnetic meter was also checked and found to be well within the 5% standard. Two tipping-bucket rain gauges (located on site) allowed the daily rainfall amount to be measured. The Irish Meteorological Office supplied potential evapotranspiration values as monthly totals.

WATER QUALITY MONITORING
    A refrigerated autosampler (WaterSam, Rottenburg, Germany) was set at the wetland inlet to produce a composite daily sample based on discrete hourly samples of 100 mL. Grab water samples were taken from four points throughout the wetland (at the inlet to the system and at the outlet of the first, second, and third cells). Composite daily samples, collected from the autosampler at the inlet to the system, allowed the nutrient loading into the wetland to be quantified within 2d of collection. Testing for total N, total P, and chlorophyll a was performed by the Irish Environmental Protection Agency. Testing for fecal coliforms, chemical oxygen demand (COD), TSS, ammonia, nitrate, and inorganic phosphorous was performed in the Department of Engineering Hydrology, National University of Ireland, Galway, and BOD[sub5], COD, and TSS measurements were performed by Galway County Council. All analyses were performed in accordance with standard methods (American Public Health Association, 1995).

NITROGEN REDUCTION BY THE WETLAND
    Nutrient loading into and from the wetland were calculated by accounting for inflow and outflow discharge with the electromagnetic pressure transducer and the V-notch weirs. The product of the discharge data and the nutrient concentration gave the nutrient loading. The retention time within each cell was calculated by dividing the water volume retained within that cell by the incoming flow.
    Nitrate depletion rates (denitrification) and particulate organic N settlement were calculated from the grab samples. Water samples were analyzed from 26 May 1998 to 17 Feb. 2000 (n = 77 data sets). Particulate organic N settlement was calculated from the difference in influent and effluent particulate organic N values. Particulate organic N was calculated from the total N minus the ammonia N and the nitrate N concentrations. As particulate organic N affects uptake and release rates from the substrate, the remineralization of the particulate component was calculated based on ammonia N production rates within the wetland cells. Apart from the limited occurrence of remineralization, the majority of the particulate organic N retained by the system was assumed to be immobilized and ultimately incorporated into the substrate. The N content of the suspended solids deposited within each cell was also quantified (after Tebbutt, 1998).
    A N mass balance was calculated for the system throughout the duration of the project, with daily measured concentrations and flows. The influent and effluent data were expressed in terms of grams per day (g d[sup-1]). The settlement of particulate organic N was calculated by subtracting the influent organic N and the effluent organic N from each cell. The N uptake by the wetland vegetation was calculated from a series of measurements, which were taken throughout the course of the study. An average denitrification rate was calculated for each cell from the total influent N minus the total effluent N, the settlement, and plant uptake.

PHOSPHORUS IMMOBILIZATION BY THE WETLAND
    Throughout the duration of the study, 77 water samples were analyzed for ortho-phosphate P (PO[sub4]-P) and, from the 24 July 1999 samples, for total reactive P (a measurement performed on the total unfiltered sample) and filtrable reactive P (American Public Health Association, 1995). The water sample was filtered through a 0.45-µm filter paper to obtain filtrable reactive P. Total reactive P minus the filtrable reactive P gives a measure of the particulate reactive P. Filtration of the water sample separates the suspended form of P and the filtrable form of P (American Public Health Association, 1995)
    The ability of the wetland substrate to function as a sink for P was also investigated at the end of the study. Before the wetland commenced operation, chemical analysis was performed on the substrate. The average initial total P concentration of the substrate was measured (266 mg kg[sup-1]), which is within the range of values found in Irish soils (Conry, 1994). The depth of the soil was measured at six points throughout the wetland (at the inlets and the outlets of the first reed bed, the retention pond, and the second reed bed, respectively). This gave the average total depth of the substrate (including the depth increase due to sediment deposition) in each cell during the lifetime of the wetland. The substrate increase (expressed as millimeters of dry matter) due to sediment deposition was calculated from the suspended solids deposition. The volumetric moisture content was calculated for the substrate, thereby allowing an estimate of the initial substrate height to be made. Two separate sets of soil samples were taken from each cell in plastic containers (which were previously rinsed in the pond water). Chemical analysis was performed on one sample set. The total P content of the mixed sample (i.e., the initial substrate plus the deposited sediment) was calculated (in mg kg[sup-1] dry weight) with Morgan's extracting solution. This allowed a calculation of the total P content of the deposited sediment to be made. Using the surface area of each cell and the total sediment mass, the P content of the sediment was calculated. (Side slopes approached 1:0, thereby simplifying calculations.) Knowing the average influent P loading, the amount of P adsorbed by the soil could then be calculated. The percentage error of the calculations could be calculated from the P content of the suspended solids (after Tebbutt, 1998).
    Adsorption isotherms are calculated for the upper layers of the wetland substrate. The isotherms are based on the method developed by Froelich (1988), which estimates the ongoing ability of the wetland substrate to adsorb phosphates. Undisturbed soil samples were collected from the surface of the wetland and sample analysis was performed immediately. The soil samples were overlain by approximately 150 mL of chemically amended water to give varying P concentrations. The samples were subsequently centrifuged for a period of one day and the overlying water was tested for P. The amount of P adsorbed by the soil could then be calculated.

NITROGEN AND PHOSPHORUS UPTAKE BY STANDING MACROPHYTES
    Total N and P uptake by the wetland plantation species in each cell is calculated from the average N and P retention value measured per shoot. During the course of the study, representative samples of both species were destructively sampled. Using dry weight calculations on a series of plant samples from each reed bed, the approximate number of shoots per meter squared, and the total reed coverage area, the cumulative N and P retention (g) was calculated. Using these data, N and P uptake rates were calculated for both species of vegetation.

RESULTS AND DISCUSSION

NITROGEN REMOVAL
    Table 3 presents results of the N mass balance for the 1998-1999 study period. The N mass balance indicates that, on average, a 51% reduction in total N occurs from the inlet to the outlet of the system. The retention pond has the poorest reduction in total N (13%), with the first and third cells recording reductions of 30 and 21%, respectively. The high standard deviations in the concentration measurements are a result of variations in the operation of the secondary treatment system.
    Common reed accounted for approximately 94 and 78% of the total-vegetative area in the first and third cells, respectively, and common cattail accounted for the remainder. Annual cumulative N retention was estimated to be, on average, 83 mg m[sup-2] d[sup-1] for common reed and 48 mg m[sup-2] d[sup-1] for common cattail. Per shoot the N content of common cattail, however, was found to be four to five times that of common reed. The total N uptake by the wetland species plays a virtually insignificant role when compared with the total loading on the system, with an areal N loading rate of approximately 2.44 g m[sup-2] d[sup-1] and cumulative plant uptake of 0.083 g m[sup-2] d[sup-1] (i.e., 33 times smaller than the loading). This is illustrated in Fig. 1.
    Denitrification is an anaerobic process that occurs in the upper layers of the substrate and is highly temperature dependent (Bachand and Horne, 2000). This was also verified in the Williamstown wetland with direct correlations between observed denitrification rates and water temperature. A plot of denitrification rate and temperature for the second reed bed illustrates this (Fig. 2). The computed annual denitrification (net nitrate N removal) rate for the wetland system based on the mass balance presented in Table 3 is 0.542 kg d[sup-1] or 198 kg yr[sup-1]. This represents an annual reduction rate of 33.5% of the influent N.
    Operating a free-water surface constructed wetland under Mediterranean conditions consisting of three cells in replicate under a mean hydraulic loading rate of 68 cm d[sup-1], Bachand and Horne (2000) calculated nitrate N removal rates of 261 to 835 mg N m[sup-2] d[sup-1], which linearly reduced to zero during the winter period (with drop in temperature). Comparatively, for this study, an average nitrate depletion rate of 528 mg N m[sup-2] d[sup-1] was computed for the overall wetland system. A standard deviation of 437 mg N m[sup-2] d[sup-1] reflects the high variability among cells. In the retention pond the denitrification rate was found to be considerably lower, with a rate of 101 +/- 211.9 g d[sup-1].
    Net settlement of particulate organic N (calculated from the difference in influent and effluent particulate organic N values) accounts for the removal (as long term storage in the substrate) of 12% of the total incoming N. Settlement was maximum in the first cell, with a value of 140 +/- 178.1 g d[sup-1]. On the basis of suspended solids deposition within the cells, sediment N, calculated after Tebbutt (1998), accounts for 77.6 and 40.3 g N d[sup-1] in the first and second reed beds, respectively.

PHOSPHORUS REMOVAL
    The variability of the influent and effluent P is shown in Table 4. The high standard deviations are due to variations in the operation of the secondary treatment system. The estimated retention of P in the cells is shown in Table 5. This comprises two components: P reduction from solution and retention in the suspended solids.
    The reduction of P by the wetland system is poor, giving an average reduction rate of 13% for total P and 26% for ortho-phosphate P. The average summer and winter reductions are 27 and -1%, respectively, for total P and 32 and 22%, respectively, for ortho-phosphate P. Mineralization of organic P in the wetland is also limited. The average ratio of ortho- to total P at the inlet is approximately 82%; however, the ratio at the outlet averages 71%. This reduction is due to phytoplankton in the retention pond assimilating the ortho-phosphate P during the summer period. During the summer period these temporary storage compartments become increasingly more important than the long-term storage compartments. Organic P increases to approximately 17% of the total P in the retention pond. In the retention pond and second reed bed, an apparent production of 41 and 37 g d[sup-1] of organic P, respectively, reflects this (Table 5). Ortho-phosphate P in suspension is also affected. An average ortho-phosphate P production of 27 g d[sup-1] occurs in the first reed bed. This suggests that organic P is being converted to ortho-phosphate P in this cell. Phosphorus retention in suspended solids in the first reed bed is almost double the retention in the second reed bed (Table 5). Due to eutrophication effects, no suspended solids deposition occurred in the retention pond. Average P loading and reductions for the entire period of analysis (1998 and 1999) and for the growing season (15 Apr. 1999 to 21 Oct. 1999) and nongrowing season (28 Oct. 1999 to 29 Mar. 2000) are illustrated in Fig. 3. The relationship between the particulate reactive P (denoted in the figure as particulate P) and the suspended solids is illustrated in Fig. 4. (The particulate content is normalized by expressing it as a fraction of total reactive P).
    The total P uptake by the wetland species is estimated to be 7 mg m[sup-2] d[sup-1]. This is virtually insignificant when compared with the total P loading on the system of 390 mg m[sup-2] d[sup-1], 55 times smaller than the loading rate. On dieback of the wetland vegetation, a large proportion of nutrients (both N and P) held in the stems and leaves retreats down to the roots where it is stored for next year's new growth.
    Emergent and submerged macrophytes play a significant role in limiting eutrophication. Due to eutrophication effects, the ortho-phosphate P is assimilated by the phytoplankton in the retention pond and is subsequently converted to organic P. During the summer period the percentage of P in suspension rises to approximately 25% of the total P (Fig. 3A).
    Removal-retention efficiency of the substrate is highly dependent on the loading rate; long-term P removal is achieved with loadings of less than 5 g P m[sup-2] yr[sup-1] (Faulkner and Richardson, 1989; Richardson and Craft, 1993). White et al. (2000) reported a 60% P immobilization with an average loading of 4.75 g P m[sup-2] yr[sup-1]. In Williamstown the average yearly P loading is 101 g P m[sup-2] yr[sup-1], which is 20 times higher than the above recommended rates. The calculated sediment P content yields retention rates of between 3 and 6% of the influent P. These calculations were verified by comparison with the P content of the measured deposited suspended solids (after Tebbutt, 1998).
    The wetland soil still exhibits good P sorption capacity, however. Adsorption isotherm data indicates that the wetland still has a capacity for continued orthophosphate P adsorption from the water column (Table 6). Adsorption isotherms indicate the continued ability of a soil to adsorb P. Its saturation concentration (the point at which the soil is unable to absorb any more P) is higher than the average dissolved ortho-phosphate P concentration in the overlying water, indicating that the soil still has available sorption sites.
    Although adsorption isotherms indicate a downward P flux into the soil, the reducing conditions that exist within the substrate mean that, while a net sorption may be occurring, significant desorption of P from the soil into the water column may also occur. As a result, the net adsorption is slowed down. Extractable aluminum, generally believed to be among the most important bonding parameters (Meyer, 1979), is very low (<5.0 mg kg[sup-1]) in comparison with laterite, which has good P binding capacity. The iron to P ratio in the soil is 1:0.09 (Table 7).
    The occurrence of iron and particularly its relationship to the redox potential within the wetland substrate is of particular importance. At redox potentials (Eh) of less than +250 mV, Fe[sup3+] will be converted to Fe[sup2+], releasing associated P. As the redox potential measurements indicate that conditions in the wetland substrate are largely anaerobic, the above reaction is likely to be important. Redox potential measurements throughout the course of the study indicated that the soil had an Eh of approximately +100 mV, indicating that the soil was largely anaerobic (Russell, 1988).

PATHOGEN REMOVAL
    Based on weekly measurements performed from September 1999 to March 2000, an average overall reduction in fecal coliforms of 99.77% (from an average influent value of 45 808 coliforms 100 mL[sup-1]) indicates that the system has excellent pathogen removal capability. Fecal coliform reduction rates in the reed beds are higher than the rates recorded in the retention pond. This indicates that the pathogen attenuation mechanisms unique to reed beds (i.e., attenuation by the dense stands of reeds, predation by protozoa, and the exposure to antibiotic excretions from the roots of the macrophytes within the reed beds) are of some significance. A retention time of 8 d will ensure high die-off rates in any case.

SUSPENDED SOLIDS REMOVAL
    An overall reduction in suspended solids of between 84 and 90% was noted in the wetland. Most settlement occurred in the first cell, where reductions of up to 94% were recorded. However, due to the intense eutrophication in the retention pond (where the water remained hypertrophic 55% of the time), an increase in suspended solids was observed. Aside from the effect of algal blooms, the reduction in TSS is a function of the hydraulic retention time.

BIOLOGICAL OXYGEN DEMAND REMOVAL
    An overall reduction in BOD of 49% (standard deviation 28%) was achieved by the wetland system during the study period. No discernable trend was evident in terms of a reduction in treatment efficiency with decreasing ambient temperature (i.e., summer growing season versus winter nongrowing season). If anything, the summer period proved to have the lower efficiencies, brought about by the formation of algal blooms in the detention pond. Higher efficiencies were also evident with higher BOD loadings. The primary treatment process is physical settlement as opposed to biological oxidation.

SUMMARY AND CONCLUSIONS
    The final effluent from the Williamstown wetland is of acceptable quality for discharge to the ground water via a percolation field, having typically a BOD of 9 mg L[sup-1] (standard deviation 6 mg L[sup-1]), TSS of 9 mg L[sup-1] (standard deviation 10 mg L[sup-1]), total N of 14.2 mg L[sup-1] (standard deviation 3.2 mg L[sup-1]), ammonia N of 0.8 mg L[sup-1] (standard deviation 1.79 mg L[sup-1]), nitrate N of 9.2 mg L[sup-1] (standard deviation 2.7 mg L[sup-1]), total P of 4.45 mg L[sup-1] (standard deviation 2.4 mg L[sup-1]), and ortho-phosphate of 3.15 mg L[sup-1] (standard deviation 1.1 mg L[sup-1]). The average percentage reduction in the wetland system over the two-year study period is 48% for BOD, 83% for TSS, 51% for total N, and 13% for total P. Pathogen removal in the wetland system is excellent having a fecal coliform removal efficiency of 99.77% (standard deviation 0.27%) and a final effluent mean concentration of 118 per 100 mL for the monitored period September to March 2000. This removal efficiency is to be expected given the long hydraulic retention time of 8 d. A conservative T[sub90] of 72 h (night time) would achieve this removal efficiency through die-off alone. This wetland system has presently 6.8 m[sup2] PE[sup-1] (based on a present day PE of 147). This is significantly above the Irish Environmental Protection Agency-recommended design standards of 1 m[sup2] PE[sup-1] for tertiary treatment (Irish Environmental Protection Agency, 2000). This does not necessarily mean that the design of wetlands should be downsized, though, as it is very evident from this study that the primary treatment process in the wetland system for suspended solids, BOD, N, and the limited P removal is physical settlement of the particulates. A total hydraulic retention time of 8 d allows ample time for the finer particulates to-settle out. However, the formation of algal blooms in the retention pond during the growing season reduces the efficiency of this process significantly.
    The design of an integrated retention pond in a FWS constructed wetland needs to be questioned. The retention pond in Williamstown wetland provides an ideal environment for eutrophication to occur (i.e., deep water levels, low flushing rates, no plant growth). Eutrophication is detrimental to the P processing capacity of the wetland. The retention pond is adequate in its N processing efficiency, albeit in the presence of a substrate and an adequate retention time, N removal will occur irrespective of plant coverage or water depth. However, its inclusion in the treatment system has no functional relevance other than to increase the retention time. The incorporation of a subsurface reed bed or an extra reed bed would better serve the system.
    The nutrient uptake by the emergent wetland vegetation does not play a significant role, having annual uptake rates of 30 g m[sup-2] N and 2 g m[sup-2] P. These uptake rates represent only 3.4% of the total N load and 1.8% of the total P load. Kadlec (1989) cites annual uptake rates for wetland vegetation (Typha spp.) treating secondary effluent in Michigan of 60 g m[sup-2] N and 12 g m[sup-2] phosphorous. He also cites that in natural wetlands, where there is a lesser availability of nutrients, uptake rates can be three to five times lower. Based on crop densities at Williamstown, vegetation development was two-thirds complete by the end of the study period, requiring one more growing season to achieve full crop density. Aside from nutrient uptake, the vegetative crop is necessary for summer shading to eliminate algal blooms, shelter against wind shear turbulence (resuspension of sediments), and aid settlement by filtration. The treatment efficiencies for the individual cells clearly show that inclusion of an extra reed bed in place of the retention pond, which promotes eutrophication in summer due to the lack of vegetative cover, would better serve the system. The difficulty with including floating vegetation cover in the pond is the exposure of the site to the prevailing wind direction, which has the tendency to pile floating vegetation to one side or to remove it completely.
    In situ redox measurements indicate that oxygen is limited and that conditions in the substrate and at the water interface are generally anaerobic, favoring denitrification as opposed to nitrification. Under these conditions decomposition and mineralization of settled organic matter is attenuated and, therefore, litter fall and settled organic particulate matter can contribute significantly to peat accretion.
    Overall, the performance of the Williamstown wetland system is without doubt very encouraging in promoting the use of FWS wetlands as a tertiary treatment system for protecting sensitive receiving waters. Ample hydraulic retention time ensures good physical treatment through settlement and also allows sufficient time for pathogen die-off. The system performed well in N reduction in comparison with such treatment systems through good summer denitrification rates and long-term or permanent storage of settled organic N as litter in the substrate. However, in terms of P reduction, wetland systems are of limited value unless they are designed to be excessively large.
    Importantly, sufficient sizing of wetlands allows their safe use as retention in the case of breakdown and/or malfunction of the secondary treatment plant. They have the added advantage of smoothing out variations in the influent quality by virtue of their storage volume.
ADDED MATERIAL
    M. Healy, Dep. of Civil Engineering, and A.M. Cawley, Dep. of Engineering Hydrology, National University of Ireland, Galway, Ireland. Corresponding author (Mark.Healy@nuigalway.ie).

ACKNOWLEDGMENTS
    The authors wish to thank Mr. Paul Ridge, Senior Executive Engineer, Galway Co. Council for initiating this project and providing the necessary funding. They would also like to thank Dr. Michael Flanagan, Regional Inspectorate, Irish Environmental Protection Agency, Castlebar and Mr. John Heneghan, Tuam Water Laboratory, Galway County Council for their invaluable assistance in the water quality measurements. They would also like to express their special appreciation to Mr. John Mulqueen, Teagasc, NUI, Galway for reviewing this manuscript. The helpful comments by the anonymous reviewers are also appreciated.
    Table 1. Details of combined wetland configuration.

                         Cell dimensions
Cell               Length     Width     Depth     Area        Volume      HRT(FN+) (mean flow 55 m[sup3] d[sup-1])  HLR(FN++)
                               m                   m[sup2]       m[sup3]               d                       cm d[sup-1]
First cell          28         10        0.3       277         83                  1.51                       20
Retention pond      38.5        8        0.8       300        240                  4.36                       18.7
Second cell         47        9-12       0.3       448        134                  2.44                       12.7
Combined                                          1025        457                  8.3                         5.4

FOOTNOTES
+ Hydraulic residence time.
++ Hydraulic loading rate.
    Table 2. Seasonal and annual wetland water balance.

                                 Effluent                             Potential
                               inflow rate       Rainfall         evapotranspiration
Summer inflow, m[sup3] d[sup-1]  41.5 +/- 12.8      3.1                    2.1
1998 and 1999, mm d[sup-1]      40.5 +/- 12.5      3.02                   2.03
Winter inflow, m[sup3] d[sup-1]  66.7 +/- 28.9      4.7                    0.3
1999, mm d[sup-1]               65.1 +/- 28.2      4.54                   0.3
Annual inflow, m[sup3] d[sup-1]  55.5 +/- 23.3      3.9                    1.2
1998 and 1999, mm d[sup-1]      54.1 +/- 22.7      3.8                    1.2

    Table 3. Mean influent nitrogen values +/- standard deviation and nitrogen mass balance.

                                                                     Influent
Parameter                        First reed bed                   Retention pond
                         mg L[sup-1]         g d[sup-1]       mg L[sup-1]        g d[sup-1]
Total N                 29.1 +/- 9.7    1614 +/- 537      20.5 +/- 5.9    1152 +/- 575.4
No[sub3]-N               19.1 +/- 7.1    1060 +/- 693.3    13.3 +/- 4.4     747 +/- 257.4
TKN(FN+)                10.0 +/- 5.5     555 +/- 186.5     7.2 +/- 2.1     405 +/- 547.1
NH[sub3]-N                2.2 +/- 3.5     122 +/- 144.5     2.0 +/- 3.1     112 +/- 182.9
Denitrification(FN++)                        300                               101
Settlement(FN§)                        140 +/- 178.1                      31 +/- 98.1
Plant uptake                              22 +/- 10.3                            0

Parameter                      Second reed bed                         Effluent
                         mg L[sup-1]                      g d[sup-1]
Total N                 17.9 +/- 5.0        1020 +/- 233.8            826 +/- 195.8
NO[sub3]-N               12.2 +/- 3.7         695 +/- 303.8            535 +/- 314.4
TKN(FN+)                 5.7 +/- 2.2         325 +/- 99.6             290 +/- 110
NH[sub3]-N                1.1 +/- 2.5          63 +/- 152.3             47 +/- 144.1
Denitrification(FN++)                            141
Settlement(FN§)                             19 +/- 33.7
Plant uptake                                  34 +/- 11.9

FOOTNOTES
+ Total Kjeldahl nitrogen.
++ Denitrification = total N (in) -- total N (out) -- settlement -- plant uptake.
§ Settlement = organic N (in) -- organic N (out).
    Table 4. Variability in influent and effluent phosphorus (+/- SD) in wetland cells.

                                                             Influent
Parameter               First reed bed                    Retention pond                  Second reed bed          Effluent
                mg L[sup-1]        g d[sup-1]        mg L[sup-1]       g d[sup-1]        mg L[sup-1]              g d[sup-1]
Total P        5.1 +/- 3.0      283 +/- 167.7    4.6 +/- 1.9     259 +/- 108.8    4.35 +/- 1.6     245 +/- 85.0   258 +/- 119
PO[sub4]-P     4.17 +/- 1.6      232 +/- 67.5     4.6 +/- 1.5     259 +/- 84.1     3.58 +/- 1.2     204 +/- 67.6   180 +/- 65.1
Organic P     0.93 +/- 0.96      51 +/- 27.9       0 +/- 1.27      0 +/- 73.6     0.72 +/- 0.93     41 +/- 36.4    78 +/- 43.9

    Table 5. Phosphorus retention in the wetland cells.

                     P reduction from solution(FN+)
                                                                  P in retained
Cells               Total P     PO[sub4]-P     Organic P         suspended solids(FN++)
                                                g d[sup-1]
First reed bed        24         -27             51                   15.5
Retention pond        14         -55            -41                    0
Second reed bed      -13          24            -37                    8.1

FOOTNOTES
+ Calculated from the difference in water solution phosphorus concentrations.
++ Calculated from the P content of the suspended solids (after Tebbutt, 1998) retained by each cell.
    Table 6. Phosphorus flux within the wetland substrate column.

                                    Average water
                    Equilibrium     column PO[sub4]-P      Saturation PO[sub4]-P
Location              PO[sub4]-P      concentration         concentration(FN+)
                                         mg L[sup-1]
Inlet                  <3               3.83                    4.85
First reed bed          2.6             4.1                     6.03
Retention pond          6               2.9                    >6
Second reed bed         2.3             3.11                   >7

FOOTNOTE
+ Saturation PO[sub4]-P concentration refers to the maximum difference between ortho-phosphate added and ortho-phosphate sorbed.
    Table 7. Chemical analysis and comparison study of wetland substrate.

                    Total    Total     Extractable      Extractable
Location             Fe       P            Al               Ca           CEC(FN+)
                              mg kg[sup-1]                mg L[sup-1]
Inlet                7600     700         <5.0            3377           16.64
First reed bed       7860     800         <5.0            2578           16.25
Retention pond       7880     800         <5.0            5142           11.47
Second reed bed      8760     900         <5.0            9471            9.44

FOOTNOTE
+ Cation exchange capacity.
Fig. 1. Total nitrogen entering the wetland cells versus the amount retained by the cumulative vegetation. (Cell 1 and Cell 3 refer to the first and second reed beds, respectively.)
Fig. 2. Relationship between denitrification rate and temperature in the second reed bed.
Fig. 3. Variation in phosphorus concentration in the wetland cells. (A) Summer season (15 Apr. 1999 to 21 Oct. 1999). (B) Winter season (28 Oct. 1999 to 29 Feb. 2000). (C) Averaged throughout study.
Fig. 4. Relationship of suspended solids to particulate reactive phosphorus in suspension.

REFERENCES
    American Public Health Association. 1995. Standard methods for the examination of water and wastewater. 19th ed. Am. Public Health Assoc., Washington, DC.
    Bachand, P.A.M., and A.J. Horne. 2000. Denitrification in constructed wetland free-water surface wetlands: II. Effects of vegetation and temperature. Ecol. Eng. 14:17-32.
    Conry, M. 1994. Environmental impact statement. Galmoy Mine Project. Eolas, Dublin.
    Coombes, C. 1990. Reed bed treatment systems in Anglian water. p. 223-234. In P.F. Cooper and B.C. Findlater (ed.) Constructed wetlands in water pollution and control. IAWPRC, London.
    Cooper, P.F. 1996. Reed beds and constructed wetlands for wastewater treatment. Water Res. Centre Publ., Swindon, UK.
    Daly, D. 1985. Groundwater in County Galway with particular reference to its protection from pollution. Rep. 2.2.7/2.11.7. Geol. Survey of Ireland, Dublin.
    Faulkner, S.P., and C.J. Richardson. 1989. Physical and chemical characteristics of freshwater wetland soils. p. 41-72. In D.A. Hammer (ed.) Constructed wetlands for wastewater treatment. Lewis Publ., Chelsea, MI.
    Froelich, P.N. 1988. Kinetic control of dissolved phosphate in natural rivers and estuaries: A primer on the phosphate buffer mechanism. Limnol. Oceanogr. 33:649-668.
    Irish Environmental Protection Agency. 2000. Waste water treatment manuals. Treatment systems for small communities, leisure centres and hotels. Irish Environmental Protection Agency, Johnstown Castle.
    Kadlec, R.H. 1989. Decomposition in wastewater wetlands. p. 459-468. In D.A. Hammer (ed.) Constructed wetlands for wastewater treatment. Lewis Publ., Chelsea, MI.
    Kadlec, R.H. 1997. An autobiotic wetland phosphorus model. Ecol. Eng. 8:145-172.
    Knight, R.L., R.W. Ruble, R.H. Kadlec, and S. Reed. 1993. Wetlands for wastewater treatment performance database. p. 35-58. In G.A. Moshiri (ed.) Constructed wetlands for water quality improvement. Lewis Publ., London.
    Luederitz, V., E. Eckert, M. Lange-Weber, A. Lange, and R.M. Gersberg. 2001. Nutrient removal efficiency and resource economics of vertical flow and horizontal flow constructed wetlands. Ecol. Eng. 18:157-171.
    Meyer, J.L. 1979. The role of sediments and bryophytes in phosphorus dynamics in a headwater stream ecosystem. Limnol. Oceanogr. 24:365-375.
    Richardson, C.J., and C.B. Craft. 1993. Effective phosphorus retention in wetlands: Fact or fiction. p. 271-282. In G.A. Moshiri (ed.) Constructed wetlands for water quality improvement. Lewis Publ., London.
    Russell, E.W. 1988. Russell's soil conditions and plant growth. 11th ed. Longman Sci. & Tech., New York.
    Shatwell, T., and I. Cordery. 1999. Nutrient storage in urban wetlands. p. 339-347. In B. Ellis (ed.) Impacts of urban growth on surface water and groundwater quality. Publ. 259. IAHS, Wallingford, UK.
    Sundby, B., C. Gobeil, N. Silverberg, and A. Mucci. 1992. The phosphorus cycle in coastal marine sediments. Limnol. Oceanogr. 37:1129-1145.
    Tebbutt, T.H.Y. 1998. Principles of water quality control. 5th ed. Butterworth-Heinemann, Oxford.
    White, J.S., S.E. Bayley, and P.J. Curtis. 2000. Sediment storage of phosphorus in a northern prairie wetland receiving municipal and agro-industrial wastewater. Ecol. Eng. 14:127-138.
    Wood, A., and L.C. Hensman. 1989. Research to develop engineering guidelines for implementation of constructed wetlands for wastewater treatment in South Africa. p. 581-589. In D.A. Hammer (ed.) Constructed wetlands for wastewater treatment. Lewis Publ., Chelsea, MI.