Subsurface flow constructed wetland performance at a Pennsylvania campground and conference center

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.

ABSTRACT
A constructed wetland treatment system consisting of subsurface flow (SSF) wetland cells, sand filters, and final effluent wetlands was found to be effective in removing carbonaceous biochemical oxygen demand (CBOD) and total suspended solids (TSS) to below 30 and 10 mg L[sup-1], respectively. Removal efficiency of total nitrogen (TN) loads improved from 60.1 to 88.5% over the 2-yr study, primarily due to increased vegetation densities in the SSF wetland cells. In both years, parallel wetland treatment cells had significantly different (p < 0.001) plant densities of broadleaf cattail (Typha latifolia L.) and softstem bulrush [Schoenoplectus tabernaemontani (K.C. Gmel.) Palla], with significantly more TN removed from the more densely vegetated cell. Overall, the assimilation of N by plants removed less than 25% of the TN load, regardless of plant density, indicating that the primary role of deeply rooted macrophytes is supporting sequential nitrification-denitrification within the anaerobic wetland substrate. More than 99% of the dissolved phosphate (PO[sup3-[sub[sub4]-P) was removed within the entire system in both years, but removal efficiencies within the wetland cells decreased from 91.2% the first year to 66.1% the second year, indicating that adsorption sites for PO[sup3-[sub[sub4]-P may be saturated despite increased plant assimilation. Experimental manipulation of waste applied to the sand filters demonstrated that a header-type distribution system promoting horizontal flow was more effective at nitrifying ammonium (NH[sup+[sub[sub4]-N) discharged to the sand filters than the surface application of waste promoting vertical flow.
    Abbreviations: SSF, subsurface flow; SF, surface flow; CBOD, carbonaceous biochemical oxygen demand; TSS, total suspended solids; TN, total nitrogen.
    CONSTRUCTED wetland systems designed to treat domestic waste are typically surface flow (SF), SSF gravel bed, or combined SF-SSF systems (Cole, 1998). Wetland systems can serve as attractive alternatives to conventional treatment processes because they can often be constructed, operated, and maintained at lower costs than conventional systems. Aside from their primary role in removing organic compounds and nutrients through physical, chemical, and biological processes, constructed wetlands provide secondary benefits such as wildlife habitat, ground water recharge, and water reuse (Gearheart and Higley, 1993; USEPA, 1999; House et al., 1999). The multiple benefits of treatment wetland systems have given rise to their use in treating wastes from single family homes to large municipal systems, although treatment wetlands are often most cost-effective for small communities in rural areas where land is available and relatively inexpensive.
    Rural residences and communities are often served by decentralized, on-site domestic waste treatment systems. On-site treatment systems frequently have failed due to placement in poorly drained or shallow soils, and new development in rural areas is often hindered by these same soil limitations. Restrictions on the direct discharge of wastewater effluent into high-quality waters have also limited wastewater treatment solutions in many rural areas. Properly designed and constructed wetland systems may treat domestic waste to secondary standards or better and can also promote infiltration or reuse of treated waste, thus offering a cost-effective alternative to conventional waste treatment in these rural areas. The effectiveness of treatment wetlands in removing CBOD and TSS is well documented and design criteria related to these constituents are emerging (USEPA, 1988, 1999; Reed et al., 1995; Kadlec and Knight, 1996). However, the mechanisms and processes that control nitrogen and phosphorus removal in treatment wetlands are still poorly understood, and there are no specific design criteria to ensure the removal of nutrients within these alternative treatment systems (USEPA, 1993; Reed et al., 1995).
    This study investigated the nitrogen and phosphorus removal capabilities for a full-scale SSF wetland treatment system at a campground and conference center in Centre County, PA. The wetland treatment system was designed to remove CBOD and TSS from the camp's waste stream. In addition, the system was designed to enhance infiltration of system effluent into shallow ground water, thereby limiting the direct discharge of effluent into a nearby high-quality, cold water headwater stream. Our goals in this study were to: (i) monitor the treatment effectiveness of the wetland system, especially for nitrogen and phosphorus, for two years following construction in 1997; (ii) evaluate the establishment of vegetation within the system and its role in nitrogen and phosphorus removal; and (iii) experimentally manipulate waste distribution onto two parallel sand filters to assess their effectiveness at transforming and removing nitrogen and phosphorus.

METHODS

TREATMENT SYSTEM AND SAMPLING SITES
    The Krislund Campground is located in a headwater mountain valley near Madisonburg, Pennsylvania (40°57'08" N, 77°32'42" W). During the summer months, approximately 200 to 250 campers use the campground on a daily basis, while winter use is limited to weekend meetings and retreats. The campground has a bathhouse, dining hall, and toilet facilities available for camper use. Prior to construction of the wetland treatment system, waste from the campground facilities was treated by septic tanks and on-lot leachate fields. However, these systems failed over time due to the shallow, poorly drained colluvial soils on the site. To address the failed on-site disposal and to prevent surface discharge to an adjacent high-quality trout stream, the wetland treatment system (Fig. 1) was constructed in the early summer of 1997. Wastewater from camping areas and dining hall and conference buildings was collected in septic tanks and piped by gravity flow to a collection tank adjacent to the treatment wetland cells. The two parallel wetland cells were identical in design and construction, with wastewater equally dosed to both treatment cells by sump pumps located in the collection tank. The wetland cells (430-m[sup2] surface area, each 12.8 m long × 16.8 m wide × 0.46 m deep) were slightly oversized according to USEPA design criteria (USEPA, 1988):
    A[subs] = Q × (lnC[sub0] - lnC[sube]) / K[subT] × d × n
    where A[subs] = wetland surface area (401 m[sup2]), Q = design flow (17.03 m[sup3] d[sup-1], C[sub0] = influent BOD[sub5] (150 mg L[sup-1]), C[sube] = effluent BOD[sub5] (20 mg L[sup-1]), K[subT] = temperature-dependent decay constant (0.62 d[sup-1] at 18°C), d = wetland design depth (0.46 m), and n = substrate porosity (0.3).
    The medium for the two SSF wetland cells was washed river gravel (approx. 2 cm in diam.); both cells were lined with synthetic liners. The two cells were planted during the 1997 summer with bare root stock of broadleaf cattail and softstem bulrush at approx. 0.61- × 0.61-m spacing. Water levels in the treatment wetlands were separately set at the design depth (0.46 m) by two vertical standpipes.
    Waste discharge from the two wetland standpipes was combined in a collection box and piped by gravity to a siphon tank (1.2-m[sup3] capacity). When full, the siphon tank alternately emptied onto two unlined, unvegetated sand filters (19.5 m long × 16.8 m wide × 0.61 m deep). At the beginning of the 1998 camp season, we modified the sand filter distribution system for experimental purposes. The PVC pipe distribution network for the west sand filter discharged only to the head of the filter through a single perforated lateral (header-type sand filter, Fig. 1). The eastern pipe distribution network discharged water to the entire sand filter surface (surface-distributed sand filter, Fig. 1). The water level in both sand filters was also controlled by separate vertical standpipes to contain the waste within the sand layer. The overflow from the sand filter standpipes then discharged into two small surface-flow wetlands (final effluent wetlands, Fig. 1).
    Wetland influent samples (Site 1) were obtained from the collection tank located at the head of the two parallel wetland cells, and at the standpipes discharging waste from each wetland cell (Sites 2 and 3). Wastewater samples were also collected from each sand filter discharge (Sites 4 and 5) to determine differences in treatment effectiveness due to the different distribution systems. A grab sample was also collected from the eastern final effluent wetland (Site 6), and from an intermittent channel (Site 7) draining the eastern final effluent wetland and flowing into Roaring Run, a second-order headwater stream. Grab samples were also collected from Roaring Run at an upstream and downstream location so tha any influence of the wetland treatment system on the nutrient concentrations within the stream could be compared with an unaffected upstream location.

SAMPLING FREQUENCY AND ANALYTICAL TECHNIQUES
    Wastewater samples were collected and analyzed at weekly to biweekly intervals from mid-June through early September 1998, and from early June through early September 1999. All samples were collected in 60-mL polypropylene syringes and immediately placed on ice. All sample filtration (0.45 µm) was conducted immediately in the field. Samples for anion analysis were frozen, while samples for NH[sup+[sub[sub4]-N analysis were acidified with 0.5% concentrated sulfuric acid and stored at 4°C. Dissolved nitrate (NO[sup-[sub[sub3]-N) was determined by ion chromatography (American Public Health Association, 1995) or second derivative spectroscopy (Crumpton et al., 1992; Ferree and Shannon, 2000); PO[sup3-[sub[sub4]-P was determined by ion chromatography or the ascorbic acid method (American Public Health Association, 1995). Ammonium was determined by the automated phenate method on a Technicon (Tarrytown, NY) autoanalyzer (American Public Health Association, 1995), while TN was determined by persulfate digestion followed by second derivative spectroscopy (Crumpton et al., 1992; Ferree and Shannon, 2000). All analyses were conducted with periodic quality control check samples traceable to NIST standards. Additional sand filter effluent samples were collected periodically by Krislund Campground and state Department of Environmental Protection personnel for CBOD[sub5] and TSS concentrations, pH, and fecal coliform determinations.
    In late June 1999, we installed a pressure transducer in the siphon tank, and daily flow rates were determined by multiplying the water storage volume of the siphon tank times the number of siphon discharges per day. Direct determinations of flow from the treatment wetland cells were used along with daily water usage and camp occupancy to determine daily flows in 1998. Nutrient loads were calculated from the product of concentrations at sampling points in the system and flows obtained from the siphon tank. We did not measure flows at other points in the system; thus our load calculations for the sand filter effluent and final effluent wetland are conservative because of infiltration and evaporation in these components of the treatment system.

PLANT BIOMASS AND NUTRIENT MEASUREMENTS
    We determined plant biomass and nutrient uptake in 1998 and 1999 for both treatment wetlands to assess how the dominant plants affected N and P transformations and removal in the system. Due to the patchiness of growth over the first two growing seasons, wetland plant densities for the 1998 and 1999 growing seasons were determined by replicate stem counts of the two dominant species, broadleaf cattail and softstem bulrush. Stem counts were conducted in each of the two treatment wetlands after plant harvest at the end of each growing season. At the height of the growing seasons (late August), we randomly harvested 10 to 30 plants of the dominant species. Aboveground dry biomass (g m[sup-2]) for each species was determined for each wetland cell and year by multiplying the annual stem count per wetland surface area (stems m[sup-2]) times the dry biomass for a known sample (10-30 stems) of harvested plants (grams per 10-30 stems). Beggar-ticks (Bidens vulgata Greene) invaded the eastern wetland cell in 1999, so the annual herbaceous plant was only harvested from the east cell at the end of the 1999 growing season. The distribution of beggarticks in the east wetland cell was rather uniform, so plant biomass was determined by harvesting several randomly selected 0.5-m[sup2] plots. The entire aboveground portion of the harvested species were dried at 60°C after cutting the plants into ˜10-cm pieces to facilitate drying.
    The entire aboveground portions of the dried plants were also ground, homogenized, and analyzed to determine N and P accumulation in each species for the growing season. Plant tissues were ground to pass through a 1.0-mm screen and subjected to sulfuric acid digestion. Total N and total P were determined by standard Kjeldahl and molybdenum blue methods, respectively.

STATISTICAL ANALYSES
    Statistical comparisons of nutrient removal among parallel treatment units were conducted with paired sample t-tests, as were comparisons of influent and effluent nutrient loads within individual units of the treatment system. Two-factor analysis of variance was used to determine the significance of differences in plant biomass, with year (1998 or 1999) and wetland cell (east and west) as the two factors. All statistical tests were conducted with alpha = 0.05.

RESULTS
    Waste flows for the 1998-1999 summer seasons were not significantly different (p > 0.05), with mean values of 10.18 m[sup3] d[sup-1] for 1998 and 9.97 m[sup3] d[sup-1] for 1999. Corresponding hydraulic loading rates for 1998 and 1999 were 2.34 and 2.32 cm d[sup-1], respectively. Flows generally increased from June through early August of both years, and occasionally exceeded the design flow for CBOD removal on several days during significant rainfall events (Fig. 2). The system was effective at removing CBOD and TSS, with CBOD concentrations in the sand filter effluent less than 30 mg L[sup-1] and TSS less than 10 mg L[sup-1] on all sampling dates (Table 1). Circumneutral pH values between 7.05 and 7.70 were observed, while fecal coliform counts were less than 140 cfu per 100 mL in the sand filter discharges.

SPATIAL AND TEMPORAL TRENDS IN NITROGEN AND PHOSPHORUS

TREATMENT WETLANDS
    Total N in the wetland influent and effluent was comprised primarily of NH[sup+[sub[sub4]-N, with lesser amounts of organic N. For both years, NH[sup+[sub[sub4]-N was between 87 and 90% of the influent TN and 92 and 99% of the effluent TN. Dissolved nitrate concentrations were below detection limits (0.01 mg L[sup-1]) in the treatment wetland influent and effluent on every sampling date, so the small differences between TN and NH[sup+[sub[sub4]-N were attributed to organic N. There did not appear to be a seasonal trend in influent or effluent NH[sup+[sub[sub4]-N (or TN). Ammonium concentrations in the treatment wetland effluents were significantly lower (p < 0.001) than influent concentrations throughout the summers of 1998 and 1999 (Fig. 3). Influent concentrations ranged between 20 to 64 mg L[sup-1], while effluent concentrations ranged between 5 and 55 mg L[sup-1]. In addition, NH[sup+[sub[sub4]-N concentrations in effluent discharged from the two treatment wetlands (Sites 2 and 3) were significantly different (p < 0.001) throughout the 1998 and 1999 summers (Fig. 3). Mean removal of TN loads for the 1998 July-August period of heaviest camp use was 22.1% for the poorly vegetated eastern wetland and 38.7% for the well-vegetated west wetland. Removal efficiencies increased in 1999 for both cells, although TN removal in the west cell (57.7%) remained significantly higher than in the eastern cell (49.5%). For both years, differences in TN loads removed in the two wetland cells converged as flows increased in August. Total N loads, unlike TN and NH[sup+[sub[sub4]-N concentrations, exhibited seasonal trends consistent with camp water use, generally increasing from low values early in June, peaking in mid-summer, and decreasing in late August (Fig. 4).

SAND FILTERS
    The amount of nitrification of NH[sup+[sub[sub4]-N in wetland effluent that occurred in the sand filters was dependent on the type of distribution system used in the sand filters. The surface-distributed sand filter effluent (Site 4), had high NH[sup+[sub[sub4]-N concentrations that peaked at ˜12 mg L[sup-1] in late July-early August of both years (Fig. 5a). In contrast, NO[sup-[sub[sub3]-N concentrations were consistently below 3 mg L[sup-1] in 1998, and generally remained below 10 mg L[sup-1] during 1999, indicating that the effluent from the surface-distributed sand filter was only partially nitrified. The high NO[sup-[sub[sub3]-N and low NH[sup+[sub[sub4]-N concentrations observed in June and early July of 1999 were due to a broken outlet pipe, which allowed free drainage of water from the sand filter without storage. In contrast, NO[sup-[sub[sub3]-N was consistently greater than NH[sup+[sub[sub4]-N in the effluent from the header-type sand filter (Fig. 5b). Dissolved NO[sup-[sub[sub3]-N concentrations peaked above 20 mg L[sup-1] in late July-early August of both years, while NH[sup+[sub[sub4]-N was always below 3.5 mg L[sup-1]. The sand filters were effective at removing 10.8 to 18.7% of the TN load to the system over the two summers.

FINAL EFFLUENT WETLANDS, INTERMITTENT STREAM, AND ROARING RUN
    The final effluent wetlands further removed NH[sup+[sub[sub4]-N and NO[sup-[sub[sub3]-N from the waste. Dissolved NO[sup-[sub[sub3]-N concentrations in the final effluent wetland (Site 6) were below 10 mg L[sup-1] with the exception of several dates in August of both years when concentration peaked at 11 to 12 mg L[sup-1] (Fig. 6a). Ammonium concentrations at Site 6 were below 3.0 mg L[sup-1] in 1998 and 1.0 mg L[sup-1] in 1999. The mean efficiency of TN removal for the entire treatment system through the final effluent wetlands increased from 60.1% in 1998 to 88.5% in 1999.
    Surface water concentrations of NH[sup+[sub[sub4]-N and NO[sup-[sub[sub3]-N in the intermittent stream draining the final effluent wetlands were always less than the final effluent wetlands (Fig. 6a). In 1998, NO[sup-[sub[sub3]-N concentrations peaked at 3.4 mg L[sup-1] in early August, while NH[sup+[sub[sub4]-N peaked at 1.7 mg L[sup-1] in mid-August. Concentrations of NO[sup-[sub[sub3]-N and NH[sup+[sub[sub4]-N in the intermittent stream during 1999 were consistently lower than in 1998, with NO[sup-[sub[sub3]-N reaching 2.5 mg L[sup-1] on one sampling date in late June and NH[sup+[sub[sub4]-N remaining below 0.5 mg L[sup-1]. By late August of both years, there was no flow in the intermittent stream due to decreased flows and water loss from the system through evapotranspiration and infiltration from the unlined sand filters and effluent wetlands.
    Mean concentrations of NO[sup-[sub[sub3]-N in Roaring Run upstream and downstream from the wetland treatment system were 0.13 and 0.14 mg L[sup-1], respectively (Fig. 6b). The highest downstream concentrations (0.40-0.70 mg L[sup-1], which occurred in August of both years, coincided with similar upstream concentrations immediately following rainfall events. Mean upstream and downstream NH[sup+[sub[sub4]-N concentrations were 0.03 and 0.05 mg L[sup-1], respectively.

DISSOLVED PHOSPHORUS
    Influent concentrations of PO[sup3-[sub[sub4]-P ranged between 4.3 and 13.6 mg L[sup-1] in 1998 and 1999, and were significantly higher (p < 0.001) than PO[sup3-[sub[sub4]-P in the treatment wetland effluent at Sites 2 and 3 (Fig. 7). The west wetland (Site 3) removed significantly (p < 0.001) more PO[sup3-[sub[sub4]-P than the east wetland (Site 2) in both years. Dissolved phosphate loads were reduced by 85.1 and 97.3% in the east and west wetlands cells, respectively, in 1998, while 1999 load reductions were 55.4 and 71.3% in the east and west wetlands cells, respectively. There were not significant differences in the PO[sup3-[sub[sub4]-P loads removed between the two sand filters in 1998, with loads reduced by 99.2 to 99.3% in the discharge from Sites 4 and 5. However, in 1999 there were significant differences (p = 0.04) between PO[sup3-[sub[sub4]-P loads removed through Sites 4 (95.3%) and Site 5 (97.1%), as well as a significant decrease (p < 0.001) between 1998 and 1999 in PO[sup3-[sub[sub4]-P loads removed through discharge from the sand filters. Concentrations of PO[sup3-[sub[sub4]-P in the final effluent wetland were usually below detection limits (0.01 mg L[sup-1]) in both years, with removal efficiencies of 99 to 100%.

PLANT BIOMASS AND NUTRIENT UPTAKE
    Plant biomass in the treatment wetlands was highly dependent on year and wetland cell. The differences were attributable to preferential grazing by deer (Odocoileus virginianus) of the small cattail and bulrush shoots in the eastern wetland before establishment of an adequate root-rhizome system. Deer-proof fencing was erected around the perimeter of the wetland system during the fall of 1997, and plant growth was unaffected by grazing during the subsequent two growing seasons.
    Broadleaf cattail and softstem bulrush biomass in the west wetland cell was significantly greater (p < 0.001) than in the east wetland cell for both growing seasons. Broadleaf cattail biomass increased significantly (p < 0.001) between 1998 and 1999 in both the west and east cells (Table 2). However, while softstem bulrush biomass in the west cell increased between 1998 and 1999, biomass in the east cell decreased significantly between 1998 and 1999. The 1999 decline in softstem bulrush biomass in the east cell coincided with an invasion of beggar-ticks only in the east cell, with biomass accumulating to 634 g m[sup-2].
    Mean aboveground tissue concentrations of N and P for broadleaf cattail were 16.4 and 2.2 g kg[sup-1], and for softstem bulrush were 13.9 and 2.0 g kg[sup-1], respectively (Table 2). Wetland location (east vs. west) and year did not have a significant effect on plant nutrient levels. In contrast with softstem bulrush and broadleaf cattail, beggar-ticks accumulated tissue concentrations of N (44.2 g kg[sup-1]) and P (6.5 g kg[sup-1]) that were two to three times higher than those found in broadleaf cattail or softstem bulrush.
    Seasonal aboveground plant uptake of N and P was a small percentage of TN and P removal in the treatment wetlands. In 1998, aboveground biomass of broadleaf cattail and softstem bulrush in the east wetland only removed 3.6% of the TN load and 1.7% of the PO[sup3-[sub[sub4]-P load to the wetland, whereas the two species in the more densely vegetated west wetland removed 7.9 and 3.8% of the TN and PO[sup3-[sub[sub4]-P loads, respectively. In 1999, the TN and PO[sup3-[sub[sub4]-P removal attributable to plants increased as plant densities continued to increase in the 2-yr-old system. The rapid invasion of beggar-ticks into the east cell in 1999, combined with high tissue concentrations, caused the percentage of TN and PO[sup3-[sub[sub4]-P removed by plants to increase to 25.5 and 25.0%, respectively. Plant uptake in the west cell, which was not invaded by beggar-ticks in 1999, increased to 11.1% of the TN load and 10.3% of the PO[sup3-[sub[sub4]-P load to the wetland.

DISCUSSION
    Our results indicate that the Krislund wetland treatment system, through the final effluent wetlands, was effective at removing nitrogen, and treatment effectiveness improved as the system matured. While 60.1% of influent TN was removed in 1998, 88.5% was removed in 1999 (Fig. 8a). The improvement in TN removal appeared to be due to increased plant densities in both treatment cells, although the mechanisms supporting nitrogen loss in the two cells differed dramatically. It is unlikely that NH[sup+[sub[sub4]-N volatilization had a significant role in N removal at the circumneutral pH values observed in the system.
    In the west treatment wetland, TN removal by the combined aboveground tissues of broadleaf cattail and softstem bulrush accounted for only 7.9 and 11.1% of TN inputs in 1998 and 1999, respectively. However, TN removal by all processes in the west cell improved from 38.7% in 1998 to 57.7% in 1999, indicating that plants removed 19 to 20% of the TN removed by all processes in the west treatment wetland in both years. Thus, 80 to 81% of the nitrogen was removed by processes other than plant assimilation, suggesting that the dominant process removing nitrogen in the west cell in both years was nitrification-denitrification. Conversely, TN removal by plants in the east cell increased from 3.6% in 1998 to 25.5% in 1999, while TN removal from all processes in the east cell increased from 22.1% in 1998 to 49.5% in 1999 (Fig. 8a). In 1998, the aboveground tissues of broadleaf cattail and softstem bulrush still accounted for 16% of the TN removed in the east wetland cell, whereas in 1999, the high biomass and tissue N concentrations of beggar-ticks increased TN removal by plants to 52% of the TN removed in the east cell (Fig. 8a).
    Our results indicate that the short-term consequence of competition by shallow-rooted annuals like beggar-ticks is that significantly more N and P are temporarily assimilated into plant tissues from the shallow root zone. However, plant assimilation is a permanent net sink for nutrients only if the biomass is harvested and permanently removed from the wetlands, and this is not a practical management strategy in most treatment wetlands. The long-term consequence of replacing deeply rooted perennials with shallow-rooted annuals is a decrease in net N removal because nitrification and subsequent denitrification are reduced. Therefore, the more important role of deeply rooted emergent macrophytes like broadleaf cattail and softstem bulrush in full-scale treatment wetlands is in transporting and leaking oxygen to the rhizosphere to support aerobic decomposition of organic matter and nitrification (Bendix et al., 1994; Kludze and Delaune, 1996; Jespersen et al., 1998).
    Several studies have shown that plant assimilation can be a significant mechanism for N and P removal in batch-fed, small-scale experimental systems (Breen, 1990; Rogers et al., 1990, 1991). However, our results for wetlands with deeply rooted macrophytes are in agreement with measurements of plant productivity and nutrient assimilation in larger systems with realistic loading rates. These studies indicate that the proportion of nutrient assimilation by plants seldom exceeds 20 to 25% of the TN load, and that the fraction of TN load assimilated by plants is inversely proportional to the load (Gersberg et al., 1986; Frankenbach and Meyer, 1999; Tanner et al., 1998).
    In the Krislund system, oxidation of NH[sup+[sub[sub4]-N within the rhizosphere was rapidly followed by denitrification within the anoxic environment of the bulk wetland substrate, because NO[sup-[sub[sub3]-N was never detected in the effluent of either treatment wetland. Others (Tanner et al., 1998; Maschinski et al., 1999) observed similar relationships between NH[sup+[sub[sub4] and NO[sup-[sub[sub3], suggesting that nitrification is the rate-limiting step controlling N removal in the anoxic medium of SSF wetlands, and that NO[sup-[sub[sub3] formed through nitrification is rapidly denitrified, keeping NO[sup-[sub[sub3] at low concentrations despite significant TN removal.
    Mean TN removal rates for the 1998 summer were 0.31 g m[sup-2] d[sup-1] and increased to 0.73 g m[sup-2] d[sup-1] in 1999, and were comparable with other studies. Maschinski et al. (1999) obtained rates of ˜1 g m[sup-2] d[sup-1] in a small system planted to native southwestern U.S. plants, while Tanner et al. (1995, 1998) obtained mean TN removal rates ranging from 0.14 to 2.23 g m[sup-2] d[sup-1] for New Zealand pilot-scale wetlands treating dairy wastewaters with similar hydraulic loading and nitrogen concentrations. Our results indicate that TN removal from the treatment wetlands is significantly related to TN loading (Fig. 9). Kadlec and Knight (1996) reported a similar positive relationship for loading rates up to 3 g m[sup-2] d[sup-1] in a wide variety of SF and SSF wetlands. However, the removal efficiency decreases at higher loading rates.
    Emergent macrophytes have been shown to reach maximum densities as high as 2000 to 3000 g m[sup-2] in other high-nutrient wetlands (Dubbe et al., 1988; Martín and Fernández, 1992; Tanner, 1994, 1996). Our biomass values for the west cell in 1999 (1103 g m[sup-2]) indicate that maximum densities have not been attained after two growing seasons and that removal of nitrogen through nitrification-denitrification should continue to increase as deeply rooted plant densities increase. However, shallow-rooted plants like beggar-ticks, once established, may replace or slow the growth of desirable species, thus preventing rhizosphere oxidation and limiting nutrient loss to plant assimilation. Tanner (1996) also found, in an investigation of several wetland macrophytes, that weed management and early spring planting was important for the establishment of softstem bulrush. Our research supports these findings and indicates that rapid establishment of dense stands of desirable but slow-spreading perennial species can be an effective management strategy by preventing the invasion of competitive annuals that spread by rapid seed dispersal.
    Processes other than plant uptake dominated PO[sup3-[sub[sub4]-P removal in the Krislund system. Removal of PO[sup3-[sub[sub4]-P in the 2-yr study was significant, but removal efficiencies for the treatment wetlands decreased despite an increase in plant biomass and total P assimilation by plants over the period (Fig. 8b). Treatment wetland studies have demonstrated substantial P removal in newly constructed wetlands (Tanner et al., 1995; Maschinski et al., 1999), while other studies have documented that P removal efficiency decreases with time (Mann, 1990; Tanner et al., 1998). The Krislund wetland system appeared to follow a similar P removal trajectory. Removal of PO[sup3-[sub[sub4]-P is primarily by adsorption to substrate exchange sites, and it is unlikely that the sorption capacity of the substrate can be sustained with time at the loading rates associated with full-scale treatment systems. The trend in decreasing PO[sup3-[sub[sub4]-P removal will probably continue in the treatment wetlands as available adsorption sites in the gravel substrate are saturated with time and more dissolved P is exported from the system.
    While the sand filters only removed a minor amount (10.8-18.7%) of the TN load to the system, they played an important role in nitrifying NH[sup+[sub[sub4]-N discharged from the treatment wetlands. The surface-distributed sand filter was significantly less effective at nitrifying waste because of the short vertical flow path to the saturated zone of the filter bed and reduced detention time for waste within the sand filter, while the header-type distribution network allowed for horizontal flow and more contact time with aerobic sand layers within the filter. Gross et al. (1995) demonstrated that sand filters can provide effective nitrification of high NH[sup+[sub[sub4]-N waste discharged from SSF wetlands, while Tanner et al. (1999) have shown that frequent water level fluctuations can also accelerate nitrification in anoxic waste. The sand filters in the Krislund system were not vegetated for experimental reasons, but establishment of vegetation on the sand filter beds could also be used to enhance further denitrification, as well as to provide another source of wastewater reduction through evapotranspiration during the summer period of highest camp use. Tong and Sikora (1995) have shown that dense plantings of water-tolerant species such as reed canarygrass (Phalaris arundinacea L.) can enhance nitrification-denitrification in filter beds by leaking carbon from roots to the substrate in carbon-limited systems. While nitrification in the sand filters of this study was effectively supported, denitrification within the lower saturated layers of the filters may have been limited by the availability of carbon, as demonstrated by the low CBOD values discharged from the sand filters.
    Our study demonstrates that full-scale treatment wetlands can effectively remove nitrogen and phosphorus, and that proper management of vegetation is a key component equal to design in importance. Invasive, shallow-rooted annuals may assimilate more nitrogen and phosphorus on a seasonal basis, but this short-term benefit is offset by the loss of long-term nitrification-denitrification potential provided by oxygen leakage from deeply rooted perennials. The performance of the system could be further improved to optimize nitrification by fully converting the sand filter distribution network to a header-type system. The final effluent wetlands could also be improved by additions of compost or other suitable carbon amendments to enhance further denitrification of sand filter effluent.
ADDED MATERIAL
    Robert D. Shannon,* Oscar P. Flite III, and Michael S. Hunter
    R.D. Shannon, Dep. of Agricultural and Biological Engineering, 233 Agric. Eng. Bldg.; O.P. Flite III, Environmental Pollution Control Program, 204 Agric. Sci. and Industry Bldg.; and M.S. Hunter, Environmental Resource Management Program, 206 Agric. Sci. and Industry Bldg., The Pennsylvania State Univ., University Park, PA 16802. Received 25 Feb. 2000. *Corresponding author (rds13@psu.edu).

ACKNOWLEDGMENTS
    The authors wish to thank Mark Stephens and Jason Fellon of the Pennsylvania Dep. of Environmental Protection, Jody Quimby and Tom Whittle of CET, Inc., Steve Cort and Kent Rishel of Krislund for their assistance on the project, and Elisa Brown and Michelle Ferree for assisting in the laboratory and field. Funding for the project was provided by Penn State University's College of Agricultural Sciences through the undergraduate research grants program, and through the Joan Luerssen Faculty Enhancement Fund and Horace T. Woodward Faculty Development Fund.
    Table 1. Concentrations of constituents from water samples collected by camp personnel and analyzed by an independent laboratory. Samples were collected from the effluent standpipes from the sand filters.

Date           CBOD[sub5](FN+)        TSS(FN++)   pH     Fecal coliforms
                                mg L[sup-1]              cfu per 100 mL
20 July 1998     9.4                  6.7        7.70         89
2 Aug. 1999     11                    3.9        7.66        111
11 Feb. 1999     4.3                 <1.0        7.32         56
12 Apr. 1999    21                    2.8        7.45         39
7 June 1999     <1.0                  6.0          -          40
24 June 1999     7.2                  1.0        7.05         97
12 July 1999    15                    1.8        7.25         22
8 Aug. 1999     28                    9.9        7.25        140

FOOTNOTES
+ Carbonaceous biochemical oxygen demand.
++ Total suspended solids.
    Table 2. Aboveground biomass and nutrient concentrations in tissues for dominant plant species in west and east wetland treatment cells. Biomass values are means (+/- 1 SE) of three stem counts for broadleaf cattail and softstem bulrush, or harvests from three 0.5-m[sup2] plots for beggar-ticks. Nutrient concentrations in tissues represent means (+/- 1 SE) of triplicate analyses of composite samples.

                          West wetland biomass                   East wetland biomass
                       1998                1999               1998              1999          N in tissues     P in tissues
                                                 g m[sup-2]                                         g kg[sup-1]
Broadleaf cattail   484 (+/- 12)       1009 (+/- 9)        176 (+/- 2)       521 (+/- 10)     16.4 (+/- 1.2)   2.2 (+/- 0.2)
Softstem bulrush   38.0 (+/- 1.8)      94.7 (+/- 1.8)     23.9 (+/- 2.1)    10.9 (+/- 0.4)    13.9 (+/- 0.1)   2.2 (+/- 0.1)
Beggar-ticks            -                   -                  -            634 (+/- 13)      44.2 (+/- 2.7)   6.5 (+/- 0.5)

Fig. 1. Site layout of Krislund wetland treatment system.
Fig. 2. Flow from treatment wetlands for 1998-1999.
Fig. 3. Ammonium (NH[sup+[sub[sub4]-N) concentrations in treatment wetland influent and effluent for 1998-1999.
Fig. 4. Total N loads in treatment wetland influent and effluent for 1998-1999.
Fig. 5. Ammonium (NH[sup+[sub[sub4]-N) and dissolved nitrate (NO[sup-[sub[sub3]-N) concentrations- in effluent of (a) surface-distributed sand filter and (b) header-type sand filter.
Fig. 6. Ammonium (NH[sup+[sub[sub4]-N) and dissolved nitrate (NO[sup-[sub[sub3]-N) concentrations in (a) final effluent pond (Site 6) and intermittent stream (Site 7) and (b) upstream and downstream sampling sites in Roaring Run.
Fig. 7. Dissolved phosphorus concentrations at Sites 1 through 5 in the wetland treatment system.
Fig. 8. Removal efficiencies for (a) total N and (b) dissolved PO[sup3-[sub[sub4]-P through the wetland treatment system for 1998-1999. Plant uptake percentages are shown for treatment wetland cells (Sites 2 and 3).
Fig. 9. Relationship between total N removed and total N mass loading to treatment wetland cells.

REFERENCES
    American Public Health Association. 1995. Standard methods for the examination of water and wastewater. APHA, Washington, DC.
    Bendix, M., T. Tornbjerg, and H. Brix. 1994. Internal gas transport in Typha latifolia L. and Typha angustifolia L.I. Humidity-induced pressurization and convective throughflow. Aquat. Bot. 49:75-89.
    Breen, P.F. 1990. A mass balance method for assessing the potential of artificial wetlands for wastewater treatment. Water Res. 24:689-697.
    Cole, S. 1998. The emergence of treatment wetlands. Environ. Sci. Technol. 32:218A-223A.
    Crumpton, W.G., T.M. Isenhart, and P.D. Mitchell. 1992. Nitrate and organic N analyses with second-derivative spectroscopy. Limnol. Oceanogr. 37:907-913.
    Dubbe, D.R., E.G. Garver, and D.C. Pratt. 1988. Production of cattail (Typha spp.) biomass in Minnesota, USA. Biomass 17:79-104.
    Ferree, M.A., and R.D. Shannon. 2000. Evaluation of a second derivative UV/Visible spectroscopy technique for nitrate and total nitrogen analysis of wastewater samples. Water Res. (in press).
    Frankenbach, R.I., and J.S. Meyer. 1999. Nitrogen removal in a surface-flow wastewater treatment wetland. Wetlands 19:403-412.
    Gearheart, R.A., and M. Higley. 1993. Constructed open surface wetlands: The water quality benefits and wildlife benefits--City of Arcata, California. p. 561-567. In G.A. Moshiri (ed.) Constructed wetlands for water quality improvement. Lewis Publ., Boca Raton, FL.
    Gersberg, R.M., B.V. Elkins, S.R. Lyon, and C.R. Goldman. 1986. Role of aquatic plants in wastewater treatment by artificial wetlands. Water Res. 20:363-367.
    Gross, M.A., R. Burson, and A. Brown. 1995. Sand filtration of constructed wetland effluent. p. 597-606. In K.L. Campbell (ed.) Versatility of wetlands in the agricultural landscape. ASAE, St. Joseph, MI.
    House, C.H., B.A. Bergmann, A.M. Stomp, and D.J. Frederick. 1999. Combining constructed wetlands and aquatic and soil filters for reclamation and reuse of water. Ecol. Eng. 12:27-38.
    Jespersen, D.N., B.K. Sorrell, and H. Brix. 1998. Growth and root oxygen release by Typha latifolia and its effects on sediment methanogenesis. Aquat. Bot. 61:165-180.
    Kadlec, R.H., and R.L. Knight. 1996. Treatment wetlands. CRC Press, Boca Raton, FL.
    Kludze, H.K., and R.D. DeLaune. 1996. Soil redox intensity effects on oxygen exchange and growth of cattail and sawgrass. Soil Sci. Soc. Am. J. 60:616-621.
    Mann, R.A. 1990. Phosphorus removal by constructed wetlands: Substratum absorption. p. 97-106. In P.F. Cooper and B.C. Findlater (ed.) Constructed wetlands in water pollution control. Pergamon Press, Oxford, UK.
    Martín, I., and J. Fernández. 1992. Nutrient dynamics and growth of a cattail crop (Typha latifolia L.) developed in an effluent with high eutrophic potentíal--Application to wastewater purification systems. Bioresour. Technol. 42:7-12.
    Maschinski, J., G. Southam, J. Hines, and S. Strohmeyer. 1999. Efficiency of a subsurface constructed wetland system using native southwestern U.S. plants. J. Environ. Qual. 28:225-231.
    Reed, S.C., R.W. Crites, and E.J. Middlebrook. 1995. Natural systems for waste management and treatment. 2nd ed. McGraw-Hill, New York.
    Rogers, K.H., P.F. Breen, and A.J. Chick. 1990. Hydraulics, root distribution and phosphorus removal in experimental wetland systems, p. 87-90. In P.F. Cooper and B.C. Findlater (ed.) Constructed wetlands in water pollution control. Pergamon Press, Oxford, U.K.
    Rogers, K.H, P.F. Breen, and A.J. Chick. 1991. Nitrogen removal in experimental wetland treatment systems: Evidence for the role of aquatic plants. J. Water Pollut. Control Fed. 63:934-941.
    Tanner, C.C. 1994. Growth and nutrition of Schoenoplectus validus in agricultural wastewaters. Aquat. Bot. 47:131-153.
    Tanner, C.C. 1996. Plants for constructed wetland treatment systems--A comparison of the growth and nutrient uptake characteristics of eight emergent species. Ecol. Eng. 7:59-83.
    Tanner, C.C., J.S. Clayton, and M.P. Upsell. 1995. Effect of loading rate and planting on treatment of dairy farm wastewaters in constructed wetlands: II. Removal of nitrogen and phosphorus. Water Res. 29:27-34.
    Tanner, C.C., J. D'Eugenio, G.B. McBride, J.P.S. Sukias, and K. Thompson. 1999. Effect of water level fluctuation on nitrogen removal from constructed wetland mesocosms. Ecol. Eng. 12:67-92.
    Tanner, C.C., J.P.S. Sukias, and M.P. Upsell. 1998. Relationships between loading rates and pollutant removal during maturation of gravel-bed constructed wetlands. J. Environ. Qual. 27:448-458.
    Tong, Z., and F.J. Sikora. 1995. Ammonium and nitrate removal in vegetated and unvegetated gravel bed microcosm wetlands. Water Sci. Technol. 32:219-228.
    USEPA. 1988. Design manual: Constructed wetlands and aquatic plant systems for municipal wastewater treatment. EPA/625/1-88/022. USEPA Office of Research & Development, National Risk Management Research Laboratory, Cincinnati, OH.
    USEPA. 1993. Subsurface flow constructed wetlands for wastewater treatment: A technology assessment. EPA 832-R-93-008. USEPA, Washington, DC.
    USEPA. 1999. Draft: Guiding principles for constructed treatment wetlands: Providing water quality and wildlife habitat. Interagency Workgroup on Constructed Wetlands. USEPA, Washington, DC.