| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Dep. of Biological and Agric. Engineering, North Carolina State Univ., Box 7625, Raleigh, NC 27695-7625
b Dep. of Biological and Agric. Engineering, North Carolina State Univ., Box 7625, Raleigh, NC 27695-7625
c RSMT, 3919 Fisher Ferry Rd., Vicksburg, MS 39180
d Dep. of Soil Science, North Carolina State Univ., Box 7619, Raleigh, NC 27695-7619
e Dep. of Biological and Agric. Engineering, North Carolina State Univ., Box 7625, Raleigh, NC 27695-7625
f Dep. of Statistics, North Carolina State Univ., Box 8203, Raleigh, NC 27695-8203
* Corresponding author (mike_burchell{at}ncsu.edu)
Received for publication January 13, 2006.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONOBJECTIVESMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: BOD, biological oxygen demand • CDF, confined disposal facility • DM, dredged material only treatment wetland • DMB18%, dredged material blend treatment wetland with 18% organic matter • DMB22%, dredged material blend treatment wetland with 22% organic matter • DMWs, dredge material wetlands • DO, dissolved oxygen • Eh, substrate redox • ET, evapotranspiration • OM, organic matter • RSMT, recycled soil manufacturing technology • SFCW, surface-flow constructed wetlands • SS, site soil only treatment wetland • SSB11%, site soil blend treatment wetland with 11% organic matter • SSB16%, site soil blend treatment wetland with 16% organic matter • SSWs, site soil wetlands
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONOBJECTIVESMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
A typical surface-flow constructed wetland (SFCW) is constructed similar to a natural freshwater marsh. Along with water and plants, wetland soil, often referred to as substrate, is a crucial component of SFCWs. Not only do substrates provide physical support for emergent wetland plants, but they also provide large surface areas for microbial attachment and for complexing ions and other compounds (Hammer and Bastian, 1989). Reed et al. (1995) add that substrates supply wetland plants with the majority of nutrients required for plant growth. In a SFCW, it is the water–substrate interface where most of the critical microbial transformations of pollutants occur (Steiner and Freeman, 1989). Wetland plant litter, which accumulates in wetland substrate, provides a renewable source of organic matter, provides sites for material exchange and microbial attachment, and helps in maintaining low redox conditions that are crucial in many important biological transformation reactions (Kadlec and Knight, 1995).
Denitrification is considered the major pathway of N removal from aquatic sediments since it can completely remove nitrogen from the system (Reddy et al., 1989). For denitrification to proceed within a system such as a constructed wetland, certain conditions must exist: presence of NO3–, anoxic conditions, low redox conditions, acceptable temperature and pH conditions, and an adequate carbon source (Knowles, 1982; Beauchamp et al., 1989; Reed et al., 1995).
Carbon is important for optimizing denitrification rates because it supports requirements for both energy and cellular synthesis for the heterotrophic bacteria that are considered to be most responsible for utilizing nitrogen oxides as electron acceptors in the absence of oxygen (Knowles, 1982). Stoichiometrically, 2.47 g of methanol (commonly used in conventional wastewater treatment systems, CH3OH), or equivalent, is required to reduce 1 g of nitrate to nitrogen gas, including cellular synthesis (Kadlec and Knight, 1995). However, reduction of nitrate in aquatic systems is usually enhanced when carbon/nitrogen ratios exceed these theoretical levels, due in part to small levels of dissolved oxygen that allow some aerobic degradation of the carbon source. McCarty et al. (1969) expressed the carbon requirement as methanol, Cm, based on the concentrations of nitrate, nitrite, and dissolved oxygen (DO):
 | [1] |
Several researchers have attempted to quantify the effect of organic matter (OM) in the enhancement of nitrate removal from water. In a two-part study, Engler and Patrick (1974) found that nitrate removal in a submerged saltwater marsh soil with between 200 and 250 g kg–1 (20 to 25% dry wt.) OM was greater than that in a swamp soil with 70 g kg–1 (7% dry wt.) OM. In the second part of that study, a positive correlation in nitrate removal was found with the addition of varying amounts of rice straw to the surface of a mineral soil that originally contained less than 10 g kg–1 (1% dry wt.) OM. They argued that the rice additions decreased the distance of nitrate diffusion to an anaerobic layer, and increased the availability of a microbial energy source, both of which would enhance denitrification. Davidsson and Ståhl (2000) found a weak correlation between nitrate loss and OM in cores studied with 50 to 640 g kg–1 (5 to 64% dry wt.) OM, but also stated that all of the cores were suitable for removal of nitrate. However, a microcosm study by Phipps and Crumpton (1994) found no significant increase in nitrate loss with increased OM additions to sediment cores they were studying, but the cores contained between 90 and 280 g kg–1 (9 and 28% dry w.) OM before the additions. Studies on constructed wetlands by Gersberg et al. (1984), Baker (1998), and Ingersoll and Baker (1998) found C/N ratios of 4:1 to 5:1 may maximize nitrate removal in those treatment systems.
Surface-flow constructed wetlands receiving nitrified influent waters that are low in dissolved carbon or biological oxygen demand (BOD) must receive a carbon supply from outside additions or from within the wetland itself to maximize its denitrification potential. In a survey of the substrate of several natural wetlands, Faulkner and Richardson (1989) reported organic matter levels between 170 and 770 g kg–1 (17 and 77% dry wt.), the result of a gradual buildup of detrital matter from decaying wetland plants. Mineral soils used as substrate in constructed wetlands can often limit denitrification due to low levels of OM (Nichols, 1983), a limitation much more prevalent in constructed wetlands than in older natural marshes (Bachand and Horne, 2000b). Constructed wetlands often take years to mature and accumulate levels of organic matter comparable to natural marsh systems. Supplementing the wetlands with an external carbon source such as methanol can stimulate denitrification, but also can substantially increase operating costs. Additions of alternative carbon sources that are easily degraded such as mulch, grass clippings, or harvested wetland plants have been shown to be an effective substitute to methanol in wetlands (Gersberg et al., 1983). However, as stated earlier, additions should be higher than the theoretical methanol/nitrogen ratios due to losses of the carbon fraction to aerobic decomposition, as well as resistance to degradation of the biomass lignin fraction (Gersberg et al., 1983, 1984).
Manipulation of Wetland Substrate to Enhance Denitrification
Despite the evidence that OM enhances denitrification rates in constructed wetlands, the importance of the substrate is often overlooked in the design of these systems. Addition of OM and nutrients to mineral substrate to be used in a SFCW may serve to enhance denitrification in the first few years of the wetland by providing a head start in the accumulation of available carbon. The increase in OM will essentially age the SFCWs so that they more quickly reach the OM levels found in natural marshes. Abundance of nutrients, particularly nitrogen, in the substrate may also serve to increase the overall production of biomass (Broome et al., 1975; Allen et al., 1989), which will later become the major source of renewable carbon to the SFCW.
Enhancement of the substrate in a SFCW can be accomplished using recycled soil manufacturing technology (RSMT), an innovative approach in using contaminated or noncontaminated soils and/or sediments to engineer inexpensive soils with increased stability, fertility, and functionality. The soil or sediment is mixed with cellulose (originating from sawdust, yard waste, waste paper products, etc.) and biosolids to provide a fertile soil that has many beneficial uses (Lee and Sturgis, 1996; Lee et al., 1998; Sturgis et al., 2001). Some of these uses have included topsoil used in landscaping, as well as landfill, Superfund, and mining site covers.
|
OBJECTIVES |
|---|
|
TOPABSTRACTINTRODUCTIONOBJECTIVESMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONOBJECTIVESMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
|
Six substrate conditions and a control were created in wetland mesocosms, in triplicate, for a total of 21 wetland mesocosms (Table 1). The treatments were placed in a randomized complete block design, with each row (A, B, and C) containing one replicate of each of the substrate conditions. The designed substrate depth in each mesocosm was 45 cm, corresponding to a volume of approximately 0.91 m3 of substrate in each mesocosm. For the substrate blends, additions of cellulose (straw or Phragmites compost) and BionSoil biosolids to each of the mineral soils were calculated on a volume basis to increase the organic matter content in the soils to approximately 50 g kg–1 and 100 g kg–1 above its original content. The substrate in each mesocosm was blended individually using a custom-made 220 V pug-mill before addition to each mesocosm. Organic matter percentage was controlled by the addition of straw and Phragmites compost to the Cape Fear loam and dredged material, respectively. Addition of the BionSoil to the blended substrate was constant, and comprised only 5% of the total volume of the blend.
|
Wetland Planting and Establishment
In early May 2000, the mesocosms, except for the control treatment, were planted with a monoculture of soft-stemmed bulrush (Schoenoplectus tabernaemontani, a.k.a. Scirpus validus) obtained from a greenhouse (Naturescapes, Suffolk, VA). Each bulrush plug contained 3 to 6 shoots that were 30 to 50 cm in height. The bulrush was planted on 15-cm centers, resulting in 72 plants per mesocosm. During the remainder of 2000, and into the spring of 2001, the plants were allowed to establish, using well water to maintain flooded conditions.
Batch Studies
Eight wetland batch studies were conducted from May 2001 through July 2002. The mixing protocol in 2001 resulted in inconsistent NO3––N concentrations in the wetlands at the beginning of the four batch studies during that year. Beginning NO3––N concentrations in the wetlands deviated by as much as 20 mg L–1 from the initial target concentrations of 25 to 50 mg L–1 in several of these early studies. The mixing protocol in 2002 was improved substantially by premixing outside the wetland mesocosms before loading rather than in the mesocosms themselves, resulting in deviations of only about 1 to 3 mg L–1 from the initial target NO3––N concentrations of 30 to 120 mg L–1. The NO3––N concentrations in both years were normalized to percentage NO3––N removal relative to initial levels to account for this variability about the targeted levels before they were analyzed statistically. Once the data were normalized, similar trends were observed in the NO3––N removal data from batch studies in 2001 and 2002, but because the initial NO3––N concentrations were much more consistent in 2002, only the methods and NO3––N removal data from 2002 will be discussed in this article.
Wetland batch studies were conducted in February (Batch 1), May (Batch 2), June (Batch 3), and July 2002 (Batch 4), each continuing from 15 to 22 d. Drainage water stored for the studies contained NO3––N concentrations <2 mg L–1, so technical-grade calcium nitrate decahydrate (Ca (NO3)2 x 10 H2O) was used to amend the feed water to various NO3––N concentrations for each batch. Initial NO3––N concentrations in Batch 1 and Batch 2 were 30 mg L–1, 60 mg L–1 in Batch 3, and 120 mg L–1 in Batch 4. Before loading, each mesocosm was drained to a depth of 1 to 2 cm. Drainage water and calcium nitrate were added to a 470-L tank and mixed with a 1/6 hp submersible pump for 5 min before loading of each wetland. Therefore, equal volumes of drainage water and calcium nitrate were added to each wetland mesocosm during each batch study, and produced very consistent initial NO3––N concentrations in each of these batches. Because of the variations on the substrate surfaces as well as variances in plant density, the depth of water in each wetland mesocosm varied from 25 to 35 cm. After loading of all of the mesocosms, well water was added to the ET tanks and the batch study was begun.
During each batch study, each wetland mesocosm was sampled for water quality analysis from the water recirculation system on Day 0, 1, 3, 5, 7, 10, 15, and 22 (if needed). Samples were collected in 500-mL plastic bottles, acidified with H2SO4 to pH 2 to 3, and then frozen until analyzed. Water was analyzed by the Soil Science Analysis Laboratory at North Carolina State University using a LACHAT Quickchem 8000 instrument for NO3––N, NH4+–N, and PO4––P using standard USEPA methods described in Burchell (2003). Precipitation was measured using a manual rain gauge at the site so that the nutrient data could be adjusted to account for dilution from rainfall.
During each batch study, water temperature was monitored continuously on an hourly basis, using HOBO temperature loggers in watertight enclosures. Temperature was monitored in a control mesocosm from one of the rows, and in a treatment mesocosm from each of the other two rows, for a total of three measurements during each batch. Measurements were made at the substrate–water interface, and were downloaded at the completion of each batch study.
On water quality sampling days, Eh, water temperature, salinity, DO, and pH measurements were taken from each wetland mesocosm. Platinum redox (Eh) electrodes were constructed using methods described by Kunickus (2000). Two Eh probes were installed in the substrate of each mesocosm; one at a depth of 2 cm and one at 15 cm. The Eh was measured with an Accumet pH/mV meter (model AP63; Fisher Scientific) connected to an Ag/AgCl reference electrode. Water temperature, salinity, conductivity, DO, and pH were measured using a YSI–600 multiparameter water quality probe (Model 600R-25-C-T-pH-DO, Yellow Springs International) connected to a YSI–610D hand-held microcomputer (Model 610D, Yellow Springs International), at 2.5 cm above the substrate–water interface in each wetland mesocosm.
Wetland Biomass Sampling
At the conclusion of the 2000, 2001, and 2002 growing season, a sample of the emergent vegetation was taken from each mesocosm and analyzed for nutrient content (TKN, TP, %C, %N) on a dry weight basis, with methods described above for substrate analysis. After senescence of the emergent vegetation, the entire stand of bulrush was harvested from each mesocosm at 5 cm above the substrate surface. This biomass was transported back to the lab where it was dried and weighed to estimate the biomass production of each treatment. Nutrient uptake by the wetland plants could then be estimated. After analysis of the biomass was complete, 30% of the dry weight of the biomass from each wetland mesocosm was transported back to the study site and added back to its respective mesocosm. This was done to simulate natural litter accumulation. Only 30% was added back to prevent the entire mesocosm from being filled with litter during the course of the study, limiting the amount of drainage water that could be added.
Data Analysis
Statistical analysis on the NO3––N removal of each of the wetland treatments included multiple regression analysis of the treatment curves that had been normalized to percentage NO3––N removal, to account for the slight variation in initial concentrations within each replicate, to determine treatment effects. This methodology is described in more detail in the Results and Discussion section under Nitrate Removal Studies. For biomass production, a randomized complete block design was used. After detecting treatment and year effects in the analysis of variance of the biomass in both site soil and dredged material-based wetlands, pairwise comparisons were performed using Fisher's multiple pairwise comparison method (Rao, 1998). Both of these analyses were conducted using SAS statistical software (SAS Institute, 2005).
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONOBJECTIVESMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
The dredged material treatment (symbolized as DM) represented wetlands that could be constructed near a CDF such as Eagle Island in Wilmington, NC, with the dredged material alone as the substrate. It contained about 120 g kg–1 (12% dry wt.) OM, and had a particle size distribution of 15% sand, 38% silt, and 47% clay. The dredged material was blended with biosolids and low and high additions of Phragmites plant organic matter to produce wetland substrate blends with organic matter levels of 180 g kg–1 (18% dry wt.) (symbolized as DMB18%) and 220 g kg–1 (22% dry wt.) (symbolized as DMB22%).
As shown in Table 1, the dredged material contained more organic matter, total nitrogen, and total phosphorus than the Cape Fear loam, but because it was screened after excavation, its bulk density as added to the wetland mesocosms was less. Additions of the straw and biosolids to the Cape Fear loam increased the OM, total nitrogen, and the total phosphorus in these wetland substrates. The additions of the straw significantly decreased the bulk density of the substrates. The additions of the Phragmites compost and biosolids to the dredged material also increased the OM, total nitrogen, and the total phosphorus in these wetland substrates. Little change in the bulk densities of the dredged material blends was observed because the Phragmites compost added was similar to the original bulk density of the dredged material.
Nitrate Removal Studies
Batch studies were scheduled to measure NO3––N removal in the various seasons to incorporate various water temperatures and wetland plant growth stages. In analyzing the experimental data, NO3––N treatment curves from the site soil-based wetlands were compared only to the control and other site soil-based wetlands. Similarly, NO3––N treatment curves from the dredged material-based wetlands were compared only to the control and other dredged material-based wetlands. A multiple regression model, which took the percent NO3––N removal to be quadratic in time (Engler and Patrick, 1974) with normally distributed errors, provided a reasonable fit. Percent removal was utilized in analyzing the data to normalize the data due to slight variations in initial NO3––N concentrations in the batch studies.
If the mean % NO3––N removal for observation i at time ti is denoted by mnri. Then the model is denoted by
 | [2] |
 | [3] |
Table 2 provides a summary of the initial conditions of all 4 batch studies discussed in the next sections.
|
Figure 3a shows the NO3––N treatment curves for the site soil wetlands (SSWs) based on the average concentrations within each treatment. Treatment in this batch was very slow. It took over 3 wk for NO3––N to be reduced below 20 mg L–1. Statistical analysis revealed that all of the SSWs performed better than the control in removing NO3––N, but there were no differences between them.
|
Treatment efficiency of all of the wetlands was severely hindered by the cooler temperatures (average of 7.5°C) and the lack of actively growing biomass.
Batch 2
Batch 2 was conducted from 7–22 May 2002, with a target initial NO3––N concentration of 30 mg L–1. Approximately 50% of the bulrush above-ground biomass for the growing season was present at the onset of the batch. Water temperature averaged 17.8°C in the wetland treatments during Batch 2.
Figure 4a shows the NO3––N concentration data for the SSWs. Concentrations of NO3––N were reduced to 5 mg L–1 in 10, 7, and just over 8 d in the SS, SSB11%, and SSB16%, respectively. Analysis of percentage NO3––N reduction data revealed significant increases in NO3––N treatment in all of the SSWs when compared to the control. The site soil blends (SSB11% and SSB16%) resulted in significantly more rapid NO3––N loss than the SS treatment. However, NO3––N treatment in the SSB11% and SSB16% wetlands were not significantly different.
|
Despite early season cooler water temperatures (approximately 18°C), and less than 100% of the seasonal above-ground bulrush stand established, the wetlands performed much better in NO3––N treatment when compared to the Batch 1 winter study.
Batch 3
Batch 3 was a 15-d summer batch study with an initial target NO3––N concentration of 60 mg L–1. The bulrush stand was almost fully established at the time of this study. Average water temperature in the wetland mesocosms was 23.2°C. Large reductions in this elevated level of NO3––N was observed in all of the wetland treatments.
Figure 5a shows the SSWs NO3––N treatment curves. The NO3––N levels were reduced below 10 mg L–1 in slightly over 13 d, in slightly under 9 d, and in slightly under 11 d in the SS, SSB11%, and SSB16% wetland treatments, respectively. Analysis revealed that the SSB11% wetlands outperformed all of the other SSWs and the control. The NO3––N removed in the SSB16% wetland treatment exceeded that in the SS treatment and the control, whereas the SS wetland treatment exceeded the NO3––N removal in the control.
|
Batch 4
This final batch study was conducted from 2–24 July 2002. The initial target NO3––N concentration was once again doubled compared to the previous batch to 120 mg L–1. The average water temperature in this batch was the warmest observed in the entire study at 24.3°C.
Figure 6a gives the NO3––N removal curves for the SSWs. Treatment exceeded that of the control in all of the SSWs during this batch. The NO3––N concentrations were reduced to 10 mg L–1 in 19.5 d, 15 d, and in slightly over 17 d in the SS, SSB11%, and SSB16% wetland treatments, respectively. Analysis of the data indicated significant increases in NO3––N reduction in the SSWs over the control. The SSB11% wetlands outperformed the SS wetlands in NO3––N removal, but no difference in treatment efficiency was found when compared to the SSB16% wetlands. The SSB16% wetlands showed treatment that was slightly higher than the SS wetlands (P = 0.0341).
|
Discussion of Nitrate-N Removal Studies
Days Required for Ninety Percent Nitrate-N Removal
A simple way to visualize and assess the magnitude of the effect of OM addition to substrate on wetland NO3––N removal is to determine the average number of days required for each wetland treatment to remove 90% of the initial NO3––N concentration. This is also helpful in reinforcing the idea that fewer days required for treatment may imply a reduction in the wetland area required. Figures 7a and 7b show the average days (rounded up to the next 0.5 d) required to remove 90% of the NO3––N in Batches 1 through 4 for the SSWs and the DMWs, respectively. In three of the four batches, the SSB11% wetlands achieved 90% removal in the fewest days of all the SSWs. By averaging the days required for 90% removal in all batches (except in Batch 1 where 90% removal was not reached), the SS, SSB11%, and SSB16% wetlands averaged 15.7, 11, and 13 d, respectively. In the DMWs, 90% NO3––N removal was achieved in the fewest days by the DMB22% treatment in all of the batches (neglecting Batch 1). There was little difference in the overall average number of days required to achieve 90% reduction in the DM and the DMB18% wetlands (about 14 d), but the DMB22% averaged only 8.7 d.
|
Improved treatment was not observed the in the SSB16% wetlands when compared to the SSB11% wetlands (P values > 0.05). In fact, in Batch 3, NO3––N removal rates in SSB11% were greater than SSB16% (P = 0.0078), so there appears to be no advantage in increasing OM contents in this wetland substrate to 160 g kg–1 (16% dry wt.). On the other hand, when an addition of OM to the DMWs reached 220 g kg–1 (22% dry wt.), an advantage in treatment was observed. The DMB22% wetlands outperformed the DM and DMB18% wetlands in Batches 2 through 4 (P values < 0.0001), and DMB18% in the winter Batch 1 (P = 0.0181).
Denitrification within the Treatment Wetlands
The emphasis of the discussion so far was that NO3––N removal rates were enhanced by the addition of OM, providing a carbon source so that denitrification rates were not limited. However, denitrification was not directly measured, so its role in NO3––N removal within the wetlands has been assumed based on previous research and the conditions measured within the wetlands. Measurements taken within the wetlands during each of the sampling studies indicated that conditions were most favorable for denitrification to occur in Batches 2 through 4. Temperature, dissolved oxygen, redox, and pH were all measured near the substrate–water interface where denitrification reactions are most likely to occur.
Denitrification rates increase with temperature up to a maximum, decreasing rapidly thereafter. The most rapid NO3––N removal occurred in the spring and the summer, when warmer air temperatures increased wetland water temperatures. Low temperatures decrease denitrification rapidly, but some activity has been measured at 0 to 5°C (Knowles, 1982). The effects of cold water temperatures were obvious in Batch 1, where the greatest NO3––N removal was only about 50% in 3 wk. The colder water temperatures slowed activity of the denitrifying microbes within the substrate. In addition, colder water has a greater capacity to hold DO. This effect, coupled with reductions in aerobic microbial activity that deplete oxygen, resulted in higher DO concentrations which likely hindered denitrification during the winter.
Low dissolved oxygen is critical for denitrification to occur near the substrate–water interface. Knowles (1982) cites studies that indicate that DO in the range of 0.03 to 0.45 mg L–1 O2 represses biological denitrification, and levels below 1 mg L–1 are used in wastewater engineering design of denitrification basins (Metcalf and Eddy, 1991). The control treatment in this study was consistently supersaturated with DO down to the substrate–water interface, a result of reaeration and algal growth because of the lack of plants. This severely hindered denitrification, as evidenced by the slow NO3––N removal observed in this treatment. During warm conditions, the SSWs and the DMWs consistently averaged 0.9 to 2 mg L–1 DO at 2.5 cm above the substrate–water interface (proximity to the substrate was limited by the DO probe), whereas DO during the winter months (Batch 1) averaged 3 to 6 mg L–1. The values in the summer months were much closer to those critical DO levels, providing further support for denitrification as the primary mechanism for NO3––N loss.
Nitrate reduction generally begins at Eh values below 350 mV, with complete nitrate disappearance below 220 mV (Ponnamperuma, 1972; Engler and Patrick, 1974; Kadlec and Knight, 1995). Substrate redox measurements were taken in the substrate at 2-cm and 15-cm depth. Average Eh values at both depths, during all of the studies in all of the treatments, were below the critical values of 350 mV. The average Eh at 2 cm was –19 mV, with a range of –133 to 141 mV. The average Eh at 15 cm was –102 mV, with a range of –203 to 18 mV. The Eh values at 2 cm were slightly higher due to a higher potential for oxygen at that depth, as well as an increased density of bulrush roots in the wetland treatments that maintain some aerobic microsites near roots in that zone. Addition of OM to the wetland substrate did not appear to strongly influence the reduction of Eh values below the critical level for denitrification. However, wetlands with increased OM had a larger pool of carbon critical for maximizing denitrification rates at these Eh values.
Denitrification is positively correlated with pH, with optimum pH in the range of 7 to 8 (Knowles, 1982; Kadlec and Knight, 1995). The pH in all of the wetland treatments averaged between 6.0 and 6.5, just below the optimum level for denitrification. The process of denitrification tends to increase pH in treatment systems because of the formation of OH– (Metcalf and Eddy, 1991). Formation of OH– results in increased alkalinity in the water column, but this was not measured. Increases in pH were observed in all of the wetland treatments during the batch studies. Increases were generally on the order of 0.5 units through each batch study. These increases were not helpful in evaluating the magnitude of denitrification that was occurring within the wetland treatments, but this indicates that some denitrification products were being produced.
Denitrification or Plant Uptake?
Plant uptake is generally regarded as a minor factor in NO3––N removal in constructed wetlands, but can account for as much as 16 to 75% of total N uptake in a wetland (Reddy and DeBusk, 1987). To determine the contribution of plant uptake in this study, an estimate of plant uptake was conducted for 2002. With an estimate of plant uptake known, then the remaining NO3––N loss may be attributed in large part to denitrification.
Percentage N was measured in the above-ground shoots of the soft-stemmed bulrush in each mesocosm before the end of the growing season. When the above-ground biomass was collected at the end of the growing season and weighed, the mass of N in the above-ground biomass was calculated. The below-ground biomass was not collected, because rhizomes formed in 2002 could not be distinguished from those established in previous years. Above-ground biomass for all of the treatments in 2002 averaged 2400 g m–2, and there was no significant difference between treatments. Likewise, N content within the above-ground biomass was not different between treatments, ranging from 15.2 to 16.5 g kg–1 (1.52 to 1.65% dry wt.) N. Calculations for plant uptake assumed a constant N uptake rate for each replicate depending on the mass of N accumulated over a 180-d growing season, and averaged 0.42 g N d–1. The details of the calculations can be found in Burchell (2003).
On average across all replicates, a maximum of 22 to 35% of the NO3––N removal observed in Batches 2 through 4 was estimated to be due to plant removal. The estimate of plant uptake of 22% neglects uptake by the rhizomes, which were not measured. Assuming a root/shoot ratio for soft-stemmed bulrush of 0.6, based on values presented by Busnardo et al. (1992) of 0.46 and DeBusk et al. (1995) of 0.62 to 0.66, this allowed an estimate of average total biomass (rhizomes +shoots) to be calculated as 3840 g m–2. Including rhizomes, the estimate of maximum biomass uptake during the batch studies was 35%. In either case, 65 to 78% of the NO3––N remained unaccounted for; an amount that was most likely lost through denitrification. This estimate, along with the physical measurements of the conditions within the wetlands discussed above, further supports the hypothesis that the majority of the losses of NO3––N in these wetlands was due to denitrification—the goal in the design of these wetlands. This estimate may actually be conservative, because substantial N removal from the wetland substrate was measured during the course of the study (Burchell, 2003). Wetland plants usually utilize ammonia as an N source, because NO3––N is not available for uptake in highly reduced soils. Additionally, in nitrate-rich waters, additions of carbon, as were introduced to these wetlands, can shift the balance from plant uptake to denitrification (Kadlec and Knight, 1995).
Wetland Biomass
Providing OM and a nutrient source to the substrate to support increased denitrification rates early in the life of a wetland may also be beneficial in enhancing wetland plant growth, the ultimate renewable carbon source. Besides providing carbon when the plants decay, increased biomass also may help nitrate diffuse into the subsurface by increasing permeability of the upper reaches of the wetland substrate, and by translocating water into the substrate during transpiration. Increased biomass can also translate into increased microbial attachment sites below the substrate surface and in the decaying litter produced through the years, which can increase microbial activity. Increased amounts of biomass within a treatment wetland will also inhibit algae growth through shading. This will decrease reaeration possibly resulting in an increase of anaerobic processes. By providing nutrients from OM and biosolids to the wetland substrate before planting, wetland plant growth may be accelerated to reach the production more commonly observed in older wetlands. If this is accomplished, the wetland will function more efficiently in removing NO3––N at an earlier age.
Figures 8a and 8b show the yearly mass of above-ground soft-stemmed bulrush in the wetland treatments in 2000, 2001, and 2002. Average growth for bulrush is about 2000 g m–2 in full-scale wetlands systems (Reed et al., 1995). This level was not reached in 2000, the year used to establish the bulrush, in any of the wetland treatments. However, the above-ground biomass produced in all of the wetlands was significantly greater in 2001 and 2002, and all of the wetland treatments exceeded 2000 g m–2 on average. The most striking observation is the increase in biomass from 2000 through 2001. In 2001, all of the SSWs at least doubled the biomass produced in 2000, while each of the DMWs nearly doubled their biomass. There was no significant difference in biomass growth in 2002 when compared to 2001.
|
Only the DMB18% wetlands had significantly more biomass than the DM wetlands in 2000 (P = 0.0064) and 2002 (P = 0.0317), but both DMB18% and DMB22% wetlands had more biomass than the DM wetlands in 2001 (P = 0.0049 and 0.0021, respectively). It should be noted that nutrient levels within the DM treatment were greater than in the SS treatment at the beginning of the study (see Table 1), so nutrient enhancement may have had a greater effect on biomass production in the SSWs than in the DMWs.
Engineering a wetland substrate, in relation to wetland plants, could prove to be beneficial with respect to economics and downstream water quality. The high nutrient content of the manufactured substrates reduces the amount of fertilizers sometimes used to maintain growth of wetland vegetation. This could help protect downstream water quality because fertilizing vegetation in flooded systems is difficult, and improper application can lead to larger export of nutrients from a constructed wetland (Allen et al., 1989). Looser wetland soils, like those we created by adding cellulose, encourage root and rhizome penetration (Allen et al., 1989), and may encourage faster and denser wetland plant colonization. This could be cost-effective because fewer plants may be required in the initial planting of a constructed wetland system. Care must be taken, however, in the amount of organic material added, because soils with high organic matter are not well suited to physically support macrophytes (Allen et al., 1989). The lower biomass production of the SSB16% wetland when compared to the SSB11% wetlands, even though the SSB16% blend was higher in nutrients, could be related to the low bulk density of the blend [0.34 g cm–3 (Table 1)]. The plants in the SSB16% had difficulty remaining upright in the loose soils, which most likely delayed their establishment in 2000. This could be a serious engineering drawback when considering manufacturing substrate for full-scale wetland systems, so too much cellulose should not be added.
Implications of Research
Wetland Sizing
By improving wetland treatment with OM addition, the time required for treating NO3––N to a particular level is reduced. Therefore, less area would be required for a wetland that performs more efficiently. Reduced wetland area lowers the cost of property acquisition and construction. This section attempts to compare the wetland area required to treat drainage water from a watershed based on the results observed for each of the wetland treatments evaluated in this study.
Most researchers regard NO3––N reduction in wetlands as a first-order process (Reed et al., 1995; Kadlec and Knight, 1995) expressed mathematically as
 | [4] |
 | [5] |
Batch 1 and Batch 2 were chosen to calculate kNO3 for NO3––N reduction in each of the wetland treatments in cold and warm temperatures, respectively. The initial NO3––N concentration was about 30 mg L–1in both of these batches. The value of kNO3 was estimated using the following equation, which represents the slope of ln(Ce/Co) vs. t.
 | [6] |
To be first-order, the ln(Ce/Co) vs. t relationship should be linear. A linear regression was fitted to the NO3––N removal data collected for all of the wetland treatments in Batches 1 and 2. The R2 values exceeded 0.92.Values for kNO3 ranged from 0.19 to 0.55 d–1 in Batch 2 (spring) and from 0.012 to 0.02 d–1 in Batch 1 (winter) (Table 3). More details of these calculations can be found in Burchell (2003).
|
Table 3 shows the areas required for each of the wetland treatments studied to reduce NO3––N from 5 to 3 mg L–1, based on the kNO3 values calculated. Water temperature averaged about 18°C in Batch 2 (spring). The wetland area required for treatment of the water originating from the 50-ha watershed in the spring ranged from 0.7 to 2.1 ha. Wetland area required to achieve the treatment goal was reduced by OM addition in both the SSWs and the DMWs. The area required for a wetland constructed like SSB11% was 48% smaller than a wetland constructed with site soil alone. Similarly, a wetland constructed with DMB22% would require 56% less area than one constructed with DM alone as the substrate. As stated earlier, Batch 1 was used to determine kNO3 during the winter. The average water temperature for this batch was 7.5°C. There was a huge increase in wetland area required for treatment at these temperatures, and there appeared to be no advantage gained in wetland area reduction by OM addition to the substrate during the winter.
Dredged Material in Treatment Wetlands
The use of dredged material to construct wetlands is not a new idea, and has been implemented on many sites. The use of dredged material in treatment wetlands is an idea that is just now receiving attention. Results of this research indicate that wetlands constructed with dredged material, with and without OM addition, are very efficient in the treatment of NO3––N. The U.S. Army Corps of Engineers spend millions of dollars each year maintaining and building CDFs to store material dredged from waterways in the U.S. Alternative uses for this material could save a substantial amount of money. Dredged material has been used as an ingredient in topsoil, landfill covers, and in construction materials to name a few, but treatment wetlands appear in this research to be feasible as another alternative. Locations near dredging operations would be ideal for the use of dredged material as a substrate in constructed wetlands. Incorporation of Phragmites as an OM source not only would help the wetlands improve treatment efficiency, but removing Phragmites from the CDF opens space for more storage of dredged material. In addition to agricultural drainage, wetlands of this sort could be used for tertiary treatment of municipal wastewater, as well as in the treatment of stormwater, wastewater from small, isolated developments, landfill leachates, and golf course runoff.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONOBJECTIVESMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Above-ground biomass, the wetlands' ultimate renewable C source critical in maximizing denitrification rates, was evaluated for enhancement by substrate OM and biosolids addition. After establishment in 2000, growth of soft-stemmed bulrush in all of the mesocosms exceeded 2000 g m–2 in 2001 and 2002. Increased OM addition and biosolids to the site soil and dredged material blends significantly increased biomass growth in 2001 when compared to no OM addition. This supports the hypothesis that enhancement of nutrient-poor soils used for wetland substrate may reduce the time required for wetlands to reach production levels common in more mature wetlands. Since the DM soils were initially richer in nutrients, the improvement in wetland soil quality was probably more important to the site soil blends.
Increased OM content of a treatment wetland substrate may help jump-start nitrate removal efficiencies in the early years of establishment, providing C and reduced conditions critical for denitrification. Denitrification was most likely the major contributor of NO3––N removal in these batch studies, based on conditions measured in the wetland substrate and the substrate–water interface, and on an estimate of plant N uptake. Improving NO3––N removal efficiency reduced the time required for treatment, which when applied to full-scale systems, may reduce area requirements for wetlands, making their use more economically feasible. Treatment wetlands with a combination of substrate organic matter and increased biomass production, which will serve as a renewable C source to the wetland, may help maintain ideal conditions for denitrification and subsequently high nitrate removal during the life of these systems.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONOBJECTIVESMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
