| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
JORDFORSK (Norwegian Centre for Soil and Environmental Research), Frederik A. Dahls vei 20, N-1432 Aas, Norway
* Corresponding author (bent.braskerud{at}jordforsk.no)
Received for publication May 23, 2000.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIAL AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: CW, constructed wetland
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
As a result of the undesired effects, financial support for land leveling and the closing of streams was discontinued in 1989. At the same time, sedimentation ponds were suggested as an alternative to mitigate soil particle transport from the watersheds. However, available literature at that time favored large ponds, in order to create low hydraulic loads for significant retention of clay particles (e.g., Chen, 1975; Novotny and Chesters, 1981). Retention of clay particles is of major concern, due to their influence on water turbidity and their ability to adsorb phosphorus (Sharpley, 1980), heavy metals (Kabata-Pendias and Pendias, 1984), and pesticides (Leonard, 1990). On the other hand, ponds requiring a large land area were considered impossible to implement in the small-scale Norwegian agriculture. The literature suggested that the ponds should be shallow, in order to utilize the positive effect of a shorter vertical settling distance (Chen, 1975). As a result, four small, shallow sedimentation ponds were built in first- and second-order streams in 1990. However, due to concern for resuspension or new erosion of sediments under storm runoff situations, we planted emergent aquatic vegetation. Hence, the term constructed wetland (CW) seemed more accurate.
After six years of investigations, results showed that the retention of clay particles in the CWs exceeded model estimates by a factor of 2.5 to 8.2 (Braskerud et al., 2000). The clay content in the sediment reflected the clay content in the arable fields (Braskerud, 1998). A plausible explanation was that clay was entering the CWs as soil aggregates. However, vegetation had probably affected the retention of soil particles, too.
Several investigations credit vegetation for increased sedimentation. The main reason is a combination of reduced turbulence (Barko et al., 1991; Hosokawa and Furukawa, 1992) and reduced water velocity (Petticrew and Kalff, 1992; Sand-Jensen, 1998; Svendsen, 1992). According to Kadlec and Knight (1996), the mechanisms of macrophyte filtration are (i) particles flow into plant stems and leaves; (ii) particles stick to the biofilm on macrophytes; and (iii) random processes (such as Brownian motion and bioturbation) move particles to immersed surfaces. Kadlec and Knight (1996) claim that the filtration effect of macrophytes is negligible for water velocities commonly associated with CWs. However, vegetation may shelter trapped sediments from (iv) resuspension. Moreover, it is possible that aggregates are formed through (v) flocculation in CWs via (a) small-scale turbulent water, (b) differences in particle sedimentation rates, and (c) organisms producing sticky organic matter, which stimulates flocculation (Eisma et al., 1991). Vegetation is involved in a and c. Vegetation and litter are covered with biofilm, which produces organic matter, and eddies are created downstream from stems and leaves when water passes emergent vegetation. In this way, plant stems may increase particle collisions and stimulate floc formation.
Negative effects of macrophytes are also reported. Decreased sedimentation in wetlands was observed when hydraulic loads were high (up to 0.07 m d-1; Brueske and Barret, 1994) and as plant stem density increased. Stem density influences the water flow resistance, because water flow prefers sparsely vegetated areas. As a result, vegetation is short circuiting the water flow through the CWs, and the detention time is reduced (Fennessy et al., 1994).
This article focuses on the effect of vegetation on the soil particle retention efficiency. Does vegetation create flow short-circuiting or preferential flow, decrease sedimentation of soil particles, hinder resuspension of sediment in CWs, or stimulate retention of clay particles? This article aims at providing data in order to optimize the construction and use of wetlands under high hydraulic loads (e.g., when CW surface area is small compared with the watershed area).
|
MATERIAL AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
|
Constructed Wetlands
Four CWs were built in the streams of Watersheds A, B, C, and D (Fig. 1). Each was 0.4 to 0.5 m deep, except at the inlet, where the sedimentation basins were 1 m deep at low summer flow. The CWs consisted of a delta near the inlet, a sedimentation basin, and a wetland filter (Fig. 2). All CWs were rectangular, with the exception of CW A, which had a slight bend of approximately 25° to the right between Sedimentation Traps 3 and 4 (Fig. 2).
|
Vegetation cover was estimated annually in mid-September by summing the area covered by each species as if it were in monoculture. This estimation was always done by the author, who divided the CWs into sections limited by the traps (Fig. 2). Estimates are expressed as percent of the total surface area. Submerged spices, like Callitriche hamulata Kütz, were hard to estimate correctly and were rounded to the nearest ten percent.
Vegetation Year
The growing season in this area is usually May through September. However, emergent aquatic vegetation cannot been seen in the CWs before mid-June. Hence, the influence of the vegetation of a specific year lasts from mid-June one year to mid-June the next. As a result, summer is divided into two periods; mid-June through October the first year and May through mid-June the following year. Winter is considered to cover the period from November through April.
Water and Sediment Flux Monitoring
Sampling Systems
Runoff was measured by a V-notch weir, located at the outlet of every CW (Fig. 2). The overflow was connected to a data logger (Campbell Scientific [Logan, UT] CR10), which also controlled the flow-proportional sampling system in the inlet and outlet. On average, 11 subsamples were collected daily and stored in a single container. A subsample was usually 120 mL. A 1-L sample was taken from the composite sample container, usually at 9- to 12-d intervals. Flow-proportional composite sampling started in autumn 1991 for CWs A and B, and in summer 1992 for CWs C and D. For all CWs, the latest samples included in the results were taken in summer 1996. However, several events were lost during the first year due to problems with mice and frost. Constructed Wetlands B and D were only monitored in the frost-free part of the year, due to poor frost protection. No preservative was added to the composite sample container.
Sedimentation traps (0.65-L cylinders) were placed throughout the entire length of each CW (Fig. 2). Usually there were three traps in the sedimentation basin and three traps in the wetland filter. Constructed Wetland A had four traps in the wetland filter, in addition to another six in transects perpendicular to Sedimentation Traps 4, 5, and 6. Traps were constructed of two plastic cups (Fig. 3). An external holder was placed in the wetland sediment, and a sampler was located in the holder. The sampler could be retrieved from a rubber boat with a line attached to it. Sampling was carried out three or four times a year (April–May, mid-June, September, and November).
|
Traps and plates complement each other: traps give the potential (gross) retention and plates the actual (net) retention. Sampling of the sediment traps and plates started in summer 1991 for all CWs, and includes the summer 1996. Samples with malfunctions in either a trap or plate are omitted from the data set (5 and 3, respectively, out of 65 observations; four of these were from the 1993–1994 season). Malfunctions could be caused by ice cover lifting traps out of position or a macrophyte leaf covering part of a trap opening. Imperfectly replaced plates resulted in an uneven sediment distribution. Sediment texture on plates was analyzed according to the pipette method (Elonen, 1971).
Sedimentation in the delta area was measured in the summer of 1996 by measuring sediment depth in transects perpendicular to water flow. Experimental sites and procedures are detailed in Braskerud et al. (2000).
Hydraulic Load
Hydraulic load (q) was calculated by dividing average runoff (Q) in a sampling period by the wetland surface area (A). Because the CW side slopes were 1:1.5–2.0, A changed with Q. However, A was adjusted as Q changed to estimate a correct q.
Laboratory Analyses
For measuring suspended solids, 500 to 800 mL from each composite sample was filtered through a dried and preweighed Whatman (Maidstone, UK) GF/C glass fiber filter. The filter was dried at 105°C for 24 h. The amount of suspended solids deposited on the filter was expressed as mg L-1 water.
For measuring the amount of sediments in sedimentation traps, 50 mL of the shaken sediment suspension was poured into a ceramic crucible, dried at 105°C for 24 h, and weighed. Depending on sedimentation rates, one-eighth to three-quarters of plate sediments were transferred to a 1-L bottle in the field. It was easy to divide the sediment into smaller fractions, and an evaluation of the sampling method showed that plates estimated actual sedimentation correctly (Braskerud et al., 2000). The volume of the bottle was registered and the sediment dried at 105°C for 6 d, then weighed.
For measuring grain size and texture, the sediment was treated with H2O2 and HCl to remove organic matter and oxides, respectively. A sample of 10 g (dry wt.) was analyzed according to the pipette method (Elonen, 1971), using sodium-pyrophosphate as a dispersion agent.
Sedimentation Rates
Retention (percent) was found by subtracting the particle flux at the outlet from the particle flux to the inlet. Errors in the water budget were minimized in the experimental wetlands. Input of external water to the CWs through rainfall, ground water, and surface runoff was prevented by (i) small CW surface area, (ii) drainage pipes in the surrounding fields, (iii) location of CWs in clay soils with low hydraulic conductivity, and (iv) terrain barriers. The effect of evapotranspiration on particle retention was underestimated by a maximum of 2% in the period May through August. The concentration of suspended solids in each composite sample was multiplied with the water volume during the sampling event.
Retention (by weight) was estimated by summing up sedimentation in traps and plates for the sedimentation basin and the wetland filter, respectively. Since sedimentation rates from the delta area were measured only once, a division between the five years of investigation was needed. The sum of five annual sedimentation rates in the sedimentation basin was set to 100. Hence, the individual years were fractions of the total sedimentation. Each year was assigned the corresponding fraction of the total sedimentation rate in the delta, and the total annual sedimentation in the individual CWs was possible to estimate.
Statistical Analysis
Annual variations in runoff and erosion intensities may influence several of the hypotheses under investigation, especially when the tested object changes over time (e.g., vegetation cover). Use of control CWs without macrophytes makes it possible to mitigate the effect of year. The main reasons for omitting controls were too little space available for the experimental sites, and possible problems of keeping the control areas free of vegetation without disturbing traps and plates. Vegetation cover growth varied between the sites. Years with approximately the same vegetation cover could serve as replicates for the effect of vegetation.
Statistical analyses were computed using JMP 3.1.5 for Macintosh (SAS Institute, 1994). Statistical hypotheses were tested at the 5% level. Variables were tested with the Shapiro–Wilk W test for normal distribution before statistical analyses. Two variables were not normally distributed, hydraulic load and annual sediment retention. Log transformation normalized the data, but had little influence on the results of t-tests, analysis of variance (ANOVA), or multivariate analyses, and was therefore excluded. A factor analysis was performed to evaluate the correlation structure between hydrologic load, vegetation cover, sediment retention by weight, clay content in CW sediment, and resuspension of sediment. In the statistical analyses of the clay fraction and resuspension, only data from the sediment in the wetland filter was available.
Methodological Errors
Possible Errors in Composite Sampling
Loss of soil particles from a watershed changes continuously. It is obvious that sampling has to be frequent for mass balance studies in small, hydrologically dynamic systems (Johnson, 1992). The conditions for mass balance studies are fulfilled when sampling frequency is high, and the sampling period sufficiently long. As a result, the inlet samples are connected to the outlet sample. Compared with spot or grab samples, flow-proportional composite samples may take several hundred subsamples under storm runoff. Hence, composite sampling is the best way to measure mass balance (Øygarden and Botterweg, 1998). Field investigations have two drawbacks: automatic sampling systems have to be used to cover all events and sampling depth is fixed. Ideally, samples should be taken in strongly turbulent water (Bogen, 1992). In small streams, however, water depth can change dramatically within a short time. The inlet samplers were placed in the inlet part of the CWs to stay submerged under low flow periods (Fig. 2). This reduced the turbulent mixing. However, sediment input increased under storm runoff situations (Braskerud et al., 2000), when turbulence was largest. The output samplers were situated up to 7 m from the V-notch weir. In 1993, a comparison with sampling a few centimeters from the V-notch in CW A showed that the concentration of suspended solids was underestimated by as much as 25% (Braskerud, 1995). Due to the findings, the composite sampler was moved closer to the outlet. No adjustments of the previous data were attempted, because we know too little about the error. However, the error probably has little influence on the discussion of the effect of vegetation, because the same method was used every year.
Possible Errors in Traps and Plates
Traps should, by definition, contain more sediments than plates. Two errors are associated with the use of traps. First, according to Håkanson and Jansson (1983), traps estimate sedimentation in lakes correctly if they are cylindrical in shape, the opening diameter is larger than 4 cm, and the height to depth ratio is larger than 3. Sediment traps used in this study fulfilled the first two requirements. The height to depth ratio, however, was only 2.4. Hence, there may be more resuspension. However, plates are potentially more erodable compared with traps. In conclusion, resuspension from traps is unlikely to explain the difference in accumulation rates observed in this study.
The second type of error, disturbances of the water flow around the trap opening by vortex shedding (Tritton, 1988) and other hydrodynamic phenomena related to water turbulence, are more reasonable factors than resuspension from traps (James and Barko, 1993). A similar underestimation of precipitation is well known for mechanical rain recorders.
Could bed load interfere with traps and plates? Traps in this study were excluded from the bed load, because the openings were approximately 4 cm above the CW bottom (Fig. 3). It is not likely that bed load influenced sedimentation plates either, because bed load usually includes particles larger than 0.18 mm (Sundborg, 1956). No sand was found in the wetland filter (Braskerud et al., 2000), with the exception of the extreme episode in CW C (see Results section, below).
According to Braskerud et al. (2000), sedimentation plates give good estimates of the net sedimentation rates. Resuspension from traps should not interfere with the results, because it is most likely that traps underestimate sedimentation rates systematically.
Possible Errors in Hydraulic Load and Sediment Load
The first years after construction are critical for understanding the effect of vegetation on soil particle retention. Hence, it is important to know whether the hydrological load and the sediment load differ in this period.
If runoff is high, we could expect a significant difference between traps and plates due to more resuspension on the plates. The annual hydraulic load for the 5-yr period in this study was 2.0, 1.6, 2.3, 2.5, and 1.0 m d-1, respectively. Hence, the first year was close to the average (1.9 m d-1) for the period.
The annual sediment load was not known for the first years, due to insufficient composite sampling (Fig. 5). However, sedimentation rate in the sedimentation basin could act as an indicator of sediment load. The vegetation cover had a minor influence in this part of the CWs. The maximum sediment load was recorded in Year 5 and was set to the relative number 100. Starting with the first year of sampling, the annual relative retention for the period was 65, 64, 88, 100, and 48. Hence, none of the two first years seem irregular in any of the CWs.
|
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
|
Sediment Flux through the Constructed Wetlands
The average seasonal changes in soil particle flux for CW A are shown in Fig. 5. Sedimentation decreased the soil particle concentration in the CW outlet. The difference between inlet and outlet was usually largest for high hydraulic load situations. The same pattern was observed in the other CWs (data from CW C are presented in Braskerud et al., 2000). The annual hydraulic load ranged from a minimum of 0.5 m d-1 in CW B, to a maximum of 3.7 m d-1 in CW D.
Autumn 1991 was the first season with a complete set of composite samples. However, the retention of soil particles was only 1% in CW A. Later seasons never showed such low retention (Fig. 5). A similar observation was made in CW B, with a retention of 5%. The phenomenon was not observed in CW C or D, because composite sampling started 1 yr later.
Constructed Wetlands A and B were compared in autumn 1991 and autumn 1993. Results from CW A are used for illustration and are shown in Fig. 6. Autumn was quite similar in both years regarding runoff and erosion (126 and 106 mm, and 9300 and 8700 kg soil km-2 watershed, respectively). The main difference between the seasons was vegetation cover (Fig. 4). In autumn 1991 vegetation cover in CW A was 12%, and overall retention was 1%. Two years later, vegetation cover had increased to 52% and overall retention to 66%. In the first autumn, five out of nine composite sampling episodes indicated a net particle loss from the CW; in the latter, all episodes resulted in a net retention. The same development in retention was observed in CW B. In CWs C and D no samples were taken in autumn 1991.
|
|
Soil particle retention from Fig. 7 is summarized in Table 3. Results are divided between summer and winter. Only observations where vegetation cover exceeded 50% of the CW surface area are included, in order to enable the comparison of summer and winter. Retention was neither statistically different between the CWs (P < 0.85), nor between the warm and cold part of the year (P < 0.42).
|
|
|
Hydraulic Efficiency
Hydraulic efficiency influences retention, because hydraulic load changes with the active surface area (e.g., Chen, 1975; Kadlec and Knight, 1996; Persson et al., 1999). Only CW A was regarded wide enough to study the effect of vegetation on the hydraulic efficiency. Constructed Wetland A had traps in the wetland filter placed along three transects perpendicular to water flow (Traps 4–6 in Fig. 2). If the hydraulic efficiency is high, sedimentation rates should be similar in each transect. That is, the trap in the center of the wetland filter would receive as much sediment as the traps on either side. Sedimentation rates during the summer-half-year and winter-half-year, through a 4-yr period (1992–1996), are presented in Table 4. Results are standardized for each transect to give each episode equal weight. In each sampling episode, the individual traps were divided by the average retention in each transect and multiplied by 100. Table 4 shows the result of the average retention in each sedimentation trap, and the standard deviation. On an annual basis, however, less than one-third of the sedimentation accumulates in the summer.
|
Sediment Resuspension
The influence of CW age on sediment resuspension was investigated by comparing sediment accumulation in sedimentation traps with accumulation on plates. It is assumed that the difference in sedimentation rate between traps and plates estimated resuspension (Fig. 10).
The observation from Year 2 is statistically different from three of the other observations, indicating a change in resuspension from the second to the third year. However, a deviation between the sampling systems is seen the last 2 yr, since plates estimated higher sediment retention than traps (Fig. 10). The last 2 yr indicate no resuspension. If no resuspension occurred the last year, traps underestimate sedimentation by approximately 20%. It is possible to adjust the relative resuspension for previous years. Since plates measure net retention, they were used as a relative scale of trap retention. If 20 percent units are added to all data in Fig. 10, the resulting data is shown in Fig. 11. It shows the relative decrease in the difference between the two methods with time. As the CWs aged and vegetation cover increased, resuspension decreased.
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Resuspension decreased approximately 40% in 4 yr, and was negligible in 5-yr-old CWs. However, individual differences between watersheds and CWs influenced the annual resuspension significantly (cf. the confidence intervals in Fig. 10).
As CWs aged and resuspension decreased (Fig. 11), a similar change in retention was observed in to the composite samples (e.g., A and B in Fig. 7a). Hence, these results indicate a limit of the effect of vegetation on sediment retention. When vegetation cover exeeds approximately 50%, a level is probably reached where other factors, such as sediment load and hydraulic load (Braskerud et al., 2000), have a greater influence on the retention performance than vegetation.
Changes in sedimentation rates on the sedimentation plates support the hypothesis of an upper limit of the effect of vegetation (Fig. 9). Constructed Wetland C experienced the first drop in the sedimentation rate after reaching a 95% vegetation cover in Year 3. In the other CWs, however, sediment accumulation continued to increase for one more year (Fig. 9). Declining sedimentation rates in the last year are probably a result of decreased sediment loads.
The hydraulic load in CW D was more than twice the load in CWs A, B, and C (Table 1). High hydraulic load was probably the main reason for the lack of a statistically significant increase in the effect of vegetation cover in this CW (Fig. 7a). This indicates that the effect of macrophytes on sediment retention also has an upper limit regarding hydraulic load; for example, in the extreme episode the retention dropped from an average of 50% (Table 3) to 30%.
In a comparative study of Nordic CWs and ponds as phosphorus traps, the performance in CWs was approximately twice the retention in ponds covering the same surface area to watershed area ratio (41 and 17% phosphorus retention, respectively). Ponds were sparsely covered with vegetation. Macrophytes were probably an important factor for higher retention in CWs (Uusi-Kämppä et al., 2000).
Vegetation in CWs should not be harvested. Vegetation makes it possible to use the positive effect of short particle settling distance in shallow ponds (Chen, 1975), due to hindered resuspension.
Hydraulic Efficiency
Water flows in areas with the least resistance. There will always exist both sparsely and densely vegetated areas in a wetland. At low flow rates, a preferential flow pattern through the wetland filter may be created (Fennesy et al., 1994). As a result, the hydraulic efficiency decreases. In storm runoff situations, however, the preferential flow channels have too little capacity, and water uses more of the wetland filter width. Hence, the hydraulic efficiency increases.
The maximum annual difference between traps in each transect averaged 18 to 42% (Table 4). Hence, sediments are well distributed, and vegetation seems to decrease short-circuit flow. This conclusion is supported by a 2.5D-model simulation, where vegetation increased the hydraulic efficiency twice (Persson et al., 1999). Moreover, based on modeling results, Jadhav and Buchberger (1995) concluded that when flow velocities are low, detention time decreases with vegetation density. However, vegetation increases detention time when flow velocities increase. This may be one of the reasons for the reduction of resuspension by vegetation.
On the other hand, macrophytes decay during the cold and rainy part of the year, and the resistance against preferential flow weakens. Hence, a meandering water flow was observed in the winter (Table 4). Still, some resistance is offered, because last year's stems and leaves cover the sediments as 5- to 40-cm-long bristles. After all, water flow enters CW A's wetland filter in a meandering manner, and only the mid-transect had a statistically different sedimentation rate.
Flocculation
Vegetation cover did not increase the clay concentration in the sediment (Fig. 12). This indicates little flocculation of clay particles in CWs with short detention time. However, it is hard to differentiate between flocs and aggregates in practice (Droppo and Stone, 1994). This experiment did not distinguish between aggregates formed by aggregation processes in the fields and flocs formed through flocculation in the CWs. Since the former process may be of major importance in this experiment (Braskerud et al., 2000), the effect of vegetation on flocculation should not be excluded before more detailed experiments have been carried out.
Effect of Hydraulic Load
According to the factor analysis, hydraulic load increases sediment retention in a CW (Fig. 12). This is the opposite of what could be expected according to the retention model presented in the introduction (e.g., Haan et al., 1994). This phenomenon was explained as a result of the selective erosion and transportation processes in watersheds in Braskerud et al. (2000). These processes may increase sedimentation, despite lower detention time, because the size of particles and aggregates increases. This has also been observed for phosphorus from arable land (Braskerud, 2000). Sedimentation was the most important retention process, because phosphorus was closely connected to soil particles. To some degree, resuspension also increases with increasing hydraulic load. Factor 1 indicates that the clay content in the sediment decreases when the hydraulic load increases (Fig. 12). This is as expected since hydraulic load affects the detention time for settling particles. Even though aggregation causes the clay fraction to behave as coarser particles, some aggregates will be small and affected by the CW detention time. There was no significant difference in retention of suspended solids between the CWs (Table 3), probably a result of too small of differences in hydraulic loads (Table 1).
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. D. Ziegler, J. Negishi, R. C. Sidle, P. Preechapanya, R. A. Sutherland, T. W. Giambelluca, and S. Jaiaree Reduction of stream sediment concentration by a riparian buffer: filtering of road runoff in disturbed headwater basins of montane mainland southeast Asia. J. Environ. Qual., January 1, 2006; 35(1): 151 - 162. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. C. Braskerud, K. S. Tonderski, B. Wedding, R. Bakke, A.-G. B. Blankenberg, B. Ulen, and J. Koskiaho Can Constructed Wetlands Reduce the Diffuse Phosphorus Loads to Eutrophic Water in Cold Temperate Regions? J. Environ. Qual., November 7, 2005; 34(6): 2145 - 2155. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. E. Jordan, D. F. Whigham, K. H. Hofmockel, and M. A. Pittek Nutrient and Sediment Removal by a Restored Wetland Receiving Agricultural Runoff J. Environ. Qual., July 1, 2003; 32(4): 1534 - 1547. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. K. Harter and W. J. Mitsch Patterns of Short-Term Sedimentation in a Freshwater Created Marsh J. Environ. Qual., January 1, 2003; 32(1): 325 - 334. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
