JEQ Grow Your Career With ASA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Related articles in JEQ
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via ISI Web of Science (5)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Gao, S.
Right arrow Articles by Peters, D. W.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Gao, S.
Right arrow Articles by Peters, D. W.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Gao, S.
Right arrow Articles by Peters, D. W.
Related Collections
Right arrow Irrigation
Right arrow Water Quality
Right arrow Wetlands and Aquatic Processes
Right arrow Water Pollution
Journal of Environmental Quality 32:1557-1570 (2003)
© 2003 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America

TECHNICAL REPORTS
Wetlands and Aquatic Processes

Selenium Removal and Mass Balance in a Constructed Flow-Through Wetland System

S. Gao*,a, K. K. Tanjia, Z. Q. Linb,c, N. Terryb and D. W. Petersa,d

a Dep. of Land, Air and Water Resources, Univ. of California, One Shields Ave., Davis, CA 95616
b Dep. of Plant and Microbial Biology, Univ. of California, 111 Koshland Hall, Berkeley, CA 94720
c Dep. of Biological Sciences & Environmental Sciences Program, Southern Illinois Univ., Edwardsville, IL 62026-1651
d U.C. Hansen Agricultural Learning Center, 14292 W. Telegraph Rd., Santa Paula, CA 93060

* Corresponding author (sugao{at}ucdavis.edu)

Received for publication July 1, 2002.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
A field study on the removal of Se from agricultural subsurface drainage was conducted from May 1997 to February 2001 in the Tulare Lake Drainage District (TLDD) of San Joaquin Valley, California. A flow-through wetland system was constructed consisting of ten 15- x 76-m unlined cells that were continuously flooded and planted with either a monotype or combination of plants, including sturdy bulrush [Schoenoplectus robustus (Pursh) M.T. Strong], baltic rush (Juncus balticus Willd.), smooth cordgrass (Spartina alterniflora Loisel.), rabbitsfoot grass [Polypogon monspeliensis (L.) Desf.], saltgrass [Distichlis spicata (L.) Greene], cattail (Typha latifolia L.), tule [Schoenoplectus acutus (Muhl. ex Bigelow) Á. Löve & D. Löve], and widgeon grass (Ruppia maritima L.). One cell had no vegetation planted. The objectives of this research were to evaluate Se removal efficiency of each wetland cell and to carry out a mass balance on Se. The inflow drainage water to the cells had average annual Se concentrations of 19 to 22 µg L-1 dominated by selenate [Se(VI), 95%]. Average weekly water residence time varied from about 3 to 15 d for Cells 1 through 7 (target 7 d), 19 to 33 d for Cells 8 and 9 (target 21 d), and 13 to 18 d for Cell 10 (target 14 d). Average weekly Se concentration ratios of outflow to inflow ranged from 0.45 to 0.79 and mass ratio (concentration x water volume) from 0.24 to 0.52 for year 2000, that is, 21 to 55% reduction in Se concentration and 48 to 76% Se removal in mass by the wetland, respectively. The nonvegetated cell showed the least Se removal both in concentration and in mass. The global mass balance showed that on the average about 59% of the total inflow Se was retained within the cells and Se outputs were outflow (35%), seepage (4%), and volatilization (2%). Independent measurements of the Se retained in the cells totaled 53% of the total Se inflow: 33% in the surface (0–20 cm) sediment, 18% in the organic detrital layer above the sediment, 2% in the fallen litter, <1% in the standing plants, and <1% in the surface water. Thus, about 6% of the total Se inflow was unaccounted for in the internal compartments.

Abbreviations: TLDD, Tulare Lake Drainage District


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
UNDER THE SUPPORT OF the University of California Salinity/Drainage Program, Tulare Lake Drainage District (TLDD), and the California Department of Water Resources, an experimental flow-through wetland system was established in 1996 near Corcoran, California to investigate the potential of varying combinations of plants and water residence times for the removal of Se from saline subsurface drainage waters from croplands before discharge into evaporation ponds (Fig. 1) . The agricultural drainage water in the west side of San Joaquin Valley contains elevated concentrations of Se that have caused toxicity to waterfowl in Kesterson Reservoir when the irrigation drainage water was impounded (Tanji et al., 1986; Benson et al., 1993; Schuler et al., 1990). Currently, the collected subsurface drainage waters are either disposed into evaporation ponds (Evaporation Pond Technical Committee, 1999) or reused as irrigation of salt-tolerant crops and halophytes (Drainage Water Treatment Technical Committee, 1999). In 1987, the USEPA (1987) set the national chronic criterion for Se at 5 µg L-1 in aquatic systems. Later, Hamilton and Lemly (1999) recommended a national water quality criterion of 2 µg L-1 Se based on toxic effects on biota from recent studies. The TLDD wetland project was to test if Se concentration from agricultural drainage water can be reduced to <=2 µg L-1 before disposal into evaporation ponds to minimize toxic effects on waterfowl. The research team consisted of personnel from the University of California at Berkeley and Davis campuses, Tulare Lake Drainage District, and California Department of Water Resources.



View larger version (29K):
[in this window]
[in a new window]
  		Fig. 1. Schematic diagram of the layout of the Tulare Lake Drainage District (TLDD) flow-through wetland system.

 
The benefits of using wetlands to remove a wide range of water-borne contaminants have been long recognized, especially for heavy metals (Sobolewski, 1999; Zhu et al., 1999; Kadlec and Knight, 1996). The anoxic environment and organic matter production in wetlands promote chemical and biological processes enhancing metal removal from the impounded waters. A few cases using field-scale wetlands for Se removal from wastewaters have been reported (e.g., Hansen et al., 1998). Additional studies have been done on the biogeochemistry of Se, such as immobilization by reduction into elemental Se, the association of Se with organic phase, volatilization, and plant uptake (e.g., Zhang and Moore, 1996; Zawislanski and Zavarin, 1996; Banuelos et al., 1997; Frankenberger and Engberg, 1998).

Volatilization is considered an important pathway of Se removal in aquatic systems, including wetlands and evaporation ponds (Frankenberger and Karlson, 1995, Hansen et al., 1998). Selenium volatilization converts toxic Se into less toxic volatile forms such as dimethylselenide (Frankenberger and Karlson, 1989; Terry and Zayed, 1994). Volatilization of Se in aquatic systems is affected by many environmental factors such as microbial community, macrophytes, Se speciation, organic matter amendment, and other physiochemical conditions (e.g., Azaizeh et al., 1997; Zhang and Moore, 1997a, b; Frankenberger and Karlson, 1989; Thompson-Eagle and Frankenberger, 1990). The role of volatilization in Se removal in various wetland systems is not readily predictable. In a Se(IV)-dominated wastewater, biological volatilization accounted for 10 to 30% of the total Se removed by a wetland system (Hansen et al., 1998). In contrast, data on volatilization from Se(VI)-dominated inflow water into a wetland system are not available. However, when Se(VI) water was interacted with a wetland sediment under laboratory conditions, a much lower rate was reported where volatilization accounted for <0.1% of the total loss (Zhang and Moore, 1997b).

Aquatic macrophytes play an essential role in creating and maintaining wetland conditions favorable for contaminant removal, for example, contribution of organic detrital matter producing reducing conditions. Phytoaccumulation also contributes to Se removal in wetlands. The uptake of Se by aquatic plants varied greatly (Pilon-Smits et al., 1999; Zayed et al., 1998, Qian et al., 1999). Many wetland species have a great potential for Se uptake, such as parrot's feather [Myriophyllum aquaticum (Vell.) Verdc.], iris-leaved rush (Juncus xiphioides E. Mey.), cattail, and sturdy bulrush (Pilon-Smits et al., 1999). But plant uptake by itself seems to account for only a minor proportion of the total mass removed (Sobolewski, 1999; Hansen et al., 1998).

The sediment sink mechanism in Se removal in wetlands has been recognized in various studies. The surface sediment accumulates most of the Se by reduction of Se oxyanions to elemental Se and Se associated with the organic phase (Zhang and Moore, 1996; Zawislanski and Zavarin, 1996), including the TLDD system (Gao et al., 2000). These studies provide valuable information for understanding Se removal in wetland systems. However, there are several essential questions that have not been comprehensively addressed, especially on Se mass balance and partitioning in wetland systems. In particular, previous studies have not fully taken into account various interacting factors such as climate, vegetation type, residence time, seepage, and evapotranspiration.

The objectives of this research were to evaluate Se removal efficiency in the wetland cells of the TLDD constructed flow-through system and to carry out a mass balance on Se in a semiarid environment. The Se removal efficiency was evaluated based on the data collected from May 1997 to December 2000. Selenium mass balance was performed for the period of May 1997 to September 2000 when a final comprehensive sampling was performed on standing water, sediment pore water, fallen litter, organic detrital layer, plants, and mineral sediment. These findings are valuable for evaluating flow-through wetland systems for Se removal from agricultural drainage water or other wastewaters.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Tulare Lake Drainage District Flow-Through Wetland System and Sampling
The TLDD flow-through wetland consisted of ten 15- x 76-m unlined cells with eight types of vegetation planted either singly or in combination (Fig. 1). Plant types selected include sturdy bulrush, baltic rush, smooth cordgrass, rabbitsfoot grass, saltgrass, cattail, tule, and widgeon grass. The experimental facility was constructed in TLDD near Corcoran, California in May 1996. This site has a Mediterranean climate with average rainfall of about 145 mm yr-1. Detailed information on the construction and design of the wetland is described in Terry (1997). Cattail, tule, baltic rush, smooth cordgrass, widgeon grass, and saltgrass were transplanted by hand in the cells. Sturdy bulrush and rabbitfoot grass were seeded. Planting was completed by the end of June 1996. To establish the vegetation, fresh water was initially applied followed by blended drain water. Later in April 1997 the supply water for the wetland cells consisted of only saline subsurface drainage that contained selenium from an adjacent tile-drained farm.

Water flow monitoring and sampling in the TLDD wetland began in May 1997. Water inflow was monitored with a totalizing flow meter fitted with terminal plastic orifice pressure regulators and the outflow rate was measured with a 90-degree V-notch weir on weekly to twice-weekly basis. Inflow and outflow water samples were taken once a week. Total Se concentration in the inflow and outflow waters was determined weekly and water Se speciation was determined several times during each year. Selenium volatilization was measured monthly from June 1997 to September 1999. A final comprehensive sampling was conducted in September 2000. Water flows were monitored up to December 2000. The wetland system was shut down in February 2001.

Water Flow and Residence Time
The weekly inflow rate was calculated based on the total volume of water discharged into the cell divided by the time interval between meter readings. The outflow rate was calculated based on Cone's formula for a 90-degree V-notch weir: Qout = 2.49H2.48, where outflow rate (Qout) is in ft3 s-1 and H, the height of water above the V-notch, is in feet (Kulin and Compton, 1975). A metric-transformed formula was used in our computation: Qout = 6090.9(h/304.8)2.48, where Qout is in m3 d-1 and h (the water height) is in mm.

Taking the wetland cell as the control volume and treating any fluxes cutting across the boundary of the control volume as inputs and outputs, the global mass balance of water is defined by:

[1]
where {Delta}V/{Delta}t is the change in water volume per unit time (taken as duration of monitored period herein) within the cell. Since a constant water depth was maintained throughout the study, {Delta}V/{Delta}t was set to zero. The terms Qin and Qout are the measured inflow and outflow rates, respectively; Qprecip is the measured precipitation rate from a nearby weather station (Stratford); and Qet is the estimated evapotranspiration rate obtained from ET = ET0 x Kc x surface area. The term ET0, the reference mowed-grass evapotranspiration rate, was obtained from a nearby weather station and Kc, the crop coefficient, was assumed to have an annual average value of 1.2; about 0.65 in the winter for senescing plants to about 1.5 in the summer. The wetland site is located in an area surrounded by irrigated alfalfa (Medicago sativa L.) and cotton (Gossypium hirsutum L.). Seepage losses from the cell were initially measured by water head drop in a plastic column filled with water that was inserted into the sediment as well as a drop in ponded water surface by shutting off inflow and outflow at night. Since these direct methods of measuring seepage losses are laborious, seepage was obtained as a closure term in Eq. [1]:

[2]

The differences between the directly measured and estimated values using Eq. [2] were small (Tanji and Gao, 2001) and so seepage was estimated throughout the study by Eq. [2].

The residence time of surface water in the cell may be calculated (Kadlec, 1989) from:

[3]
where {tau} is in days, L and W are the length and width in m, D is the depth of water in m, {epsilon} is the substrate porosity (correcting for fraction of water column occupied by vegetation), and Q is the volumetric flow rate in m3 d-1. In actual calculations, Q was taken as the average of inflow and outflow (Kadlec and Knight, 1996). The substrate porosity was taken as 1.0 for Cells 3 and 8 due to their large open water body, 0.95 for Cells 1, 2, 4, and 9, and 0.9 for Cells 5, 6, 7, and 10 based on the estimate on vegetation cover.

The residence time in the cells can be adjusted by changing the standing water depth and/or the inflow rate. The water depth was set at about 8 cm in Cells 5 and 6 (grasses) initially (water depth of Cell 6 was later increased to approximately 15 cm), about 15 cm for Cells 1, 2, 3, 4, and 7, variable depths for Cells 8 and 9 (i.e., 15 cm for shallow areas and 60 cm for deeper areas), and 30 cm for Cell 10. The design for the sediment surface of Cells 8 and 9 followed the pattern of shallow–deep–shallow–deep–shallow by dividing the cell into five sections corresponding to various combination vegetation regimes. Cell 8 had a pattern of bulrush–widgeon (grass)– widgeon–widgeon–bulrush and Cell 9, cattail–widgeon–tule–widgeon–cattail.

The initial residence time was set at 2 to 4 d. Due to comparatively small reductions observed in Se concentration in the outflow relative to the inflow, the residence time in many cells was increased in August 1997 by decreasing the inflow rate. In some cells, residence times observed greater than 21 d did not remove additional Se as compared with 7 d. The new target optimal residence time was set at 7 d for Cells 1 to 7, 21 d for Cells 8 and 9, and 14 d for Cell 10.

Problems Encountered in Tulare Lake Drainage District Wetland Monitoring and Maintenance
The management and operation of the 10-cell TLDD flow-through system encountered various problems, especially before early 1999. Muskrat (Ondatra zibethicus) burrows were discovered in the levee banks between Cells 1 and 2 and between Cells 6 and 7 that caused horizontal crossover of water flow from Cell 1 to 2 and from Cell 6 to 7. The muskrat burrows were fixed by compaction of levees between the cells by running a heavy tractor over the burrows. The cattail in Cell 7 accidentally burned in the fall of 1998 while controlling weeds on the banks by burning, but recovered well in spring 1999. Cattail in Cell 10 died in late 1999 due to the deep standing water depth of 30 cm, preventing regeneration of new cattail plants. In June 2000, Cell 10 was drained to enable the cattail to regenerate and seeds were sown to germinate, but was not successful due to the saline surface soil. In spring 2000, the monoculture cells, particularly Cells 4, 5, and 6, were invaded by cattail and sturdy bulrush from seed dispersal. Therefore, the occurrence of these events must be taken into consideration while evaluating and interpreting the results on residence times, water balance, and Se mass balance. Particular attention will be given to the crossover flow since it directly affects mass balance and when it was necessary to combine the collected data from affected cells under certain conditions.

Selenium Mass Balance
Taking the wetland cell as the control volume and treating any fluxes cutting across the boundary of the control volume as inputs and outputs, the global mass balance on Se is defined by:

[4]
where {Delta}M/{Delta}t is the change in mass of Se within the cell over the time interval of {Delta}t and accounting for Mt, the initial mass of Se in the soil. The terms in and out are respectively the mass (concentration x volume) of Se in the inflow and outflow waters, gas is the mass of Se volatilized, and seep is the mass of Se in seepage losses.

Within the control volume (cell) there are several internal Se compartments that were measured in this study. Thus, an independent calculation of {Delta}M/{Delta}t in Eq. [4] can be obtained from:

[5]
where {Delta}Mint/{Delta}t is the change in storage of Se in the internal compartments to differentiate from the global mass balance, and sed, org, lit, pl, and sw denote the mass of Se in the sediment, organic detritus layer, fallen litter, plants, and standing water, respectively.

Comprehensive Sampling for Selenium Mass Balance
A comprehensive sampling of each of the TLDD wetland cells was conducted in September 2000 for standing water, sediment pore water, fallen litter, organic detrital layer, the underlying mineral sediments to 20- and 100-cm depths, and plants. Total Se concentration and speciation for Se(VI), Se(IV), and organic Se were determined in the standing water and total Se in samples from other internal compartments. Then, estimates on mass of Se in each compartment were performed considering their mass or volume and Se concentrations.

The detailed sampling plan is illustrated in Fig. 2 . Twenty surface water samples were taken from each cell from four rows with five samples in each row. Surface water samples were taken before any other samples. Water samples were stored in ice coolers and kept refrigerated in the laboratory until analysis.



View larger version (23K):
[in this window]
[in a new window]
  		Fig. 2. Sketch of the comprehensive sampling in wetland cells for mass balance, 20–22 Sept. 2000.

 
To estimate Se in seepage losses, three ceramic suction cup lysimeters were installed in early May 2000 in three locations in each cell, approximately one-fourth of the distance from the inlet, at the center, and approximately one-fourth of the distance from the outlet (Fig. 2). The cups were installed at depths below the rooting zone, at approximately 35 cm below the mineral sediment surface level. The rooting depths of the wetland plants were all shallow, less than 20 cm. Vacuum was applied to the lysimeters to collect pore water into previously acid-washed glass vials. Sampling of sediment pore water was conducted in May, July, and September 2000. The mass of Se in root-zone drainage was obtained from the product of the average Se concentration of the pore water and annual seepage volume estimated from water balance (Eq. [2]).

To estimate the mass of Se in fallen litter in each cell, six 3720-cm2 areas were drawn using a square top- and open-bottom aluminum box at designated locations in each cell (Fig. 2) and the fallen litter was collected by hand. The fallen litter samples were dried at approximately 55°C, weighed, ground, and analyzed for total Se. The total mass of fallen litter was calculated by multiplying the total area of the cell and the unit area sample weight. The mass of Se in fallen litter was then obtained from the product of fallen litter biomass and its Se concentration.

To estimate the mass of Se in detrital matter in the cell, a layer with brownish color and rich in fine organic materials right above the mineral sediment was collected using 30 acrylic tubes with an inside diameter of 10 cm pressed to a depth of 10 cm. These tubes were inserted into the undisturbed sediment surface after large pieces of fallen litter were carefully removed. Surface water was decanted and the brown-colored fine detrital layer was scraped off with spatulas. Three subsamples were collected from each cell by compositing each of 10 tubes, randomly chosen, into one sample. A second set of samples was collected by scrapping off the detrital layer from the six 20-cm-deep sediment core samples (described below). The mass of Se in the fine and thin organic detrital layer was obtained from the average of the two sampling methods. The processing, preparation, and analysis of detrital layer samples were analogous to that of fallen litter.

Selenium in the surface sediment was collected by six cores using 5-cm-i.d. acrylic tubes with an approximate length of 25 cm. The tubes were pulled up manually after inserting to about a 20-cm depth. The acrylic tubes were then capped on both ends. The sediment was extruded from the cores and sectioned into 0- to 5-, 5- to 10-, 10- to 15-, and 15- to 20-cm segments. After air-drying, the samples were ground and analyzed for total Se. To investigate Se concentration in deeper sediment, auger samples (approximately 2.5 cm in diameter) up to a 1-m depth were taken at three locations in each cell (Fig. 2). Sediment samples were taken in 10-cm increments, air-dried, ground, and analyzed for total Se.

The analytical methods for Se speciation in water samples and total Se analysis in water, fallen litter, detrital layer, and sediment were previously described in detail in Gao et al. (2000). Selenium speciation in water samples was performed using the method by Zhang et al. (1999). Total Se in water samples was determined following the procedure by Cutter (1982) and Yoshimoto (1992). Total Se in the fallen litter, detrital layer, and sediments was analyzed following a modified digestion procedure by Zasoski and Burau (1977). A hydride generation atomic absorption spectroscopy (AAS) technique was used for quantifying Se concentrations.

Three whole green plants were removed using a spade from each of five selected sampling quadrates (Fig. 2). Dead standing litter was also collected and included in Se mass balance. After the roots were rinsed on-site to remove sediments, all plant samples were placed in sealed plastic bags and kept at 4°C during their transport to the laboratory. The plant cover (%) in each cell was recorded, and the standing aboveground plant biomass was determined by the harvesting method. The root biomass was estimated by the ratio of the shoot biomass to the root biomass that was previously determined for each whole green plant (Terry, 2000). The plant samples were dried at 60°C, and ground in a Wiley mill to pass a 0.43-mm mesh screen. The ground samples were acid-digested with HNO3 and H2O2 (Method 3050B; USEPA, 1996). Total Se was measured with AAS with hydride generation (VGA-77; Varian, Palo Alto, CA) following Method 7742 (USEPA, 1994). Standard reference material (SRM-1567a, wheat flour) was measured as an internal quality control for analyses of Se in plant samples. Detailed information on the results of plant biomass and Se concentration can be found in Terry (2001).

Volatilization of Se was measured monthly in triplicates in Cells 1 to 7 from June 1997 to September 1999. Volatilization measurement was not done in year 2000 to minimize disturbance inside the cells for final comprehensive sampling. Volatile Se was measured using the method developed by Lin et al. (1999). Detailed information on Se volatilization and results were reported in Terry (2000). The amount of Se volatilized for each cell was obtained by integrating the measured data. For cells and the period of time when observed data were not available, an estimate was made based on the measurement of other cells in similar conditions.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Overall Performance of Wetland Cells and Selenium Removal Efficiencies
The annual average hydraulic residence times for TLDD wetland cells for years 1998, 1999, and 2000 are shown in Fig. 3 . The residence times were higher and fluctuated more in 1998 than in 1999 and 2000, due to plumbing problems. In 1999 and 2000, the residence times were relatively more stable except for cells with crossover flows. Average weekly water residence time varied from about 3 to 15 d for Cells 1 through 7 (target 7 d), 19 to 33 d for Cells 8 and 9 (target 21 d), and 13 to 18 d for Cell 10 (target 14 d).



View larger version (40K):
[in this window]
[in a new window]
  		Fig. 3. Average annual residence times. Error bars are standard deviations from weekly measurements. BUL, sturdy bulrush; BAL, baltic rush; CORD, smooth cordgrass; RAB, rabbitfoot grass; SALT, saltgrass; CAT, cattail; B–W–B, bulrush–widgeon grass–bulrush; T–W–C, tule–widgeon grass–cattail.

 
Horizontal crossover flow from Cell 1 to Cell 2 and from Cell 6 to Cell 7 from muskrat burrows occurred in fall 1999 and was fixed in late 1999 and early 2000, respectively. These crossovers are reflected in the longer residence time for Cells 1 and 6 as compared with Cells 2 and 7 in 1999. Rabbitsfoot grass (Cell 5) suffered from inappropriately high water depth and the water level was lowered during the course of this research as indicated by the decrease of residence time in 2000. Cattail in Cell 10 in which residence time was about twice of that in Cell 7 died in late 1999 from high water depth. The cell was completely drained in June 2000 in an effort to reestablish the cattail, but without success. Thus, the monitored data for Cell 10 was only up to June 2000. Considering all of these factors, the more reliable data for Se removal efficiency are those collected in 1999 and 2000.

Selenium concentrations in outflows including Cell 3, the nonvegetated cell, were all lower than those of inflow water (Fig. 4) . In year 2000, outflow Se concentrations in Cells 2, 4, and 6 were the lowest, followed by Cells 1, 8, 9, and 10, and highest in Cells 3 (nonvegetated), 5, and 7. Statistical analysis was performed on the data of year 2000 using SAS Version 8.1 (SAS Institute, 2001) to test the differences between Se concentrations of inflow and outflow as well as differences of outflows among cells. Results show that Se concentrations in the outflow are all significantly lower than the inflow (p < 0.001) for all 10 cells. Using a Bonferroni-type multiple comparison in the ANOVA model, outflows from all of the cells except Cell 5 had significantly lower Se concentrations than Cell 3 (nonvegetated) (p < 0.05). However, none of the cells achieved the target Se concentration of <=2 µg L-1 in outflow.



View larger version (44K):
[in this window]
[in a new window]
  		Fig. 4. Average annual Se concentration in inflow and outflow waters. Error bars are standard deviations from weekly measurements. BUL, sturdy bulrush; BAL, baltic rush; CORD, smooth cordgrass; RAB, rabbitfoot grass; SALT, saltgrass; CAT, cattail; B–W–B, bulrush–widgeon grass–bulrush; T–W–C, tule–widgeon grass–cattail.

 
Selenium removal efficiencies by the wetland cells are reflected by the annual average ratios of outflow to inflow in concentration as well as Se mass in year 2000 when there were the fewest problems in maintaining the wetland system as compared with previous years (Fig. 5) . In terms of Se concentration, Cells 2, 4, 6, and 10 showed better performance compared with other cells with outflow to inflow ratios of <0.5 (i.e., Se concentration was reduced by >50% of the inflow water). Other cells had a Se concentration ratio of >0.5 with lowest values (poorest removal) for Cell 3 and 5. However, in terms of mass, Cells 4 and 6 had the lowest ratio of 0.24 (i.e., the highest Se removal by 76%) and Cell 3, the nonvegetated control cell, had the highest ratio of 0.52 (i.e., the lowest Se removal). Other vegetated cells had Se mass ratio ranging from 0.31 to 0.44. Using a Bonferroni-type multiple comparison in the ANOVA model (SAS Institute, 2001), the concentration ratio of Cell 3 was significantly higher than all other cells (p < 0.05) except Cell 5 (p = 0.23). While testing on the mass ratios, the value of Cell 3 was significantly higher than Cells 1, 4, 5, 6, 8, 9, and 10 (p <= 0.01) but not Cell 2 (p = 0.06) and Cell 7 (p = 0.11). Overall, Se concentration ratio ranged from 0.45 to 0.79 and mass ratio from 0.24 to 0.52, that is, 21 to 55% in Se concentration reduction and 48 to 76% Se mass removal, respectively.



View larger version (48K):
[in this window]
[in a new window]
  		Fig. 5. Selenium concentration ratio and mass ratio of outflow to inflow in 2000. Error bars are standard deviations from weekly measurements. BUL, sturdy bulrush; BAL, baltic rush; CORD, smooth cordgrass; RAB, rabbitfoot grass; SALT, saltgrass; CAT, cattail; B–W–B, bulrush–widgeon grass–bulrush; T–W–C, tule–widgeon grass–cattail.

 
Selenium Biochemical Transformation in Wetland Cells
Selenium speciation in inflow and outflow waters for year 2000 is shown in Fig. 6 . The inflow water was dominated by the oxidized form Se(VI) (95%), with a smaller percentage of Se(IV) (approximately 5%), and negligible organic Se. In contrast, the proportion of reduced Se forms, Se(IV) and organic Se, were significantly higher in the outflows, indicating that Se reduction processes have taken place in the wetland.



View larger version (48K):
[in this window]
[in a new window]
  		Fig. 6. Selenium speciation in inflow and outflow waters in 2000. Error bars are standard deviations of four seasonal measurements during the year. BUL, sturdy bulrush; BAL, baltic rush; CORD, smooth cordgrass; RAB, rabbitfoot grass; SALT, saltgrass; CAT, cattail; B–W–B, bulrush–widgeon grass–bulrush; T–W–C, tule–widgeon grass–cattail.

 
Analysis was done on the longitudinal distribution of total Se concentration and speciation for Cells 1 to 9 with sampling locations indicated in Fig. 2. Detailed information can be found in Tanji and Gao (2001). The results show that the total Se concentration in surface water generally decreased in the direction of inflow to outflow except Cell 3. The proportion of reduced Se did increase in most cells along the direction from inlet to outlet except for Cell 3 (nonvegetated), which was most likely due to mixing from high wind conditions. The nonvegetated Cell 3 is not sterile because of volunteer growth of algae and microbes, resulting in significant Se reduction.

Selenium Accumulation in Surface Sediment
There is a distinct pattern for Se accumulation in the surface sediment for all 10 cells (Fig. 7) . The data represent the average of three longitudinal sampling points in each cell sampled in September 2000 after 4 yr of operation. The highest Se concentration was found in the surface and Se concentration dramatically dropped with increasing sediment depth. The background Se concentration in the sediments was about 0.11 ± 0.02 mg kg-1. Although there was some variation in concentrations, the trend and pattern in Se accumulation in all the wetland cells are clearly shown.



View larger version (38K):
[in this window]
[in a new window]
  		Fig. 7. Selenium concentration in sediments in 2000. Values at depth of 0 cm represent Se concentration in the fine organic detrital layer. Error bars are standard deviations of several sampling locations shown in Fig. 2. BUL, sturdy bulrush; BAL, baltic rush; CORD, smooth cordgrass; RAB, rabbitfoot grass; SALT, saltgrass; CAT, cattail; B–W–B, bulrush–widgeon grass–bulrush; T–W–C, tule–widgeon grass–cattail.

 
Sampling mineral sediment to a 100-cm depth showed that Se concentration in the depths greater than 20 cm approached background level. There were no differences in the longitudinal distribution of Se along the direction of inflow to outflow for the deeper sediments (for detailed information, see Tanji and Gao, 2001).

Selenium Distribution in the Compartments of Wetland Cells
Selenium concentrations in fallen litter, fine organic detrital layer, and surface sediment (0–5 cm) (Fig. 8) showed that fallen litter contained the highest Se concentration, followed by the detrital layer and then the surface 0 to 5 cm of the sediment. The distribution of mass of Se (Fig. 9) , however, followed an opposite direction as compared with the concentration pattern, that is, sediment > detrital layer > fallen litter, because of the smaller masses in fallen litter and detrital layer as compared with the sediment.



View larger version (37K):
[in this window]
[in a new window]
  		Fig. 8. Selenium concentration in fallen liter, organic detrital layer, and surface sediment (0–5 cm) in September 2000. Error bars are standard deviations of several sampling locations shown in Fig. 2. BUL, sturdy bulrush; BAL, baltic rush; CORD, smooth cordgrass; RAB, rabbitfoot grass; SALT, saltgrass; CAT, cattail; B–W–B, bulrush–widgeon grass–bulrush; T–W–C, tule–widgeon grass–cattail.

 


View larger version (42K):
[in this window]
[in a new window]
  		Fig. 9. Mass of Se accumulated in the 0- to 20-cm sediment, detrital layer, and fallen litter in September 2000. BUL, sturdy bulrush; BAL, baltic rush; CORD, smooth cordgrass; RAB, rabbitfoot grass; SALT, saltgrass; CAT, cattail; B–W–B, bulrush–widgeon grass–bulrush; T–W–C, tule–widgeon grass–cattail.

 
Water Balance
Water balance for each cell as well as for the total 10 cells is shown in Table 1 . Total water applied to the TLDD wetland system during the period of May 1997 to September 2000 was 386 954 m3, of which about 30% was discharged as outflow, 18% as evapotranspiration, and 51% as seepage loss, and <1% present as standing water.


View this table:
[in this window]
[in a new window]
  		Table 1. Water balance in the Tulare Lake Drainage District (TLDD) flow-through wetland cells for the period of May 1997 to September 2000.

 
When water balance is examined for individual cells, large variations were noted when crossover flow was ignored. In terms of outflow, the values range from 10 to 62% of the applied water. When Cells 1 and 2 are combined because of horizontal crossover, the average is 25%. Similarly, when Cells 6 and 7 are combined because of the crossover, the average is 28%. When crossover cells are combined the outflow ranges from 25 to 44%. Evapotranspiration ranges from 14 to 22% of the inflow water. The standing water was only a small percentage of the total inflow (<1%). The largest losses were due to the seepage ranging from 34 to 60%. In addition to soil hydraulic properties, the dense rooting system of the plants appeared to increase seepage losses.

Selenium Mass Balance
The results of global Se mass balance (Eq. [4]) for each cell based on data collected during the period of May 1997 to September 2000 are shown in Table 2 . The mass of Se in the inflow and outflow was integrated from average Se concentrations determined and water volume between monitoring dates. Selenium loss in seepage was calculated from the average Se concentration in pore water collected at a 35-cm depth and the volume of seepage water. Selenium loss by volatilization was measured monthly for over two years and estimates were made for those months or cells with no measurements, based on previously measured data for the cell in question. Estimates were made for winter when the field site was inaccessible due to the rain, and for year 2000 when no measurement was done.


View this table:
[in this window]
[in a new window]
  		Table 2. Global Se mass balance in the Tulare Lake Drainage District (TLDD) flow-through wetland cells for the period of May 1997 to September 2000.

 
The total Se introduced into the 10 cells was 8108 g, of which about 41% (3337 g) left the wetland system via outflow (35%), seepage (4%), and volatilization (2%). Thus, Se retained in the cells (or the {Delta}M/{Delta}t) was about 59% of the applied Se.

Results of mass of Se determined in the internal compartments of the 10 cells are shown in Table 3 . On the average, about 53% of the total Se from inflow was retained in the wetland cells as surface sediment (0–20 cm) (33%), organic detrital layer above the mineral sediment (18%), fallen litter (2%), standing plants (<1%), and standing water (<1%). The {Delta}Mint/{Delta}t was 4305 g Se (53% of the total inputs), which is close to the estimate from the global balance, that is, {Delta}M/{Delta}t of 4771 g Se (59% of the total inputs). When global outputs and Se in the internal compartments are considered, about 94% of the inputs are accounted for, that is, 6% of the Se was unaccounted for in the internal compartments. This percentage of error is considered quite acceptable for a field study of this magnitude and duration.


View this table:
[in this window]
[in a new window]
  		Table 3. Internal Se mass balance in the Tulare Lake Drainage District (TLDD) flow-through wetland cells for the period of May 1997 to September 2000.

 
It was necessary to combine Cells 1 and 2 and Cells 6 and 7 to minimize the crossover flow impact on Se removal evaluations, especially in outflow and seepage loss, as shown in Tables 1, 2, and 3. The largest sink of Se in the cells is mineral sediment (24–41% of the inputs) and then organic detrital layer (12–34% of the inputs). On an individual cell basis, from 86 to 110% of the Se inputs was accounted from measurement when Cells 1 and 2 and Cells 6 and 7 are combined. Selenium mass balance in the TLDD wetland system was summarized in Fig. 10 .



View larger version (14K):
[in this window]
[in a new window]
  		Fig. 10. Selenium mass balance in the Tulare Lake Drainage District (TLDD) flow-through wetland after four years of operation.

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
CONCLUSIONS
 REFERENCES
 
Selenium removal efficiency by the flow-through wetlands was appraised by either a decrease in concentration of Se from the inlet to the outlet or a decrease in the mass of Se from the inlet to outlet. Water Se concentration currently used to evaluate its toxic risk to fish and wildlife is 5 µg L-1 for surface waters (USEPA, 1987). All the wetland cells reduced Se concentration and some of the outflow waters approached Se concentration below 5 µg L-1 in some seasons but none of the cells achieved <=2 µg L-1. It has been realized lately, however, that it is more important to know the dietary exposure to Se (both water and sediment Se) in determining the risk potential for chronic effects in aquatic organisms (USEPA, 1998). The TLDD agricultural drainage water is dominated by Se(VI). Selenium removal efficiencies tend to be higher from Se(IV)-dominated water compared with Se(VI)-dominated water based on available data (Hansen et al., 1998; Zhang and Moore, 1997b). A study by Hansen et al. (1998) on Se removal in a selenite-dominated wetland system revealed that about 89% of Se was removed and Se concentration decreased from 20 to 30 µg L-1 in the inflow to <5 µg L-1 in the outflow. Similar conclusions were drawn that most of the selenium was removed by the immobilization into sediments and plants tissues, although biological volatilization may have accounted for as much as 10 to 30% of the Se removed. Selenium removal differences between the two wetland systems, however, could be due to their designs; Hansen et al. (1998) were studying a much larger wetland (36 ha) compared with the TLDD wetland system, which was smaller than 1 ha. Nonetheless, the TLDD wetland cells did reduce Se by 21 to 55% on a concentration basis and 48 to 76% on a mass basis.

Cell 3 (nonvegetated) showed the lowest Se removal efficiency. The results indicate that vegetation enhances Se removal in the wetland cells. Vegetation contributes organic materials as fallen litter and detrital matter that accumulate on the sediment surface. The organic matter then serves as the substrate and carbon sources for microbes. The decomposition of organic matter results in anoxic conditions that lead to reduction of electron acceptors including Se(VI), which is reduced to Se(IV), Se(0), and some organic forms. Elemental Se is immobile while most of the organic forms of Se reside in the solid organic phase. This may explain the higher Se removal efficiencies in the vegetated cells as compared with the nonvegetated cell. The nonvegetated Cell 3 with open surface water, however, supported voluntary growth of algae and microbes. Therefore, a reducing environment for microbial reduction of Se and immobilization occurred to some extent in Cell 3.

The highest concentrations of Se were measured in the fallen litter that was coated with microbial slime as well as in the organic detrital layer. Reduction of Se(VI) to Se(0) is driven by microbially mediated processes (Oremland et al., 1990). The abundant supply of organic carbon sources enhances Se reduction processes and accumulates Se in the surface sediments. Although detailed information about Se association with organic materials is not clear from the literature, the large fractions of organic matter–associated Se in the wetland cells (Gao et al., 2000) and other studies (e.g., Zawislanski and Zavarin, 1996; Zhang and Moore, 1996) have illustrated the important role of organic matter in Se immobilization in wetland systems.

The biochemical reduction of Se in the wetland cells is one of the major processes in Se transformation. The percentages of reduced Se forms in outflow waters increased gradually from 1998 to 2000 (Tanji and Gao, 2001). This trend indicates that as the wetland matured and accumulated organic matter, Se reduction intensified. We believe that steady state conditions have been approached in the TLDD wetland cells. Generally speaking, surface water Se concentration in September 2000 decreased along the water flow path due to sinks of Se within the cells except Cell 3. Moreover, the percentage of reduced forms of Se [Se(IV) + organic Se] increased along the water flow path except Cell 3. This increasing trend of reduced Se species denotes prominent reduction processes along the water flow path. In Cell 3, however, both total Se concentration and reduced forms of Se remained constant along the flow path in the open water, which was perhaps due to mixing from high-wind conditions at the time of sampling. This mixing was not evident in vegetated cells.

The role of organic matter in Se concentration is also shown in Fig. 8. Selenium concentrations in the fallen litter were very high, followed by the detrital layer and lower concentrations of Se in the mineral sediment. But as shown in Fig. 9, Se on a mass basis is highest in the mineral sediment followed by the detrital layer and smallest in the fallen litter because of differences in the sizes of these pools.

Previously reported sediment Se fractionation data revealed that the largest fraction of Se immobilized was due to reduction of inflow water Se(VI) to elemental Se (46%) followed by organic matter–associated Se (34%) (Gao et al., 2000). Other sinks of Se, including soluble, adsorbed, and carbonate-associated Se were 3, 10, and 3% of the total Se in the sediments, respectively. Thus, it was concluded that the major Se sink mechanisms in the sediments are the reduction of selenate to elemental Se and immobilization into the organic phase.

The internal mass balance of Se in the wetland cells from May 1997 to September 2000 (Table 3) has further shown that the two major sinks of Se are in the surface-20-cm mineral sediments (33% of inflow) and Se in the organic detrital layer (18% of inflow). Other sinks (fallen litter, standing plants, and standing water) of Se range from <1 to 2% of total inflow Se. The results obtained from the TLDD flow-through wetland cells are for site-specific conditions such as dimensional configuration (15 x 76 m), residence times (7–21 d), medium levels of Se concentration in the inflow water (mean concentration of 19–22 µg L-1), and inflow water dominated by Se(VI). Although much has been learned from a scientific point of view, additional studies are desirable to extend the current findings to engineering design and performance evaluation for a scaled-up operational flow-through system. Our research continues toward this goal with a conceptual modeling effort by using and integrating the information and database obtained.

A flow-through wetland can remove significant amounts of Se from Se-contaminated water and the removal efficiency may be improved by considering several factors. The large seepage loss in water budget was site-specific but it does not contribute much to Se loss (approximately 4% of total Se input) in this study. In considering the effects of plants on Se removal, most plant species showed enhancing Se removal rates significantly compared with nonvegetated conditions. However, rabbitfoot grass is clearly not a favorable choice due to its requirement of very shallow water surface. Widgeon grass (a submergent plant) is not recommended either because of its small biomass production. There were no distinct differences among other plants (i.e., bulrush, baltic rush, smooth cordgrass, saltgrass, tule, and cattail) in terms of biomass production, Se uptake, and removal efficiencies of vegetated wetland cells. Overall, the results from this study have shown that using flow-through wetland for Se removal is applicable for Se-contaminated water but perhaps not to a level of safe limits of some aquatic biota and waterfowl. The major concern is the potential Se environmental ecotoxicity risks. Much of the Se removed by the wetland was retained inside the wetland in the organically rich surface sediment. There is a potential of this immobilized Se to enter into the aquatic food chain. Reduction of Se in concentration and mass is accompanied by an increase in the proportions of reduced species [Se(IV) and organic Se] in the standing water and outflow that are more toxic than Se(VI) at the same concentration for many aquatic biota. Further studies are needed in this area to minimize the potential effect of reduced Se forms on wildlife, particularly waterfowl that feed on Se-contaminated macroinvertebrates and seeds of wetland plants.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
REFERENCES
 
The experimental flow-through wetland can reduce Se concentration (21–55%) and remove large amounts of Se mass (48–76%) from agricultural drainage water on the west side of the San Joaquin Valley. Selenium removal appears to be more effective with vegetation planted in the wetland, perhaps because it promotes reducing conditions in the sediment. The largest amount of Se from the flowing water was retained inside the wetland system, especially in the organic detrital layer and surface sediment, with the outflow as the second output. Selenium losses due to volatilization and seepage were one magnitude lower. Further studies on Se removal performances in scaled-up wetland systems are necessary. There is also a need to evaluate and reduce Se ecotoxicity risk in wetland systems.


   ACKNOWLEDGMENTS
 
This research has received the support of several scientists. Dr. J. Letey, director of the UC Salinity Drainage Program, served as a key science advisor; Doug Davis, manager of Tulare Lake Drainage District, played a key role in establishing and maintaining the wetland system; and Kurt Kovac of the California Department of Water Resources coordinated weekly field monitoring in the final two years. Other scientists who contributed to the wetland project include T. Fan, R. Higashi, M. Lytle, and A. Zayed.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 


Related articles in JEQ:

This Issue in Journal of Environmental Quality

JEQ 2003 32: 1167-1172. [Full Text]  




This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Related articles in JEQ
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via ISI Web of Science (5)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Gao, S.
Right arrow Articles by Peters, D. W.
Right arrow Search for Related Content
PubMed