TREATMENT WETLANDS - CHAPTER 11 ppt

Because there is generally an excess of calcium in surface water and waste-water, calcium concentration does not change appreciably in many wetland treatment systems see Table 11.7.. Bec

In addition to the pollutants discussed in earlier chapters,

wastewaters typically contain many other substances Some

of these elements can cause problems when discharged to

receiving waters, and their removal must be considered

dur-ing design These additional materials include salts, acids,

bases, macronutrients, micronutrients, and heavy metals, and

may be categorized in a number of ways Salts include

com-pounds that readily dissociate in water to form charged ions

that may or may not be used as nutrients for plant and animal

growth Common examples of salts are sodium chloride

(NaCl) and gypsum (CaSO4) Acids release a hydrogen ion

when they dissociate (e.g., hydrochloric acid, HCl), and bases

release a hydroxyl ion (e.g., ferric hydroxide—Fe(OH)3)

Spe-cific environmental conditions determine whether the cations

(positively charged ions) and anions (negatively charged ions),

formed when a salt, acid, or base is dissolved in water, are

chemically or biologically active Collectively, ionic

materi-als contribute to the electrical conductivity (EC) of the water

When ionic materials are combined with dissolved nonionic

materials, the result is the total dissolved solids (TDS)

con-tent of the water

Nitrogen and phosphorus, discussed in Chapters 9 and

10, are examples of macronutrients, which have strong

bio-geochemical cycles in a wetland Sulfur also is typically

present in variable but potentially high concentrations, and

has just as powerful influences on wetland functioning The

magnitude of these influences is just emerging as a

control-ling factor on wetland performance for a number of other

pollutants Most obvious is the role of sulfides in

immobiliz-ing trace metals

Iron, aluminum, and manganese are ubiquitous in

wet-lands, but are present at elevated concentrations in mine

drainage waters and the wetlands constructed to treat them

A trace metal can be either a required micronutrient or toxic,

depending on the concentration For example, copper and

zinc are essential elements for plants and animals at low

con-centrations, but they are toxic to some organisms at elevated

concentrations However, for some trace metals, such as

cad-mium and lead, essentiality for plants or animals has never

been found In this chapter, many of the important elements

are discussed, but the list does not include all the elements

that may be found in waters, or all those that might require

treatment The use of a wetland treatment system to modify

the concentration of elements depends on how the elements

interact with the wetland environment and on the wetland

designer’s knowledge of design factors that can enhance or

diminish these processes There is a rapidly growing body

of knowledge about how wetland treatment systems affect

specific trace elements A thorough, updated review of the scientific literature is recommended for project design

A number of substances are considered measures of water quality, but are seldom of concern as pollutants to be treated

in constructed wetlands These include common metals (e.g., sodium, potassium, calcium, and magnesium) as well

as halogens (e.g., fluorine, chlorine, bromine, and iodine) Together with sulfate, these compounds often dominate the total ion content of natural waters and wastewaters In total, they form the major part of EC and TDS Sulfate is of spe-cial importance, because of its active biogeochemical cycle, and interactions with trace metal removal In some treatment wetlands, several of these collective water quality parameters have become important in their own right

11.1 HALOGENS

Chloride and bromide are widely regarded as being vative” in wetland environments, meaning that they interact with the ecosystem to a very limited extent Therefore, they can be used as tracers of water movement in the wetland Usually, chloride is present at concentrations that preclude its use as an injected tracer, but it sometimes serves as a means

“conser-of confirming the wetland water budget Fluoride is usually

a very minor trace constituent in aquatic systems, but there are industrial effluents that contain relatively high concentra-tions The aluminum industry is one such source, including leachates from solid waste disposal sites Bromide is often present at low background concentrations, and injected bro-mide may then be used to trace internal water movements Very little is known about the fate and transport of iodine in freshwater systems

C HLORIDE AND C HLORINE

U.S EPA (2002a, 2006) sets a criteria maximum tion (CMC), which is an estimate of the highest concentration

concentra-of a material in surface water to which an aquatic community can be exposed briefly without resulting in an unacceptable effect The value for chloride is 860 mg/L, and for chlorine

is 19 µg/L U.S EPA (2002a, 2006) also sets a criterion tinuous concentration (CCC), which is an estimate of the highest concentration of a material in surface water to which

con-an aquatic community ccon-an be exposed indefinitely without resulting in an unacceptable effect The CCC value for chlo-ride is 230 mg/L, and for chlorine is 11 µg/L

The chlorine content of wetland plant tissues has not been measured often Results from two projects are shown in

404 Treatment Wetlands

Tables 11.1 and 11.2 The Oxnard, California, systems were

exposed to high chloride, and developed high leaf tissue

concentrations (5–40 g/kg dry weight, or 0.5–4.0%)

Presumably, much of the chlorine associated with the dry

matter was originally in solution in the plant water content,

which is 70–80% of the wet weight Therefore sap

concentra-tions would be about five times lower Interestingly,

Salicor-nia spp is a hyperaccumulator of chlorine (Table 11.1) It is

also notable that roots contain much less chlorine than the

shoots Standing dead and litter of Typha latifolia were found

to contain much less chlorine than live leaves at the

Hough-ton Lake, Michigan, treatment wetland (Table 11.2)

Chlorine is biologically interactive in wetland

eco-systems It is influential in the osmolality salinity balance,

but metabolic utilization does not usually cause changes in

water concentrations (Wetzel, 1983) However, there are

cir-cumstances in which utilization can be measured Xu et al.

(2004) measured chloride and sulfate profiles in vertical flow

mesocosms with Typha latifolia in sand, during growth of

the plants The hydraulic loading rate was low (0.66 cm/d),

and consequently, transpiration was an important effect (about 0.3 cm/d) Sulfate was added at concentrations far in excess of any potential plant requirements (80 mg/L SO4-S)

As the water traversed the root zone, sulfate concentrations increased to about double their inlet value, which was strictly attributed to transpiration losses No increase occurred in unvegetated controls However, the profiles of chloride were very different: virtually all of the inlet chloride was absorbed

in the mesocosms, from a starting concentration of about 5.0 mg/L Given the biomass increase of the plants, chloride removal would have produced a chloride content of the cat-tails of about 4,000 mg/kg, which is at the low end of the

range measured for Typha (Tables 11.1 and 11.2).

The more typical situation is an overabundance of ride entering the wetland Because of the relatively low biological demand for chloride, the total chloride mass is usually relatively constant between the inflows and outflows and storages of a treatment wetland (Table 11.3) Therefore, the wetland chloride mass balance can be used to confirm the water budget

chlo-TABLE 11.2

Halogen Content of Biomass Compartments for Typha latifolia in the

Houghton Lake, Michigan, Treatment Wetland in 1991

Note: Units are mg/kg for tissues, and mg/L for water.

Source: Unpublished data.

TABLE 11.1 Chlorine and Fluorine Concentrations in Plant Tissues at Oxnard, California

Note: Units are mg/kg for tissues, and mg/L for water.

Source: Data from CH2M Hill (2005) Additional testing for the Membrane Concentrate Pilot Wetlands Project.

Report to the City of Oxnard Water Division, Oxnard, California, United States.

Examples of Conservative Materials Entering and Leaving FWS Treatment Wetlands

HLR (m/yr)

EC In (µS/cm)

EC Out (µS/cm)

Chloride In (mg/L)

Chloride Out (mg/L)

TDS In (mg/L)

TDS Out (mg/L) Pass-Through

Anomalies

New Hanover, North Carolina Full scale 1 0.79 6,256 1,468 2,753 411 5,742

© 2009 by Taylor & Francis Group, LLC

406 Treatment Wetlands

Chloride can serve as a tracer of water movement,

espe-cially in the analysis of the very slow underground movement

of water For instance, the progress of a chloride front from

rapid infiltration basins, underground to a monitoring well on

the edge of the receiving wetland, and then out into that

wet-land, is shown in Figure 11.1 The wastewater treatment plant

at Genoa-Oceola, Michigan, received very high chlorides

(about 400–550 mg/L) because of the widespread use of water

softeners in the region The treated water was discharged onto

rapid infiltration basins on a hilltop adjacent to a wetland The

hydrologic gradient moved the water slowly toward the

wet-land, and the high chloride arrived in wells at the wetland edge

after about three years After about six years, the wetland

sur-face waters reflected the chloride of the wastewater, with some

dilution from the other flows in the aquifer

Disinfection: Chlorine in Wetlands

Free chlorine is toxic to most life forms, and is one of the

most frequently used wastewater disinfectants There are

implications for treatment wetlands that receive chlorinated

effluents, because the residual toxicity may negatively

influ-ence the microbial communities within the system

Some of the free chlorine added during disinfection is

converted in solution to chlorides or chloramines, the

lat-ter being regarded as an undesirable pollutant The products

that result from disinfecting water by the addition of chlorine

are:

Free residual chlorine—the portion of chlorine

remaining as molecular chlorine, hypochloride

(HOCl), or hypochlorite ion (OCl –)

Combined residual chlorine—the portion of

chlo-rine that combines with ammonia or nitrogenous

compounds, forming chloramines

Total residual chlorine (TRC)—the sum of free

residual plus the combined residual chlorine

sev-as ammonia Wetlands have enough organic matter to mote formation of trihalomethanes (THM) (Gallard and von Gunten, 2002), but other organic halides may also form Total organic halides (TOX) and halo-acetic acids (HAA)

may also form (Rostad et al., 2000) The photochemical

pro-cess is initiated by production of oxidants such as peroxides These oxidants then oxidize the chlorinated compounds Volatilization, adsorption, and interactions with aquatic plants and the soil system may also contribute to the decay

of residual chlorine

Studies of the loss of TRC were conducted in the Tres Rios, Arizona, FWS wetlands (Wass, Gerke, and Associ-ates, 2004) Disappearance was approximated by first-order behavior The rate constants were calculated to be 0.86 d−1for the Cobble C1 and Hayfield H1 wetlands, based on tran-sect data Thus there would be more than 90% reduction in TRC for a three-day detention time Surveys of organo-chlo-rine compounds in the wetlands showed decreasing gradients from inlet to outlet for TOX, THM, and HAA (Table 11.4)

B ROMIDE AND B ROMINE

Bromide is not commonly measured as a constituent of natural freshwaters or wastewaters It is a common choice for a water movement tracer, and a number of studies have therefore determined the background bromide in treatment wetland waters Example values were 0.2–0.3 mg/L at Tres Rios, Arizona; 0.13–0.18 mg/L at Orlando Easterly, Florida; 0.05–0.15 mg/L at Hillsdale, Michigan; and 0.3–0.4 mg/L

at Des Plaines, Illinois Bromine is, therefore, also found in only minor trace amounts in vegetation Concentrations of

25–46 mg/kg dry mass were measured in Typha latifolia at

the Houghton Lake, Michigan, treatment wetland Parsons

et al (2004) added a uniform sudden dose of bromide to

0 100 200 300 400 500 600

Years of Operation

WWTP Edge well Wetland

FIGURE 11.1 Progress of chloride from the infiltration beds of the wastewater treatment plant (WWTP), underground to a well on the

wetland edge, and then out into the wetland at Genoa-Oceola, Michigan (From unpublished data.)

establish an initial surface water concentration of 100 mg/L

in a prairie pothole in Saskatchewan The wetland had no

surface inflow or outflow, and thus the dose remained

con-fined to the system except for infiltration They measured

20–90 mg/kg dry mass in aboveground tissues of Heracleum

lanatum, Polygonum spp., and Carex spp., with localized

values ranging up to 12,700 mg/kg dry mass The fraction of

the dose retained in vegetation was estimated to be 8.7%, as

determined by areal averaging for the diverse species

Bromide sorbs to soils to about the same extent as nitrate

(Clay et al., 2004), which is negligible in most situations

Bromine does not have a role in plant metabolism; however,

bromide can be taken up by plants, to help satisfy the anionic

component of the charge balance in the plant internal water

In that respect, bromide competes with chloride, as

docu-mented by Xu et al (2004) in Typha and Phragmites

sys-tems Plant uptake is, therefore, presumably greatest during

the growing season, during which new plant water is building

within the wetland

F LUORIDE AND F LUORINE

Fluorine in water exists primarily in the form of sodium and calcium salts Calcium fluoride is used as a flux in steel manufacturing Sodium fluoride is used as a drinking water additive for prevention of dental cavities (tooth decay) The recommended optimum level ranges from 0.7 mg/L for warmer climates to 1.2 mg/L for cooler climates Median concentrations are 0.2 mg/L in surface water and 0.1 mg/L

in groundwater (U.S EPA, 2002a) The current Maximum Contaminant Level set by the U.S EPA in 1986 is 4 mg/L Fluoride levels typically range from 0.03 to 0.57 mg/L in

eastern U.K rivers (Neal et al., 2003a).

Fluoride differs from chloride and bromide, because it has been a target of treatment wetland design The aluminum industry relies upon molten salt electrolytes that contain fluo-rides Solid wastes from the industry are usually landfilled, and produce leachates that contain elevated concentrations of fluoride (up to 100 mg/L)

Fluoride partitions more strongly to soils and ments than do bromide and chloride The Langmuir adsorp-tion capacities of soils ranges from 100–400 mg/kg for silts and loams (Bower and Hatcher, 1967) However, the oxyhydroxides of iron and aluminum have much higher bind-ing capacities, 30,000–50,000 mg/kg This property causes

sedi-fluoride to be a poor tracer For instance, LeBlanc et al (1991)

found: “Fluoride was abandoned as a tracer early in the test because fluoride concentrations were rapidly attenuated by

adsorption…” in the mineral soils of the site under study Similarly, Jamieson et al (2002) found only 57% recovery

on a fluoride tracer test of a dairy wastewater treatment land in Nova Scotia

wet-Fluorine is taken up by plants to a moderate extent (see Table 11.1), with tissue concentrations typically in the 100–

500 mg/kg range But because it is not a macronutrient or a micronutrient, it is probable that such uptake is driven by the plant water ionic balance The result of limited uptake and sorption is a limited overall reduction of fluoride in treat-ment wetlands (Table 11.5) Data are too sparse to determine whether seasonal effects are present, or to elucidate possible differences between wetland types or plant varieties

TABLE 11.4

Inputs and Outputs of Chlorination Byproducts in the

Tres Rios, Arizona, Demonstration Wetlands (µg/L)

Example Performances of Treatment Wetlands for Fluorine (mg/L)

Imperial, California Agricultural runoff 0.54 0.56 Unpublished data

Alcoa, Tennessee Aluminum waste leachate 5.80 4.90 Gessner et al (2005)

Russelville, Kentucky Aluminum processing 15.4 8.50 Rowe and Abdel-Magid (1995)

408 Treatment Wetlands

11.2 ALKALI METALS

Sodium, potassium, calcium, and magnesium are rarely the

object of regulatory concern, because under most

circum-stances they do not pose any toxicity threat Nevertheless,

each of these has a role in wetland functioning, and can yield

valuable information about pollutant processing and the

wet-land water budget

S ODIUM

Sodium is important in plant and animal physiology Sodium

ions help to regulate osmotic pressure in cells, and therefore

affect the diffusion of all essential growth nutrients between

the external environment and the protoplasm of the living

cells A “sodium pump” fueled by the conversion of

energy-bearing adenosine triphosphate (ATP) maintains internal cell

sodium concentrations at optimal levels The sodium content

of wetland plant aboveground tissues ranges from <0.05% to

more than 1.3% dry weight (Table 11.6) The median across

the 13 species is 0.28%, or 2,800 mg/kg

Because most freshwater wetland species have low

sodium requirements, the dissolved sodium content of

waste-water passing through wetlands changes little (Table 11.7)

Thus, sodium concentrations can be used as a conservative

tracer for calculating dilution and concentration and for

track-ing groundwater discharges from wetlands For instance, the

concentration of sodium in arid land treatment wetlands is

likely to increase during the summer season due to

evapora-tive concentration

Sodium is useful as a marker for added salt (NaCl),

which may enter wastewater treatment systems because of

its use in water softening and road de-icing For example, the

Cumberland County, Pennsylvania, data in Table 11.7, show

a large increase in sodium during a spring flushing event,

which was also accompanied by a large pulse of chloride

(data not shown: Ci = 14 mg/L and Co = 140 mg/L) The gins of the sodium may have been the accumulation of road de-icing salt, contributed by the highway runoff the wetland was designed to treat At the Genoa-Oceola, Michigan, site, discussed in the section on chloride, the water softener salt also had elevated sodium (150 mg/L), compared to the wet-land background of about 5 mg/L The underground plume reached the wetland after about six years, at which time the wetland surface water sodium had increased to 95 mg/L

ori-P OTASSIUM

Ionic pumping maintains potassium levels in plants at centrations of 1.0–4.0% Potassium regulates the open-ing and closing of stomata on plant leaves Stomata are the valves that allow gases inside the plant to be exchanged with the atmosphere Potassium also is used as an enzyme acti-vator in protein synthesis in most cells Potassium typically comprises about 2.6% of the dry weight of wetland plants Potassium concentrations in water of surface flow treatment wetlands are typically between 1.0 and 40 mg/L (Table 11.7), with an average world river concentration of about 3.4 mg/L(Hutchinson, 1975) Potassium has not been the target of treat-ment wetland design In general, there is not much change in potassium from wetland inlet to outlet (Table 11.7)

con-C ALCIUM

Calcium is biologically active because it is used as a ent by invertebrates and vertebrates, and because of its role

nutri-in the carbonate cycle Calcium is required by, and present

in sizeable amounts in, angiosperm plants (Vymazal, 1995) The median concentration in a variety of wetland plants is

TABLE 11.6

Examples of Major Ion Content of Wetland Plants

Plant Type

Sodium (% dw)

Potassium (% dw)

Calcium (% dw)

Magnesium (% dw)

Note: The letter “W” denotes a treatment wetland.

Sources: Data from Boyd (1978) In Freshwater Wetlands: Ecological Processes and Management Potential Academic Press, New York, 155–167; and Vymazal (1995) Algae and Nutrient Cycling in Wetlands CRC Press/Lewis Publishers, Boca Raton, Florida, 1995

Examples of Major Cations Entering and Leaving Treatment Wetlands

HLR (m/yr)

In (mg/L)

Out (mg/L)

In (mg/L)

Out (mg/L)

In (mg/L)

Out (mg/L)

In (mg/L)

Hidden River, Florida Urban runoff 2 3.8 0.477 0.828 0.069 0.106 7.08 8.35 0.094

Norco, Louisiana Refinery, West Cell 1 17.8 360 430 8.4 9.1 140.7 46.7 23.4

Cumberland County, Pennsylvania Highway runoff 1 Event — 28.1 62.7 6.54 7.93 16.6 16.2 1.15

Mor˘ina, Czech Republic, HSSF Municipal sewage 1 9.3 127 102 19 20 98 89 21

Slavošovice, Czech Republic, HSSF Municipal sewage 1 9.8 43 16 36 18 41 17 16

© 2009 by Taylor & Francis Group, LLC

410 Treatment Wetlands

0.77% dry mass (see Table 11.6), and is similar in fresh-

water planktonic algae However, levels in floating plants

and filamentous green algae range upward to 5–7% dry mass

(see Table 11.6; Vymazal, 1995) The photosynthetic organs

of plants and algae may develop calcium carbonate (calcite)

encrustations in hard water environments Because there is

generally an excess of calcium in surface water and

waste-water, calcium concentration does not change appreciably in

many wetland treatment systems (see Table 11.7)

In some treatment wetlands, there is an iron deficiency,

and calcium biogeochemistry is dominant When this occurs,

the wetland sediments contain a high proportion of calcium

carbonate, which is referred to as calcitic mud or marl The

southern Everglades contain extensive areas of these calcitic

muds, which form under conditions of shorter hydroperiod,

as a result of calcium carbonate precipitation mediated by

periphyton These materials are very dense, low in organic

content, and are typically low in phosphorus content

Calcar-eous periphyton in the south Florida environment contributes

to high soil calcium, with concentrations ranging from 3–4%

in peats to 20–40% in calcitic wetland sediments (Reddy et

al., 1991; DeBusk et al., 2004).

Calcium is also important in constructed wetlands

receiv-ing some types of leachates Municipal landfills may contain

construction materials including gypsum wallboard (calcium

sulfate), and the waste piles from phosphate fertilizer

manu-facture contain mostly calcium sulfate as well

M AGNESIUM

Magnesium is an essential micronutrient because of its role

in phosphate energy transfer and because it is a structural

component in the chlorophyll molecule (Wetzel, 1983)

Because magnesium concentration of surface water almost

always exceeds the requirements for plant growth, elevated

magnesium concentrations are not affected when waste-

water travels through wetland treatment systems

Magne-sium is more soluble than calcium, and precipitate formation

does not occur (see Table 11.7) Plant tissue concentrations

are approximately 0.25% dry weight, but may be higher for

floating plants and filamentous algae (see Table 11.6)

11.3 COLLECTIVE PARAMETERS

H ARDNESS

Hardness measures the concentrations of divalent cations in

a water sample The prevalent divalent ions in most surface

waters are calcium and magnesium Rainwater typically has

low hardness (soft water) with a calcium concentration between

0.1 and 10 mg/L, a magnesium concentration of about 0.1 mg/

L, and a hardness value less than 30 mg/L as CaCO3 Surface

water hardness is variable, depending on the soil and rock

con-centrations of calcium and magnesium, and on the degree of

contact with rocks, soils, and pollution Inland surface water

hardness varies from 10 to 300 mg/L as CaCO3, with a calcium

concentration between 0.3 and 70 mg/L and magnesium centration between 0.4 and 40 mg/L

con-T OTAL I ON C ONTENT

Two chemical parameters are commonly used to indicate the collective concentrations of dissolved substances: TDS and specific conductance These parameters do not specify the distribution of contributing ions and organic compounds that contribute, but they are helpful in support of the wetland water budget Further, the TDS content of water is sometimes

a regulated parameter, especially in arid regions, where salt buildup is a water quality concern

Total Dissolved Solids

TDS is used to quantify the degree of pollution in many industrial wastewater effluents, including textile wastes, food processing wastes, and pulp and paper wastes When dis-charged to surface or groundwaters, these dissolved solids may represent a significant pollution source The total quan-tity of dissolved solids in a water sample is measured by filtration followed by sample evaporation This quantity con-tains both inorganic ions and organic compounds TDS is nearly as conservative in wetlands as specific conductance and chloride Because TDS concentrations are high in many wastewaters and the individual components of these solids greatly exceed the biological requirements for growth, wet-lands generally have a negligible effect on this parameter (see Table 11.3)

Electrical Conductivity

EC, also called specific conductance, of an aqueous tion is the reciprocal of the resistance between two platinum electrodes, 1 cm apart and with a surface area of 1 cm2 The reciprocal of EC is equal to resistance, and is a function of the total quantity of ionized materials in a water sample Specific conductance usually is reported at a temperature of 25°C and in units of µS/cm, or µmhos/cm Measurements can

solu-be made with pocket-portable, inexpensive meters Specific conductance is nearly proportional to the TDS in many sur-face waters and is a convenient measure of the salt content of wastewaters

Total ionic salts in wetlands, as measured by specific conductance, may be somewhat altered by biological con-ditions in wetlands, but physical processes of dilution and evaporation represent the major influences Therefore, EC is

a relatively accurate indicator of dilution and concentration effects by rainfall and runoff and evapotranspiration in wet-land treatment systems

Treatment wetlands are usually, but not always, nated by the introduced flows Rainfall and evapotrans-piration are minor in comparison, except possibly for the duration of extreme events Therefore, in the long run, EC of the inlet and outlet waters are close to the same For the 17 pass-through systems of Table 11.3, the outlet EC averages

domi-98% o 2% of the inlet EC This represents long-term mean

performance, over an average of five years Those wetlands

receive an average annual hydraulic loading of 31 m/yr,

which is far greater than precipitation or evapotranspiration,

which is about 1 or 2 m/yr

There are circumstances in which conductivity, chloride,

and TDS change from inlet to outlet, and data then present

a challenge for finding the cause Three such anomalies are

listed in Table 11.3 The first of these is data from the first

17 months of operation of the New Hanover County, North

Carolina, system This FWS wetland treats landfill leachate

that has a high EC (>6,000 µS/cm), and during the start-up

period produced an average outlet EC = 1,468 µS/cm

Chlo-ride and TDS exhibit similar large decreases The reason in

this case is a very low hydraulic loading (0.79 m/yr), coupled

with large rainfall However, the larger effect was the long

time required to replace the low conductivity water used to

initially fill the wetland Both factors produce large dilution

of the incoming leachate

The second illustration is for the Incline Village, Nevada,

wetland This arid-climate system does not receive water in

the summer, because it is used to irrigate fodder crops The

design of the wetland was for disposal, primarily (y90%) to

evaporative losses, and secondarily (y10%) to infiltration

The hydraulic loading is very small, and large

concentra-tion increases in chloride and TDS are produced as the water

moves through the sequence of cells (Kadlec et al., 1990).

The third illustration is the Ouray, Colorado, wetland,

which exhibits twofold increases in TDS over a five-year

averaging period (HDR/ERO, 2001) This is a strong signal

of secondary sources of water entering the treatment

wet-land This has been identified as a maintenance issue: “The

wetland system experiences an outside flow problem with

sulfates, which concentrate.…”

Conductivity has also been used as a diagnostic tool

for internal processes in treatment wetlands in a number

of ways For example, EC is often much higher in the pore

waters of FWS wetland soils and sediments than it is in

over-lying waters DBE (2003) measured the EC of surface waters

and pore waters in Cell 1 of the ENRP in 2001 The pore water EC was higher than that of the overlying surface waters (Table 11.8) Among the potential reasons is that rooted plants extract their transpiration requirement from pore water, but reject some or all of the associated salts The result

is an upward positive gradient in EC in the top soil layer That gradient may continue into the overlying water, and pro-duce stratification of the EC of the water column The pattern

of these results indicates an internal recycle loop, in which dissolved substances are drawn down into the root zone by transpiration or other flows, only to be rejected by the plants

to avoid buildup of TDS within their tissues This creates an upward gradient, which causes diffusive solute movements back into the water column from the pore water

Density-Induced Vertical Stratification

Conductivity is also a tool to understand the phenomenon

of density stratification in constructed and other wetlands Wetlands are typically too shallow to stratify due to thermal gradients, but the same is not true for density segregation, which may exist due to the character of incoming wastewater, rainfall, or added tracers Two examples will serve to illus-trate the potential for vertical stratification

Salt Plumes in FWS Wetlands

Various salts, such as sodium bromide and lithium chloride, are convenient tracers for water movement Considerable quantities are needed, and consequently it is tempting to add concentrated solutions, in order to deal with manageable tracer solution volumes for addition However, when dense solutions are introduced into the bottom of wetlands, there may be a strong energy barrier to vertical mixing, resulting

in the dense material remaining on the bottom of the wetland Concentrations used in a South Florida Water Management District (SFWMD, 2002) tracer study were 7.5% LiCl (density

= 1.04 g/cm3) (Söhnel and Novotny, 1985) Concentrations of sodium bromide in a Tres Rios, Arizona (Whitmer, 1998), tracer study were about 20% NaBr (density = 1.16 g/cm3)

TABLE 11.8

Pore Water Concentrations of Alkalinity, Calcium, and Electrical Conductivity in Cell 1 of the

ENRP Wetland in Florida

Note: EAV = emergent aquatic vegetation; SAV = submerged aquatic vegetation; FAV = floating aquatic vegetation.

Source: Data from DBE (2003) Assessment of hydraulic and ecological factors influencing phosphorus removal in Stormwater

Treat-ment Area 1 West Final report to Florida DepartTreat-ment of EnvironTreat-mental Protection, Contract No WM795, April 2003.

412 Treatment Wetlands

(Söhnel and Novotny, 1985) It is noteworthy that

demonstra-tion of stratificademonstra-tion in limnology laboratory courses utilizes

densities of 1.105 and 1.05 g/cm3 (Wetzel and Likens, 1991)

More directly, the work of Schmid et al (2003, 2004b) shows

that both tracer injections would lead to stable, unmixed layers

of tracer on the wetland bottom Both tracer studies showed

very poor tracer recovery, suggesting that the tracer “got

stuck” in the wetland sediments

Further evidence of vertical stratification was reported

by Chimney et al (2006), for a constructed wetland in the

Florida Everglades (Figure 11.2) In this case, a likely cause

is the back-diffusion of salts from the concentrated pore

water, as described in Table 11.8

Stratification in HSSF Wetlands

The presence of a gravel matrix in a HSSF wetland serves to

exacerbate the potential for vertical stratification It is well

known that flows through clean porous media are

suscep-tible to layering For example, Wood et al (2004) found that

density effects occurred when the invading solution tration was greater than approximately 13,000 mg/L It is not

concen-surprising that Drizo et al (2000) found that tracer bromide

at 20,000 mg/L sank to the bottom of HSSF wetlands Rash and Liehr (1999) have also reported stratification effects in HSSF wetlands

The Grand Lake, Minnesota, wetland exhibited large vertical stratification during a start-up period of two years, as evidenced by much higher EC in the bottom water samples

than in the top samples (Figure 11.3; Kadlec et al., 2003)

These wetlands were initially filled with water from an cent natural bog, which had lower EC than the incoming wastewater The stratification persisted, as a result of the very low flow over the summer and fall of 1996 The stratifica-tion was mitigated in the fall of 1996 by pumping bottom water to the surface, although it restratified the next spring but not as strongly After about two years of operation, only mild stratification existed, with 21% higher EC in the bottom water samples than in the top samples

adja-–70 –60 –50 –40 –30 –20 –10 0

Conductivity (µS/cm)

Emergent SAV

FIGURE 11.2 Vertical stratification of electrical conductivity in the ENRP, Florida constructed FWS wetland Points represent the averages

for 141 dates during May 1995 to October 1997 (Data from Chimney et al (2006) Ecological Engineering 27(4): 322–330.)

0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000

Days

Inflow Outflow Top Bottom

FIGURE 11.3 Electrical conductivity for Grand Lake, Minnesota, HSSF wetland cell #1 Introduction and removal were at the cell bottom

High conductivity water persisted at 45 cm depth for about two years, and low conductivity water persisted at 15 cm depth The influence of

snow melt can be seen in the lower conductivity values on the top in each early spring (Adapted from Kadlec et al (2003) In Constructed

Wet-lands for Wastewater Treatment in Cold Climates Mander and Jenssen (Eds.), WIT Press, Southampton, United Kingdom, pp 19–52.)

11.4 SULFUR

Sources of sulfur include geochemical weathering of

miner-als, wind-blown sea salt, and emissions from fossil-fuel

com-bustion (Wetzel, 1983) Large quantities of sulfur enter the

atmosphere from natural and industrial sources, and return to

earth as acid precipitation containing sulfates (sulfuric acid)

Treatment wetlands receive these atmospheric inputs as well

as sulfur compounds that may be included in the chemicals in

the water to be treated Municipal wastewater contains sulfur

compounds, originating from the potable water supply, and

augmented by waste products Drinking water standards are

strict for sulfide (2 µg/L), but less so for sulfate (250 mg/L)

Hydrogen sulfide is a reactive and toxic gas with problematic

side effects, including a rotten egg odor, corrosion, and acute

toxicity

The processing of sulfur in wetland ecosystems is

rep-resented by interconversions of several sulfur compounds in

the different micro-regions of the ecosystem (Figure 11.4)

Oxidized forms, such as sulfite, sulfate, and thiosulfate, are

found in the oxygenated portion of the FWS water column

Reduced forms, including sulfide, bisulfide, and elemental

sulfur, are found in the soils and sediments under conditions

of low redox potential Ionic and molecular forms are

preva-lent Hydrogen sulfide and methylated sulfur compounds are

volatile, and may be lost from the wetland to the atmosphere

Sulfate is an essential nutrient because its reduced, sulfhydryl

(-SH) form is used in the formation of amino acids Because

there is usually enough sulfate in surface waters to meet the sulfur requirement, sulfate rarely limits overall productivity

in wetland systems

Although seldom a water quality target in its own right, sulfur is an important part of the chemical processing in wet-lands From a treatment perspective, sulfur has a critical role

in the formation and storage of metal sulfides In this tion, the principal reactions of sulfur in the environment are explored, together with the treatment and storage potential.Sulfur concentrations in wetland plant tissues typically range from 0.1–0.6% dry mass, but algal concentrations may

sec-be considerably larger (Table 11.9) Belowground tissues have not often been measured, but are considerably higher than aboveground plant part concentrations Treatment wet-land sediments contain sulfur at 0.1–1.0% dry mass

D ISSIMILATORY S ULFATE R EDUCTION

Aerobic organisms excrete sulfur as sulfate However, upon death and sedimentation, heterotrophic bacteria release the sulfur in the reduced state, which can result in the accumu-lation of high levels of hydrogen sulfide in wetland sedi-ments A second process that transforms sulfate and other oxidized sulfur forms (sulfite, thiosulfate, and elemental sulfur) to hydrogen sulfide in anaerobic sediments, dissimi-latory sulfate reduction, is mediated by anaerobic, heterotro-

phic bacteria such as Desulfovibrio and Desulfotomaculum, which use sulfate as a hydrogen acceptor (Castro et al., 2002;

Aerobic Soil Layer

FIGURE 11.4 Sulfur pathways and forms in FWS wetlands (From Mitsch and Gosselink (1993), Wetlands Second Edition, Van Nostrand

Reinhold Company, New York Reprinted with permission.)

414 Treatment Wetlands

Lloyd et al., 2004) The presence of decaying organic matter

in the wetland sediments and soils depletes oxygen and

cre-ates acid pore waters Organic matter fuels sulfate reduction

Equation 11.1 is favored at low pH, while Equation 11.2

dominates at higher pH As ferrous sulfide (FeS) is highly

insoluble, hydrogen sulfide does not tend to accumulate

until the reduced iron is removed from solution When ion

concentrations are low, or when sulfate and organic matter

concentrations are high, significant hydrogen sulfide

concen-trations can occur Several other metal sulfides are also very

insoluble, including ZnS, CdS, and others

Sulfate is mildly sorbable on soils For example, Fumoto

and Sverdrup (2001) found Freundlich isotherm parameters

of 0.17 < KF < 0.44 [units: (mol/kg) × (mol/L)–n ] and n = 0.078

for mineral soils For water at 20 mg/L, the resultant sorbed amounts were measured in the range 250–500 mg S/kg

H YDROGEN S ULFIDE

Hydrogen sulfide exists in water solution as un-ionized (H2S)

or singly or doubly ionized (bisulfide, HS−, or sulfide, S2−), depending on water temperature and pH The two dissocia-tion reactions are:

Above (mg/kg)

Below

Plants

Sediments

30–50 cm (11 lakes, Poland) 16–227 — 80–2,890 Samecka-Cymerman and Kempers (2001)

Note: Water concentration (mg/L) is for sulfate.

C  oonized hydrogen sulfide concentration, mol//L

first dissociation constant, dimensio

At equilibrium, the un-ionized form is predominant at low

pH, and bisulfide is dominant at high pH in aqueous systems

(Figure 11.5) However, equilibrium is not necessarily attained

in wetland systems, because of continual influxes of sulfate,

and a large number of microbially mediated processes that

may occur In the presence of sulfate in aqueous solution,

oxi-dation prevails at Eh > −300 mV at circumneutral pH (Pankow,

1991) In wetlands, the sulfate reduction zone occupies the

range −200 < Eh < −100 mV (Reddy and D’Angelo, 1994)

Nonetheless, there may be large fractions of un-ionized H2S,

which is volatile and may be lost to the atmosphere

Volatilization of hydrogen sulfide requires mass transport

to the air–water interface, followed by transfer into the air,

and follows rules analogous to those discussed for ammonia

volatilization (see Chapter 9) The Henry’s law constant for

H2S at 25°C is 3.49 mg/L·atm, which indicates large volatility

At other temperatures, the Henry’s law constant, in units of

mol fractions in the liquid and gas, is given by Lide (1992):

¤

¦¥

³µ´

environ-Hydrogen Sulfide in Municipal Wastewater Treatment Wetlands

The Listowel, Ontario, system studied five wetlands for four years, including the H2S content of the incoming and outgoing waters (Table 11.10) (Herskowitz, 1986) The sulfate content was only infrequently monitored, but was typically in the range of 170–200 mg/L Alum additions to wetlands 1, 2, and 3 accounted for 24 mg/L of the incoming sulfate There were marked differences between wetlands of high aspect ratio (Systems 1, 3, and 4), which had lower outlet H2S, and those of low aspect ratio (Systems 2 and 5), which had much higher outlet H2S All systems produced hydrogen sulfide in the warm months

As a point of reference, the rate of emission of H2S was measured in anaerobic ponds in the Mediterranean climate at

Meze, France (Paing et al., 2003) The pond reduced sulfate

0 20 40 60 80 100

pH

Molecular H2S Bisulfide Sulfide

FIGURE 11.5 Aqueous equilibrium concentrations of sulfide and bisulfide at 25°C Based on dissociation constants from Lange’s

Hand-book of Chemistry (1985).

TABLE 11.10 Hydrogen Sulfide in the Listowel, Ontario, Constructed Wetlands, 1980–1984, in mg/L

416 Treatment Wetlands

from 165 mg/L to 57 mg S/L, but produced sulfides, with an

increase from 3.8 to 19.2 mg S/L The H2S emission rates were

found in the range of 20–576 mg S/m2·d (mean = 172) This led

to atmospheric concentrations as high as 5 ppm, which is well

above the human odor threshold of about 0.05 ppm Almasi

and Pescod (1996) also found high sulfides in ponds for warm

(25°C) and cool (10°C) conditions, in the range 15–60 mg/L,

leading to H2S concentrations of 2–12 mg S/L

O XIDATION OF S ULFUR AND S ULFIDES

When it is exposed to air or oxygenated water, hydrogen

fide may be oxidized back to sulfate This may occur via

sul-fur bacteria such as Beggiatoa, which promotes the oxidation

of hydrogen sulfide to elemental sulfur:

Photosynthetic bacteria, such as purple sulfur bacteria, use

hydrogen sulfide as an oxygen acceptor in the reduction of

carbon dioxide, resulting in partial or complete oxidation

back to sulfate:

where CH2O represents organic matter

Under some circumstances, treatment wetlands have

been observed to turn purple, as happened in a FWS

treat-ing high-strength potato processtreat-ing wastewater (P Burgoon,

personal communication)

In any case, these microbially mediated reactions suggest

that elemental sulfur may be found in treatment wetlands

Anecdotal reports of elemental sulfur have been made for the

Houghton Lake, Michigan, system; the Tres Rios, Arizona,

wetlands; the Brighton, Ontario, system; and the Nucˇice,

Czech Republic, HSSF system In extreme situations, a

whit-ish colloid or adhering whitwhit-ish precipitate is seen in the

out-flow from the treatment wetland (Figure 11.6)

Gammons et al (2000a) comment about this

phenom-enon in connection with the Butte, Montana, mine drainage

treatment wetlands:

cells created a foul smell and also resulted in unsightly

pre-cipitates of colloidal sulfur in downstream aerobic waters It

is evident from the above observations that optimal wetlands

performance is in some respects a delicate balancing act

Too much BSR [biological sulfate reduction] activity results

in an undesirable accumulation and release of H 2 S, whereas

too little results in decreased metal attenuation.

Winter and Kickuth (1989b) reported about 36% of the removed sulfur in a HSSF wetland treating textile industry wastewater was stored in the form of elemental sulfur

O RGANIC S ULFUR

Organic sulfur compounds account for a good share of the sulfur found in wetland sediments For instance, 84–88%

of the total sulfur in a New Jersey peat was organic sulfur

(Spratt et al., 1987), and over 90% in a West Virginia peat

(Wieder and Lang, 1988) In treatment wetlands, the storage

of sulfur is also in major part associated with humic als For instance, Winter and Kickuth (1989b) reported about 30% of the removed sulfur was in humic materials

materi-Additionally, there are several low molecular weight organic sulfur compounds that may be found in wastewater Methanethiol (CH3SH) and dimethyl sulfide (DMS or (CH3)2S) are perhaps the most common, and are quite volatile (Faulkner

and Richardson, 1989; Lomans et al., 2002) Both are extremely

odiferous There are several mechanisms that can produce and destroy these volatile organic sulfur compounds (see Lomans

et al., 2002) Kiene and Hines (1995) found both were formed

in natural fen peat at the same rate of 40 nmol/L·d (256 µg/m2·d

in the top 20 cm of soil) Wood et al (2000) measured DMS

removals of 80% (152–28 mg/L) in SSF wetlands treating swine wastewater, and attributed the loss to mineralization and oxidation Domestic wastewater contains lesser amounts

of DMS, and reductions are not significant for anoxic HSSF

systems Huang et al (2005b) found small removals,

averag-ing 20% (2.24 µg/L down to 1.79 µg/L) Their studies involved eight wetlands, with average redox of −35 mV, and reductions

of sulfate of 60% (72.5 mg/L down to 29.4 mg/L)

P HYTOTOXICITY

Lamers (1998) documents that sulfate has negative effects on

the growth rate of Carex nigra, Juncus acutiflorus, and Gallium

FIGURE 11.6 (A color version of this figure follows page 550)

This HSSF wetland outlet structure at Tamarack, Minnesota, has become coated with elemental sulfur.

palustre, at concentrations of 64 and 128 mg S/L Koch and

Mendelssohn (1989) report that 32 mg S/L of sulfide produced

negative effects in Panicum hemitomon and Spartina

alterni-flora The presence of sulfide is coupled with anaerobic

con-ditions in the root zone, but the effects of sulfide go beyond

mere anoxia (Koch et al., 1990) Hydrogen sulfide apparently

inhibits the activity of alcohol dehydrogenase, thereby limiting

the ability of plants to avail themselves of alternative anoxic

energy pathways This effect was confirmed by measuring a

reduced 15N uptake rate in the presence of sulfide However,

the availability of free sulfide is strongly mediated by the

pres-ence of iron, because of the formation of iron sulfides

Phytotoxicity was found to be very serious at the 45-mg

S/L level in Phragmites australis (Armstrong et al., 1996)

These authors found that aeration pathways became blocked,

interfering with the diffusive connection to the atmosphere,

and thus reducing the plant’s ability to oxygenate the

rhi-zosphere Smolders and Roelofs (1996) found for Stratiotes

aloides, an aquatic macrophyte characteristic for

mesotro-phic freshwater marshes, that levels of 320 mg S/L were toxic

to the roots) Lamers et al (2002) found root parts, growing

in 1.7–3.4 mg S/L of sulfate into the peaty sediment, clearly

showed sulfide toxicity by becoming black, slimy, and unfit

for nutrient uptake from the sediment Free sulfide could not

be detected in the surface water They concluded that only

roots in the surface water would survive Nuphar lutea did

not propagate in the sulfate-treated enclosures However, the

sensitivity of a wetland plant species to free sulfide not only

depends on the actual sulfide levels in the rhizosphere, but

also on detoxification mechanisms like radial oxygen loss

As noted above, high sulfide concentrations in freshwater

sediments may also lead to higher fluxes of volatile organic

sulfur compounds to the atmosphere due to the microbial

methylation of hydrogen sulfide

P ERFORMANCE OF W ETLANDS FOR S ULFUR R EMOVAL

Since sulfate inputs in surface flow wetland treatment systems

frequently exceed the biological requirements of wetland

biota, wetlands generally are not as effective for removal of

sulfur as for other contaminants (Wieder, 1989) Although

microbial routes provide for gaseous losses of H2S and DMS,

these require the very low redox potentials usually found

only in deeper wetland sediments Metal sulfide precipitation

often blocks much of the gaseous loss by immobilizing

sul-fides in the sediments Plant storage is minimal, for instance

plant uptake was estimated at 1.5 g S/m2·yr in a natural bog

(Hemond, 1980) Winter and Kickuth (1989a,b) reported only

1% taken up by plants in a HSSF treatment wetland

Conse-quently, the majority of sulfur removal will generally be to

organic, elemental, and metal sulfide forms in the wetland

sediments

As a result, the median-observed concentration

reduc-tion is only 14% for 32 wetlands (Table 11.11) Only a few

mine water wetlands show more than 50% reduction, and

that may be attributed to the anaerobic mode of operation

in some cases Subsurface horizontal flow wetlands also

sometimes satisfy this anoxic condition Winter and Kickuth (1989a, b) reported that a root-zone, soil-based treatment sys-tem receiving textile wastewaters from a facility in Bielefeld, West Germany, removed from 80–85% of the sulfur mass at

an hydraulic loading rate of 1.14 cm/d for a removal rate of 9.6 kg/ha·d These authors reported that the majority of this sulfur was largely stored in the wetland soil as elemental sul-fur (31%) and organic sulfur (25%), and that only a small fraction was released by volatilization to the atmosphere

Huang et al (2005b) observed 24–88% reduction for

hydrau-lic loadings of 2.0–4.5 cm/d, corresponding to fur loadings of 5–10 kg/ha·d In Table 11.11, results from five HSSF constructed wetlands in the Czech Republic are pre-sented The median removal is 51%, indicating anaerobic conditions in the beds with horizontal sub-surface flow How-ever, Vymazal and Kröpfelová (2006) pointed out that despite removal of sulfates, elimination of ammonia may occur at the same time The question remains to be answered whether this removal is due to “conventional” aerobic nitrification, which may proceed in aerobic microzones adjacent to plant roots, or due to Anammox, i.e., anaerobic ammonia oxidation

sulfate–sul-Sulfate removal should not be viewed as a process that

is independent of other wetland chemistry and processes

For instance, studies by Wiessner et al (2005a) determined

that sulfate reduction was strongly dependent on ter biochemical oxygen demand (BOD), presumably acting

wastewa-as a carbon source, in a manner analogous to tion BOD of 200 mg/L led to 100% removal of sulfate Conversely, ammonia reduction decreased from 75 to 25%

denitrifica-as sulfate reduction incredenitrifica-ased Wiessner et al (2005b)

conclude that the importance of the sulfur transformation processes inside the rhizosphere of constructed wetlands, even in the case of treatment of domestic wastewater, has been underestimated Extreme variations of removal pro-cesses in large-scale treatment wetlands may reflect this fact The sensitivity of nitrification, for example, could be due to nutrient or oxygen limitations, but could also have been additionally or exclusively caused by products of sul-

fur transformation Wiessner et al (2005b) suggest that, in

view of the interesting high application potential of simple wetland systems for the removal of metals by sulfide pre-cipitation or the treatment of sulfate-rich wastewaters such

as acid mine drainage, the dynamics of the sulfur cycle in the rhizosphere should be understood in more detail

S ULFUR -I NDUCED E UTROPHICATION

Interactions also may exist between sulfur processing and phosphorus removal in treatment wetlands In some systems, but not all, a fraction of the accreted phosphorus is bound

by iron-containing substances Other fractions are bound in calcium-rich materials, or in organic components of soils and sediments Two effects have been reported: (1) a reduction

of phosphorus uptake due to sulfide toxicity, and (2) sulfide binds iron and interferes with that component of phosphorus storage that relies upon the iron–phosphorus link (Lamers

et al., 1998).

418 Treatment Wetlands

The response of wetlands to new sources of sulfate,

and ultimately sulfide in the sediments, therefore, differs

considerably depending upon the sources and quantities

of iron in the wetland (Lamers et al., 2002) For

exam-ple, a Typha-Carex wetland at Tienhoven near Utrecht,

Netherlands, had relatively high iron, and a continuing

sup-ply from groundwater Addition of sulfate caused oxidation

of iron in response to sulfate reduction, and considerable quantities of phosphorus were released However, a second

wetland, dominated by Nuphar surrounded by Phragmites,

in Weerribben National Park, Netherlands, did not have a groundwater supply of iron Sulfate additions to that wet-land caused large quantities of sulfide formation, but no phosphorus was released

TABLE 11.11

Example Performances of Treatment Wetlands for Sulfate Reduction

Inlet (mg/L)

Outlet (mg/L)

Reduction

Mine Water

Northumberland, United Kingdom Shilbottle 1 8,000 2,000 75 Batty and Younger (2004)

Leachate

Agricultural Runoff

Industrial

Municipal Wastewater

Municipal HSSF

11.5 TRACE METALS: GENERAL

CONSIDERATIONS

T OXIC E FFECTS IN W ATER AND S EDIMENTS

A number of trace metals are essential micronutrients at low

concentrations, but some trace metals may occur in municipal

wastewaters at concentrations that are toxic to sensitive

organ-isms The probability of exceeding sensitive toxicity levels

in mine water, leachates and some industrial waters is much

higher For most trace elements, biochemical transformations

and chemical characteristics can lead to biomagnification, a

phenomenon in which increasing concentrations occur in

con-sumers along a food chain Although most trace metals are

more concentrated in biological tissues and soils than they are

in surface water, hazardous situations do not always occur

Metals in wastewater must be removed prior to final

dis-charge to protect the environment from toxic effects, but the

use of wetlands to accomplish this goal must be examined

cautiously Surface flow wetland treatment systems are open

to biota that may be exposed to potentially dangerous levels

of metals, primarily in the wetland sediments The removal

of metals can result in storage in sediments that is inimical

to the subset of wetland organisms that live or feed in those

sediments To prevent this problem from occurring, wetland

treatment system designers and regulators should consider

pretreatment to reduce influent metal concentrations Deep

water systems with floating plants send sediments to depths

that are out of reach of top feeders A second alternative is to

minimize the opportunity for ingestion of metals Subsurface

flow wetlands accomplish this purpose

In the United States, most treatment wetlands are not

con-sidered waters of the United States, and would therefore not

be required to meet water quality guidelines for the waters they are designed to protect The levels of metals that may be tolerated by sensitive organisms have been promulgated in the form of guidelines for the protection of receiving waters and associated sediments Examples of such guidelines are presented in Tables 11.12 and 11.13 These may or may not

be considered applicable to treatment wetlands, which are not necessarily themselves protected by U.S water quality guidelines

A BIOTIC M ETAL P ARTITIONING

Depositing sediments are capable of adsorbing significant quantities of trace metals directly or indirectly through the accumulation of coatings such as organic matter, iron, and manganese oxyhydroxides, which will in turn act as trace ele-ment collectors Organic matter, which may exist as a sur-face coating or as a particulate, may play an important role

in metal speciation and bioavailability Microbial position of organic matter typically results in sediments that are anoxic under a thin oxic surface layer Under these anoxic conditions, divalent cationic metals such as cadmium, copper, lead, nickel, and zinc are readily form metal sulfides

decom-As long as amorphous sulfide concentrations are in excess of the total trace-metal concentration (on a molar basis), these metals will occur predominantly as insoluble metal sulfides

In anoxic sediments, metals in excess of sulfide may complex with organic matter This buffers organisms against metal toxicity (Doig and Liber, 2006)

An appreciation of transformation rates of divalent ionic partitioning is needed to predict behavior and risk within natural environments Although divalent cationic metals are not redox-active species within soil or aquatic environments,

U.S EPA Human Health (µg/L)

Note: As noted, some are hardness-dependent The criteria maximum concentration (CMC) is an estimate of the highest concentration of a material in

surface water to which an aquatic community can be exposed briefly without resulting in an unacceptable effect The criterion continuous concentration (CCC) is an estimate of the highest concentration of a material in surface water to which an aquatic community can be exposed indefinitely without resulting in an unacceptable effect U.S EPA numbers are provided in U.S EPA (2002a, 2006) Screening quick reference tables (SQRT) are provided

by National Oceanic and Atmospheric Administration (1999).

420 Treatment Wetlands

oxidation and reduction reactions may nevertheless affect

partitioning Retention can be modified by changes in

sub-strate chemistry For instance, zinc sorbs strongly to iron and

manganese oxyhydroxides in aerated systems, and reacts with

hydrogen sulfide to yield zinc sulfide in anaerobic

environ-ments Thus, changes in redox status may shift zinc

partition-ing For example, reductive dissolution of iron and manganese

oxyhydroxides under anaerobic conditions releases zinc into the

aqueous phase; persistence of anoxic conditions may then lead

to a repartitioning of zinc into sulfide or carbonate precipitates

However, slow transformation rates and fluctuation in

condi-tions may alter these predicted phase changes Redox is affected

by wetland water depth, and depth is therefore a partial

surro-gate for redox At a contaminated wetland site in Idaho, these

speciation effects were observed (Table 11.14) (Bostick et al.,

2001) There are associated consequences for metal removal

For instance, for zinc, an increase in water depth from 0.3 to

1.0 m produced a decrease in removal from 38–18% for a fixed

detention time of one day (Gillespie et al., 2000).

where

a C

Langmuir parameter, L/mgmetal concentr

metal concentraS

Langmuir parameter

The maximum capacity for metal is CSmax = K/a, which

is only achieved at high water concentrations The

half-satu-ration water concenthalf-satu-ration, where C S = 0.5CSmax, is equal to

1/a At low water concentrations, CL << 1/a, the Langmuir

relation reduces to a linear partition equation:

Lesage et al (2006) reported that removal of Co, Ni, Cu,

and Zn by gravel and straw could be well described by the Langmuir isotherm with R2q 0.97 for gravel and R2q 0.93 for straw El-Gendy (2006) reported that sorption of heavy

TABLE 11.13

Guidelines for Metal Concentrations in Sediments

Wisconsin TEC (µg/g) PEC (µg/g)

Background Level (mg/L)

Lowest Effect Level (µg/g)

Severe Effect Level (µg/g) SQRT (µg/g)

Note: Wisconsin (WDNR, 2003) levels are a threshold effect concentration (TEC) and a probable effect concentration (PEC)

Ontario guidelines from OMOEE (1994) Screening quick reference tables (SQRT) are provided by National Oceanic and

Atmos-pheric Administration (1999).

TABLE 11.14

Speciation of Sedimentary Zinc as a Function of

Water Depth in a Metal-Contaminated Wetland

Water

Depth (cm)

ZnO (%)

Hydroxide-Sorbed Zn (%)

ZnS (%)

ZnCO3 (%)

Note: Values are for the top 5 cm of sediment cores.

Source: Adapted from Bostick et al (2001) Environmental Science

and Technology 35: 3823–3829.

metals from the landfill leachate by roots of water hyacinth

(Eichhornia crassipes) could be described by both

Lang-muir and Freundlich isoterms with respective correlation

coefficients R2 of 0.94 and 0.93 Organic sediments contain

polar functional groups such as acids and phenolics that are

responsible for cation exchange capacity (CEC), and can be

involved in chemical binding (Ho and McKay, 2000) The

metal-sediment reaction may be represented as:

2

where P represents the organic material (peat).

If Equation 11.14, cation exchange, is presumed to

repre-sent data, then the amount of metal bound is reprerepre-sented by

(Kadlec and Keoleian, 1986):

eq

HH

where the bracket notation denotes molar concentration This

suggests that the partition coefficient ([MP2]/[M2+] = CS/CL =

K) should go down markedly with decreasing pH, which is

in fact the observation of many investigators, as shown in

Kadlec and Keoleian (1986)

E QUILIBRIUM M ETAL C HEMISTRY C ALCULATIONS

There are several computer codes that have been developed to

compute the theoretical thermodynamic equilibria in water

solu-tions For instance, MINTEQA2 is an equilibrium speciation

model that can be used to calculate the equilibrium composition

of dilute aqueous solutions in the laboratory or in natural

aque-ous systems (U.S EPA, 1991) The model calculates the

equi-librium mass distribution among dissolved species, adsorbed

species, and multiple solid phases under a variety of

condi-tions A comprehensive database is included that is adequate for

solving a broad range of problems without need for additional

user-supplied equilibrium constants The model employs a

pre-defined set of components that includes free ions and neutral

and charged complexes The database of reactions is written in

terms of these components as reactants An ancillary program,

PRODEFA2, serves as an interactive preprocessor to help

pro-duce the required MINTEQA2 input files Code to achieve

sim-ilar results, PHREEQC, is available from the U.S Geological

Survey (Parkhurst et al., 1980) These equilibrium calculations

may be used to prepare solubility diagrams for metals in the

presence of a variety of anions (see, e.g., Figure 11.7)

There are, however, substantive discrepancies that occur

between predictions and observed wetland water chemistry

Solubility calculations for a mesocosm of homogenized

sedi-ment indicated supersaturation with respect to the sulfides

of iron, copper, nickel, and zinc, yet measurements

demon-strated a substantial supply of both trace metals and sulfide

from the solid phase to the pore waters (Naylor et al., 2005)

Ratios of metals measured in pore waters were consistent with

their release from iron and manganese oxides, indicating that

supply, as much as removal processes, determines the

pseudo-steady state concentrations in the pore waters The Naylor et al

(2005) observations suggest that trace metals are not ately bound in an insoluble, inert form when they are in con-tact with sulfide As a result of the complex wetland situation, forecasts from computer programs such as MINTEQA2 are not accurate representations of wetland situations For exam-ple, Frandsen and Gammons (2000) found predictions of zinc remaining in solution to be underestimated by several factors

immedi-of ten at low water concentrations, and to be strongly dent upon the assumed form of the solid sulfide (Figure 11.8).The reasons for such large discrepancies are presumably related to the complexities of the treatment wetland envi-ronment, compared to abiotic laboratory experiments with controlled water chemistry The calculations are for thermo-dynamic equilibrium conditions, which may not be satisfied for the dynamic conditions of flow through wetlands Unfor-tunately, it would appear that equilibrium calculations are of little value in predicting treatment wetland performance for metal removal

depen-D ESIGN E QUATIONS FOR M ETAL R EMOVAL

The literature does not currently contain a clear indication of what first-level design calculation should be used for treat-ment wetland sizing for metal removal The two simplest choices are a fixed load removal or a first-order areal calcula-tion The former is advocated in the mine water treatment

literature (Hedin et al., 1994; Younger, 2000; Younger et al.,

2002), and is termed the area-adjusted contaminant removal rate This is essentially a zero-order model, which fixes the removal rate per unit area of the wetland:

–400 –200

–600

PbS PbCO3PbSO4

FIGURE 11.7 Solid phases of lead in the presence of sulfate, sulfide,

and carbonate (From DeVolder et al (2003) Journal of

Environ-mental Quality, 32(3): 851–864 Reprinted with permission.)

Tarutis et al (1999) determined RA values for 35

wet-lands for iron and manganese, and found median and mean

values of 3.5 and 6.5 g/m2·d for iron and 0.24 and 0.73 g/m2·d

for manganese, respectively Younger et al (2002) found RA

median and mean values of 10 and 11.4 g/m2·d for iron (N =

20 wetlands) and 0.10 and 0.28 g/m2·d for manganese (N = 17

wetlands), respectively However, in both studies, the

coef-ficients of variation were unacceptably large, in the range of

100–200% This zero-order uptake model would correspond

to a fixed rate of supply of a precipitating reactant, such as

sulfide The subsequent sections of this chapter show that

metal removal is not constant for most wetlands, but strongly

correlated to the metal loading to the wetland

The second choice, a first-order model, presumes that

more detention time (lower hydraulic loading) will result in

greater metal removal, and that higher inlet concentrations

will result in more removal It is consistent with the concept

of mass transfer of the metal to the reactive sediment layer,

which is driven by the concentration difference between

water and sediment pore-water metals Younger states that

“However, as the design of passive systems advances, it is

likely that volume-based or retention-time based measures

of performance will prove to be more appropriate in many

circumstances.” Tarutis et al (1999) concluded that evidence

showed that the first-order model was a better choice, and

advocated the plug-flow form:

C C

kA Q

where

k areal rate constant, m/d

Data from 35 wetlands gave a mean value k = 0.29 m/d (106 m/yr) for iron, and k = 0.057 m/d (21 m/yr) for manga- nese (Tarutis et al., 1999) Younger et al (2002) reported k = 0.105 m/d (38 m/yr) for iron, and k = 0.061 m/d (22 m/yr) for

manganese, at the Quaking House system

Goulet et al (2001) examined the efficacy of the

first-order model for iron, manganese, copper, and zinc in a water wetland in Kanata, Ontario, over a two-year period These authors found that it was acceptable for some metals (e.g., zinc) but not for others (e.g., iron and manganese) They observed that such a model could not account for releases, and might also be affected by the low concentrations that

storm-existed in the incoming water Crites et al (1997) found

exponentially decreasing profiles for zinc in water along the flow direction at the Sacramento wetlands (Figure 11.9), as indicated by Equation 11.17

One of the premier uses of mesocosm experimentation is

to control external factors, in an effort to elucidate processes

The efforts of Manyin et al (1997) assist in the

understand-ing of which model may be more realistic Side-by-side cosms were fed iron solutions of varying concentrations and at varying flow rates Data fit to Equation 11.17, across three inlet concentrations and four hydraulic loadings, produced a mean

meso-k = 53 m/yr, with R2 = 0.83 Thus, when there are not artifacts

of variable pH, temperature, flow, or ancillary chemistry, the first-order model is quite effective in describing data As a cor-ollary, the use of an area-adjusted contaminant removal rate was entirely inappropriate for these controlled conditions.Another approach to investigation of removal models relies upon the accumulation of metals in new wetland sedi-ments If for the sake of simplicity, the plug flow model

is used as a basis for interpretation, it may be shown that exponential reductions in waterborne metals along the flow path result in the appearance of exponentially distributed

0.001 0.01 0.1 1 10 100 1,000 10,000 100,000

0.001 0.01 0.1 1 10 100 1,000 10,000 100,000

Bisufide Sulfur (µg/L)

Inlet Outlet Amorphous ZnS Sphalerite ZnS

FIGURE 11.8 The reduction in zinc in an anaerobic treatment wetland in Butte, Montana, compared to MINTEQA2 forecasts for

amor-phous and mineral ZnS (Adapted from Frandsen and Gammons (2000) In Wetlands & Remediation: An International Conference Battelle

Press, Columbus, Ohio, pp 423–430.)

top sediment accretions of those metals The buildup of

those new top sediments may be considered the result of

the wetland carbon cycle, which produces a stable annual

accretion for a fully developed wetland When combined

with Equation 11.17, the concentration in the new sediments

can be computed at each location The rate of deposition at

some fractional distance y along the wetland flow path is:

Then a mass balance on the top layer of deposited material

determines the concentration in the surficial solid layer:

J

kC J

kAy Q

y rractional distance along flow path, dimensiionless

A benefit of this analysis is that the temporal

variabil-ity of water concentrations is averaged by the sampling of

many months or years of accretion The Sacramento,

Cali-fornia, project sampled longitudinal profiles after 3.5 years

of operation (Nolte and Associates, 1998b), and designated

the new top sediments as the A layer Interpretation of the

Sacramento data was via the first-order areal model

(Dom-beck et al., 1998) Results indicated that exponential declines

explained a considerable amount of the variability in A layer

solids concentrations (Figure 11.10) Rate constants fall in the

range 22–71 m/yr Thus it appears that the first-order model

is applicable to field situations on a long-term average basis

More detailed models have been proposed, but these

gener-ally lack calibration to multiple systems A significant effort at

detailed process modeling was the development and calibration

of the Constructed Wetland Fate and Aquatic Transport ation (CWFATE) model for the Sacramento, California project (Jones & Stokes Associates, 1993; Nolte and Associates, 1998a, b) CWFATE attempts to describe water budgets, lead biomass cycles, and partitioning, but not chemical precipitation Flana-

Evalu-gan et al (1994) proposed a detailed model, and calibrated it

for iron and aluminum at the Lick Run, Pennsylvania, wetland, but this model has not gained general acceptance The PIRA-MID project developed a proprietary metals model for mine water applications (PIRAMID Consortium, 2003a)

In this chapter, system data analyses are presented for percent concentration reduction, areal removal rates, and first-order rate constants

S TORAGE IN P LANTS

Plants are a secondary location for metal storage, compared

to sediments Furthermore, most of the metal found in plants

is located in the roots and rhizomes (Table 11.15) quently, harvest of aboveground plant parts is not an effective means of removing metals from the wetland

Conse-S EDIMENT S TORAGE C ONCENTRATIONS

A Well-Mixed Surficial Zone

There are two ways to consider the buildup of stored metals

in a treatment wetland The first presumes that metal storage occurs throughout the root zone of the FWS wetland, with transfers to sediments and roots driven by processes such as sorption and transpiration flows

Additionally, wetland sediments contain a variety of organisms that ingest sediments over a wide range of depths, and redeposit their gut contents primarily at the sediment surface (Robbins, 1986) By such means, metals may reach the entire root zone layer, and continue to build up in con-centration As for other contaminants in wetlands, the major storage will be in the roots, soils, and sediments, rather than

in aboveground tissues Thus, this approach considers the root zone to be a well-mixed region, with ever increasing

y = 25.8 exp(–0.0099x)

R 2 = 0.884

1 10 100

Distance (m)

FIGURE 11.9 Profile of water-phase zinc along the flow direction in Sacramento, California, Cell 7B in May 1995 The corresponding

k = 45 m/yr (Data from Crites et al (1995) Removal of metals in constructed wetlands Proceedings of the 68th Annual WEFTEC Conference

Water Environment Federation, Alexandria, Virginia; and Crites et al (1997) Water Environment Research 69(2): 132–135.)

424 Treatment Wetlands

solid concentrations The metal load removed contributes

to the increase, in simple form, as:

of copper was 132 mg/m2·yr The top 25 cm of the wetland contained 50 kg/m2 of dry solids (at Rb = 200,000 g/m3), and hence the rate of concentration increase was 2.65 mg/m2·yr Therefore, the increase from 27 to 35.7 mg/kg would take 3.3 years Storage lifetimes for other metals ranged from zero (TEL exceeded for baseline condition) to 120 years for lead.This calculation presumes vertical uniformity in the root zone, which is not typically observed Further, it does not include the effects of new sediment deposition, which is very common in FWS treatment wetlands Accretion, precipita-tion of metals, and downward diffusion with sorption all tend

to create layering in the upper soil sediment horizon

Managed Peat Systems

When metal concentrations are high, accumulation on ments is also high for many metals Under high loadings, the

sedi-TABLE 11.15

Fraction of Removed Metal Load Found in Plants

after Five Years of Operation

Source: Adapted from Nolte and Associates (1998b) Sacramento

Regional Wastewater Treatment Plant Demonstration Wetlands

Project: Five Year Summary Report 1994–1998 Report to Sacramento

Regional County Sanitation District, http://www.srcsd.com/cw.html ,

Nolte and Associates, Sacramento, California.

0.01 0.1 1 10 100 1,000

Distance (m)

Zn Cu As Ag Cd Hg

FIGURE 11.10 Profiles of metals in sediments along the flow direction for Sacramento, California, Cell 7 in 1997 (Data from Nolte and

Associates (1998a) Sacramento Regional Wastewater Treatment Plant Demonstration Wetlands Project 1997 Annual Report to mento Regional County Sanitation District, Nolte and Associates, Sacramento, California.) The corresponding areal k-values are:

wetland ecosystem cannot generate enough sorption capacity

via its carbon cycle In some instances, the initial amount of

substrate may provide for a long removal lifetime For instance,

Eger and Lapakko (1989) estimated lifetimes for four wetlands

associated with nickel removal at the Dunka Mine site in

Min-nesota At hydraulic loadings of 0.12–0.60 cm/d, and nickel

concentrations of 0.9–15.4 mg/L, they estimated sorption

capacity would last 20–780 years In other cases, there may

be a need to replenish the sorbent on a more frequent basis In

the wetland environment, that means digging out the substrate

and finding appropriate means of disposal Operational and

maintenance costs so incurred must then be considered in the

evaluation of a project The alternative of using the peat as a

sorbent in a controlled, mechanical apparatus should then be

considered (Coupal and Lalancette, 1976; Sharma and Forster,

1993; Brown et al., 2001; Fine et al., 2005).

The “Layer Cake” Assumption

The second way to interpret metal storage in FWS wetlands is

to presume that all the current storage is contained in a newly

formed top sediment layer, which does not mix with

previ-ous layers (Kadlec, 1998) This is a limiting concept, because

there is very likely to be some sediment metal mixing as a

result of bioturbation and vertical chromatographic flows

Vertical stratification of FWS treatment wetland sediments/

soils usually includes a top-most floc layer, which has been

described as “a slurry of dark, decomposing, loosely structured

detrital material that pours out when the sampler is tipped” and

termed the A layer (Nolte and Associates, 1997) This layer has

been observed in a range of treatment wetland types, ranging

from those receiving secondary effluent in Sacramento,

Cali-fornia, to those receiving agricultural runoff in South Florida

(SFWMD, 2006) The thicknesses are about 5–30 cm; for

instance, 11.3 o 2.7 cm along the flow direction at Sacramento

Cell 7 in 1996 Below this floc, there may be a second layer

dominated by litter and sediment, termed the B layer by Nolte

and Associates (1997) This B layer averaged 4.6 o 1.0 cm for

Sacramento Cell 7 in 1996 Below this, there existed a C layer

of gleyed clay soil at Sacramento, in turn atop the base clays

of the wetland basins The metals content of these various ers were found to be markedly different, with 2–13-fold higher

lay-concentrations in the A layer than the C layer (Figure 11.11) It

is probable that the layering pattern in concentrations is driven

by processes of precipitation and sorption

Metal storage in the top layer creates sustained sediment concentrations that reflect the dilution of new metal deposits

by the accretion of new wetland solids Such accretions result from microbial, algal, and macrophyte detritus production, and lead to replacement of floc layers, which are depleted by consolidation In a sense, this concept amounts to plug flow

of metals and biomass downward into the top sediments, or

“conveyor-belt” deposition Under this assumption, a mass balance on the top layer of deposited material determines the concentration in the surficial solid layer:

J

S metal

sed

where

C J

S meta

top layer solid concentration, mg/g

J ssediment accumulation rate, g/m ·yr2

For example, consider Jsed = 200 dry g/m2·yr of new ments, originating from 1,000 dry g/m2·yr of biomass

sedi-production Suppose the metal removal rate is Jmetal = 10 mg/

m2·yr When this metal stream is diluted by the new ment stream, a new, top sediment concentration of 50 µg/g (0.05 mg/g) is forecast This may then be compared to the appropriate sediment quality criterion for the metal in ques-tion This example may be continued to estimate the corre-sponding water concentration:

k

For the example under consideration, if the value of k =

50 m/yr, then C = 0.2 mg/m3 (µg/L) From the magnitudes of

0.01 0.1 1 10 100 1,000

Layer A Layer B Layer C

FIGURE 11.11 The vertical layering effect on average sediment metal concentrations in the first 100 m of Sacramento, California, Cell 7

(Data from Nolte and Associates (1998a) Sacramento Regional Wastewater Treatment Plant Demonstration Wetlands Project 1997 Annual

Report to Sacramento Regional County Sanitation District, Nolte and Associates, Sacramento, California.)

426 Treatment Wetlands

the numbers in this hypothetical example, it is easily seen that

if the wetland is effective, and there is elevated metal in the

incoming water, it will be difficult to meet stringent sediment

quality standards (Kadlec, 1998)

11.6 THE OXIDE FORMERS

I RON

Iron is a metal that may occur at trace to high concentrations in

wetland surface waters and sediments It is required by plants

and animals at significant concentrations In plants, iron is an

essential element in chlorophyll synthesis, cytochromes, and

in the enzyme nitrogenase In animals, iron is important in

oxidative metabolism and is a key component in hemoglobin

Iron at low to moderate concentrations is not generally

regarded as a threat to human health or aquatic life The U.S

EPA has recommended a continuous concentration criterion of

1,000 µg/L for protection of freshwater aquatic environments,

and a drinking water human health criterion of 300 µg/L

(U.S EPA, 2002b) The province of Ontario, Canada, has a

lower standard for protection of aquatic life, also 300 µg/L

Perhaps the greater concern is for the blanketing effect of

thick deposits of iron precipitates in wetlands designed to

treat high iron concentrations (Kelly-Hooper, 1999)

High concentrations of soluble iron in surface water and

wetland systems may result from natural or artificial iron

sources, typically as seeps of ferrous iron and iron sulfides

(pyrites) from anaerobic groundwaters Iron bacteria that

pro-duce ocher, such as Leptothrix ochracea and Spirophyllum

ferrugineum, derive their energy needs from the oxidation of

reduced iron These bacteria typically occur in wetland areas

where anoxic waters meet aerated surface conditions, such

as upwelling springs or other venting groundwaters At such

locations, reddish brown flocculent deposits form

Iron in the wetland waters may be dissolved or

partic-ulate Most reported wetland studies do not specify which

forms were determined This may be a critical unresolved

issue, because there may be very large amounts of suspended

iron in wetland waters Gammons et al (2000b) report that

the iron concentrations between filtered and unfiltered

sam-ples can differ by a factor of up to 100 Their data may be

approximated as unfiltered iron equal to the square root of

filtered iron, over the range 10–5,000 µg/L

Wetland Storage and Processing of Iron

In an oxygenated environment, ferric iron is present as

insol-uble oxyhydroxides, denoted as FeOOH If there is not

suf-ficient alkalinity in the water, the reaction produces acidity:

However, if there is sufficient alkalinity, removal of iron to

precipitates is not accompanied by a decrease in pH:

31

12

The water is deemed acidic for iron removal if the ratio

of iron to CaCO3 alkalinity is greater than 1.1 (Younger et al.,

2002)

Oxidation and reduction of iron occurs relatively ily depending on redox potential and pH (Faulkner and Richardson, 1989) Fe3+, or ferric iron, is the dominant form under oxidized conditions (Eh > 0 at pH q 6.5) Fe2+, or fer-rous iron, is the dominant form under reduced conditions

eas-in wetlands and other aquatic environments Fe3+ forms stable complexes with a variety of ligands It joins with the hydroxide ion in surface waters to form reddish-brown ferric hydroxide (Fe(OH)3), which is also known as ocher Ocher

is insoluble and either settles to the bottom sediments or remains in suspension, adsorbed to living and dead organic matter (see Figure 7.10) Other important compounds formed

by ferric iron include ferric phosphate (FePO4), iron-humate complexes, and ferric hydroxide-phosphate complexes.Ferric iron is reduced to the ferrous form under anaerobic conditions The ferrous iron is more soluble, resulting in the release of dissolved iron and associated anions such as phos-phate from anaerobic sediments in wetlands The formation

of this soluble ferrous iron may be controlled somewhat by sulfide, which forms the relatively insoluble ferrous sulfide (FeS) Sulfide formation is written as:

The required HS– is microbially generated, and occurs preferentially in organic environments by the reduction of sulfate (see Equations 11.1, 11.2, and 11.3)

The role of sulfate-reducing bacteria (SRB) in the cycling

of iron and sulfur was studied in a young constructed wetland

located in Kanata, Ontario, Canada (Fortin et al., 2000)

Sedi-ments and water samples were collected over the course of one year within each of three FWS cells SRB populations were largest during the cold winter months, when the temperature

of the water was 1°C The presence of high-SRB populations also corresponded to highly anoxic conditions within the sedi-ments and to a decrease of sulfate concentrations, suggesting that cold temperature did not affect the activity of SRB The results indicated that iron and sulfur cycling in the constructed wetland was active throughout the year, especially in the cold winter months This suggests that iron removal in wetlands can be effective in temperate climates, even though the tem-perature of the water decreases drastically during the winter

Soils and Sediments

Wetland soils can contain large amounts of iron, especially when exposed to metalliferous waters (Table 11.16) On a dry basis, ferric oxide contains 70% iron by weight (700,000 mg/kg), and this represents an upper limit to sedimentary iron concentrations in oxic wetland waters Iron sulfides contain 53% (FeS2) and 66% (FeS) iron Such mineral precipitates are diluted by newly formed organic materials in the wetland

environment, and lesser concentrations are observed For

ins-tance, Doyle and Otte (1997) measured 6,000–40,000 mg/kg,

and higher values in the rhizosphere and near worm burrow

walls

Freeze-coring and analysis of the wetland substrates

indi-cated that total sulfur was present in three forms, in the following

proportions (Younger, 2000): FeS: 35%; FeS2: 31%; S°: 34%

On the basis of these observations, it was postulated that

pH rise was due to the consumption of protons via reactions

involving reduction of ferric hydroxide and precipitation of

elemental sulfur The removal of iron from solution and

for-mation of significant quantities of S° is consistent with the

following coupled reactions:

Goulet and Pick (2001) found that the presence of cattails had

little effect on the partitioning of iron in shallow wetland

sed-iments in FWS wetlands Studies at four Ontario treatment

wetlands showed total iron in the sediments of 2,000–12,000

mg/kg, with sediment organic content of 8–20% About half

of the sediment iron was in reactive forms, oxides,

monosul-fides, or sorbed on organic matter The balance was

domi-nated by forms associated with either the pyrite (one wetland)

or the silicate fraction of the sediment (three wetlands)

Wetland Plants

Metals reach plants via their fine root structure, and most

are intercepted there Some small amounts may find their

way to stems, leaves, and rhizomes Upon root death, some

fraction of the metal content may be permanently buried, but there are no data on metal release during root decomposition However, wetland plants bring oxygen to their root zone to maintain respiration, and some fraction is lost by radial dif-fusion away from the roots This creates small aerobic zones near the roots, in which iron precipitates may form These

are termed iron plaque.

Nonetheless, some iron is taken up into aboveground sues Iron occurs in wetland plants at concentrations rang-ing from about 200–2,000 mg/kg dry mass (Vymazal, 1995) Plant roots contain a much higher concentration of iron than stems or leaves (Table 11.17) Uptake by plants and algae may

tis-be for purposes of growth enhancement, or at higher metal concentrations for protective purposes Biomagnification of iron does not occur

A common concept of wetland treatment is the perceived risk of seasonal release of contaminants during winter, when wetland macrophytes die back This theoretical risk was investigated experimentally in mesocosm experiments on plant litter collected from long-established mine water treat-ment wetlands in the United Kingdom (Batty and Younger, 2002) Metals were not released from the plant litter; and iron concentrations in the litter increased after 6 months of decomposition, which was attributed to adsorption Field studies undertaken within the PIRAMID project (PIRAMID Consortium, 2003a) found that wetlands were net sinks for iron in all seasons

Performance of Wetlands for Iron Removal

Wetlands interact strongly with iron in a number of ways, and thus are capable of significant metal removal Three major mechanisms are operative:

TABLE 11.16

Iron Content of Top Sediments in a Variety of Wetlands

Iron

Panel, Ontario Cattail marsh Urban stormwater 12,000 Goulet and Pick (2001)

Monahan, Ontario Cattail marsh Urban stormwater 2,500 Goulet and Pick (2001)

Falconbridge, Ontario Cattail marsh Acid mine drainage 1,500 Goulet and Pick (2001)

Riverwalk, Ontario Cattail marsh Tailings leachate 1,500 Goulet and Pick (2001)

West Page Swamp, Idaho Cattail, Arrowhead Tailings leachate 115,500 DeVolder et al (2003)

Widows Creek, Alabama Cattail, Juncus Tailings leachate 60,000–85,000 Ye et al (2001a,b)

Show Low, Arizona Pintail marsh Municipal 30,575 NADB database (1998)

Tres Rios, Arizona Four wetlands Municipal 18,857 Wass, Gerke, and Associates (2002)

Sacramento, California Seven wetlands Municipal 17,096 Nolte and Associates (1998)

Champion Paper, Florida Pilot Pulpand paper 9,400 NADB database (1998)

Monroe Co., New York Pilot Leachate 2,560–2,720 Eckhardt et al (1997)

Poland 11 lakes Brown coal pits 115–21,500 Samecka-Cymerman and Kempers (2001)

3 Uptake by plants, including algae (however, after

the plant senescence or algal die back most iron is

leached out)

Information on the effects of wetlands on iron

concentra-tions has been reported at a low level of detail, with emphasis

on removal percentages Three principal categories of influent

waters are mine drainage, landfill leachates, and municipal

wastewater Mine drainage control dominates the applications,

with well over a hundred systems in place and reporting data by

1990 (Weider, 1989) Many landfill leachates contain enough

iron to induce monitoring of wetland systems built for

leach-ate improvement Iron is not typically monitored in municipal

wastewater wetlands, and there are but few data sets that exist

Example performances are shown in Table 11.18 Inflow

and outflow concentrations vary over 1,000-fold Removal

percentages are high for mine waters and leachates, but lower

or negative for waters with low iron content SSF systems,

due to prevailing anoxic/anaerobic character of the filtration

bed, usually do not perform well in terms of iron removal

The filtration bed becomes more anaerobic over the period

of operation and majority of iron is reduced to more soluble

ferrous compounds which are washed out of the system

If hydrogen sulfide is present as a consequence of sulfate reduction (Equations 11.1 and 11.2), iron may form insoluble ferrous sulfides (Equation 11.27) and is deposited in the bed

Coal Mine Drainage Wetlands

Information on iron removal in wetlands is available ily from acid mine drainage (AMD) wetland treatment sys-

primar-tems in the United States (Girts et al., 1987; Kleinmann and

Hedin, 1989; Hedin, 1989) However, treatment wetlands also became widely used in the United Kingdom in the 1990s Younger (2000) lists 24 wetland systems, which fall into two principal categories:

Aerobic FWS wetlands, usually vegetated by Phragmites australis These are typically used for

iron reduction in net-alkaline waters

FWS with an anoxic compost substrate, typically vegetated by Phragmites australis Because of the

organic substrate, these are called anaerobic FWS wetlands These are typically used for iron reduc-tion in more acidic waters

Of the 137 AMD wetland treatment sites reviewed by Wieder (1989), 66% had influent iron concentrations less than

50 mg/L An average total iron concentration of 60.6 mg/L was reduced to an average outflow concentration of 15.4 mg/L,

•

•

TABLE 11.17

Iron Content of Above- and Belowground Plant Parts in a Variety of Wetlands

Above (µg/g)

Below

TVA Mussel Shoals, Alabama FWS CW Typha spp. 45–142 1,011–7,437 NADB database (1998)

Widows Creek, Alabama FWS CW Typha latifolia 1,217 68,469 Ye et al (2001a,b)

Coeur d’Alene, Idaho FWS Nat Typha latifolia 200 — DeVolder et al (2003)

11 lakes, Poland FWS Nat Typha angustifolia 350 — Samecka-Cymerman and Kempers (2001) TVA Mussel Shoals, Alabama FWS CW Phragmites australis 112–161 2,533–4,547 NADB database (1998)

Nucice, Czech Republic SSF CW Phragmites australis 139 — Vymazal and Krása (2005)

New York FWS CW Phragmites australis 618–799 7,060–9,280 Eckhardt et al (1997)

Br˘ehov, Czech Republic SSF CW Phragmites australis 74 3,677 Vymazal et al (2006)

11 lakes, Poland FWS Nat Phragmites australis 1,053 — Samecka-Cymerman and Kempers (2001) TVA Mussel Shoals, Alabama FWS CW Phalaris arundinacea 89–309 2,445–8,352 NADB database (1998)

Nucice, Czech Republic SSF CW Phalaris arundinacea 323 — Vymazal and Krása (2005)

Br˘ehov, Czech Republic SSF CW Phalaris arundinacea 70 3,383 Vymazal et al (2006)

11 lakes, Poland FWS Nat Phalaris arundinacea 1,202 — Samecka-Cymerman and Kempers (2001) TVA Mussel Shoals, Alabama FWS CW Scirpus acutus 47–107 1,820–2,754 NADB database (1998)

TVA Mussel Shoals, Alabama FWS CW Scirpus cyperinus 83–723 1,185–2,228 NADB database (1998)

11 lakes, Poland FWS Nat Scirpus lacustris 430 — Samecka-Cymerman and Kempers (2001) Widows Creek, Alabama FWS CW Juncus effusus 320 41,318 Ye et al (2001a,b)

Coeur d’Alene, Idaho FWS Nat Sagittaria latifolia 220 — DeVolder et al (2003)

Note: FWS = free water surface; SSF = subsurface flow; CW = constructed wetland; Nat = natural wetland

for an average iron removal efficiency of 58.2% and a median

value of 80.9% (Wieder, 1989) This median is clearly also

close to that experienced at the U.K sites (Table 11.18) This

treatment efficiency was found to weakly correlate directly

with wetland area and inversely with wetland depth An

aver-age iron removal rate of 3,650 kg/ha•yr at pH 6 in AMD

treat-ment wetlands was given by Kleinmann and Hedin (1989)

Because the iron removal rate is correlated with iron

load-ing rate (Hedin and Nairn, 1990), lower removal rates are

expected at low influent concentrations

Landfill Leachate Wetlands

Scrap iron items discarded to landfill create an underground

source of iron, which then appears as a constituent of

leach-ate Raw leachate concentrations may be as high as 500 mg/

L, mostly in soluble form because of the anaerobic condition

of the water in the pile (McBean and Rovers, 1999) At such

high concentrations, iron precipitates pose a serious

clog-ging threat even for FWS wetlands Consequently, aeration

and precipitation are often included as pretreatment steps

(Hoover et al., 1998; Loer et al., 1999)

Oxidation/precipita-tion basins are open water impoundments designed to

pro-vide aeration for precipitation of aqueous iron, detention

time to settle precipitates, and storage volume for lating precipitate sludge These basins are a key component

accumu-in passive systems where iron is present A detention time of

at least 24 hours is recommended to produce a clear water

discharge (Hoover et al., 1998) The leachates in Table 11.18

all had lower iron than that in raw leachates due to aeration basin pretreatment or dilution

FWS wetlands produce considerable further reduction

in iron, with a median removal of 85% for the six systems

in Table 11.18 Subsurface systems may also be used For

instance, Surface et al (1993) measured an iron removal

effi-ciency of 78.6% in a HSSF wetland planted with common

reed Bulc et al (1997) reported an iron removal of 80% in

a landfill leachate HSSF system in Slovenia with outflowing concentration of 10 mg/L

Other Water Sources

A variety of other wastewaters have been subjected to wetland treatment, and in some few cases the iron content of inflows and outflows has been measured This is usually ancillary monitor-ing, which does not target regulatory requirements Byekwaso

et al (2002) found the iron reduction of 27% in a FWS-HSSF

system in Uganda treating cobalt extracting wastes The iron

TABLE 11.18

Removal of Iron in Constructed Wetlands

Inlet (mg/L)

Outlet (mg/L)

Reduction (%)

Removal (g/m 2 ·yr) Reference FWS

Eleven Systems United Kingdom Coal mine water 19.6 2.53 87 — Younger (2000)

Ten Systems Tennessee Valley Coal acid mine 58.3 9.36 82 1,028 Brodie (1990)

Albright Pennsylvania Coal mine water 2.45 0.33 87 56 Hoover et al (1998)

Springdale Pennsylvania Coal mine water 12.46 0.27 98 436 Hoover et al (1998)

Musselwhite Ontario Metal mine water 0.155 0.038 75 22 Bishay and Kadlec (2005)

Kanata Monahan Ontario Metal mine water 0.36 0.27 25 11 Goulet and Pick (2001)

Elliot Lake Panel Ontario Metal mine water 15 0.3 98 141 Goulet and Pick (2001)

New Hanover County North Carolina Leachate 1.746 0.134 92 1.3 Unpublished data

Estevan Saskatchewan Municipal lagoon 0.296 0.197 33 1 Unpublished data

Twelve Systems Arcata, California Municipal lagoon 0.60 0.55 8.3 2 NADB database (1998)

Six Systems Champion, Florida Pulp and paper 0.330 0.530 NADB database (1998)

Orange Eastern Orlando, Florida Municipal 3° 0.017 0.067 NADB database (1998)

Tres Rios Phoenix, Arizona Municipal 3° 0.200 0.233 Wass, Gerke, and Associates (2002)

HSSF

Median HSSF

430 Treatment Wetlands

content of pretreated municipal wastewater is likely to be less

than 1 mg/L in many regions, although higher in regions with

high occurrence of iron minerals (Table 11.18) For example,

Vymazal (2003) reported reduction of Fe from 1.1 to 0.72 mg/L

in a HSSF system treating municipal sewage near Prague in

the Czech Republic However, other studies from the Czech

Republic (Table 11.18) showed rather negative removal effects

for HSSF systems due to conditions discussed above When

iron concentrations are very low, the use percent removal is no

longer an appropriate measure, because very small excursions

in concentration can greatly affect the percentage

Nonethe-less, removals may or may not occur for very low iron content

waters (see Table 11.18)

Examples and Models for Iron Removal

in Treatment Wetlands

It is likely that wetland removal performance in FWS

sys-tems is area-specific (Younger et al., 2002), rather than

vol-ume-specific Three simple predictive approaches have been

suggested:

Constant percent removal Specification of a

sup-posedly constant percentage removal

Zero-order removal Specification of a supposedly

constant areal removal rate, in g Fe/m2·d This

cor-responds to zero-order removal kinetics, for which

removal is independent of the iron concentration

First-order removal The reduction rate is

pro-portional to the iron concentration (Tarutis et al.,

1999) Some of the contributing processes are

first-order, such as oxidation of ferrous iron and

bacte-rial sulfate reduction (Younger et al., 2002), as is

the process of particulate settling of precipitates

However, global removal does not necessarily

fol-low such a model

Data from 35 FWS natural wetlands receiving inputs of

fer-ruginous acid mine drainage in western Pennsylvania served

as a basis for evaluating the merits of these three approaches

•

•

•

Stark and Williams (1995) found that percentage removal was

a better index of treatment than areal removal Nevertheless, a fixed areal removal rate of 10 g Fe/m2·d remains the accepted design guideline for high levels of removal, while 20 g Fe/m2·d

is allowable to cause considerable improvement (Hedin et al.,

1994; PIRAMID Consortium, 2003a)

Tarutis et al (1999) further analyzed the 35-wetland

data set, and concluded that percentage removal should not

be used, and that areal removal did not separate the effects

of inlet concentration and flow rate They recommended the first-order model (Equation 11.17), and found a median rate constant of 0.18 m/d (66 m/yr) (N = 35) for calibrations of a plug flow version of the model

Manyin et al (1997) performed a factorial experiment

varying flow rate and inlet concentrations to wetland cosms They found that outlet iron concentrations increased with increasing inlet iron loadings Analysis of their data pro-duces a rate constant of 42 m/yr according to Equation 11.17.More complex models have been proposed, but these have not yet reached a point of usefulness in wetland design

meso-For instance, the proposed model of Flanagan et al (1994)

presupposes that removal is to sulfides, and contains no mentation step for ferric oxyhydroxides Modeling software has been developed allowing simulation wetland systems (NOAH2D) (PIRAMID Consortium, 2003b) NOAH2D sim-ulates overland flow and solute transport through reed beds, allowing for frictional resistance to flow offered by wetland vegetation, and also allows for exchange with (and hyporheic flow within) permeable wetland bed sediments This code

sedi-is a research tool, rather than software available for design The complexity of the NOAH2D code requires very long run times, and full calibration is acknowledged as unlikely by the

authors (Younger et al., 2002).

The Manyin et al (1997) findings suggest the use of a

graphical representation of wetland effluent iron versus inlet iron loading These variables are completely independent, and avoid the artifact of spurious correlations caused by the inclu-sion of a common factor in both ordinate and abscissa Infor-mation from 46 constructed wetlands was used to form such a

0.01 0.1 1 10 100

Iron Load In (g/m 2  yr)

Other Coal Mine Water Leachate

Manyin et al Mesocosms

FIGURE 11.12 Outlet iron concentrations from FWS wetlands at various loading rates.

display (Figure 11.12), together with the Manyin et al (1997)

mesocosm data There are two zones apparent on this graph:

above about 100 g/m2·yr loading, there is an increase in outlet

iron with increased iron loading; below about 100 g/m2·yr

loading, there is a much more gradual increase in outlet iron

with increased iron loading The 100 g/m2·yr loading also

represents the maximum that may be imposed without

wet-land effluent concentrations exceeding 1 mg/L The k-values

previously cited were not determined or verified for the low

loading region

In the low loading region, behavior probably reflects

localized recycling of iron to and from precipitates and

adsorbed forms, in response to localized variations in pH

and mediated by diffusion to and from the water column

Background levels may also result from suspended

particu-late forms of iron Uptake into aboveground macrophyte

plant parts is low, but larger amounts of iron are found in

roots, so that overall plant cycling can be of importance

Batty and Younger (2002) found that where dissolved iron

concentrations in wetland waters were at or below 1 mg/L,

direct uptake of iron by plants could account for 100% of

iron removal This finding explained why aerobic reed-beds

removed dissolved iron at far greater rates than would be

anticipated on the basis of the first-order kinetics of Fe2+

oxi-dation Batty and Younger (2002) also found 1 mg/L iron was

also an optimum for healthy growth of Phragmites australis:

at greater concentrations, the plants were not as productive,

while at lower concentrations were healthy

Variability of Iron Removal in Treatment Wetlands

Intersystem variability in the Stark and Williams (1995)

data sets was relatively high, with removal of 64% o 28%

(mean o SD) It was less in the Manyin et al (1997)

cosms, with removal of 89% o 13% (mean o SD) The

meso-cosms were operated in the laboratory, and had no variability

in depth, aspect ratio, weather conditions, or substrate; and

were all operated at circumneutral pH

Intrasystem variability has not been reported for

continu-ous flow wetlands, but it is high for event-driven wetlands

For example, the coefficient of variation of outlet iron from

the Hidden River wetland in Florida was 4.7 (ASCE, 2003)

Summary

Iron is effectively removed by treatment wetlands over the

high end of the concentration range, which is typical of acid

mine drainage and landfill leachates Rates are rapid, and

significant loadings may be removed At low concentrations,

and loadings below 100 g/m2·yr, iron cycling creates low

con-centrations in the water column Removal is to precipitates,

either oxyhydroxides or sulfides Accumulation of these

materials is a factor in the ecological status of the wetland

A LUMINUM

Aluminum occurs naturally in surface waters, to a small extent

in the hydrated ionic forms, and to a greater extent complexed

with silicates in a colloidal form Aluminum solubility varies with pH It is least soluble at a pH of 7 and increases in solu-bility as Al3+, Al(OH)2+, and Al(OH)2 ions at 4 < pH < 5; and

as aluminate ion Al(OH)4 at pH > 9 (Gensemer and Playle, 1999) Aluminum precipitates as amorphous Al(OH)3, which

may slowly form the mineral Gibbsite (Berkowitz et al.,

2005) Precipitation is rare in natural waters, but is of interest

in treatment processes that rely upon addition of aluminum chlorides or sulfates for purposes of phosphorus removal

Wetland Processing and Storage of Aluminum

Aquatic systems typically contain low concentrations of total aluminum In the Adirondacks region of the eastern United States, 203 lakes had a mean pH = 6.3, and a mean total Al =

138 µg/L In Florida, 168 lakes had mean pH = 6.3 and a mean total Al = 89 µg/L Lakes are typically net sinks for aluminum, with 10–50% retention for low pH, and 70% for circumneutral conditions (Gensemer and Playle, 1999) Organic complex-ation occurs in natural waters, as binding to humic substances Trivalent cations such as Al3+ are more susceptible to binding

to organic ligands than divalent cations Peat based wetlands often provide the conditions of low pH that foster Al3+, and hence dissolved organic matter, or dissolved organic carbon, is

an important factor in wetland aluminum water chemistry.Aluminum is toxic to many species of algae, with effects being observed over a range of a few hundred to a few thousand µg/L total aluminum However, aquatic invertebrates are much less affected, and do not biomagnify aluminum Wetland mac-rophytes are tolerant of high aluminum concentrations, in the thousands of µg/L total aluminum Salmonid fish are suscep-tible to a variety of effects for concentrations of a few hundred

to a few thousand µg/L U.S EPA (1988a) ambient water ity standards are 87 µg/L (chronic) and 750 µg/L (acute), but research is currently in progress to improve these numbers

qual-Aluminum uptake by emergent and floating plants (Typha and Lemna) was found mainly into the roots and rhizomes (Goulet et al., 2005) In contrast, there was no clear pattern for submerged plants (Utricularia and Potamogeton) Roots also

had the highest aluminum concentrations at the Tres Rios, Arizona, treatment wetlands (Table 11.19; Wass, Gerke, and Associates, 2002) Sediments generated in treatment wetlands are often high in aluminum, with values in treatment wetlands ranging from 1.4% (Tres Rios, Arizona) to 4% (Sacramento, California) (Nolte and Associates, 1998b)

The process of phosphorus adsorption onto aluminum hydroxides has seen extensive application to water treatment and lake eutrophication management as a means of control-ling excess dissolved phosphorus Alum or polyaluminum chloride additions are designed to form a floc of insoluble Al(OH)3, which in turn adsorbs phosphorus In wetlands, without coagulation, this floc settles slowly or not at all, leaving particulate aluminum and phosphorus in suspension (Bachand and Richardson, 1999) However, alum addition is

a pretreatment step for waters sent to wetlands for further polishing Wetlands are also the recipients of water treatment

backwash and sludge (Kaggwa et al., 2001) Consequently,

432 Treatment Wetlands

aluminum is sometimes a contaminant in treatment wetland

influents, either by intentional or accidental discharges

Aluminum is a strong determinant of the phosphorus

adsorption capacity in wetland soils (Reddy and D’Angelo,

1994) The phosphorus adsorption process in treatment

wet-lands is a temporary mechanism, which is exhausted when

all antecedent soil sorption sites are used However, the

amount of phosphorus that may be bound to wetland soils

forms an important part of the wetland start-up capacity for

phosphorus removal A phosphorus removal project has been

implemented using aluminum sludge from a water treatment

plant as a soil amendment (SAIC, 2005)

There are two main groups of projects that have

mea-sured aluminum removal in wetlands:

FWS wetlands, which typically receive low

con-centrations and loads of aluminum

SSF wetlands, which, for aluminum, typically

treat acid mine waters with quite high

concentra-tions and loads

Aluminum Removal Processes

Mine waters with pH less than 4 commonly contain high

concentrations of aluminum (>10 mg/L) (Younger et al.,

2002) Aluminum is present as trivalent Al, and the currently

accepted removal reaction is:

3

The precipitate is amorphous, white, and of low density It

may later crystallize to gibbsite These solids are common

components of many natural soils, and pose no problem as

new wetland sediments and soils (Younger et al., 2002) In

the presence of dissolved sulfate, hydroxysulfate precipitates

may form as well

Performance of Wetlands for Aluminum Removal

Examples of wetland reductions in concentration and mass

removals are shown in Table 11.20 Performance is variable

•

•

across systems, with a median concentration reduction of 50% Wieder (1989) examined 20 acid mine wetlands for reductions, and found a median of 47.7%, and a 75th percentile of 78.9% Wieder (1989) also found one quarter of the acid mine wet-lands had zero reduction Vymazal and Krása (2003) reported decrease of aluminum concentration from 451 μg/L to less than 40 μg/L in a 62-m long HSSF constructed wetland in the Czech Republic The major decrease (from 451 μg/L to 65 μg/L) occurred within the first 15 m of the bed Other results from HSSF wetlands treating municipal wastewater (Table 11.20) exhibited good removal of aluminum (53–90%)

Removal is to accretion in sediments, which form a source of potential return fluxes of aluminum, most likely

as particulates In a mesocosm study, Wieder et al (1990)

found that although aluminum was initially removed in a wetland, this process decreased with time, and the wetland

began to export aluminum to downstream waters Wieder et

al (1990) found aluminum concentrations of 10 mg/L were

found to be toxic to cattails, leading to their mortality and

release of sorbed aluminum Wieder et al (1988) found the

total aluminum content of the peat increased from 2,375 to 13,634 mg/kg dry mass, with the majority bound as organic and oxide compounds

Flanagan et al (1994) proposed a model for wetland

treatment of iron and aluminum in acid mine drainage, but its utility has apparently not been tested for aluminum (Mitsch

and Wise, 1998) Regarding aluminum, Younger et al (2002)

conclude that much remains to be investigated in detail

Example Treatment Wetlands for Aluminum Removal

There are but few data sets on the removal of aluminum from types of incoming water other than acid mine drainage (Table 11.20) Bachand and Richardson (1999) conducted field mesocosm studies on aluminum dosing to foster phosphorus removal in treatment wetlands in south Florida The “pin flocs” generated did not settle effectively Another FWS wetland study tested the concept of phosphorus removal via aluminum dosing

of agricultural runoff, but with coagulation before sending the

TABLE 11.19

Aluminum Concentrations in the Tissues of Wetland Plants in a Municipal Wastewater Treatment Wetland

(mg/kg)

Roots (mg/kg)

Reference

Three square bulrush Scirpus olneyi 12 214 Wass, Gerke, and Associates (2002)

Pennywort Hydrocotyle spp. 28 254 Wass, Gerke, and Associates (2002)

Soft stem bulrush Scirpus validus 11 168 Wass, Gerke, and Associates (2002)

Common reed Phragmites australis 39 2,177 Vymazal et al (2006)

Common reed Phragmites australis 103 — Samecka-Cymerman and Kempers (2001)

Reed canarygrass Phalaris arundinacea 1,578 — Samecka-Cymerman and Kempers (2001)

Reed canarygrass Phalaris arundinacea 43 2,584 Vymazal et al (2006)

Narrow-leaved cattail Typha angustifolia 54 — Samecka-Cymerman and Kempers (2001)

Giant bulrush Scirpus lacustris 82 — Samecka-Cymerman and Kempers (2001)

Examples of Aluminum Removals in Treatment Wetlands

In (µg/L)

Out (µg/L)

Reduction (%)

In (kg/ha·yr)

Out (kg/ha·yr)

Removal (kg/ha·yr)

Removal (%) Free-Water Surface

South Florida STC Dosed Al-Dosed agricultural runoff 18,834 19 100 6,187 6 6,181 100 South Florida NTC Dosed Al-Dosed agricultural runoff 31,536 51 100 10,360 17 10,343 100 Friendship Hill, Pennsylvania Acid mine water 50,000 35,000 30 — — — —

Subsurface Flow

Keyser’s Ridge, Maryland Acid mine water 49,600 9,020 82 — — — —

© 2009 by Taylor & Francis Group, LLC

434 Treatment Wetlands

water to FWS wetlands (CH2M Hill, 2001a) Aluminum was

passed to the wetlands in varying amounts, and accumulated

in the sediments Good removal from the water was observed

for the year-plus duration of the study, but a moving sediment

front appeared to have developed Undosed control wetlands

had very low levels of aluminum both entering and leaving

Water treatment alum sludge entered a wetland in Uganda,

and was effectively removed to the sediments (Kaggwa et al.,

2001) Sediment accretion was selectively higher in the inlet

region of the wetland

The aluminum industry employs FWS treatment

wet-lands in Kentucky to treat a mixed effluent from a metal

pro-cessing plant (Rowe and Abdel-Magid, 1995) Over 90% of

the aluminum is removed, from an entering level of 1.5 mg/L

Metal mine water was stripped of small quantities of

alu-minum in a FWS wetland in Ontario (Bishay and Kadlec,

2005), with reduction from 34 to 4 µg/L About two thirds

of the aluminum was removed from urban stormwater by a

FWS in Florida (Harper et al., 1986) Over 90% of the

alu-minum in a municipal wastewater was removed in a HSSF

wetland in the Czech Republic (Vymazal and Krása, 2005)

Another coal industry wastewater is the leachate

gener-ated from combustion products waste piles These waters

have much less aluminum than acid mine drainage (0.6–0.9

mg/L), and only slightly acidic pH Studies supported by the

Electric Power Research Institute (EPRI) in Pennsylvania

found that 70–90% of aluminum was removed in FWS

wet-lands (Hoover et al., 1998; Rightnour and Hoover, 1998).

Based on this limited data, it appears that aluminum

typi-cally is removed from surface water passing through wetlands

M ANGANESE

Manganese is an essential element that is chemically similar

to iron in its behavior in surface waters Manganese is vital

to plant photosynthesis and is used as an enzyme cofactor for

respiration and nitrogen metabolism by plants and animals

Although manganese is toxic to some organisms at elevated

concentrations, this situation occurs infrequently, typically

with mining wastes Manganese concentrations greater

than 2 mg/L were found to be toxic to algae in laboratory

experiments (Goldman and Horne, 1983) Manganese is not

observed to bioconcentrate in the wetland food chain

Manganese may exist in oxidation states ranging from

−3 to +7, but the manganous (+2) and manganic (+4) forms

are the most important in aqueous systems (Wetzel, 1983)

At low redox potentials and low pH the predominant form

is Mn(II) (Figure 11.13) The mineral forms are pyrolusite

(MnO2), hausmanite (Mn3O4), rhodochrosite (MnCO3), and

manganite (MnOOH) Manganous manganese can form

soluble complexes with bicarbonate, sulfate, and organic

compounds Under reducing conditions, manganese forms

insoluble complexes with carbonate, sulfide, and hydroxide

Wetland Processing and Storage of Manganese

In an oxygenated environment, manganese is primarily

removed from solution by oxidation and hydrolysis (Sikora

et al., 2000; Younger et al., 2002):

21

Abiotic oxidation of Mn(II) is very slow, and the process

is generally considered to be biologically mediated, by ria, fungi, and algae Carbonate may be an intermediate, but likely as substituted dolomite (Ca,Mg,Mn)CO3 rather than

bacte-the mineral rhodocrosite (Younger et al., 2002) Because

manganese is soluble at acidic pH, there is not a ity for manganese precipitation in acidic waters As a result, most of the removed manganese is extractable by weak acid

possibil-(e.g., 85%; Wildeman et al., 1993b).

The oxidation of manganese may be inhibited in the presence of large amounts of iron, because iron exerts a pref-erential claim on available oxygen (Hedin and Nairn, 1993):

be above neutral for this process (Figure 11.13) (Wildeman

et al., 1993b).

–200 0 200 400 600 800

MnCO3

Mn3O4MnOOH

FIGURE 11.13 Approximate distribution of manganese species

Second Edition, Academic Press, Philadelphia; and Sikora et al (2000) Water Environment Research 72(5): 536–544 Reprinted

with permission.)

A Freundlich isotherm was fit to sorption data for river

gravel and limestone substrates (Sikora et al., 2000) They

reported n = 0.43 and KF = 22.6 for river gravel, and n = 0.43

and KF = 6.6 (see Equation 11.13)

Manganese is found in wetland plants, algae, and

sedi-ments Concentrations in sediments exposed to mine waters

may be very high, up to 10,000 µg/g For municipal

waste-water treatment wetlands, sediment manganese is typically

200–500 µg/g (Table 11.21) Manganese concentrations in

plant tissues are of the same order of magnitude, and

above-ground values do not differ much from belowabove-ground values

(Table 11.22)

The general conception is that in the aerobic, surface

waters of a wetland, oxidized forms are abundant, Mn(IV),

while in the anaerobic soils and sediments, reduced forms

prevail, Mn(II) (La Force et al., 2002) As a consequence,

both manganese forms in sediments and manganese cycling

are driven by redox patterns For example, at the Cataldo,

Idaho, mine drainage wetlands, winter represented aerobic

conditions, while spring had deeper waters and anaerobic

conditions in the sediments (La Force et al., 2002) As a

result, oxides were 34% of the total manganese in winter, but

only 3% in spring In general, the ratio of oxidized to reduced

manganese species was 1:1 in spring and 1:8 in winter

The storage of manganese in wetlands entails little or no risk

for waters other than acid mine drainage, because sediment

con-centration standards are typically quite high (see Table 11.21)

It is possible to estimate the sediment concentrations created

by sustainable removals of manganese, in terms of the dilution

of the manganese storage, by the accretion of new wetland

sol-ids For example, consider 200 dry g/m2·yr of new sediments,

originating from 1,000 dry g/m2·yr of biomass production

If the sediment concentration is to be kept below some limiting

level, such as the 40,000 µg/g severe effects level of the Ontario

guideline, for that Ontario guideline, deposition of 8 g Mn/m2·yr could be sustainably tolerated Meeting sediment standards on a sustainable basis is easily achieved for all concentrations

Performance of Wetlands for Manganese Removal

Manganese is typically removed in FWS wetlands (Table 11.23) For numerous systems, with inlet manganese ranging from 0.1–38,000 µg/L, the median concentration reduction

is 54% There is an increasing wetland exit concentration in response to increases in wetland manganese loading (Fig-ure 11.14) Outlet concentrations are below 1.0 mg/L for wetlands receiving leachates and other waters, but coal mine drainage wetlands are much more heavily loaded, and can have exit concentrations up to 20 mg/L

Horizontal subsurface constructed wetlands, where tion beds are usually anoxic or anaerobic, may release man-ganese over the period of operation This is due to dissolution

filtra-of manganese oxyhydroxides precipitates as a consequence

of low redox potential If redox conditions become very low and sulfate is present, manganous ions may precipitate with hydrogen sulfide from sulfate reduction to form insoluble sulfides Vymazal and Krása (2003) reported a substan-tial reduction of Mn in a HSSF constructed wetland in the Czech Republic; average inflow concentration of 278 μg/L was reduced to 53 μg/L However, results from three other HSSF systems in the Czech Republic exhibit substantial Mn release (Table 11.23) Vertical flow constructed wetlands due

to higher oxygenation of the bed exhibit good removal of manganese (Table 11.23)

There are two popular methods of interpreting performance data for metal removal in wetlands: the areal load removal and the first-order removal model These are both explored in detail for iron and manganese in a 35-wetland data set by Tarutis

TABLE 11.21

Manganese Content of Top Sediments in a Variety of Wetlands

Panel, Ontario Cattail marsh Urban stormwater 35 Goulet and Pick (2001)

Monahan, Ontario Cattail marsh Urban stormwater 50 Goulet and Pick (2001)

Falconbridge, Ontario Cattail marsh Acid mine drainage 15 Goulet and Pick (2001)

Riverwalk, Ontario Cattail marsh Tailings leachate 55 Goulet and Pick (2001)

West Page Swamp, Idaho Cattail, Arrowhead Tailings leachate 12,500 DeVolder et al (2003)

Widows Creek, Alabama Cattail, Juncus Tailings leachate 200–400 Ye et al (2001a,b)

Show Low, Arizona Pintail and telephone Municipal 521 NADB database (1998)

Tres Rios, Arizona Four wetlands Municipal 314 NADB database (1998)

Champion Paper, Florida Pilot Pulp and paper 285 NADB database (1998)

Sacramento, California Pilot Secondary 206–415 Nolte and Associates (1998b)

Leadville, Colorado Salix and Carex Mine water 9,500–1,790 August et al (2002)

Cataldo, Idaho Typha and Scirpus Mine water 1,600 Hansel et al (2002)

Monroe County, New York Phragmites FWS Landfill leachate 113 Eckhardt et al (1999)

Monroe County, New York Phragmites FWS Landfill leachate 303 Eckhardt et al (1999)

Keyser’s Ridge, Maryland Peat bed Highway runoff 236 Wieder et al (1988)

Lick Run, Ohio Mushroom compost Acid mine 395 Mitsch and Wise (1998)

436 Treatment Wetlands

et al (1999) The former presumes a fixed removal per square

meter of wetland, regardless of inlet concentrations and flows,

and is basically a zero-order model The latter separately

includes the effects inlet concentration and flow rate

A problem of the zero-order model is that it cannot be

transferred from high loadings to low loadings The

recom-mended design removal rates are low: 0.5–1.0 g/m2·d (Hedin

et al., 1994); 0.5 g/m2·d (PIRAMID Consortium, 2003a)

However, this removal far exceeds the inlet loading to

non-mine water wetlands, which is only half that value at a

maxi-mum (Figure 11.14), and removal is not complete for those

lightly loaded systems

Miretsky et al (2004) determined first-order volumetric

rate constants for batch floating plant mesocosms For initial

concentrations of 1–4 mg/L, 87–98% of the manganese was

removed in less than one day for Pistia stratiotes Resulting

values of kV were 5.8 h−1 for Pistia stratiotes and 28.9 h−1 for

Spirodela intermedia Sikora et al (2000) also found very

high plug flow rate constants for SSF systems, with 10–60

m/yr in river gravel wetlands and 100–600 m/yr in limestone

wetlands Tarutis et al (1999) took data at 35 FWS coal mine

drainage wetlands over a one-year period Plug flow rate stants averaged 21 m/yr

con-Plants are a minor repository for removed manganese

The Ye et al (2001a) Widows Creek, Alabama, study mined that Typha latifolia and Juncus effusus contained

deter-just 0.95% of the annual amount of manganese entering the system Vymazal and Krása (2005) found 78% of the added manganese in sediments and belowground plant parts in a HSSF wetland treating domestic wastewater, and 1.73% in aboveground plant parts The majority of removed man-ganese is therefore associated with wetland sediments, in sorbed or chemically precipitated forms Long-term sustain-able removal requires continuing maintenance of oxidizing conditions

Example Treatment Wetlands for Manganese Removal

Quaking Houses, County Durham, United Kingdom (Younger et al., 2002)

Leachate from an abandoned colliery, containing up to 15 mg/L manganese and 30 mg/L of iron, was discharging into

TABLE 11.22

Manganese Content of Above- and Belowground Plant Parts in a Variety of Wetlands

Water (µg/L)

Above (µg/g)

Below (µg/g) Reference

TVA Mussel Shoals, Alabama SSF CW Scirpus acutus 100 19 23 Behrends et al (1996)

TVA Mussel Shoals, Alabama SSF CW Scirpus atovirens 100 10 17 Behrends et al (1996)

TVA Mussel Shoals, Alabama SSF CW Scirpus cyperinus 100 11 14 Behrends et al (1996)

TVA Mussel Shoals, Alabama SSF CW Phalaris arundinacea 100 20 48 Behrends et al (1996)

TVA Mussel Shoals, Alabama SSF CW Phragmites australis 100 28 62 Behrends et al (1996)

TVA Mussel Shoals, Alabama SSF CW Typha spp. 100 12 38 Behrends et al (1996)

Widows Creek, Alabama FWS CW Typha latifolia 7,600 1,097 200 Ye et al (2001a, b)

Coeur d’Alene, Idaho FWS Nat Typha latifolia — 1,300 — DeVolder et al (2003)

Nucice, Czech Republic SSF CW Phragmites australis 278 450 — Vymazal and Krása (2005)

Poland FWS Nat Phragmites australis 502 122 — Samecka-Cymerman and Kempers (2001) Nucice, Czech Republic SSF CW Phalaris arundinacaea 278 41 — Vymazal and Krása (2005)

Poland FWS Nat Phalaris arundinacea 898 619 — Samecka-Cymerman and Kempers (2001)

Widows Creek, Alabama FWS CW Juncus effusus 7,600 312 117 Ye et al (2001a, b)

Coeur d’Alene, Idaho FWS Nat Sagittaria latifolia — 830 — DeVolder et al (2003)

Bielkowo, Poland SSF CW Glyceria maxima — 241 545 Obarska-Pempkowiak et al (2005)

Bielkowo, Poland SSF CW Typha latifolia — 540 — Obarska-Pempkowiak et al (2005)

Bielkowo, Poland SSF CW Phragmites australis — 158 — Obarska-Pempkowiak et al (2005)

Chaco-Pampa, Argentina FWS FAP Pistia stratiotes 1,000 1,319 — Miretsky et al (2004)

Chaco-Pampa, Argentina FWS FAP Spirodela intermedia 1,000 5,038 — Miretsky et al (2004)

Cataldo, Idaho FWS Nat Typha latifolia 135 3,400 790 Hansel et al (2002)

Cataldo, Idaho FWS Nat Phalaris arundinacea 135 1,200 1,500 Hansel et al (2002)

Cataldo, Idaho FWS Nat Scirpus microcarpus 135 1,800 1,900 Hansel et al (2002)

Cataldo, Idaho FWS Nat Equisetum arvense 135 1,500 1,000 Hansel et al (2002)

Monroe County, New York FWS CW Phragmites australis — 265 567 Eckhardt et al (1999)

Note: FWS = free water surface; SSF = subsurface flow; CW = constructed wetland; Nat = natural wetland; FAP = floating aquatic plants.

a stream (Stanley Burn), where it killed all aquatic life in its

path A 45-m2 pilot wetland was found to be effective,

com-prised of a shallow pond with a substrate of stable waste

(manure and straw) Subsequently, a full-scale, four-cell FWS

series wetland was implemented in 1997 The first two cells,

totaling 440 m2, were the heart of the system, with two

follow-on cells regarded as partly cosmetic The two working cells

contained baffles and islands to prevent short-circuiting The full-scale system substrate was a mixture of municipal com-post and manure The average hydraulic loading was about

20 cm/d The reduction in manganese was 26.4% over the first 27 months of operation, from 4.4 to 3.2 mg/L The areal removal rate was 0.26 g/m2·d (950 kg/ha·yr), and the first-order

areal removal rate constant was k = 22 m/yr (0.061 m/d).

TABLE 11.23

Removal of Manganese in Constructed Wetlands

Inlet (mg/L)

Outlet (mg/L)

Reduction (%)

Removal (g/m 2 ·yr) Reference FWS

35 systems Western Pennsylvania Coal mine water 10.0 8.4 16 266 Tarutis et al (1999)

10 systems TVA (Eastern United States) Coal acid mine 9.6 5.0 33 153 Brodie (1990)

124 systems Eastern United States Coal acid mine 37.7 24.0 34 — Wieder (1989)

Musselwhite Ontario Metal mine water 0.103 0.04 61 11.96 Bishay and Kadlec (2005)

6 Systems Champion, Florida Pulp and paper 0.600 0.318 47 5 NADB database (1998)

New Hanover County North Carolina Leachate 0.208 0.026 88 0.1 Unpublished data

Estevan Saskatchewan Municipal lagoon 0.113 0.175 55 -0.66 Unpublished data

HSSF

VF

Wapserveen 1 The Netherlands Municipal + dairy 0.22 0.02 91 0.11 Unpublished data

438 Treatment Wetlands

#40 Gowen, Gaines Creek Watershed,

Oklahoma (Nairn, 2003)

Acid mine drainage, in the form of an artesian flow, was

add-ing unacceptable quantities of acidity, iron, and manganese

to Gaines Creek In 1998, the University of Oklahoma

con-structed a four-cell treatment wetland system to demonstrate

the ability of a passive system to treat this discharge The first

and third cells were vertical flow, with a layer of compost

above a limestone drainage layer (see Figure 11.15) The

sec-ond and fourth cells were horizontal aerobic units, and all were

planted with Typha latifolia A flow of 20 L/min (29 m3/d)

was directed to the wetlands, each of which had an area of

185 m2, resulting in a hydraulic loading of about 4 cm/d The

incoming water had pH = 3.4, Fe = 250 mg/L, Al = 36 mg/L,

and Mn = 14 mg/L At the system outlet, over the 8-month

study period, the pH was increased to 7.7, and Fe = 0.77 mg/L,

Al = 0.05 mg/L, and Mn = 5.8 mg/L The manganese load

removed was 3,796 kg/ha·yr (1.04 g/m2·d) Observations

dur-ing water quality sampldur-ing events indicated considerable

wildlife use of the treatment wetlands Several species of

amphibians (e.g., bullfrogs, leopard frogs, and salamanders),

reptiles (e.g., snapping turtles, garter snakes), birds (e.g., red winged blackbirds, killdeer, great blue herons), and mammals (e.g., moles, voles, coyotes) used the site Biological assess-ments in the summer of 2000 indicated healthy populations

of fish and macroinvertebrates in three of the four cells roinvertebrate community structure indicates a trend from tolerant to less tolerant species with flow through the wetland system In the final FWS cell, 314 bluegill fingerlings and

Mac-7 adults were seined (Nairn, 2003)

11.7 HEAVY METALS

C OPPER

Copper is an essential micronutrient for plants and animals because it is used for protein synthesis and in blood pigments Plants and animals require minimal amounts of this element, and deficiencies in nature are rare (Wetzel, 1983) The aver-age copper concentration in the world’s lakes and rivers is about 10 µg/L

Typically, copper is present in surface waters as chelated compounds of Cu(II) The ratio of free ionic to total dissolved

Flow out

Flow in

Organic material Limestone

Limestone

Flow out

Flow in

Organic material

Flow out

FIGURE 11.15 Three types of constructed wetlands used in acid mine drainage treatment.

copper is often less than 1% and decreases with increasing

organic loads and pH above neutrality (Morel and Hering, 1993)

When chelated with organic compounds, copper may remain

relatively soluble Depending upon the nature of the

complex-ing ligand, the soluble complex may not be bioavailable

U.S EPA (2002a) sets criteria for freshwater

concentra-tions of copper according to hardness At hardness of 100

mg/L, the maximum concentration allowable is 13 µg/L, and

the continuous concentration allowable is 9 µg/L, as dissolved

copper U.S EPA (2002a) provides formulae for the limits at

other hardnesses, with soft waters having lower criteria The

maximum level for human ingestion is 0.61 mg/L

Sediment standards for copper have been set by the

Prov-ince of Ontario, and have been adopted in some of the states

in the United States These set a lowest effect level of 16 µg/g,

and a severe effects level of 110 µg/g (Persaud et al., 1993;

OMOEE, 1994) OMOEE (1994) has also identified the level

for presettlement deposits as 25 µg/g in Great Lakes

sedi-ments Wetzel (1983) reports sediments in Lake Schöhsee,

Germany, were 95 µg/g, at a water concentration of 1.0 µg/L

Copper is a biocide that is commonly used to control

algae and other organisms, such as schistosomes (swimmer’s

itch), which require snails and waterfowl as intermediate

hosts Copper has relatively low toxicity to benthic

macro-invertebrates and to fish, which may be unaffected to

con-centrations as high as 500 µg/L The sensitivity of algae and

the tolerance of fish and benthos to copper have resulted in

the widespread use of copper sulfate as an algaecide and

molluscicide in lakes

Wetland Processing and Storage of Copper

The storage of copper in wetlands entails a risk of creating

potentially impaired sediments It is possible to estimate the

sediment concentrations created by sustainable removals of

copper, in terms of the dilution of the copper storage by the

accretion of new wetland solids This is a mass balance on

the top layer of deposited material (Equation 11.22)

This analysis may be extended to estimate sediment

con-centrations For example, consider 200 dry g/m2·yr of new

sediments, originating from 1,000 dry g/m2·yr of biomass

production If the sediment concentration is to be kept below

some limiting level, such as the 110-µg/g severe effects level

of the Ontario guideline, Equation 11.22 can be solved for the

allowable removal flux For that Ontario guideline, deposition

of 22 mg Cu/m2·yr could be sustainably tolerated Meeting

rigorous sediment standards on a sustainable basis is not

eas-ily achieved for any but trace concentrations, but has been

accomplished for systems like Savannah River, South

Caro-lina But for high concentrations, like those entering White

Cedar Bog, Minnesota (620 µg/L), a plume of high sediment

concentration (100–3,000 µg/g) developed (Eger et al., 1980).

Copper is found in wetland plants, algae, and sediments

Concentrations in algal tissues are typically higher than in

plants and sediments, with concentrations of 100–1,000 µg/g

dry weight for a number of species (Vymazal, 1995) In

contrast, copper concentrations in plants are much lower,

approximately 1–20 µg/g dry mass (Table 11.24) However,

if the water concentration is increased to high levels, more

copper is accumulated (Qian et al., 1999; Manios et al.,

2003) Aboveground plant parts are of lowest concentration, and belowground organs are somewhat higher Wetland sedi-ments have the highest levels, and are therefore the major repository of stored copper

Copper Removal Processes

Constructed wetlands have the potential to trap and remove metals contained in wastewater Long-term removal is expected to occur by accumulation and burial in the plant detritus in a manner similar to the removal of phosphorus

Chemical Precipitation

Copper forms very insoluble compounds with sulfur, ing both cupric (Cu2+) and cuprous (Cu+) sulfides (Sobolewski, 1999):

Sulfide and bisulfide are formed by SRB in the anaerobic zones of treatment wetlands If a source of sulfate is pres-ent in the incoming water, the wetland can be configured to provide a sustainable supply of sulfide (Sobolewski, 1996) Copper also forms insoluble hydroxides and carbonates (Morel and Hering, 1993) However, they are not of impor-tance in the presence of the more insoluble sulfide

The removal of copper can also take place through precipitation with iron and manganese oxides However, this mechanism is contingent on considerable supplies of iron, which are typically present only in a fraction of cases Al- though iron and manganese oxides are generally excellent scavengers for other metals, they are unstable under anoxic

co-conditions (Knox et al., 2004).

where P represents the organic material (peat).

If Equation 11.38, cation exchange, is presumed to sent system performance, then the amount of copper bound

repre-is represented by (Kadlec and Keoleian, 1986)

where the bracket notation denotes molar concentration

This suggests that the partition coefficient ([CuP2]/[Cu2+] =

440 Treatment Wetlands

TABLE 11.24

Copper Content of Wetland Plants and Sediments

Wetland Plant and Part Reference

Water (µg/L)

Solid (µg/g) Plant Shoots

Plant Roots

Plant Rhizomes

Sediments

Note: The corresponding water concentrations are approximate.

a Roots and rhizomes.

CS/CL) should go down markedly with decreasing pH, which

is in fact the observation of many investigators, as shown in

Kadlec and Keoelian (1986) Because of the exponent of two

on hydrogen ion concentration in Equation 11.39, halving the

pH decreases the partition coefficient by a factor of 4 The

concentration of available exchange sites, [HP], is related to

the CEC of the peat Brown et al (2001) found that the

cop-per sorption potential of various peats was linearly correlated

with their CEC Uptake ranged from 10–60 mg/g to 100–200

meq/100 g (R2 = 0.90)

Equation 11.39 postulates that there are a finite number of

exchange sites, and hence the uptake of copper should reach a

maximum The Langmuir isotherm provides for such

satura-tion, and fits peat sorption data (Chen et al., 2001; see

Equa-tion 11.11)

The maximum capacity for copper is CSmax = K/a Brown

et al (2001) found capacities from 10–60 mg/g for 11 Irish

peats, whereas Chen et al (2001) found 13 mg/g for a New

Zealand peat Freundlich isotherms are also often used, and fit

nearly as well (Kadlec and Rathbun, 1984) (Equation 11.13)

Kadlec and Rathbun (1984) reported n = 0.4 and KF =

3.4 at pH = 6.5, and 2.0 at pH = 5.2 Chen et al (2001)

reported n = 0.145 and KF = 5.6 at 3.1 < pH < 3.6 Whichever

models are used, peats have the ability to store very large

quantities of copper By extrapolation, organic substrates in

general have similarly large capacities According to

Equa-tion 11.13, peats can store 5–10 mg/g for water concentraEqua-tions

of 1–10 mg Cu/L and circumneutral pH

Performance of Wetlands for Copper Removal

Copper is effectively removed in FWS wetlands (Table 11.25)

For 26 systems, with inlet copper ranging from 1 to 7,300

µg/L, the median concentration reduction is 66% There is a

possibility that first-order models of copper removal would

be appropriate (Tarutis et al., 1999), but there are no

cali-brations for existing systems It is probable that for overland

flow wetlands, the rate-limiting step is the transfer of

cop-per from the water to sorption and precipitation sites in the

sediments Traditional mass transfer modeling would utilize

a first-order model for that process (see Chapter 6) It is also

likely that percentage reductions in Table 11.25 would be

greater if the hydraulic efficiencies were higher, e.g., closer

to plug flow The Sinicrope et al (1992) data correspond

to plug flow k-values from 18 to 47 m/yr for a SSF system

Vymazal and Krása (2003) reported the k-values of 17 m/yr

and 15.5 m/yr for HSSF systems in Nucˇice and Morˇina, the

Czech Republic However, in the HSSF system at Brˇehov,

copper was washed out of the system This corresponds well

with increase concentrations of iron and manganese at the

outflow from this system The k-value for this system was

−5.8 m/yr Kadlec and Srinivasan (1995) report copper

uptake in FWS cattail microcosms of 25 < kPF < 120 m/yr

The Savannah River, South Carolina, results give a

first-order areal plug flow removal rate constant kPF y 75 m/yr

These are consistent with estimates of mass transfer

coef-ficients (Kadlec and Knight, 1996)

Quite complicated models have been written for copper removal The effort of Lung and Light (1996) was exercised, but not calibrated or verified The most significant effort was the development and calibration of the CWFATE model for the Sacramento, California, project (Jones & Stokes Associates, 1993; Nolte and Associates, 1998a,b) CWFATE attempts to describe water budgets, copper biomass cycles, and partitioning, but not chemical precipitation

Plants are a minor repository for removed copper The

Murray-Gulde et al (2005) Savannah River, South Carolina, study determined that Scirpus californicus took up 6.76%

of copper entering the system In the Sinicrope et al (1992)

study, the removed copper was found mostly in the soil (91%), with smaller amounts in fine roots (8%), and the bal-ance in coarse roots, rhizomes, and shoots Vymazal and Krása (2005) found 76% of the added copper in sediments and belowground plant parts in an HSSF wetland treating domestic wastewater, and 6.3% in aboveground plant parts.The majority of removed copper is therefore associated with wetland sediments, in sorbed or chemically precipitated forms These may be categorized according to sequential

extraction schemes (Morea et al., 1989; Sobolewski, 1996)

The principal forms are associated with iron and manganese oxides, sulfides, and exchange sites on organics (Figures 11.16 and 11.17) In each case, long-term sustainable removal requires a continuing supply of secondary materials

Anaerobic Wetlands

Many wetland systems have been constructed using post or other organic waste to generate an anaerobic envi-ronment and provide a source of organic carbon (Skousen,

com-1997; Younger et al., 2002; Eger and Wagner, 2003) (see

Figure 11.15) These wetlands include various flow patterns, including fully flooded VF downflow and shallow FWS over-land flow

Reduction of organic matter over sulfate generates gen sulfide (Equations 11.1 and 11.2), which may react with copper to form insoluble copper sulfides (Equations 11.35 and 11.36) Lifetime estimates based on the total amount of carbon in these systems suggested that the substrate would last for several decades, but data indicate that these predic-tions were overly optimistic Eger and Wagner (2003) suggest that the amount of available substrate carbon is about 10%

hydro-of the theoretical Decaying wetland plants are a potential carbon source Bioavailable carbon production in wetlands is about 60 g C/m2·yr (see the denitrification section of Chapter 9) According to Equation 11.2, this would support sulfide production of 80 g S/m2·yr, a factor of 50 less than the typical

design rate recommended by Wildeman et al (1993b) The

corresponding copper removal rate is 160 g Cu/m2·yr, which

is also far less than the design value of 3,650 g Cu/m2·yr tentatively advanced by the PIRAMID Project of the Euro-pean Union (PIRAMID Consortium, 2003a) for anaerobic wetlands Results from HSSF wetlands (Table 11.25) indi-cate generally high removal of copper (81–84%), but in one instance copper was released from the system