TREATMENT WETLANDS - CHAPTER 25 doc

The project captures the feedlot runoff, stores it in a holding pond, and then treats the wastewater in constructed wetland cells operated during approximately 150 days each year from mi

25

The strategy of this book up to here has considered individual

wastewater constituents by analysis of the available data and

in terms of the types of wetlands that may be employed to

reduce concentrations of those individual chemicals That

might be termed a horizontal approach because removal of,

say, biochemical oxygen demand (BOD), is a crosscutting

issue that affects many types of effluents Primary emphasis

has been on the reduction of specific contaminants,

includ-ing BOD, total suspended solids (TSS), nutrients, metals, and

hydrocarbons

In this chapter, a “vertical” approach is taken, in which

particular treatment wetland applications are considered as

a group This is done because some technology growth areas

are very application specific It is not necessary to repeat the

many details already presented in two key application areas—

metals and hydrocarbons—because each has had an entire

chapter already (Chapters 11 and 13) Likewise, the general

application area of stormwater runoff has been singled out for

an entire chapter (Chapter 14) There remain several

applica-tions that deserve additional attention because they present

unique features that can influence design These include

wet-land treatment of animal waste, food wastes, wet-landfill leachate,

and chemical industry effluents

25.1 ANIMAL WASTES

CONFINED ANIMAL OPERATIONS

Liquid animal wastes consist of manure, flushwater, and

rainfall The Agricultural Waste Management Field

Hand-book (USDA-NRCS, 1992) provides information on

vol-umes of manure (feces and urine) and mass excretion rates of

nutrients for different types of livestock Average quantities

of nitrogen, phosphorus, and BOD5 in manure are given in

Table 25.1 Solid wastes consist of manure and sometimes

bedding material such as straw These amounts are then

diluted with water, and the amount of water added to the

sys-tem usually exceeds the amount of manure For cattle and

dairy facilities in the United States, 160–200 L/d per cow are

used for flushing requirements for freestall alleys (CH2M

Hill and Payne Engineering, 1997) Swine flushing quantities

are typically 60–95 L/d per adult animal However, Tanner

and Kloosterman (1997) use a lower water rate of 50 L/d per

cow, and the French use an even lower range, from 5–15 L/d

per cow (Liénard et al., 2002).

Operators use a variety of methods to manage wastewater,

including lagoons, ponds, storage structures, compost areas,

filters, and sediment basins Liquid wastes are typically lected and treated in one or more lagoons before being land applied The organic matter and nutrients in manure are impor-tant resources that are recycled to the land, if possible, to foster crop production Solid and slurry wastes usually have very high nutrient concentrations and are good soil amendments, but, in many cases, it is not feasible to collect and contain all animal waste Some portions are leached from loafing lots, and spills occur In other cases, the availability of cropped land does not accommodate the entire waste production

col-A constructed wetland offers another option for polishing wastewater before it is recycled as flushwater, land applied,

or ultimately discharged As for other wastewater types, the goal is to prevent manure from being discharged to receiving waters In natural recipients, the organic matter and nutrients promote algal growth and deplete dissolved oxygen, or in other words, promote eutrophication

Wetland treatment can be implemented on farms in cold climates Systems have been evaluated across Canada, from Alberta (Reimersma, 2001) to the Atlantic Provinces (Smith,

2003; Smith et al., 2006) Other cold-climate applications

have been reported in Germany (Kern and Carlow, 2000; Kern, 2003) These low-technology, solar-driven systems are passive and user-friendly, and thus farmers do not need to acquire the skills of wastewater treatment plant operators The operation of a wetland system is closely related to cur-rent farming practices The wetland system requires sunlight, nutrients, and water, but does not need to be harvested The treatment wetland plants can tolerate relatively high concen-trations of nutrients, but full-strength wastes should not be applied directly to wetlands because many wetland plants cannot tolerate the high nutrient levels

The use of wetlands for treating concentrated animal wastes gained a measure of popularity in the early 1990s The Brenton cattle operation in Iowa, which started in 1930, was the oldest system in North America found by CH2M HILL (1997) Other early sites include Sand Mountain, Alabama, started in 1989, and the Newton and Hattiesburg, Mississippi, systems, also started in 1989 In the mid-1990s, data collec-tion efforts catalogued the performance of 68 wetland sys-tems in North America (DuBowy and Reaves, 1994; CH2M

Hill and Payne Engineering, 1997; Knight et al., 2000)

Design procedures are offered, which generally fall in the two categories of loading methods and rate constant methods (CH2M Hill and Payne Engineering, 1997)

Examples are included here to illustrate the breadth of applicability for several types of animal operations

Example: Auburn Poultry

Three systems of free water surface (FWS) constructed

wet-lands were prepared at the Alabama Agricultural Experiment

Station’s (AAES) Poultry Science Unit at Auburn University

in Auburn, Alabama, in June 1992 (Rogers et al., 1995; Hill

and Payton, 1998) Each system consisted of two cells in

series, one planted with Phragmites and Scirpus, a second

with Sagittaria lancifolia, while the third series acted as an

unvegetated control Each wetland cell was 5.5 m wide, 30.5 m

long with an average depth of 30 cm, and lined with 10 cm

of bentonite clay Subsystems of wooden dowels were tested

within the open water system The wetlands influent was taken

from the first cell of the lagoon system fed by the caged layer poultry operation, and the wetland effluent flowed to a hold-ing pond nearby A hydraulic loading rate of 3.1 cm/d was applied to each system, which resulted in chemical oxygen demand (COD) and total Kjeldahl nitrogen (TKN) loading rates of 145 and 30 kg/ha·d, respectively In order to deter-mine plant porosity blockage and thus also to assist in calcula-tion of detention time, water column displacement tests were performed The measured values were 10.7% and 7.0% for

Phragmites and Sagittaria, respectively The detention time

was calculated to range from 11.1 to 11.6 days The tration reductions are shown in Table 25.2, and ranged from 40–60% for the principal contaminants for the vegetated sys-tems It was expected that temperature would have positively

concen-TABLE 25.1 Approximate Waste Quantities for Animals on the Basis of 100 kg Weight

Nitrogen (g/d·100 kg)

Phosphorus (g/d·100 kg)

Beef cattle Forage feeder 31 11 137

Source: Adapted from Rogers et al (1995) In Versatility of Wetlands in the Agricultural Landscape Campbell (Ed.),

American Society of Agricultural Engineers, St Joseph, Michigan; and Hill and Payton (1998) Transactions of the

American Society of Agricultural Engineers 41(2): 393–396.

but treatment showed little correlation to water temperature

(Hill and Payton, 1998)

C ATTLE

Wetlands are used to manage feedlot runoff, dairy cattle waters,

and pasture waters Two fairly distinct situations may be found

in the use of treatment wetlands for cow operations: pastures

and feedlots versus dairies Pastures and feedlots are primarily

event-driven, runoff control systems, whereas dairies are

con-cerned with water originating in milking parlors, comprised of

cleaning waters The two are considered separately here

Pasture runoff has only been recently considered for

treat-ment The studies of Tanner and coworkers in New Zealand

have utilized modern design concepts and made extensive

measurements, which form the beginnings of a performance

knowledge base (Tanner et al., 2003; Tanner et al., 2005b)

Because some pastures are tile-drained, there is sometimes

the possibility of directing tile drainage to constructed

wet-lands (seeFigure 14.8, Chapter 14) The density of animals

on the contributing watershed is typically low, and therefore

the concentrations of nutrients and other contaminants is

rel-atively low (e.g., 10–20 mg/L nitrogen) Further, speciation

of nitrogen favors nitrate as the dissolved form, rather than

ammonia, because the infiltration through the vadose zone

provides for nitrification

Feedlots present a different situation because the density

on the contributing watershed is high Nitrogen is almost all

in the form of organic and ammonia (TKN), and

concentra-tions of all pollutants are high in the direct runoff One of

the earliest constructed wetlands was a two-cell FWS system

built in 1930 at the Brenton Brothers feedlot in Dallas County,

Iowa, which serves a 26-ha feedlot that typically has 7,000

head of cattle The first wetland cell, of extremely irregular

shape, comprises 25 ha and receives rain-driven runoff from

the feedlot Over a two-year study period, there were

reduc-tions of 87% BOD, 60% TSS, and 58% TKN This system

was one of the two included in the North American Livestock

Database (CH2M Hill and Payne Engineering, 1997)

Subse-quently, other wetland systems have been tested, with good

results For instance, runoff from Iowa State University’s

380 head beef cattle concrete feedlot was treated via solids

settling, soil infiltration, and a small constructed wetland,

in series (Lorimor et al., 2003) Concentrations and

nutri-ent mass flows were reduced by over 97% for TKN, 94%

for total phosphorus, and 93% for TSS No buildups in the

system were observed over five years The authors concluded

that the passive treatment system could very effectively

pro-tect surface-water quality below open beef feedlots In their

review of vegetated systems for cattle runoff control, Koelsch

et al (2006) found that virtually all of the existing

informa-tion was from non-CAFO (Confined Animal Feeding

Opera-tions) situations, i.e., dairies These authors concluded that

critical design factors included pretreatment, sheet flow,

dis-charge control, siting, and sizing, together with maintenance

of a dense vegetation stand

The Canadian livestock farming community is generally not required to meet the strict surface-water discharge regulations imposed on municipalities and industries However, farming operations generate high levels of nutrients and bacteria that originate from manure storage and application areas, manure storage tank overflows, and feedlot runoff, and there is a need

to intercept and reduce these contaminant loads tion of treatment wetland technology was the prime objective

Demonstra-in constructDemonstra-ing demonstration wetland systems Demonstra-in Manitoba

in 1996 Full-scale wetland projects were constructed at two cattle feedlot operations in the Interlake Region of Manitoba (Pries and McGarry, 2001) The second constructed wetland, described here, was at an 1,800-head feedlot bordering Lake Manitoba in the Interlake area, commissioned in 1998 The treatment wetland was constructed to reduce the nutrient and solids loading to Lake Manitoba

The project captures the feedlot runoff, stores it in a holding pond, and then treats the wastewater in constructed wetland cells operated during approximately 150 days each year from mid-May through mid-October The 0.25-ha hold-ing pond collects and stores contaminated runoff from the feedlot area during the late fall, winter, and early spring, and has a working volume of approximately 3,300 m3 The aver-age hydraulic retention time (HRT) in the pond during the wetland operating period is approximately two months The stored water is pumped to the wetland cells; the topography does not allow for gravity flow

The treatment wetland consists of two 0.5-ha cells ating in parallel with an average depth of 0.3 m and a work-ing volume of 3,000 m3 Flow to each of the wetland cells

oper-is controlled by gate valves For normal operation, the gate valves at the weir box are opened to allow a maximum flow

of about 100 L/min to each cell The flow enters the land in the center of the width of an influent distribution deep zone, which distributes the water across the width of each cell Treated water is discharged at the effluent level control structure with a weir for flow measurement In dry years there may be no outlet discharge due to water removal through high rates of evapotranspiration

wet-The holding pond provided considerable treatment prior

to discharging the contaminated stormwater to the wetland Table 25.3 presents the average annual data from monthly samples for two years of operation The wetland effluent BOD5 was 15 mg/L or less throughout both years, and the TSS was also consistently less than 15 mg/L Inflow ammonia averaged 2.1 mg/L, with excursions to 4.8 mg/L, but the efflu-ent ammonia-nitrogen was less than 0.5 mg/L Phosphorus reduction was consistent each year with lower concentrations

in the effluent than was measured in the influent

D AIRY O PERATIONS

Dairy operations can be major contributors of nutrients to the watersheds in which they are located For instance, in a nutri-tion study, it was found that cows excreted 88.2% of phosphorus

consumed daily: 68.6% in feces, 1.0% in urine, and 30.3%

in milk (Morse et al., 1992) Current and abandoned dairy

systems in South Florida’s Lake Okeechobee watershed have

a demonstrated phosphorus pollution potential (Nair et al.,

1995) In Ireland, about 25% of the land area is devoted to

dairy farming, and it is estimated that the percentage of

pol-lution attributed to agriculture is approximately 32% in rivers

and streams that are slightly to moderately polluted

Most dairies produce two types of waste streams: “white”

water, which is defined as the wastewater produced during

cleaning and sterilization of the milking equipment, and

“green” water, which is wastewater resulting from the

wash-down of the manure-spattered walls and floors of the milking

parlor and the associated holding pen or staging areas The

combination of these two waste streams produces the

“typi-cal” milkhouse wastewater Because each dairy is different,

the ratio of white water to green water will vary Some farms

may have existing manure treatment lagoons, which may be

used as pretreatment device for the wetland treatment

sys-tem A typical dairy waste treatment schematic is shown in

Figure 25.1

Constructed wetlands have been demonstrated to be

effective add-ons to supplement the treatment afforded by

farm dairy pond systems Farm dairy ponds provide for the

return of a significant proportion of the nutrient loading to

the pasture via periodic desludging, but the water discharges

are not sufficiently good to protect receiving water quality

(Table 25.4; see also Table 25.3) This application has spurred

considerable research over the past decade (Geary and Moore,

1999; Karpiscak et al., 1999; Newman et al., 2000; Hunt and

Poach, 2001; Shamir et al., 2001; Kowalk, 2002; Ibekwe et

al., 2003; Forbes et al., 2004a; Dunne et al., 2005c; Smith et

al., 2006; Healy et al., 2007) Cronk (1996) identified 40 dairy

treatment wetlands in North America, established between

1989 and 1993 A database compilation in 1996 identified 60

constructed wetlands at 38 sites in North America (CH2M

Hill and Payne Engineering, 1997; Knight et al., 2000)

How-ever, a large number of systems have been built, starting in the late 1990s For instance, by 2004 there were 12 FWS wetlands in the Anne valley of County Wexford in Ireland (Figure 25.2) (Carroll et al., 2005) In general, these systems

have large surface areas, averaging 120 m2/cow, and provide excellent water quality improvement (Table 25.4)

Although dairy wetlands have primarily been FWS, zontal subsurface flow have also been implemented (Tanner

hori-et al., 1995a; Tanner hori-et al., 1995b; Mantovi hori-et al., 2003; Kern,

2003) Tanner and Kloosterman (1997) recommend HSSF

as an add-on to FWS, presumably because of the potential for clogging of a gravel bed with direct lagoon discharges The New Zealand wetland design guidelines suggest about 1.0 m2/cow for this post-FWS application (Tanner and Kloos-terman, 1997)

However, if a HSSF system is to be used as a alone” device, more area is required to achieve good pollut-ant reductions In the United Kingdom, a HSSF wetland was used to treat influent with an average BOD5 of 1,192 mg/L

“stand-on a dairy farm in Droint“stand-on (Cooper et al., 1996) An area of

250 m2 for 160 cows (1.6 m2/cow) proved unacceptable, and vertical flow wetlands were subsequently installed ahead of the HSSF system A 95% reduction in BOD5 was achieved (Figure 25.3) In Germany, an area of about 10 m2/cow pro-duced reductions of 91% COD, 86% total nitrogen (TN), and 99% fecal coliforms (FC), with better nutrient reductions in winter (Kern, 2003) A HSSF system in Italy dimensioned at 1.84 m2/cow (at a measured 80 L/d per cow) was successful

in reducing BOD from 451 mg/L to 28 mg/L, for a settled

influent from Imhoff tank pretreatment (Mantovi et al., 2003)

Total nitrogen was reduced from 65 to 33 mg/L, due to alization of organic nitrogen, because ammonia removal was nonexistent It should be noted that European dairy wetlands

miner-TABLE 25.3

Manitoba Interlake Site 2: Average Annual Concentrations in the Wetland Inflows and Outflows

Average Value 1999 Average Value 2000

Source: From Pries and McGarry (2001) Constructed Wetlands for Feedlot Runoff Treatment in Manitoba Proceedings of the 2001

Water Environment Association of Ontario Technical Conference; Toronto, Ontario Reprinted with permission.

do not usually have a lagoon for pretreatment, although that

is the norm in North America and New Zealand Therefore,

direct comparisons are difficult

Research is being directed to more highly oxygenated

systems, such as forced aeration or vertical flow wetlands

(Whitney, 2003) Vertical flow wetlands have high rates of oxygen transfer and are capable of handling high oxygen

demands of milk-house wastes (Green et al., 2002)

Rec-ommended vertical flow wetland criteria for dairy waters in France total 0.22–0.44 m2/cow in two stages in each of two

Stream

FIGURE 25.1 Concept for a farm dairy treatment wetland system (From Tanner and Kloosterman (1997) Guidelines for Constructed

Wetland Treatment of Farm Dairy Wastewaters in New Zealand NIWA Science and Technology Series No 48, National Institute of Water

and Atmospheric Research (NIWA): Hamilton, New Zealand Reprinted with permission.)

In (mg/L)

Out (mg/L)

In (mg/L)

Out (mg/L)

In (mg/L)

Out (mg/L)

Source: Data from Carroll et al (2005) In Nutrient Management in Agricultural Watersheds: A Wetland Solution Dunne et al (Eds.), Wageningen Academic

Publishers, Wageningen, The Netherlands.

trains (Liénard et al., 2002) As for other VF systems, dairy

VF wetlands are pulse loaded to induce hydrodynamic

move-ment of air through the wetland bed

The practitioner can find several design guidelines

docu-ments (CH2M Hill and Payne Engineering, 1997; Tanner and

Kloosterman, 1997; Forbes et al., 2004a) However, the

sup-porting data is in the process of being strengthened at this

time, as the various authors note, and improvements are to be

expected as more operating data become available Rate

con-stants for dairy FWS can be found in the literature, in Payne

Engineering and CH2M HILL (1997) In general, these are

within the distributions presented in earlier chapters of this

book, and indeed the dairy data were included in those

deter-minations McGechan et al (2005) give rate constants for

HSSF and VF dairy systems for nitrogen processing for

meso-cosms for dilute farm influents However, loading tions are more typically used to interpret data and for design

specifica-As an example, one set of median-loading recommendations for FWS systems is shown in Table 25.5 It should be noted that those loadings correspond to fairly large wetland areas, comparable to those used in the Irish systems in Table 25.4.Those dairy loading recommendations are comparable to, but somewhat less than, those recommended by U.S EPA for municipal wastewaters

Nitrogen reductions are in general less than for BOD and TSS Dairy wastewaters differ from most municipal and domestic wastewaters in that they contain a relatively high amount of organic nitrogen For instance, the database used in this book shows a median inlet organic nitrogen of

37 mg/L for animal treatment wetlands, whereas the median

FIGURE 25.2 Farm wetland in the Anne valley near Waterford, Ireland This FWS system takes advantage of a downhill location from

the barns and yards to employ gravity flow.

FIGURE 25.3 Performance of the first and second iterations of the Drointon, U.K., reedbeds for farm waste treatment In the first case, the

BOD loading to the HSSF wetland was 255 kg/ha·d, and in the second case it was 35 kg/ha·d (Based on Cooper et al (1996) Reed Beds and Constructed Wetlands for Wastewater Treatment WRc Publications, Swindon, United Kingdom.)

BOD: 2,450 mg/L TSS: 1,281 mg/L

BOD: 1,214 mg/L TSS: 633 mg/L

608 mg/L

506 mg/L

BOD: 2,158 mg/L TSS: 865 mg/L

for municipal systems is 7 mg/L As a consequence, there is

a need to mineralize this organic nitrogen before it can be

reduced by oxidative and other processes This additional

processing step slows overall nitrogen removal and is partly

to blame for low removal rates Mineralization of the organic

nitrogen can increase the ammonia content of the water prior

to its nitrification Ultimately, the loss of total nitrogen

sug-gests an implied oxygen supply that is limiting for nitrogen

processing The internal ammonia load created by

mineral-ization of organic nitrogen adds to the incoming ammonia

The rates of nitrogen removal in animal wetlands are strongly

suggestive of mechanisms other than traditional nitrification–

denitrification Oxidized nitrogen is rarely a significant

com-ponent of influents or effluents from dairy water wetlands

This is speculatively due to an adequate carbon supply and

rapid denitrification, and perhaps also due to Anammox or

similar alternative processing mechanisms (see Chapter 9)

Example: Dairy Farm Operations

A study on a three-cell integrated FWS system in County

Wexford, Ireland, was conducted over a 2.5-year period

(Dunne et al., 2005a; Dunne et al., 2005b) Water from

farm-yard management and dairy operations for a 42-cow facility were collected in a central storage tank before discharge to the wetlands The three cells had a combined area of 4,265 m2

(102 m2/cow) and were maintained at a 30–40-cm depth

A 250-m2 sedimentation basin preceded the treatment marshes The vegetative cover was 80–90% during the grow-

ing season, and comprised of a mixture of Carex riparia, Typha latifolia, Phragmites australis, Sparganium erectum,

and other emergent species The flow rate to the wetlands averaged 7.6 m3/d (180 L/cow·d), and was greatly augmented

by rainfall (1,142 mm/yr) to produce an outflow rate of 31.1 m3/d As a result, there were considerable differences between mass reductions and concentration reductions (Table 25.6) Part of the observed concentration reduction was attributable to dilution Nonetheless, this lightly loaded system produced very good removals, but with lesser per-formance in winter for ammonia and phosphorus It is note-worthy that the implied maximum oxygen supply was about 4.8 gO/m2·d, which is a feasible reaeration rate It is also the case that the nitrogen supply places these wetlands in the group with agronomic control BOD and TSS are reduced to what

TABLE 25.5 Median-Loading Recommendations for FWS Animal Wastewater Wetlands (kg/ha·d)

Source: Adapted from data in CH2M Hill, Payne Engineering (1997) Constructed wetlands for animal waste treatment Report to the Gulf of Mexico Program Nutrient Enrichment Committee, U.S EPA: Stennis Space

Center, Mississippi; and U.S EPA (2000b) Constructed wetlands: Agricultural wastewater EPA 843/F-00/002,

U.S EPA Office of Water.

TABLE 25.6

Summary Performance of the Teagasc FWS Dairy Wetlands, Located at the Teagasc Research Center, Johnstown Castle, Wexford, Ireland

Flow (m 3 /d)

SRP (mg/L)

NH 4 -N (mg/L)

BOD 5

(mg/L)

TSS (mg/L)

SRP (g/m 2 ·yr)

NH 4 -N (g/m 2 ·yr)

BOD 5

(kg/ha·d)

TSS (kg/ha·d)

appear to be background concentrations because there was

little change in concentrations after the first two wetlands

SWINE

Swine wastewaters are strong effluents that received only

minimal treatment in past years A very popular system in

the southeastern United States is an anaerobic lagoon

fol-lowed by a spray irrigation system In North Carolina, there

are 4,500 active and 1,700 inactive swine waste lagoons

(Humenik et al., 2004), but these are no longer permitted

because of problems with overflow and leakage Constructed

wetlands are a logical add-on, and therefore investigations

on their effectiveness were initiated, approximately in 1990, at

Auburn University’s Sand Mountain Agricultural Experiment

Station in DeKalb County, Alabama (Hammer et al., 1993a;

McCaskey et al., 1994) Another facility followed soon after

(1991) at the Pontotoc, Mississippi, Experiment Station

(Cathcart et al., 1994) Based on early swine wetland data

and the performance of municipal treatment wetlands, design

recommendations quickly proliferated (USDA-NRCS, 1992;

Hammer, 1994) Loading numbers of approximately 60–70

kg/ha·d for BOD were suggested In fact, FWS swine

wet-lands are at the poor performance edge of the municipal

con-centration-loading band (Figure 25.4), and this loading now

appears likely to produce 60–80 mg/L BOD in the wetland

outflow This may be due to the much higher concentrations

in swine wastewater (Cronk, 1996) Research intensified in

the mid-1990s, with projects in North Carolina and Indiana

Investigations were conducted from 1992 to the present at a

continuous marsh site in the eastern coastal plain in Duplin

County, and at a marsh–pond–marsh site in Greensboro at

North Carolina A&T State University The site descriptions

and operational procedures are reported in a number of

pub-lications (Hunt et al., 1994; Reddy et al., 2001; Stone et al., 2002; Poach et al., 2003; Stone et al., 2004) The Purdue Uni-

versity system at West Lafayette involved 16 FWS wetland

cells operated in a designed experimental array (Reaves et al., 1994) These studies all showed significant reductions in

BOD, TSS, and total nitrogen, but the removal percentages were not outstanding (Table 25.7)

Because of the high ammonia levels in swine lagoons and wetlands, there has been concern that ammonia volatil-ization could be a problem A study was conducted to address the concern that nitrogen removal was due to high ammonia

volatilization Poach et al (2002; 2004) documented that less

than 10% of the input nitrogen was lost by ammonia tilization (seeFigure 9.7, Chapter 9) However, as for other animal wastes, the organic nitrogen content is high, and a mineralization step must precede ammonia conversions The process of denitrification is greatly assisted by the availability

vola-of organics and carbon, with the result that oxidized nitrogen

is not present in the waters, and by the fact that redox

poten-tials are generally quite low (Hunt et al., 2003; Szögi et al.,

2004) Nitrogen processing in swine wetlands is quite perature dependent, as for other FWS systems However, that does not preclude their use in cold climates during the unfro-zen season, even as far north as Alaska (Maddux, 2002).Subsurface flow wetlands have also been used to treat swine effluents in North America (Sievers, 1997) and else-

tem-where around the world (Lee et al., 2004b) However, the

high TKN concentrations also cause problems with respect

to oxygen supply in HSSF systems Therefore, vertical flow

systems have also been piloted (Kantawanichkul et al., 1999; Sezerino et al., 2003) As an alternative to vertical flow,

a fill-and-drain mode has been employed with good success

(Behrends et al., 2003; Rice et al., 2005) This concept

involves continuously pumping wastewater back and forth

FIGURE 25.4 The position of swine treatment wetlands compared to all other FWS systems.

between adjacent cells on a two-hour cycle This vertical

flow design provides aeration of the gravel substrate and

exposes the internal biofilms to atmospheric oxygen During

the “drain phase” of the cycle, atmospheric oxygen causes

enhanced oxidation of ammonia and organic matter This

mode of operation is described further in Chapter 24

Example: Swine

In 1995, six wetland systems to treat swine lagoon

wastewa-ter were constructed at the North Carolina A&T State

Uni-versity farm near Greensboro, North Carolina (Reddy et al.,

2001; Stone et al., 2004) The wetland systems were

config-ured into a marsh section, a central pond section, and another

marsh section (marsh–pond–marsh) The marsh sections were

approximately 10 m r 10 m, and the pond section was 10 m r

20 m The marsh sections were planted with Typha latifolia

and Schoenoplectus americanus in March 1996 Beginning

in September 2000, the wetlands were loaded with specific

total nitrogen at rates of 5, 14, 23, 32, 41, and 50 kg/ha·d For

one year, the wetland hydraulic loading rates were held

approximately equal, with only the TN loading rate varying

The operating depths of the constructed wetlands were 15 cm

for marsh sections and 75 cm for pond sections

Mean TN and NH4-N concentration reductions were 35

and 25%, respectively However, the nitrogen performance

was on the poor side of the scatter of performances for other

FWS wetlands (Figure 25.5) The calculated first-order

plug-flow rate constants (k20) for TN and NH4-N were 3.7–4.5 m/d

and 4.2–4.5 m/d, respectively, which are considerably lower

than the central tendencies given in Chapter 9 These are at

about the 10th percentile for FWS wetlands However, no

allowance was made for the internal mineralization of organic

nitrogen in the Stone et al (2004) analysis.

Z OOS

The Wuhan Zoo, Wuhan City, China, is investigating the use

of sedimentation systems and wetlands to reduce nutrient

losses from the site to a lake The project is conducted by the Research Centre for Eco-Environmental Sciences, Chinese Academy of Sciences, Beijing (D’Arcy, 2005) A reed bed has been operating for some years at Chester Zoo in the U.K., treating runoff from its Asian elephant enclosure (D’Arcy, 2005) A vertical flow treatment wetland is used at the Saint-Felicien zoo in Quebec (Pries, 1994) The seals of the Rhe-ine zoo in Westphalia, Germany, have their water recycled back to their basin to cut costs that otherwise would arise

by replacing used water with fresh drinking water (Dartmann

et al., 2000) The seals produce large amounts of residue that

cloud the water, capable of impairing the visitors’ view, cially in deep-water zones A horizontal flow wetland system

espe-is used to filter and clarify the water

A QUACULTURE

Fish farming is an increasing activity worldwide, as wild plies dwindle and dietary preferences change Constructed wetlands have been explored as a means of controlling the associated pollution, notably for catfish, trout, and shrimp operations However, other fish species have also been the

sup-object of treatment wetlands, such as milkfish (Chanos nos) (Lin et al., 2002a; Lin et al., 2002b).

cha-Catfish

Channel catfish (Ictalurus punctatus) farming is the

larg-est segment of aquaculture in the United States Most of the nation’s catfish is raised in Alabama, Arkansas, Louisiana, and Mississippi on 70,000 ha of catfish farms (Tucker, 1999) More farm-raised catfish is produced and sold in these states than all other U.S aquacultured species combined In fact, the farm-raised catfish industry enjoyed phenomenal growth over the past 37 years, from 2.6 million kilograms produced

in 1970 to more than 275 million kilograms in 2006 (Harvey, 2006) Nearly all commercial aquaculture in the southeastern United States is conducted in earthen ponds Good produc-tion from ponds is encouraged by using manufactured feeds

Inlet (mg/L)

Load (kg/ha·d)

Outlet (mg/L)

Inlet (mg/L)

Load (kg/ha·d)

Outlet (mg/L)

Inlet (mg/L)

Load (kg/ha·d)

Outlet (mg/L)

or increasing the availability of natural foods by adding

fertil-izers Less than 30% of the nitrogen and phosphorus added in

feed or fertilizer is recovered at harvest The remainder of the

nutrient load is left in the pond and may be discharged when

it rains or when ponds are drained between fish crops

In the United States, aquaculture effluents are regulated

under the National Pollutant Discharge Elimination System

(NPDES) as part of the Federal Water Pollution Control Act

of 1972 A study by the Southern Regional Aquaculture

Cen-ter characCen-terized catfish pond effluents from 45 ponds over a

two-year period (Tucker, 1999) Effluent quality was poorest

in the summer when fish-feeding rates and water

tempera-tures are highest Catfish pond effluents generally have higher

concentrations of nutrients and organic matter than natural

stream waters, but much lower concentrations than municipal

and industrial wastewater

As a result of successful demonstrations, the use of

constructed wetlands for finfish production was thoroughly

examined by Posadas and LaSalle (1997; 1998) Their report

concluded that the use of constructed wetlands was not

eco-nomically viable because of the added costs Kouka and Engle

(1994) estimated that wetland treatment would add $0.18/kg

to the price of the fish, which is on the order of $1.75/kg

(2006 USD) (Harvey, 2006)

Example: Catfish

In 1992, a two-cell, 2,350-m2 test wetland was implemented

near Greensboro, Alabama (Schwarz and Boyd, 1995) Water

from a production pond was passed through the constructed

wetlands at HRTs of 1, 2, 3, and 4 days, with hydraulic

loading rates of 77–91 L/d·m2 Cell one was planted with

California bulrush (Scirpus californicus) and giant cutgrass (Zizaniopsis miliacea), and cell two with maidencane (Pani- cum hemitomon) Concentrations of potential pollutants

were much lower in effluent from the wetland than in ent from the channel catfish ponds The mean mass removals were total nitrogen, 61%; biochemical oxygen demand, 62%; suspended solids, 87%; and total phosphorus, 75% Overall performance of the wetland was not correlated with HRT, but that was clearly because hydraulic loading was held constant while depth was varied Thus, this data supports the notion that removal is area specific rather than volume specific The researchers observed that although pond overflow and water exchanges could be treated with reasonably sized wetlands, complete pond draining usually proceeds rather rapidly and would create unacceptably high hydraulic loads

influ-Tilapia

“Tilapia” is the generic name of a group of cichlids endemic

to Africa, consisting of three aquaculturally important genera

(Oreochromis, Sarotherodon, and Tilapia) Worldwide

har-vest of farmed tilapia has now surpassed 800,000 metric tons, and tilapia are second only to carp as the most widely farmed freshwater fish in the world (Pompa and Masser, 1999) Posi-tive aquacultural characteristics of tilapia are their tolerance

to poor water quality and the fact that they eat a wide range

of natural food organisms Indeed, tilapia have been ered as part of a system to improve swine water quality, used

consid-in conjunction with wetlands (Dontje and Clanton, 1999) Biological constraints to the development of commercial tilapia farming are their inability to withstand sustained water temperatures below 10 to 15°C The U.S imports most of its

FIGURE 25.5 Total nitrogen loading graph for the marsh–pond–marsh system at North Carolina A&T University Farm FWS swine

wet-lands at Greensboro, North Carolina, compared to the general FWS loading data The line that represents their swine data (from Stone et al (2004) Ecological Engineering 23(2): 127–133) is on the upper edge of the cluster of data.

demand, but tilapia farms are beginning to appear, and with

them the need to improve water quality

In the mid-1990s, constructed wetlands were tested for

treatment of tilapia wastewater at New Mexico State

Univer-sity, Las Cruces, New Mexico (Zachritz and Jacquez, 1993)

Brown et al (1999) used salt-tolerant plants (Suaeda esteroa,

Salicornia bigelovii, and saltbush Atriplex barclayana) to

remove nutrients from saline aquaculture wastewater in

Tuc-son, Arizona These were grown in sand in draining lysimeters

and were irrigated to meet evapotranspiration and produce

a 0.3 leaching fraction using aquaculture effluent generated

from an intensive tilapia culture system The system removed

98% and 94% of total and inorganic nitrogen, respectively

Behrends et al (1999; 2002) utilized fill-and-drain SSF

wet-lands to treat wastewater from an experimental tilapia-rearing

(Oreochromis niloticus) facility in Muscle Shoals, Alabama,

with good success

Salmonids

Salmon and trout (family Salmonidae) are also farmed

inten-sively in more northerly climates and have the same general

water quality issues as catfish In the Province of Quebec,

for example, trout farm production increased from 40 tons/

yr in 1976 to 1,900 tons/yr in 1996 (Comeau et al., 2001)

Trout farms effluents are typically 20 times more diluted than

medium-strength municipal wastewaters and even below

municipal secondary treatment criteria (Comeau et al., 2001;

Schulz et al., 2003a) With respect to

receiving-water-qual-ity objectives, the most constraining element to remove from

freshwater fish farms is phosphorus

Evaluation of treatment wetlands to treat effluents began

in the mid-1990s, with work at the Freshwater Institute in

West Virginia (Summerfelt et al., 1999), and at Chisholm,

Minnesota (Axler et al., 1996) Subsequently, projects

pro-liferated, involving both FWS systems (Michael, 2003) and

HSSF systems (Comeau et al., 2001; Naylor et al., 2003;

Schulz et al., 2003a; Ouellet-Plamondon et al., 2004; Schulz

et al., 2004; Lymbery et al., 2006) In general, findings

indi-cate that constructed wetlands are an ecologically attractive and economical method for treating salmon and trout farm effluents to reduce solids and phosphorus Examples of wet-land performance are given in Table 25.8

Shrimp

The production of crustaceans in pond culture is a major cultural enterprise In the United States, this is a fledgling industry, with 120 farms growing only about 3,800 metric tons

agri-of saltwater shrimp and freshwater prawns (USDA-NASS, 2006) In contrast, 647 farms grow 16,000 metric tons of crawfish U.S imports of shrimp totaled 560,000 metric tons

in 2006, because shrimp farming is primarily a tropical ity, with major centers in China, India, Thailand, Indonesia, and South America (Harvey, 2006) Effluent waters from shrimp aquaculture can contain elevated levels of phospho-rus, ammonia, nitrate, and organics Regulation of the water quality of shrimp pond discharges has begun in many loca-tions Constructed wetlands offer an ecologically beneficial, low-cost treatment alternative, and, as a result, exploratory

activ-studies began in the mid-1990s (Sansanayuth et al., 1996)

It was observed by Gautier et al (2001) that an engineered

mangrove wetland produced good water-quality improvement for a shrimp farm effluent in Colombia However, the use of natural wetlands is not always allowed, and thus research mostly focuses on constructed wetlands

A 7.7-ha mesohaline (3–8 ppt) FWS-constructed land was used to treat 13,600 m3/d of effluent from 8.1 ha of intensively farmed shrimp ponds near Port Mansfield, Texas,

wet-on the coast of the Gulf of Mexico (Badrinarayanan, 2001;

Tilley et al., 2002; Staff, 2006) The depth of water was

main-tained between 0.15 and 0.45 m, which provided a lic retention time (HRT) in the wetland on the order of one day Originally, the wetland was planted with ten species of salt-tolerant wetland plants, but by the second year of opera-

hydrau-tion, the wetland had undergone self design Typha latifolia

Inlet (mg/L)

Outlet (mg/L)

Inlet (mg/L)

Outlet (mg/L)

Inlet (mg/L)

Outlet (mg/L)

Inlet (mg/L)

Outlet (mg/L)

Chisholm, Minnesota Axler et al (1996) Autumn 1995 5.60 15.1 1.8 10.1 0.9 31.60 5.91 13.30 0.07

became the overwhelmingly dominant plant species,

occu-pying nearly two thirds of the wetland area Concentrations

of total phosphorus (TP), total suspended solids (TSS) and

inorganic suspended solids (ISS) were reduced by 31, 65, and

76%, respectively, during recirculation, and maintained

satis-factory water quality Parameterization of the k–C* model for

TSS or TP showed that mean target levels could be achieved

when the ratio of pond surface to wetland surface was 12:1

Tilley et al (2002) concluded that constructed wetlands can

perform satisfactorily as recirculation filters in large-scale

shrimp farms

Similar results were obtained in a combination system

in Taiwan (Lin et al., 2003b; 2005) A FWS unit was

fol-lowed by a HSSF unit, and integrated into a

commercial-scale recirculating aquaculture system for intensive shrimp

culture This study investigated performance of the treatment

wetlands for controlling water quality Removals were TSS

(55–66%), BOD5 (37–54%), total ammonia (64–66%), and

nitrite (83–94%) for recirculating water under high hydraulic

loading rates (1.57–1.95 m/d) The area ratio of the

demon-stration wetlands to culture tank was 0.43 It was concluded

that a constructed wetland was technically and economically

feasible for managing water quality of an intensive

aquacul-ture system

25.2 FOOD-PROCESSING WASTEWATERS

S UGAR R EFINING

Wastewaters from sugar-processing plants have been treated

in natural wetlands (Gambrell et al., 1987) The form of the

BOD in these wastewaters is sucrose and its precursors and

derivatives It is therefore not surprising that removal follows

the pattern established for BOD in municipal wastewaters

The k-C* model does a good job of fitting the batch

disap-pearance of soluble TOC (Figure 25.6) The rate constant

is similar to that obtained from BOD data from numerous domestic wastewater wetlands (see Chapter 8)

American Crystal Sugar’s Hillsboro, North Dakota, sugar beet refinery, built in 1973, generates about 260,000 m3

of wastewater during the eight-month processing season The plant uses a 64-ha FWS treatment wetland for advanced secondary since 1993 (Anderson, 1996) The major emergent

macrophtye species include cattails (Typha latifolia, Typha angustifolia), bulrushes (Scirpus acutus), and common reed (Phragmites australis) American Crystal Sugar subsequently

built a second 64.8-ha FWS wetland at their Drayton, North Dakota, plant in 1993

Vertical subsurface flow wetlands were tested in the British Isles (Morris and Herbert, 1997) Three first-stage beds were followed by four second-stage beds, totaling

400 m2 for the whole pilot system Warm wastewaters during the beet processing season showed good removals of 87, 88, and 80% for COD, TSS, and ammonia nitrogen, respectively,

at 2.5–7.4 cm/d hydraulic loading Cooler temperatures in the nonprocessing season gave removals of 74, 88, and 93%, at 17.4–18.6 cm/d hydraulic loading

Constructed vertical upflow SSF wetlands are used by Usina Costa Pinto S.A in Piracicaba, Brazil, to treat waste-water from a sugar mill Rice is planted in the several cells

in series, situated in terraces on a modestly steep slope (Figure 25.7) The strong influent has BOD up to 10,000 mg/L, which is reduced by about 50%

P OTATO P ROCESSING

Potato processing produces wastewaters with high starch content This material is highly biodegradable, and has been the target of constructed wetlands treatment since 1990

(de Zeeuw et al., 1990).

Lamb-Weston’s Connell, Washington, facility uses free water surface treatment wetlands as part of its wastewater

0 20 40 60 80 100 120 140 160 180 200

Time (hr)

Low High Model Low Model High

FIGURE 25.6 Disappearance of soluble carbon in wetland microcosms dosed with sucrose Data represent two levels of starting

concentra-tion, two levels of stirring, and two levels of supplementary nutrient addition Model lines are for k = 36 m/yr and C* = 20 mg/L (Data from Gambrell et al (1987) Journal of the Water Pollution Control Federation 59(10): 897–904; graph from Kadlec and Knight (1996) Treatment Wetlands First Edition, CRC Press, Boca Raton, Florida.)

treatment system to handle the approximately 5,300 m3 of

wastewater it produces each day (Burgoon et al., 1999) The

goal of the project is to remove TSS and COD and to reduce

nitrogen to the amount needed for growing an irrigated fodder

crop Two pilot projects were conducted (Kadlec et al., 1997)

The first investigated FWS wetlands only and produced 60–

70% reduction in COD with two to four days’ detention The

second was a prototype for the full-scale system, consisting

of a FWS–VSSF–FWS combination

In the full-scale project, clarified wastewater is pumped

to a 10-ha lined FWS treatment wetland (Figure 25.8),

fol-lowed by a 4-ha vertical SSF flow system, then to a 2-ha

denitrification FWS treatment wetland, and finally ends in a

500-mL storage pond (Burgoon et al., 1999) Treated water

is then used for seasonal irrigation on 250 to 500 ha of

fod-der crops The first FWS wetland provides TSS and COD

removal, whereas the VSSF component is highly effective at

nitrification The final FWS wetland provides denitrification

utilizing a small amount of raw influent as a carbon source

(Burgoon, 2001) The holding pond provides storage so that

the water may be irrigated only during the growing season The design hydraulic retention time is 6.6 days (summer) and ten days (winter)

achieved 86–89% reduction, corresponding to k = 48 m/yr and C* = 4.2 mg/L for an assumed PTIS = 5 (see Chapter 8)

A relatively smooth reduction occurred along the flow tion (see Figure 8.7, Chapter 8)

direc-The Eichten Cheese, Center City, Minnesota, system sists of a septic tank, subsurface flow constructed wetland,

con-FIGURE 25.7 Vertical flow rice wetlands at Piracicaba, Brazil Very strong sugarmill effluents (up to 10,000 mg/L BOD) flow upward

through the root zone.

FIGURE 25.8 Potato wastewater wetland at Connell, Washington.