INFORMATION TO USERS

INFORMATION TO USERS
INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI films
the text directly from the original or copy submitted. Thus, some thesis and
dissertation copies are in typewriter face, while others may be from any type of
computer printer.
The quality of this reproduction is dependent upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality illustrations
and photographs, print bleedthrough, substandard margins, and improper
alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete manuscript
and there are missing pages, these will be noted.
Also, if unauthorized
copyright material had to be removed, a note will indicate the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and continuing
from left to right in equal sections with small overiaps.
Photographs included in the original manuscript have been reproduced
xerographically in this copy.
Higher quality 6" x 9" black and white
photographic prints are available for any photographs or illustrations appearing
in this copy for an additional charge. Contact UMI directly to order.
ProQuest Information and Learning
300 North Zeeb Road, Ann Arbor, Ml 48106-1346 USA
800-521-0600
ElEMOVAL OF VIRUSES AND POLLUTION INDICATORS IN
CONSTRUCTED WETLANDS
by
Juan Antonio Vidales Contreras
A Dissertation Submitted to the Faculty of the
DEPARTMENT OF SOIL, WATER AND ENVIRONMENTAL SCIENCE
In Partial Fulfillment of the Requirements
For the Degree of
DOCTOR OF PHILOSOPHY
In the Graduate College
THE UNIVERSITY OF ARIZONA
2001
UMI Number: 3010253
®
UMI
UMI Microform 3010253
Copyright 2001 by Bell & Howeai Information and Learning Company.
All rights reserved- This micxroform edition is protected against
unauthorized copying under Title 17, United States Code.
Bell & Howell Informatiion and Learning Company
300 North Zeeb Road
P.O. ©0x1346
Ann ArboB-, MI 48106-1346
2
THE UNIVERSITY OF ARIZONA ®
GRADUATE COLLEGE
As members of the Final Examination Conmiittee, we certify that we have
Juan Antonio Vidales Contreras
read the dissertation prepared by_
REMOVAL OF VIRUSES AND POLLUTION INDICATORS IN CONSTRUCTED
entitled
WETLANDS
and recommend that it be accepted as fulfilling the dissertation
requirement for the Degree of
Doctor of Philosophy
Jo-'OI
Arthur W. Warrick
Date
^ I
Kevin E. Lansey
Date
Ian L. Pepper
//
/ ro /o /
"Datey
Martin M. Karpiscak
^
/•g / g./
Date
Charles P. Gerba
V //g /O(
Date
Final approval and acceptance of this dissertation is contingent upon
the candidate's submission of the final copy of the dissertation to the
Graduate College.
I hereby certify that I have read this dissertation prepared under my
direction and recommend that it be accepted as fulfilling the dissertation
requirement.
Dissertation Director
Date^
3
STATEMENT BY THE AUTHOR
This dissertation has been submitted in partial fulfilment of requirements for an
advanced degree at The University of Arizona and is deposited in the University Library
to be made available to borrowers according to the rules of the Library.
Brief quotation from this dissertation are allowable without special permission,
provided that accurate acknowledgment of the source is made. Requests for permission
for extended quotation firom or reproduction of this manuscript in whole or in part may
be granted by the head of the major department or the Dean of the Graduate College
when in his or her judgement the proposed use is in the interests of scholarship. In other
instances, however, permission must be obtained from the au
SIGNED
4
ACKNOWLEDGEMENTS
I want to express my appreciation to CONACYT for its financial support during
my Ph. D. studies at the University of Arizona. I am deeply grateful to Dr. Charles P.
Gerba and Dr. Martin M. Karpiscak for their guidance through the development of this
research work. Extended thanks are also due to Mrs. Susan Hopf for her valuable
collaboration during my experiments. I wish to acknowledge to Dr. Carlos Enriquez for
his invaluable comments to this manuscript and to thank the others members of my
committee Drs. Kevin E. Lansey, Arthur W. Warrick, and Ian L. Pepper.
I dedicate this dissertation to Elizabeth, Omar, and Priscila for their love and
endless support, especially during the most difficult moments of my Ph. D. program.
5
TABLE OF CONTENTS
[.ABSTRACT
9
2. INTRODUCTION
10
2.1 Constructed Wetlands
10
Classification
10
Wetland Substratum
11
Wetland Plants
12
Common reed (Phragmites australis)
12
Cattail (Typha spp)
13
Bulrush {Scirpus spp)
14
Microbial Populations
2.2 Pathogens in Municipal Wastewater
14
15
Pathogenic Bacteria
15
Human Enteric Viruses
16
Parasites
16
2.3 Inactivation in the Environment
17
Temperature, Sunlight, and Metal Ions
17
Microbial Activity, Organic Matter, and pH
19
2.4 Indicator Microorganisms
20
6
Removal in Wetlands
2.5 Virus Models
Removal in Wetlands
2.6 Problem Definition
22
23
24
25
3. PRESENT STUDY
26
4. REFERENCES
28
5. APPENDIX A:
RELATIVE TRANSPORT OF Br AND PRDl IN A SURFACE FLOW
CONSTRUCTED WETLAND
38
Abstract
39
Introduction
40
Materials and Methods
42
Results and Discussion
52
Conclusions
61
Acknowledgments
62
References of Appendix A
63
7
6. APPENDIX B:
VIRUS
REMOVAL
FROM
WASTEWATER
IN
SUBSURFACE FLOW CONSTRUCTED WETLAND
A
MULTI-SPECIES
81
Abstract
82
Introduction
83
Materials and Methods
84
Results and Discussion
92
Conclusions
101
Acknowledgments
103
References of Appendix B
104
7. APPENDIX C:
REMOVAL OF MICROBIAL AND CHEMICAL INDICATORS OF POLUTION
IN A SURFACE FLOW CONSTRUCTED WETLAND
116
Abstract
117
Introduction
118
Materials and Methods
119
Results
123
Discussion
134
Conclusions
140
Acknowledgments
References of Appendix C
9
1. ABSTRACT
Tracerstudies using Br" and bacteriophage PRD1 in both surface and subsurface flow
constructed wetlands were conducted to analyze their hydrodynamic behavior and
efficiencies in removing viruses from wastewater. A survival test in situ was also conducted
to analyze the persistency of PRD 1 in wetland environments. Concurrently, a sampling
program for microbial and chemical indicators in the surface flow wetland for a period of
16 months was conducted. The tracer studies revealed a reduction of 99 and 84 percent in
the subsurface and surface flow wetland, respectively. Bromide recovery at the outlet of
both wetland systems was about 75 percent. The Convective-Dispersion Equation was able
to predict the observed PRD I and Br" breakthrough curves obtained during the tracer study
in the surface flow wetland. The monitoring program of pollution indicators showed that
biochemical oxygen demand and total suspended solids can be reduced efficiently, reaching
the tertiary effluent standard of 10 mg L"' required by The Arizona Department of
Environmental Quality. This sampling program suggested that coliphages may be a better
indicator of fecal contamination than total and fecal coliforms in surface flow wetlands.
I
10
2. INTRODUCTION
2.1 Constructed Wetlands
Constructed wetlands (CW) have been actively used to reduce suspended solids,
chemical pollutants and pathogenic microorganisms from wastewater. They are designed
to maximize efficiency of
natural self-purification processes carried out by plants,
microbial populations, and environmental factors. CW can be established almost at any
location to treat a wide diversity of wastewater with low cost and minimum trained
personnel. Basically, these systems are artificial channels, slightly inclined, and often lined
to avoid water seepage. The channels may be filled or not with some type of coarse material
like gravel or sand, and vegetated with a single or multiple plant species. Wastewater must
pass through the rooted-basin, above the media or through the media resulting in microbial
degradation, pathogenic inactivation, and chemical processes. After passing through the
wetland, water is typically used for landscape irrigation or is discharged into receiving water
bodies or recharge basins for additional soil aquifer treatment.
Classiflcation
Constructed wetlands are classified by Kadlec and Knight (1996) as Surface (SW)
and Subsurface (SSW) Flow Wetlands based on whether the flow of water occurs over or
below the surface of the wetland substratum. On the other hand, Vymazal et al. (1998)
11
recently proposed a classification based on the type of macrophytes planted in the wetland.
In this classification wetlands are labeled: a) free-floating macrophyte systems; b)
submerged macrophyte-based systems; or c) rooted emergent macrophyte-based systems.
The terminology proposed by Kadlec and Knight (1996) was adopted for this study.
Surface flow wetlands are typically operated at less of 0.4 m of depth and hydraulic loading
rates ranging from 4 to 4 mm dayGenerally, subsurface flow wetlands are channels filled
with gravel or sand, operated at less than 0.6 m of saturated thickness and hydraulic loading
rates ranging from 10 to 60 mm day"' (Geller, 1997).
Wetland Substratum
Wetland substratum used for rooting can be soil, sand or gravel. In SSW, the initial
hydraulic conductivity should be adequate to support porosity reduction resulting from
growth of plant roots and deposition of suspended solids from wastewater entering the
system. Tanner et al. (1998) observed a reduction of detention time that was positively
related with the level of organic loading received by the wetland. Sanford et al. (1995), also,
found a considerable reduction of the hydraulic conductivity in a sand-and-gravel bed after
a year of landfill leachate addition while beds filled with coarser materials experienced little
reduction in hydraulic conductivity.
12
Wetland Plants
Emergent aquatic macrophytes are plants adapted to grow in water or on a substrate
with excessive water content. These macrophytes provide surface area for attacliment of
microbial populations and the physical matrix for wastewater filtration. Also, they provide
oxygen by passive diffusion and convective flow from the atmosphere to their buried parts
using their internal lacunal system (Vymazal et al. 1998). During internal gas transport,
oxygen is supplied to aerobic and facultative anaerobic populations attached to or adjacent
to the external root surface. These organisms in tum are responsible for reducing the
biochemical oxygen demand (BOD5) of wastewater (Brix, 1994). The primary aquatic
species used in constructed wetlands are common reed {Phragmites australis), cattail {Typha
spp) and bulrush {Scirpus spp).
Common reed {Phragmites australis)
The common reed can grow under permanent inundation and readily tolerates up
to 45 g L"' of salt content, and pH from 2 to 8 (Reed et al., 1988). Its deep root system
creates a great volume of active rhizosphere. Geller (1997) found that in German speaking
countries SSW planted with common reed were effective in reducing BOD5, Chemical
Oxygen Demand (COD), Phosphorous (P), and Nitrogen (N). Brix (1987) also observed
removal efficiencies over 88 percent for BOD5, P, and N
in SSW, vegetated with
Phragmites australis. In France, this type of ecosystem operated at short residence times
13
(1 to 4 h) removing 93 percent of the inflow helminth eggs; however, the system was
ineffective in reducing COD, N, and P. In the U.S. common reeds have been used
successfully in treatment wetlands; however, they are preferentially used in Europe (Reed
et aL, 1995).
Cattail {Typlta spp)
Typha angiistifolia (narrow-leaf cattail) and T. latifolia (broad-leaf cattail) are the
most typical plants found in constructed wetlands in the USA. Narrow-leaf plants tolerate
salinity levels up to 30 g L"' compared to the much lower tolerance of the broad-leaf plants
of less than I gL"'. Root penetration ofcattail is relatively shallow reaching only about 0.30
m of depth (Reed et aL, 1995). Experimental gravel beds based SSW at Richmond, South
Wales, Australia, planted with Typha latifolia were effective in removing 80, 66,91 and 90
percent of the influent load of suspended solids, COD, and fecal and total coliforms,
respectively. According to Kadlec and Knight (1996) these plants
have a superior
performance for wastewater treatment compared to bulrushes in SSW. BChatiwada and
Polprasert (1998) reported that SW systems planted with Typha angustifolia were capable
of removing some 99.5 percent of the initial load of fecal coliforms.
14
Bulrush (Scirpus spp)
In Arizona, bulrush and cattail have been used in most of the major demonstration
and research wetland projects: Constructed Ecosystems Research Facility (CERF), Tres
Rios Demonstration Wetlands, Sweetwater Wetlands and Recharge Project, and Dairy
Constructed Wetland Demonstration and Research Project (Karpiscak et aL, 1999). Bulrush
species used in constructed wetlands include: Scirpus acutus, S. cypernius, S. fluviatilis, S.
robiistus, S. validus, S. laciistris, and S. olneyi. Optimum pH for most of these plants is from
4 to 9 and salinity tolerance ranges from 0 to 5 g L"'. Constructed SSW and SW vegetated
with Scirpus have shown the ability to remove about 99 percent of viruses from wastewater
(Chendorain etal., 1998; Gersbergef a/., 1987). Tanner era/. (1998) reported that chemical
pollutants such as CBOD, total N, and P have been removed in SSW with average
efficiencies up to 91, 65, and 38 percent, respectively.
Microbial Populations
Bacteria are the primary microbial population carrying out pollutant removal from
wastewater in wetlands. They are found at concentrations ranging from 10® to lO' cfu g"' of
substratum in wetlands vegetated with Typha or Phragmites, with higher densities in the
rhizosphere (Gray and Biddlestone, 1995). In this zone, oxygen is available for aerobic
bacteria that degrade organic matter to simpler compounds. Ammonium is transformed to
nitrate by nitrifying bacteria such as Nitrosomonas and Nitrobacter. In anaerobic regions.
15
fermentative, and methanogenic bacteria transform organic matter to methane using a carbon
source such as methanol, acetate, or formate. Methane is emitted to the atmosphere directly
or through plant leaves and steams.
Other organisms found in wetlands include fungi, algae, viruses, and protozoa. Fungi
are found at high concentrations (1.6 x10® cfu g'') in the substrate areas of the rizhosphere
(Gray and Biddlestone, 1995). They play an important role in degradation of recalcitrant
organic compounds such as cellulose and lignin (Atlas and Bartha, 1993). Algae are
photoautotrophic organisms responsible for daily fluctuations of Oj and CO, as well as pH
because of their photosynthetic activity. They have been linked to the secretion of
substances that are toxic to E. coli and Vibrio cholera (Maynard et aL, 1999). Coliphages
are viruses that occur in the intestines of warm-blooded animals infecting coliform bacteria.
These viruses are discharged into wetlands by wastewater effluents or wild animals where
may be inactivated by bacterial enzymes and algae (Oliver and Herman, 1972; Sobsey
and Cooper, 1971). Protozoa are invertebrate organisms and are believed to contribute to
the removal of virus and bacteria in treatment facilities (Britton, 1994).
2.2 Pathogens in Municipal Wastewater
Water borne pathogens are frequently transmitted to healthy or immunocompromised
hosts by the ingestion of food or contaminated water. The potential risk can be appreciated
16
by the approximately 76 million people that annually become ill by the ingestion of
contaminated food in the USA (CDC, 2000). During 1971 to 1998, groundwater and surface
water were responsible for 40 and 50 percent, respectively, of waterbome disease outbreaks
related to drinking water (EPA, 1999).
Pathogenic Bacteria
Bacterial diseases caused hy Salmonella, Shigella, Vibrio cholerae, Escherichia coli.
Yersinia and Campylobacter have been reported as the major causes of waterbome outbreaks
related to wastewater (Bitton, 1994). Salmonella, alone, is believed responsible for diarrhea
in 2 to 4 million people, annually in the USA. These bacteria infect healthy individuals with
the elderly, infants, and AIDS infected persons showing more severe symptoms of
salmonellosis (Rusin era/., 2000). In 1999, V. cholerae caused 254,310 cases of cholera
in sixty one countries including the USA (WHO, 1999). Campylobacter and Salmonella
were the primary cause of illness caused by contaminated food from 1996 to 1999 (CDC,
2000). In the USA, drinking water for the period from 1971 to 1996 caused 13 percent of
bacterial disease outbreaks (EPA, 1999)
Human Enteric Viruses
The most common enteric viral agents of gastroenteritis are rotavirus, adenovirus,
Norwalk virus, astrovirus, and calicivirus (Gerba and Enriquez, 1999). Enteric viruses are
17
obligate intracellular parasites ranging in size from 20 to 70 nm. Infected individuals mayexcrete 10" viral particles per gram of feces (Rusin et al., 2000). These viruses are
responsible for 5 to 10 million deaths worldwide (Walsh and Warren, 1979). In the USA
210,000 children with a gastrointestinal infection are hospitalized annually at a cost of one
billion dollars (Ho et aL, 1998). Rotavirus has been reported as the most common cause of
infection in the world (Gerba and Enriquez, 1999).
Parasites
In recent years, Giardia lamblia and Cryptosporidium parvum have been protozoa
parasites of major health concern. C. parvum was responsible for the largest waterbome
outbreak ever documented in the United States (MacKenzie et al., 1994). Whereas, G.
lamblia is believed to be the agent of 27 percent of waterbome diseases caused by drinking
water between 1991 and 1992 (EPA, 1999). These pathogens are more resistant than viruses
and bacteria to chlorination; therefore, they may be found in reclaimed wastewater and
sediments. Persons infected with G. lamblia, may discharge up to 10^ cysts per day into the
environment (Rusin et al., 2000).
18
2.3 Inactivation in the Environment
Temperature, Sunlight and Metal Ions
Over time pathogens are inactivated in the environment. Commonly, their
inactivation follows a first order model with a decay rate that depends on environmental
factors such as temperature, sunlight, and metal ions.
Temperature has shown to be a key parameter. Enriquez et al. (1995) found a
consistent rate of inactivation of poliovirus 1, and adenovirus 40 and 41 by raising tap water
temperature from 4 to 23 °C. O'Brien and Newman (1977) observed an inactivation rate for
poliovirus of one log-unit reduction per 25 h, when the virus was incubated in situ at a
temperature between 23 and 27 °C. This same one log reduction was observed after 19 and
7 h for poliovirus 3 and coxsackievirus A-13, respectively. A longer time was observed for
an equivalent inactivation of poliovirus 1 and coxsackievirus B-1 at temperatures from 4
to 8 °C. Inactivation of PRDl and MS2 was similar at 7°C; however, PRDl was 7 to 10
times more persistent than MS2 between 10 and 23 °C (Yahya et al, 1993). Bacteria are
also highly sensitive to temperature which is commonly considered in mathematical models
that predict kinetic removal of coliform in the environment (Auer and Niehaus, 1993;
Khatiwada and Polprasert, 1998; Marias, 1974).
Sunlight is also lethal for many microorganisms. Inactivation rates were twice as
high for CB380 and CB7<1> bacteriophages incubated in situ in experimental microcosms
19
than the decay rate observed under dark condition or at 1 m of depth in a water body
(Wommack et ai, 1996). Noble and Fuhrman (1997) reported that solar radiation always
increased the decay rates of marine bacteriophage which ranged from 6.6 to 11.1 percent h"'.
In tertiary lagoons, Parhad and Rao (1972) observed a direct correlation of E coli
inactivation and light intensity; however, algae growth was also positively correlated with
light intensity and was the primary mechanism of E. coli removal.
Metal ions are capable of bacterial and viral inactivation. Sagripanti et al. (1993)
observed that five type of viruses whether enveloped or noenveloped, single or double
stranded, DNA or RNA were inactivated by the addition of cupric or ferric ions. The
addition of peroxide increased significantly the antiviral effect of copper. Apparently, the
oxidation potential of the metal ion is responsible for viral inactivation. Metal ions bind to
functional molecules or groups and subsequent reductions by superoxide radicals and HjO,
generate hydroxide radicals causing multiple damages on the binding molecule (Thurman
andGerba, 1988).
Microbial Activity, Organic Matter, and pH
Contradictory results have been reported about the effect of organic matter on virus
inactivation (Yates and Yates, 1991). Berry and Noton (1976) found that T2 inactivation was
about 7 log in 14 days of incubation in laboratory experiments. However, no antiviral
activity was observed in filtered or autoclaved seawater. Ward et al. (1986) examined the
20
inactivation mechanisms of enteric viruses in fresh water. Their results suggested that
proteolytic enzymes from bacterial isolates inactivated echovirus particles by exposition of
RNA to nuclease enzymes after cleavage of capsid proteins. Survival of enteric adenoviruses
was observed to be longer in tap water than in wastewater while polio 1 was practically
unaffected (Enriquez et al., 1995). On the other hand. La Belle and Gerba (1982) showed
that organic matter protect enteroviruses from antiviral factors in seawater.
Pathogen inactivation is also related to pH. In wetlands with open water areas or
oxidation ponds, algae proliferation may induce an increase of pH up to 10.4 that might
actively increase virus and bacterial inactivation (Kadlec and Knight, 1976; Yates and
Yates,1991 ; Maynard er a/., 1999).
2.4 Indicator Microorganisms
Detection of pathogenic microorganisms in environmental samples is a difficult,
expensive, and time consuming process that requires trained personal. Therefore, indicator
microorganisms are commonly used for detection of fecal contamination and indirectly as
a presumptive test for the presence of pathogenic microorganisms in food and water. An
ideal indicator has the following characteristics (Britton, 1994): 1) it is a member of the
intestinal tract of warm-blood animals; 2) it is detectable when enteric pathogens are present
and undetectable in uncontaminated samples; 3) it is usually present in higher densities than
21
pathogens in environmental samples; 4) it is more resistant to environmental stress and
disinfection processes than are pathogens; 5) it is incapable of multiplying in the
environment; 6) it is inexpensive, easily, and quickly detected; and 7) it is not pathogenic.
The most widely use indicator of fecal contamination in water is the coliform group.
Microbiological standards to meet public health protection goals are based on total and fecal
coliform concentrations in drinking and wastewater (Gerba, 2000). The coliform group
includes: E. coli, Enterobacter, Klebsiella, and Citrobacter which are aerobic, facultative
anaerobic, gram-negative, non-spore forming, rod-shape bacteria that ferment lactose with
gas production at 35 °C within 48 h. Fecal coliforms, a subgroup of coliforms, are
characterized by their ability to ferment lactose producing acid and gas within 24 h at 44.5
°C and include E. coli and the genera Klebsilla (Bitton, 1994).
Coliphages is another indicator of water contamination by fecal material since its
size, morphology, and structure are similar to enteric viruses (Gerba, 2000). Stetler (1984)
observed that enetroviruses were better con-elated with coliphages than total coliforms,
fecal coliforms, and fecal streptococci during drinking water treatment.
Fecal streptococci which is also used as an indicator grows in 6.5 percent sodium
chloride, at a pH of 9.6, and a temperature of 45 °C. The ratio of fecal coliform concentration
versus fecal streptococci is used as an indication of human fecal versus
animal
contamination. A ratio of 0.7 is an indicator of animal fecal contamination while a ratio
22
greater than 4 is used as an indicator of human contamination (Geldreich and Kenner,
reported in Gerba, 2000).
Removal in Wetlands
Studies conducted in CW have shown a large coefficient of variation of total and
fecal coliform populations in the wetland outflowing water. Often, averaged removal
efficiencies over 90 percent of the influent load are obser\'ed (Bahlaoui et ah, 1997; Perkins
and Hunter, 2000). This efficiency is inversely related to the hydraulic loading rate (Green
etal., 1997; Tanner era/., 1998) and may decrease considerably if vegetation density is low
(Gersberg er a/., 1989).
In constructed wetland systems, coliforms could be an unreliable indicator of the
abilit>' of the system to remove pathogens from wastewater because coliforms can occur
naturally in the environment. Van Donsel et al. (1967) observed regrow of coliform bacteria
associated with an increase of temperature and rainfall in soils amended with nutrients. A
similar condition was also found by Elmund et al. (1999) in a treatment facility. In this
case, the high numbers of Klebsilla coincided with an increased of carbohydrates in the
incoming wastewater. In subtropical environments, Solo-Gabriele et al. (2000) reported
several orders of magnitude growth of E coli in sediments. Moorhead et <2/.(1998) also
found high densities of an "unclear origin" of E coli and total coliforms in urban water
bodies. The presence of wild animals might explain these high densities of total and fecal
23
conforms often found in the environment (Gould and Fletcher, 1978; Have, 1973; Levesque
et ai, 1993). However, specialized studies should be conducted to identify their origin in
natural water systems (Griffin et ai, 2000).
2.5 Virus Models
Bacteriophages have been used to simulate transport of enteric viruses in the
environment since they are not pathogenic, their analysis is inexpensive, easy, rapid, and
highly concentrated stock solutions can be prepared in the laboratory. Many experiments
have been conducted using bacteriophage in the laboratory (Bales et al., 1991; Jin et aL,
1997; Powelson et ai, 1991; Wang et aL, 1981; Yahya et aL, 1993) and field (Deborde et
aL, 1999; Schijven et aL, 1999). Moreover, their performance in the environment has been
compared to animal viruses (Gerba and Goyal, 1981; Goyal and Gerba, 1979; Yates et aL,
1985). For example, the bacteriophage MS2 is ssRNA, 23 run in size, similar to astroviruses,
enteroviruses, and Norwalk (Rusin et aL, 2000). This bacteriophage has shown little
adsorption at pH between 7.5 and 8 in soils under unsaturated (Powelson er <3/., 1990) and
saturated (Jin et aL, 1997) conditions. In groundwater MS2 survival is similar to animal
viruses (Yates et aL, 1985) but shorter than another bacteriophage PRDl (Yahya et aL,
1993), which is also commonly used in tracer studies.
24
The coliphage PRD1 is dsDNA virion and measures 62 nm with a protein capsid that
surrounds an internal lipid-protein membrane (Caldentey et al., 1990) resembling the human
rotaviruses. It is positively charged at pH below 4.5 (Bales, et aL, 1991) comnpared with
MS2 which is positively charged only below pH 3.9 (Zreda reported in Gerba, 11984). Both
MS2 and PRX)1 viruses do not adsorb on to silica particles at 7 pH (Bales et aL., 1991). At
lower pH, PRDl is more hydrophobic than MS2 and consequently a greater abosorption of
PRD1 is observed.
Other virus models have also been used in transport experiments. One of these is
0X174 which is 27 nm in size and isoelectric between pH 6.6 and 6.8 (Dowd ef aL, 1998;
Fujito and Lytle, 1996). This virus is less hydrophobic than MS2; however, eslectrostatic
interactions might be the primary mechanism of adsorption in soils at a neutra .l pH (Jin et
aL, 1997). Another bacteriophage f2, similar to MS2, and 0X174, belongs to tLhe group of
F-specific RNA bacteriophages. Goyal and Gerba (1979) described it as a virus of very low
tendency for attachment to soils particles.
Virus Removal in Wetlands
Virus removal studies in wetlands have been conducted with different aapproaches.
Scheuerman et aL (1990) and Karpiscak et aL (1995) quantified virus removal by the ratio
of the inflow and outflow viral load while Gersberg et aL (1987) and Vinlfluan (1996)
conducted in situ virus survival studies. Tracer studies using MS2 as a bacteriopohage model
25
in a surface flow constructed wetland were used by Chendoraine et ai, (1998) to determine
virus removal and model its transport. These studies have reveled that virus removal
efficiencies of 99 percent and removal rates of 0.44 day"' can be achieved in wetlands.
Apparently, reduction of the inflow virus load is unaffected by climatic conditions; however,
it is enhanced by vegetation density and filtration processes through the wetland (Gersberg
etal, 1987).
2.6 Problem Definition
Constructed wetlands have been actively used in wastewater treatment making
effective use of the processes that occur naturally in the environment for water treatment.
They can be applied in a wide diversity of localities, climatic conditions, and wastewater
treatment demands. Additionally, their operation costs are lower than traditional wastewater
treatment facilities (Kadlec and BCnight, 1996). However, it is recognized that their design
faces engineering uncertainties because under different circumstances of operation the
removal efficiency of the system may vary widely (Unsoeld, 2000).
In the current study, the hydraulic behavior of subsurface and surface flow wetlands
have been related to their viral removal efficiency observed during tracer studies and in situ
survival tests. Removal of bacteriophage PRDl and transport of a conservative tracer in a
surface flow wetland were modeled and information about PRDl movement in the vadose
zone was analyzed. In addition, the spatial behavior of chemical and microbial indicators in
26
a wetland system operated primarily with backwash water was monitored for a 16 monthperiod. The usefulness of total and fecal bacteria and indigenous coliphage as indicator
microorganisms to evaluate bacteria and virus removal in constructed wetland systems was
discussed.
3. PRESENT STUDY
The methods, results, and conclusions of this study are presented in the appendices
included in this dissertation. The most important findings and a brief discussion are
presented in the following summary.
SUMMARY
The tracer test conducted in the subsurface flow constructed wetland indicated a high
degree of plug flow in the hydrodynamic performance of the, 6 year old, gravel bed. During
the study, most of the PRD1 was recovered within the following four days after the tracer
injection. At the end of the test, the average PRDl removal was 98.8 percent and a mass
recovery of 75 percent for Br was observed. The PRDl reduction was associated with
removal and inactivation rates of -1.17 and -0.16 day"', respectively. Apparently, virus
removal in subsurface flow wetlands is mostly due to adsorption processes with a small dieoff contribution.
27
An additional tracer study was conducted in a surface flow constructed wetland that
consists of pool-vegetated riffle zone sequences designed to remove suspended solids from
backwash water before aquifer recharge. During the study, samples from the surface and
subsurface water were obtained. In the vadose zone, PRDl was observed after the
appearance of Br in the collected samples while in the surface, the concentration-time
distributions of Br' were simulated by a plug flow model with a high longitudinal dispersion.
The convective-dispersion equation and the first order reaction approach were capable of
simulating PRD1 displacement and removal through the wetland. At the outlet of the system,
recovery of Br" and PRDl after 7.3 days was 86 and 16 percent, respectively. PRD I decay
rate was estimated to be 0.3 day"' by regression analysis and mathematical fitaess to the
experimental data. The findings of these studies suggest that contaminant transport in
wetlands can be modeled with the convection-dispersion model (CDE) including the first
order reaction approach.
The sampling program was conducted in the east polishing subsystem v/hich consists
of a settling basin, a small artificial stream, and the latter surface flow wetland where the
tracer study was conducted. Microbial populations of coliphage, total coliforms and fecal
coliforms were quantified in samples collected at five sampling sites along the system. In
addition to this microbial indicators, chemical and physical parameters such as total
28
suspended solids (TSS), biochemical oxygen demand (BOD5), chloride (CI"), chlorine (CU),
sulphate (SOJ temperature and pH also were quantified.
Coliphage removal in the system ranged from 46 to 93 percent during the 16 month
of sampling. Whereas, the population of TC and FC were consistently higher at the outlet
of the polishing wetland than in the water entering the system. In contrast, BOD5, and TSS,
were reduced efficiently below the secondary effluent standard of 30 mg L"' required by The
Arizona Department of Environmental Quality. The results of this study suggest that
coliphages are a better indicator than coliforms in wetlands since bacterial populations may
increase by regrowth or animal fecal contributions. This study also revealed that the stream
and wetland sections of the east polishing system are unable to reduce significantly BOD5,
TSS, and turbidity.
28
4. REFEEIJENCES
Atlas R.M. and Bartha R. (1993) Microbial Ecology. Benjamin Cummings, New York.
AuerM.T.andNiehaus S.L. (1993) Modeling fecal coliform bacteria -I. field and laboratory
determination of loss kinetics. Wat. Res., 27, 693-701.
Bahlaoui M.A., Baleux B. and Troussellier M. (1997) Dynamic of pollution-indicator and
pathogenic bacteria in high-rate oxidation wastewater treatment ponds. PVat. Res.,31,
630-638.
Bales R.C., HinkerS. R., KroegerT. W., Stocking K. and GerbaC. P. (1991) Bacteriophage
adsorption during transport through porous media: chemical perturbations and
reversibility. Environ. Sci. TechnoL, 25, 2088-2095.
Berry S.H., and Noton B. G. (1976) Survival of bacteriophage in sea-water. Wat. Res., 10,
323-327.
Britton G. (1994) Wastewater Microbiology. Wiley-Liss, New York.
Brix H. (1987) Treatment of wastewater in the rhizosphere of wetland plants-the root-zone
method. Wat. Sci. Tech., 19, 107-118.
Brix H. (1994) Functions of macrophytes in constructed wetlands. Wat. Sci. Tech, 4, 71-78.
Caldentey J., Bamford J.K.H. and Bamford D.H. (1990) Structure and Assembly of
bacteriophage PRD1, an Escherichia coli virus with a membrane. J. Struc. Biol., 104,
44-51.
CDC (2000) Preliminary footnet data on the incidence of foodbome illnesses. CDC,
Morbidity and Mortality Weekly Report, 49,
201-205 [On line]. Available:
http://w\vw.cdc.gov/epo/mmwr/preview/mm\vrhtml/mna49 lOal.htm
[2000,
December 10]
Chendorain M., Yates M. and Villegas F. (1998) The fate and transport of viruses through
surface water constructed wetlands. J. Environ. Oual., 27, 1451-1458.
Cliver, D.O. and Herrman
J.E. (1972) Proteolytic and microbial inactivation of
enteroviruses. Wat. Res., 6, 797-805.
Deborde D.C., Woessner W. W., Kiley Q. T. and Ball P. (1999) Rapid transport of viruses
in a floodplain aquifer. Wat. Res., 33, 2229-2238.
Dowd S.E., Pillai S. D., Wang S. and Corapcioglu M. Y. (1998) Delineating the specific
influence of virus isoelectric point and size on virus adsorption and transport through
sandy soils. Appl. Environ. Microbiol., 64, 405-410.
Elmund G.K., Allen M. J. and Rice E. W. (1999) Comparison of Escherichia coli, total
coliform, and fecal coliform populations as indicators of wastewater treatment
efficiency. Water Environ. Res., 71, 332-339.
Enriquez C.E., Hurst C. J. and Gerba C. P. (1995) Survival of the enteric adenoviruses 40
and 41 in tap, sea, and waste water. Wat. Res., 29, 2548-2553.
30
EPA, Office of Water (1999) Drinking water criteria document for enteroviruses and
hepatitis a: an addendum. EPA, Washington, DC.
Fujito B.T., and Lytle C. D. (1996) Elution of viruses by ionic and nonionic surfactant. Appl.
Environ. Microbiol., 62, 3470-3473.
Geller G. (1997) Horizontal subsurface flow systems in the german speaking countries:
summary of long-term scientific and practical experiences; recommendations. Wat.
Sci. Tech, 35, 157-166.
Gerba C.P. (1984) Applied and theoretical aspects of virus adsorption to surfaces. y4(^v. Appl.
Environ. Microbiol., 30, 133-168.
Gerba C.P. (2000) Indicator Microorganisms. In Maier, R.M., Pepper I. L. and Gerba C. P.
(ed.) Environmental Microbiology. Academic Press, San Diego, CA, pp. 491-503.
Gerba C.P., and Enriquez C. E. (1999) Health risks of enteric viral infections in children.
EPA, Office of Water, Washington, D. C.
Gerba C.P., and Goyal S. M. (1981) Quantitative assessment of the adsorptive behavior of
viruses to soils. Environ. Sci. TechnoL, 15, 940-944.
Gersberg R.M., Lyon S. R., Brenner R. and Elkins B. V. (1987) Fate of viruses in artificial
wetlands. Appl. Environ. Microbiol., 53, 731-736.
Gould D.J., and Fletcher M. R. (1978) Gull droppings and their effects on water quality.
Appl. Environ. Microbiol., 12, 665-672.
31
Goyal S.G.andGerba C.P. (1979) Comparative adsorption of human enteroviruses, simian,
rotaviruses, and selected bacteriophages to soil. AppL Environ. Microbiol., 38, 241247.
Gray K.R., and Biddlestone A. J. (1995) Engineered reed-bed systems for wastewater
treatment. TIBTECH, 13, 248-252.
Green M. B., Griffin P., Seabridge J. K. and Dhobie D. (1997) Removal of bacteria in
subsurface flow wetlands. Wat. Sci. Tech., 35, 109-116.
Griffin D.W., Stokes R., Rose J. B. and Paul III J. H. (2000) Bacterial indicator occurrence
and the use of an F+ specific RNA coliphage assay to identify fecal sources in
Homosassa Springs, Florida. Microb. Ecol., 39, 56-64.
Have M.R. (1973) Effects of migratory waterfowl on water quality at the Montezume
National Wildlife Refiige, Seneca county, New York. Jour. Research U.S. Geol.
Survey, 1, 725-734.
Ho M., Glass R. I., Pinsky P. F. and Anderson L. (1998) Rotavirus as a cause of diarrheal
morbidity and mortality in the United States. J. Infect. Dis., 158, 1112-1116.
Jin Y., Yates M. V., Thompson S. S. and Jury W. A. (1997) Sorption of viruses during flow
through saturated soil columns. Environ. Sci. TechnoL, 31, 548-555.
Kadlec R.H.and Knight R.L. (1996) Treatment Wetlands. Lewis Publishers, Boca Raton,
Florida.
Karpiscak M.M., Gerba C. P., Watt P. M., Foster K. E. and Falabi J. A. (1995) Multi-species
plant systems for wastewater quality improvements and habitat enhancement. In
Angelakis, A., Asano T., Diamadopoulos E., and Techobanoglous G. (ed.) Second
international symposium on wastewater reclamation and reuse. lAWQ, Iraklio,
Greece, pp. 37-42.
Karpiscak M.M., Wass R. D., Freitas R. J. and Hopf S. B. (1999) Constructed wetlands in
Southern Arizona [Online]. Available: http://ag.arizona.edu/OALS/ALN/aIn45/
wetlands.html.[2001, January 7]
Khatiwada N.R., and Polprasert C. (1998) Kinetics of fecal colifomi removal in constructed
wetlands. The sixth international conference on wetland systems for water pollution
control, Aguas de Sao Pedro, Brazil, Vol. IL
LaBelle R., and Gerba C. P. (1982) Investigations into the protective effect of estuarine
sediment on virus survival. Wat. Res., 16, 469-478.
Levesque B., Brousseau P., Simard P., Dewailly E., Meisels M., Ramsay D. and Joly J.
(1993) Impact of the ring-billed gull {Lams delawarensis) on the microbiological
quality of recreational water. Appl. Environ. Microbiol., 59, 1228-1230.
MacKenzie W., Hoxie N., Proctor M., Gradus M., Blair K., Peterson, D., Kazmierczak, J.,
Addiss D., Fox K., Rose J. and Davis J. (1994) A massive outbreak in Milwaukee of
33
Cryptosporidium infection transmitted through the public waterr supply. N. Engl. J.
Med., 331, 161-167.
Maier R.M. (2000) Biogeochemical Cycling. In Maier, R.M., Pepper L. L. and Gerba C. P.
(ed.) Environmental Microbiology. Academic Press, San Diegoo, CA, p. 585.
Marias G.R. (1974) Fecal bacterial kinetics in stabilization ponds. J. En~v. Eng., ASCE, 100,
119-139.
Maynard H.E., Ouki S. K. and Williams S. C. (1999) Tertiary lagoons: ai review of removal
mechanisms and performance. Wat. Res., 33, 1-13.
Moorhead D.L., Davis W. S. and Wolf C. F. (1998) Coliform densitiecs in urban water of
West Texas. Environmental Health, 14-18.
Noble R.T., and Fuhrman J. A. (1997) Virus decay and its causes in cooastal waters. Appl.
Environ. Microbiol., 63, 77-83.
O'Brien R.T.and Newman .J.S. (1977) Inactivation of polioviruses and • coxsackieviruses in
surface water. Appl. Environ. Microbiol., 33, 334-340.
Parhad N., and Rao N. U. (1972) Effect of pH on survival of E. coli. J. Wat. Pollut. Contr.
Fed.,44, 149-161.
Perkins J., and Hunter C. (2000) Removal of enteric bacteria in a surfa».ce flow constructed
wetland in Yorkshire England. Wat. Res., 34, 1941-1947.
34
Powelson D.K., Simpson J. R. and Gerba C. P. (1991) Effects of organic matter on virus
transport in unsaturated flow. Appl. Environ. Microbiol., 57, 2192-2196.
Powelson D.K., Simpson J. R. and Gerba C. P. (1990) Virus transport and survival in
saturated and unsaturated flow through soil columns. J. Environ. Oual., 19,396-401.
Pretorius W.A. (1962) Some observations on the role of coliphage in the number of £. coli
in oxidative ponds. J. Hyg., 60, 210.
Reed S.C., Middlebrooks E. J. and Crites R. W. (1988) Natural Systems for Waste
Management and Treatment. McGraw Hill Co., New York, N. Y.
Rusin P., Enriquez C. E., Johnson D. and Gerba C. P. (2000) Environmentally Transmitted
Pathogens. In Maier, R.M., Pepper I. L. and Gerba C. P. (ed.) Environmental
Microbiology. Academic Press, San Diego, CA, p. 585.
Sagripanti J., Routson L. B. and Lytle C. D. (1993) Virus inactivation by copper or iron ions
alone and in the presence of peroxide. Appl. Environ. Microbiol, 59, 4374-4376.
Sanford W.E., Steenhuis T. S., Parlange J. Y., Surface J. M. and Peverly J. H. (1995)
Hydraulic conductivity of gravel and sand as substrates in rock-reed filters. Ecol.
Eng., 4, 321-336.
Scheuerman P.R., Bitton G. andFarrah S.R. (1990) Fate of Microbial Indicators and Viruses
in a Forested Wetland. In Hammer D. A. (ed) Constucted Wetlands for Wastewater
Treatment. Lewis Publishers, Michigan.
Schijven J.F., Hoogenboezem W., Hassanizadeh M. and Peters J. H. (1999) Modeling
removal of bacteriophages MS2 and PRDl during dune recharge at Castricum,
Netherlands. Water Resour. Res., 35, 1101-1111.
Sobsey M.D.and Cooper R.C. (1971) Laboratory studies on the survival of poliovirus in
algal-bacterial wastewater treatment systems. 13 th Water Quality Conference.
Engineering Publications, University of Illinois, Urbana., Illinois, pp. 137-147.
Solo-Gabriele H. M., Wolfert M.A., Desmarais T. R. and Palmer C. J. (2000) Sources of
Escherichia coli in a costal subtropical environment. Appl. Environ. Microbiol., 66,
230-237.
Stetler R.E. (1984) Coliphages as indicators of enteroviruses, zl/?/?/. Environ. Microbiol., 48,
668-670.
Tanner C.C., Sukias J. P. S. and Upsdell M. P. (1998) Organic matter accumulation during
maturation of gravel-bed constructed wetlands treating farm dairy wastewater. Wat.
Res., 32, 3046-3054.
Thurman R.B., and Gerba C. P. (1988) Molecular mechanisms of viral inactivation by water
disinfectants. Adv. Appl. Environ. Microbiol., 33, 75-105.
Unsoeld K. (1999) Constructed wetlands for wastewater treatmwnt. [on line]. Available:
http://www.olywa.net/speech/march99/unsoeld2.html [2000, February 15].
Van Donsel D.J., Geldreich E.E. and Clark, N.A. (1967) Seasonal variations in survival of
36
indicator bacteria in soil and their contribution to storm-water pollution. Appl.
Microbiol., 15, 1362-1370.
Vinluan E.A. (1996) Survival of microbial indicators in constructed wetlands. Soil, Water,
and Environ. Sci. University of Arizona, Tucson, AZ., p. 65.
Vymazal J., Brix H., Cooper P. F., Haberl R., Perfler R. and Laber J. (1998) Removal
mechanisms and types of constructed wetlands. In Vymazal J., B.H., Cooper P. F.,
Green M. B. and Haberl R. (ed.) Constructed Wetlands for Wastewater Treatment
in Europe. Backhuys, Leiden, The Netherlands, pp. 17-66.
Walsh J.A., and Warren K. S. (1979) Selective primary health care: an interim strategy for
disease control in developing countries. N. Engl. J. Med., 302, 967-974.
Wang D., Gerba C. P. and Lance J. C. (1981) Effect of soil permeability on virus removal
through soil columns. Appl. Environ. Microbiol., 42, 83-88.
Ward R.L., Knowlton D. R. and Winston P. E. (1986) Mechanism of inactivation of enteric
viruses in fresh water. Appl. Environ. Microbiol., 52, 450-459.
WHO
(1999)
Cholera
and
Epidemic
Dysentery
[Online].
Available:
http://www.who.int/emc/diseases/cholera/index.html [2000, December 10].
Wommack K.E., Hill R. T., Muller T. A. and Colwell R. R. (1996) Effects of sunlight on
bacteriophage viability and structure. Appl. Environ. Microbiol., 62, 1336-1341.
Yahya M.T., Galsomies L., Gerba C. P. and Bales R. C. (1993) Survival of bacteriophages
MS-2 and PRD-1 in ground water. Wat. Res., 27, 409-412.
Yates M.V., and Yates S. R. (1991) Modeling microbial transport in the subsurface: a
mathematical discussion. In Hurst, C.J. (ed.) Modeling the Environmental Fate of
Microorganisms. ASM Press, Washington, D.C., pp. 48-76.
Yates M.V., Gerba C. P. and Kelley L.M. (1985) Virus persistence in groundwater. Appl.
Environ. Microbiol., 49, 778-781.
38
APPENDIX A
RELATIVE TRANSPORT OF BrAND PRDl IN A SURFACE-FLOW
CONSTRUCTED WETLAND
Juan A. Vidales*Charles P. Gerba', and Martin M. Karpiscak^
'Departament of Soil, Water and Environmental Science, University of Arizona, Tucson
AZ 85721." Office of Arid Lands Studies, University of Arizona, Tucson AZ 85719.
*Corresponding Author
39
Relative Transport of Br" and PRDl in a Surface-Flow Constructed Wetland
Juan A. Vidales', Charles P. Gerba', and Martin M. Karpiscak"
'Department of Soil, Water, and Environmental Science, University of Arizona. "Office
of Arid Lands Studies, University of Arizona.
Abstract." A tracer study using Bromide (Br) and bacteriophage PRD 1 was conducted in
a 3 ha-wetland
planted with species of bulrush {Scirpus spp) and cattail {Typha
domingensis) at the Sweetwater Recharge Facility in Tucson, AZ. Transport of the tracers
through both the wetland and vadose zone was determined. The constructed wetland
consists of pool-vegetated riffle zone sequences designed to remove suspended solids from
mixed media filter backwash water before aquifer recharge. During the tracer study, flow
rate was maintained at approximately 1.8 m^ min"' of secondarily treated wastewater
effluent. In vadose zone samples, PRD 1 was observed after the appearance of Br" penetrating
to a depth of 1.5 m. In the surface, concentration-time distribution of Br" were simulated by
a plug flow model with a high longitudinal dispersion at both the outlet and an internal
sample site. The convective-dispersion equation (CDE) and the first order reaction approach
were capable of simulating PRDl displacement and removal through the wetland. At the
outlet of the system, recovery of Br" and PRDl after 7.3 days was 86 and 16 percent,
respectively. PRD 1 decay rate was estimated to be 0.3 day"' by regression analysis and
model calibration. The findings of this study suggest that CDE, including high dispersion
and first order reaction approach, is suitable for modeling virus removal in wetlands.
40
INTRODUCTION
Water reclamation and reuse preserve and increase water availability in communities where
water resources are naturally limited or became limited by population growth. Even, after
treatment, however, reclaimed wastewater may contain pathogenic viruses which if used for
aquifer recharge, irrigation, or recreational activities increases the risk of waterbome
diseases.
Constructed treatment wetlands are a low-cost emergent technology that simulates
natural ecosystems to treat diverse types of polluted water. Wastewater treatment occurs
through a combination of physical, chemical and microbiological processes with varying
degrees of success. Most research in constructed wetlands has been focused on water quality
indicators (Bowmer, 1987; Frankenbach and Meyer, 1999; Geller, 1997; Karpiscak et ai,
1994; Mandi et al, 1996) such as biological and chemical oxygen demand, total suspended
solids, nitrates, and phosphate. Virus removal has been only sporadically studied despite the
fact that virus transmission by contaminated water is a health and economical problem
recognized worldwide. Annually, one billion dollars is spend in the US in hospitalization of
210,000 children infected by viral agents of gastroenteritis (Ho et al., 1998). Worldwide
some 5-10 million die (Walsh and Warren, 1979).
Studies conducted in constructed wetlands have reported virus removal efficiencies
as high as 98 percent in subsurface flow wetlands (Gersberg et al., 1987; Karpiscak et al..
41
1995). In a multi-species surface flow wetland, Vinluan (1996) found 90 percent removal
of an indigenous phage, bacteriophage MS2, and PRDI after 2.05, 3.67, and 13.38 days of
incubation in situ, respectively. Whereas, only a 38 percent reduction in coliphage was
observed at the outlet of a duckweed {Lemna spp.) system (Karpiscak et al., 1995). A
modeling study conducted in surface flow wetlands found that the first order reaction model,
widely used to describe virus inactivation in the environment, was unable to describe MS2
removal because virus removal was considerably higher close to the inlet of the system
(Chendorain etal., 1998). Detection of pathogenic viruses is expensive, time consuming and
not always possible. Bacteriophage is a suitable option in virus studies because it can be
prepared in high concentrations, does not infect humans, and its analysis is relatively easy,
inexpensive, and rapid. In this study, bacteriophage PRD1 was used as a human enteric
virus model. It is a dsDNA virion, 62 nm in size (Olsen et al., 1974), isoelectric between pH
of 3 and 4 (Loveland et al., 1996) and has a protein capsid that surrounds an internal lipidprotein membrane (Caldentey et al., 1990) similar to the human rotaviruses.
This paper reports the findings of a tracer study conducted to analyze PRD 1 removal
and the hydrodynamic performance of a surface flow constructed wetland designed to
remove suspended solids from filter backwash water at a wastewater reclamation plant.
42
MATERIALS AND METHODS
Site Description
The area of study was the East polishing wetland at the wetland wastewater treatment
system, in operation since July of 1997. It is located at the Sweetwater Recharge Facility in
Tucson, AZ. The wetland is the terminal component of the East polishing wetland subsystem
and is preceded by a settling basin and an artificial stream. It has been supplied for about two
years of operation with different proportions of secondary effluent and backwash water.
The backwash water comes from the nearby Tucson Water Reclamation Plant during
periodic backwashing of pressure mixed media filters used to treat activated sludge effluent
prior to its distribution as reclaimed water for use in landscape irrigation. The surface area
of the East wetland cell is about 3 ha, tlie approximate length is 329 m, and the width varies
from 72 to 112 m. Pool and vegetated riffle zone sequences is the geometric configuration
of the basin. Islands of different size are distributed in the cell mostly in the open water
zones where the water depth is approximately 1.2 m. Emergent vegetation consists of
bulrush species (Scirpus spp.) and cattail {Typha domingensis). At the outlet, a calibrated
weir is used to measure flow rates and at the inlet a manifold, placed in a small deep zone,
distributes the influent water along the east edge of the basin. The volume for water storage
was estimated to be 1.74 x 10" m^.
43
In the area of the wetland, 4.5 to 6 m of a natural clay overbank deposit is generally
found. Therefore, addition of clay to prevent wastewater seepage was unnecessary, only the
native clayey soil was compacted to some degree during the east wetland construction. Prior
to basin were vegetated with the emergent vegetation in the spring of 1997, four paired
suction lysimeters (Soil Measurement Systems, Model SW-071, Tucson AZ) of 0.27 m of
length and 0.025 m diameter were installed at the Southeast comer of the polishing cell in
the first emergent area. The 8 samplers were splitted in two arrays denominated East and
West array and placed into individual wells at 0.3,0.76, 1.5, and 3 m below the soil surface
with a separation of 1 m between adjacent lysimeters. Latter, the installation wells were
refilled; only the outlet tubing for pore water samples extraction and vacuum/pressure
application were brought to the surface. A constant vacuum source induced by a vacuum
pump conduces the pore water sample into the lysimeter through a porous stainless steel
membrane of 0.09 m of length located on its upper section.
Production and Enumeration of PRDl
Salmonella typhimiirium was used as the PRDl host bacterium to produce 3.2 L of
stock suspension by the agar overlay method (Adams, 1959) with a final concentration of
2.14 X 10'° plaque forming units (pfu) ml"'. To recover PRDl from the agar overlay, 6 ml
of Tris buffered saline, stock solution [Trizma base, Sigma Chemical Co., St. Louis MO
44
(63.2 g; NaCl, 163.6 g; KCI, 7.46 g; and 1.13 g Na2HP04 anhydrous) dissolved in 1,600 ml
of distilled water] diluted 1:10, were placed on the agar surface and incubated for 3 h, at
room temperature. The suspension was collected in a 250-ml bottle, and centrifliged at
22,095 X g for 10 min. In addition to further removal bacteria debris the supernatant was
filtrated through a 0.45 iim pore size cellulose acetate membrane filters (Costar, Cambridge,
MA).
PRDl was assayed by the double layer agar method described by Adams (1959). A
1-ml aliquot from a S. typhimurium culture, previously incubated at 37 °C for 24 h in
Trypticase Soy Broth (TSB, Difco, Detroit, MI), was combined with 4 ml of molten agar
(TSA, Trypticase Soy Agar; Difco, Detroit, MI) and 1 ml of sample or sample dilutions
added. The agar was then poured onto a layer of TSA, and incubated at 37 °C for 18 h in
order to enumerate the PRX)1.
Tracers
The study was conducted from February 12 to March 18, 1999. Chlorinated secondarily
treated wastewater at a flow rate of 1.84 m^min"' was entering into the secondary splitter box
which is separated from the east wetland cell by the artificial stream and the settling basin.
Background concentration of PRD1, Br", and Chlorine (CI2) were determined in influent and
outflow water samples from the basin for one week prior of the start of the tracer study.
45
Bromide (Br) and bacteriophage PRD1 were the conservative and virus tracers used
in this study. On March 24, a solution of 340 L of Br' was prepared in the field adding 47,
34, and 34 preweighted individual l-kg packages of Sodium Bromide (NaBr) into three
containers withl40,100 and 100 L ofwastewater, respectively. The final tracer solution was
prepared by adding 180 ml of PRDl stock suspension, and approximately 2 g of sodium
thiosulphate to 20 L of solution. This final tracer solution was then pumped at a rate of 2.5
L min'' at the end of the stream; just prior to discharge into the polishing wetland cell. The
input (Co) concentration to the basin was 354 mg L"' of Br'and 2.56 x 10^ pfii ml"' ofPRDl.
Sampling Sites and Collection
The sampling sites were situated at the inlet (1), outlet (5), lysimeter area (2), the
viewing area (3) located on the North edge of the first open water area and an internal site
at the South island from the second open pond (3) (Figure 1). After tracer addition, a surface
grab water sample was obtained daily from site I for the following three days. At site 2, a
surface grab sample and seven pore water samples firom the unsaturated soil were collected
daily during the study. From the lysimeter at 3 m of depth, in the West lysimeter array, only
a few milliliters of water were collected because problems associated with its extraction.
The lysimeters were operated at 65 kPa of vacuum by using a pressure/vacuum hand pump
(Soil Moisture Equipment, Santa. Barbara, CA.).
46
Three pneumatic samplers ofO.45 m of length and 0.013 m of diameter were placed
vertically below water surface for monitoring tracer concentrations at 0.076, 0.53, and 0.99
m of depth at the sampling site 3. They were constructed to collect water samples applying
positive pressure through an outlet tubing for closing an one way steel valve on the bottom
of the sampler. The sample was conducted to the surface by tubing connected on the
sampler well realizing the internal pressure at the time of collection. Then positive pressure
was again applied into the sampler to bring another fraction of water to the surface. After a
extraction of 0.5 L of water, a 20 ml-sample was collected.
An automated sampler (American Sigma, Model 702; New York, NY) was installed
at the sampling site 4. It was programmed every day to collect 0.6 L of sample per hour into
each of 24-polyethylene bottles previously labeled to distinguish their position in the
sampler. A second automated sampler (Isco, Model 3700; Lincoln, NE) was set up at the
outflow of the wetland where a water level recorder (Leupold & Stevens, Type Fl,
Beaverton, OR) was also placed for monitoring outflow rates.
After sample collection, 24 clean and disinfected bottles were installed into each of
the automated samplers and crushed ice was placed on the base sections. The samples were
placed on ice and transported to The University of Arizona for analysis. An ion
chromatography system (Dionex 500 Chromatographic System; Sunnyvale, CA) and the
agar overlay method were used to analyze Br' and PRDl, respectively.
47
Modeling
The experimental concentration-time distributions were simulated by using the one
dimensional convective-dispersion equation (CDE) including first order removal:
—= D
dt
dX-
V
dC
kC
dX
an analytical solution of this equation (Harris, 1963) for an instantaneous injection over the
cross sectional area satisfying the following initial and boundary conditions (Unice et al.,
2000):
C(.v,0) = 0
C(0,/) = ^^(/)
C(oo,0 = 0
is given by:
C(x,t), = a
mp=rexp(
2Ac4^t I
ADt J
exp(-/fO.
48
where
5 (t) = Dirac delta function
a = conversion factor (1 x 10'® and 1 x 10^ for PRDl and Br-, respectively)
m = mass added into the system (Kg or pfu's)
Ac = flow cross-sectional area (m")
V = convective velocity (m day"')
D — longitudinal dispersion (m~ day"')
X = longitudinal distance (m)
t = time (day)
k = removal rate constant (day"')
C = tracer concentration (mg L"' or pfli ml"').
The average flow cross-sectional area (Ac) was estimated by drawing an axis, on a copy
of original blueprints (horizontal scale 1:40, English units) of the east polishing wetland,
from the east edge of the basin to the outlet in order to trace perpendicular cross-sections
at different distances from the inlet (Table 1). Each cross-sectional area was calculated
considering 0.30 m water depth in the emergent zones and 1.2 m in the open water areas.
Using this method, the average cross-sectional area was estimated to be 41.47 m.
49
Mathematical fitness of CDE to experimental breakthrough curves was conduced
using the last square method in Mathcad professional 2000 (Mathsoft, Cambridge MA).
Mass recovery was calculated from the outlet breakthrough curve and plugged into CDE
solution to determine longitudinal dispersion, and convective velocity. Thereafter, these
parameters were used to check the amount of PRDl added and its decay rate by
mathematical fitness. With the outlet Br' mass recovery, CDE analytical solution was fitted
to experimental data collected at the island to calculate the convective velocity and
dispersion. These were used together with the previous amount of PRDl, estimated by
model calibration, to simulate the experimental breakthrough curve at the island site.
Water Balance
It was assessed by measuring inflow and outflow flow rates, evaporation and
precipitation. Evapotranspiration and infiltration were estimated. Wetland evapotranspiration
was calculated by:
Et = 0.65 X Ev
where Et is evapotranspiration (mm day"'), and Ev is evaporation (mm day"') from a class
A evaporation pan located at The Constructed Ecosystem Research Facility, on the North
side of the wetland. The 0.65- factor was determined by averaging the values recommended
50
by the Arizona Meteorological Network (AZMET, 2000) to estimate evapotranspiration
from open water bodies during Winter (0.7) and Summer (0.6). Seepage losses were
calculated by taken the difference between inlet and outlet average flow rate minus Et in m^
min"' for the period of study.
Mass Balance
Mass recovery was estimated from the zero moment of the outlet breakthrough curve
and the average outlet flow rate:
M=
where M is mass recovery (kg), ^9 is a conversion factor (1.44), Q is the average outlet flow
rate (m^ min"') and t is time (day"') .
Mass losses by seepage were estimated by integration of the outlet breakthrough
curve using the following expressions:
a) mass in the system (Ms) at any time (r) is given by:
Ms{t)=
f
[C(0 dt)
I r
m
51
where the ratio between integrals is the fraction of mass living the system with a value from
0 to 1, Ms is the mass in the system (kg).
b) assuming a homogeneous distribution of Ms in the system, losses of mass (/) by seepage
are given by:
I=
MsiO dt
where 1 is mass losses by seepage (kg), q is the seepage rate (m^ day*'), and V is the wetland
volume for water storage (m^).
The normalized first moment or detention timeof the system is defined as following:
where
is detention time (day"'). The variance of the detention time distribution with
respect to the detention time can be calculated by:
52
where
cr is the variance of the tracer respond breakthrough curve (days').
Dimensionless variance is an indicator of the extension of back-mixing and it is
determined by the ratio of the variance and the square of the tracer detention time :
cr. =
^
a
t ~
'•d
RESULTS AND DISCUSSION
Background Conditions
Background concentration for Br was 0.55 mg L"' and for PRD 1 less than one pfu
ml''. During the tracer study, chlorinated secondary effluent (containing 1.19 and 0.14 mg
L"' of total and free chlorine, respectively, on February 19, 1999) was received into the
secondary splitter box (Figure 1). This water then flowed through the small stream into the
East Polishing Basin. Total and free chlorine levels were below 0.1 mg L'' in samples from
the tracer injection point at the end of the stream. Relative concentration of Br' was 13
percent lower than PRD1 a few minutes before the end of the tracer injection at the influent
end of the wetland. PRD 1 and Br" concentrations, for the first three days of the study, are
reported in Figure 3. Apparently, PRDl performance at the sampling site 1 was similar to
53
Br" showing that the bacteriophage inactivation was unimportant at th_is initial stage of the
study.
Hydrologic Conditions
Flow conditions were close to a steady state in the basin during the study (Table 2).
Hydraulic loading rate at the outlet of the system was 84 percent of imfluent water. Water
losses by seepage and evapotranspiration were estimated to be 8.2 and 3.5 mm day"',
respectively. Kadlec and Knight (1996) suggest that wetlands can be consider similar to an
open water body for estimating their evapotranspiration losses. These losses can be
represented by about 70 to 80 percent of class A pan evaporation from an adjacent open site.
Arizona Meteorological Network recommends a factor of 0.6 for : Summer and 0.7 for
Winter to estimate reference evapotranspiration of a large water bod^^ like a pond or lake;
however, factors such as temperature, turbidity, depth, size of the po -nd, and surface area
may drastically modify this relationship (AZMET, 2000).
Lysimeter Area
The highest PRDI concentration observed in surface grab sam j)les from 2 was 3185
pfli ml"' on February 25, the day after injection of the tracers. In Fig;jures 4 and 5 Br" and
PRD1concentrations were plotted versus lysimeter depth. Each curve corresponds to the date
in which Br" or PRDl concentration was the highest one observed duiring the 21-day study
54
in one of the four lysimeters at the East and West site. Because the absolute concentration
of PRDl is much higher than Br", only the curve for the conservative tracer maximum
concentration noticed in surface samples was plotted.
Conservative Tracer
The higher Br'concentration in the vadose zone (Figures 4 and 5A) than in the
surface grab sample, taken on February 25, revels a short residence time of the main cloud
of Br' that passed over the lysimeter field. In the vadose zone, the uneven concentration
profile of Br, mostly in the West array of samplers, and its rapid displacement, in contrast
with the calculated seepage rate, appears to indicate preferential flow and a different degree
of water content surrounding the lysimeters at identical depths. A major disadvantage of
using lysimeters for water sampling from the vadose zone is their installation. Soil
conditions are altered because a hole has to be dug and refilled after sampler installation
promoting the possibility of preferential flow.
Bacteriophage
PRDl was observed to arrive later than Br' (Figures 4 and 5B) in the soil water
samples reaching, apparently, a maximum depth of 1.5 m. A similar virus performance was
reported in a tracer study conducted in an artificial recharge basin where hydrophobic
55
interactions between virus and organic colloids from secondary effluent, used in the study,
were apparently responsible of such behavior (Powelson et aL, 1993). Several previous
studies (Bales, 1991; Jin, et aL, 1997; Powelson et aL, 1991; Powelson and Gerba., 1994)
have found that hydrophobic and electrostatic interactions as well as unsaturated conditions
are mostly responsible for virus removal in soils. Since July of 1997, the East Polishing
Basin had been flooded with different proportions of secondary effluent and backwash
water. Vadose zone accumulation of organic matter from wastewater may have been
preferentially deposited in the shortcut flow paths resulting in virus hydrophobic
interactions (virus adsorption and retardation) in spite of preferential flow.
PRD1 was detected at very low concentrations (1 pfu ml"') after 10 days of study at
a depth of 0.76 m in the West array. Unsaturated conditions were higher at this location than
at the other samplers. The importance of unsaturated conditions for virus removal was shown
by Powelson and Gerba (1994) who found that removal rates were greatly increased by
decreasing soil water content. In the East array, PRD I was not observed at depths of 1.5 and
3 m, probably do to a longer contact time and a greater degree of unsaturated conditions,
in contrast with the lysimeters affected by preferential flow, limited movement of PRD1.
56
Deep Water
During the first day of sampling both Br" and PRDl were observed in samples
collected from 3 (Figure 6). After an apparent initial distribution Br concentration decreased
significantly, the following day. The highest concentration observed was at a depth of 1 m.
Bacteriophage was better mixed than Br" one day after the tracer addition ; later, its
concentration decreased drastically.
Modeling
Recovery of Br' at 5 was about 76 percent of the mass originally added into the
system; therefore, the CDE was fitted to the experimental data assuming an initial injected
mass of 67 kg, steady state conditions, and a 100-percent of the tracer recovery. Calibration
of the model was conducted as described above, the input and calculated parameters are
presented in Table 3.
Conservative tracer Br"
The breakthrough curves at the sampling sites 4 and 5 were adequately simulated by
CDE. (Figure 7). A steeper profile and higher velocity were shown by Br" for its arrival at
the island than at the outlet which is consistent with the convective velocity and longitudinal
57
dispersion calculated by model calibration. At the outlet, Br" depicted a longer tailing
probably because of contraction of the cross-sectional area that induced dead zones
formation at the end of the basin.
Concentration of Br'evidenced a nearly uniform evolution over time; although, the
dispersion of experimental Br" concentrations in contrast to CDE curve is higher at site 4
than 5 (Figure 7), probably because of its closer position to the inlet. Near the site of solute
addition, spatial movement of a tracer is three dimensional. When CDE is reduced to one
dimensional form the non-uniform displacement is averaged in a longitudinal dispersion
coefficient. Gradually, the equation will show a better fitness to experimental data
depending on how well the tracer is mixed across flow sectional area. This is a function of
transport time or distance and is negatively related to the solute diffusion coefficient.
Transport time or distance from the inlet to the island was not enough to reach flill mixed
conditions in the flow cross-sectional area.
Similarly to the findings of other studies (Chendorain et al., 1998; Kadlec, 1994), a
high longitudinal dispersion was observed through the basin (Levenspiel and Smith, 1957).
In the present study, it may be attributed to non-uniform advection induced by islands and
pool-riffle sequences as reported by Seo (1990) in laboratory studies, vegetation
(Coutanceau and Bouard, 1977; Nepf et al., 1997), and flow cross-section variation. This
intensifies the lateral and vertical concentration gradients.
58
Bacteriophage PRDl
The one dimensional convection-dispersion equation was capable of simulating
PRDl concentration-time distribution at both the island and outlet (Figure 8 and 10).
Similar to other tracer studies conducted in surface flow wetlands (Chendorain et ai, 1998),
both figures show a wide variation of PRDl concentrations suggesting a high analytical
variability or that concentration gradients of PRDl can change greatly in short distances
within wetlands because of insufficient mixing.
PRDl decay rate was estimated to be -0.30 day'' (r^ = 0.98) by regression analysis
between the normalized zero moment of the breakthrough curve at the inlet (pulse added into
the system), island, and outlet, and the time at their center of mass (Figure 9). This value was
practically equal to the decay rate calculated by mathematical fitness of CDE to the outlet
experimental curve. Likewise, m obtained by CDE calibration was only 3.6 percent higher
than the PRDl in the 3.2 L of stock suspension added initially into the influent water. The
first order model included into CDE as well as convective velocity and longitudinal
dispersion obtained by mathematical fitness were on average consistent with PRD 1 transport
conditions through the wetland. Apparently, the calculated decay rate described adequately
PRD 1 removal over time at the outlet and the internal site. This differs somewhat from MS2
removal studies conducted in surface flow wetlands by Chendorain et al. (1998). In these
studies, MS2 decay rate was a site dependent coefficient ranging from 0.076 to 5.81 day"'.
59
Detention Time and Mass Recovery
The wetland detention time for Br" was 7.3 and 6.8 days calculated from the
experimental and CDE breakthrough curve, respectively (Table 4). Also for PRDl, both
first moments were practically equal. However, these were shorter than those for Br' because
virus removal reduces the PRDl breakthrough curve and consequently its first moment
becomes shorter than the first moment of Br" breakthrough curve. Mass recovery and losses
by seepage for Br'at the outlet, based on outflow rates, were estimated to be 76.6 and 10
percent, respectively. Mass recovery based on model simulation was 74 percent.
Tracer studies in treatment wetlands are characterized by a highly variable mass
recovery (Table 5). Factors such as microbial degradation, sorption, inadequacies of
analysis and sampling, flow conditions, and inaccurate measurement of outflow rates have
been responsible of this variability. Probably, small contributions from sampling, analysis,
and flow rates may account for the mass balance underestimation.
Bacteriophage PRD1 was removed 84 percent during the 7.3 days-detention time in
the East Polishing Wetland. Dimensionless variance (cfg^) and arrival time (t,o/t) of Br- at the
outlet of the East polishing wetland were estimated to be 0.23 and 0.52, respectively. These
parameters were 0.18 and 0.58 in a tracer study conducted in a multi-species subsurface flow
wetland where 98.24 percent removal of PRD 1 was observed. In this multi-species wetland,
the removal and inactivation rate (in dialysis bags) were estimated to be -1.17 and -0.16 day'
60
respectively. Apparently, adsorption promoted by filtration was the primary mechanism
of PRD1 fixation on the wetland substratum followed by a longer inactivation processCompared with MS2 removal reported by others, this efficiency is low. For example,
Gersberg et ai (1989) found 91.5 percent removal of MS2 in a surface flow wetland and
Chendorain et al. (1998) reported 93 to 98 percent also in surface flow wetlands. Studies
conducted by Yahya et al. (1993) and Vinluan (1996) reveled that PRDl is more persistent
in the environment than MS2; however, the latter study conducted in multi-species
subsurface wetland and other conducted by Gersberg et al. (1987) suggest that survival of
MS2 and PRD I in wetland environment receiving municipal wastewater is comparable.
Only few research studies have been conducted to determine the ability of
constructed wetlands to remove animal enteroviruses. Karpiscak (1995) found human
enterovirus removal of 98 percent and Thurston (1997) over 83 percent in the same multispecies subsurface wetlands. Other studies conducted in forested and subsurface flow
wetlands have observed removal rates for indigenous coliphage to be smaller than for
enteroviruses (Scheuerman, et al.,\991). For example, Gersberg et al. (1987) reported that
indigenous FRNA coliphage was removed more slowly than polioviruses and MS2 in
vegetated gravel beds receiving municipal wastewater. Comparable findings were reported
comparing indigenous coliphage to human enteroviruses in forested wetlands (Scheuerman
et al., 1997).
61
CONCLUSIONS
Preferential flow of Br- was observed along the soil profile below the wetland.
Movement of PRDl was delayed in contrast to Br and largely removed, although, it was
detected at 1.5 m of depth. Apparently, hydrophobic interactions and unsaturated conditions
limited soil penetration of PRDl at the lysimeter sampling site. One dimension CDE was
capable of simulate acceptably Br" and PRD1 transport performance in the basin. Although,
a higher longitudinal dispersion has to be included into CDE to simulate Br' and a first order
model to predict PRD 1 removal. In contrast with Br', PRD 1 traveled without retardation
through the basin suggesting that virus removal was mostly irreversible. A 7.3 day detention
time for Br- resulted in a 84 percent reduction of PRDl. The removal rate of PRDl in the
current study was several times smaller than removal rates reported for MS2 and PRD1 in
subsurface flow wetlands (Gersberg et <z/.,l987; Vidales et al., 2000). Given the high
persistency of PRDl in the envirorunent, survival does not appear to be the primarily
removal mechanism. Apparently, its removal in wetlands is related to a filtration process and
probably sedimentation of virus colloidal carriers (LaBelle and Gerba, 1979 and 1980).
According to this study and results reported by Gersberg et al. (1989), Yaya et al.
(1993), Vinluan (1996), and Chendorain et al. (1998) PRDl may be abetter virus model for
human virus than MS2 in wetland systems.
62
The East Polishing Basin at the Sweetwater Wetland showed a lower effectiveness
in removing PRD1; however, a higher removal efficiency might be achieved by reducing
inflow in order to promote a more ideal plug flow conditions and a greater contact time.
ACKNOWLEDGMENTS
The authors wish to acknowledge technical support from Sue Hopf and Glenn France
from the University of Arizona's Office of Arid Land Studies. Financial support was
provided by Tucson Water, City of Tucson Arizona; Sanitation Districts of Los Angeles
County; United States Environmental Protection Agency; American Water Works
Association Research Foundation; USDA Water Conservation Laboratory; Pima County
Department of Wastewater Management; Water Environment Research Foundation; City of
Phoenix, Arizona; The Subregional Operators Group (Phoenix-area cities); Water
Replenishment District of Southern California; City of Riverside, California; and City of Los
Angeles Department of Water and Power.
DISCLAIMER
The American Water Works Association Research Foundation and the other agencies
listed above not had opportunity to review and comment on this paper, therefore, none of
these agencies necessarily endorse the findings presented here.
63
REFERENCES
Adams, M. H. (1959) Bacteriophage. Interscience Publishers, Inc., New York.
AZMET, Arizona Meteorological Network. Available: http://ag.arizona.edu./azmet/etl.htm.
(2000, Oct. 23)
Bales R. C., Hinkle S. R., KroegerT. W. and Stocking K. (1991) Bacteriophage adsorption
during transport through porous media: chemical perturbations and reversibility.
Environ. Sci. Technol. 25, 2088-2095.
Bowmer K. H. (1987) Nutrient removal from effluent by an artificial wetland: influence of
rhizosphere aeration and preferential flow studied using bromide and dye tracers.
Wat. Res. 21, 591-599.
Caldentey J., Bamford J. K. H. and Bamford D. H. (1990) Structure and Assembly of
bactreriophage PRD1, an Escherichia coli virus with a membrane. J. Stnic. Biol. 104,
44-51.
Chendorain M., Yates M. and Villegas F. (1998) The fate and transport of viruses through
surface water constructed wetlands./. Environ. Qual. 27, 1451-1458.
Coutanceau M.and Bouard R. (1977) Experimental determination of the main features of
the viscus flow in the wake of a circular cylinder in uniform translation. Part 1.
steady flovv. J. Fluid Mech. 79, 231-256.
64
EPA, Office of Water. (1999) Drinking water criteria document for enteroviruses and
hepatitis A: an addendum. EPA, Washington, DC. pp 2-10
Frankenbach R. Land Meyer J. S. (1999) Nitrogen removal in a surface-flow wastewater
treatment v/etland. Wetlands, 19,403-412.
Gersberg R. M., Lyon S. R., Brenner R. and Elkins B. V. (1989) Fate of viruses in artificial
wetlands. Appl. Environ. Microbiol. 53, 731-736.
Ho M., Glass R. L, Pinsky P. F. and Anderson L. (1998) Rotavirus as a cause of diarrheal
morbidity and mortality in the United States. J. Infect. Dis. 158, 1112-1116.
Jin, Y., Yates M. V., Thompson, S. and Jury W. A. (1997) Sorption of viruses during
through saturated sand columns. Environ. Sci. Technol. 31, 548-555.
Kadlec R. H. (1994) Detention and mixing in free water wetlands. Ecol. Eng. 3, 345-380.
Kadlec, R. H. and Knight R. L. (1996) Treatment Wetlands. Lewis Publishers Boca Raton,
Florida, p 893.
Karpiscak M. M., Foster K. E., Hopf S. B., Bancroft J. M. and Warshall P. J. (1994) Using
water hyacinth to treat municipal wastewater in The Desert Southwest. Water.
Resour. Bull. 30, 219-227.
Karpiscak M. M., Gerba C. P., Watt P. M., Foster K. E. and Falabi J. A. (1995) Multispecies plant systems for wastewater quality improvements and habitat enhancement.
65
In Angelakis, A., Asano T., Diamadopoulos E., and Techobanoglous G. (ed.) Second
international symposium on wastewater reclamation and reuse. lAWQ, Iraklio,
Greece, pp. 37-42.
ECing A. C., Mitchell C. A. and Howes T. (1997) Hydraulic tracer studies in a pilot scale
subsurface flow constructed wetland. Wat. Sci. Tech. 35, 189-196.
LaBelle R. and Gerba C. P. (1979) Influence of pH, salinity, and organic matter on the
absorption of enteric viruses to estuarine sediments.
Envir. Microbiol. 38,93-
101.
LaBelle R. and Gerba C. P. (1982) Investigations into the protective effect of estuarine
sediment on virus survival. Wat. Res. 16, 469-478.
Levenspiel O. and Smith W. K. (1957) Notes on the diffusion-type model for the
longitudinal mixing of fluids in flow. Chem. Eng. Sci. 6, 227-233.
Loveland J. P., Ryan J. N., Amy G. L. and Harvy R. W. (1996) The reversibility of virus
attachment to mineral surfaces. Colloids SuifacesA: Physicochem. Eng. Aspects 107,
205-221.
Mandi L., Houhoum B., Asmama S. and Schwartzbrod J. (1996) Wastewater treatment by
reed beds an experimental approach. Wat. Res. 30, 2009-2016.
66
Nepf H. M., Mugnier, C. G. and Zavistoski R. A. (1997) The effects of vegetation on
longitudinal dispersion. Estuarine, Coastal and Shelf Science 44, 675-684.
Netter R. (1997) Flow characteristics of planted soil filters. fVat. Sci. Tech. 29, 37-44
Olsen R.H., Siak J. and Gray R. (1974) Characteristics of PRDl, a plasmid-dependent
broad host range DNA bacteriophage./. Virol. 14, 689-699.
Powelson D.K., Gerba C. P. and Yahya M. T. (1993) Virus transport and removal in
wastewater during aquifer recharge. Wat. Res. 27, 583-590.
Powelson D.K., Simpson J. R. and Gerba C. P. (1991) Effects of organic matter on virus
transport in unsaturated flow. Appl. Environ. Microbiol. 57, 2192-2196.
Powelson D. K. and Gerba C. P. (1994) Virus removal from sewage effluent during
saturated and unsaturated flow through soil columns. Wat. Res. 28, 2175-2181.
Seo LW. (1990) Laboratory and numerical investigation of longitudinal dispersion inopen
channels. Water Resour. Bull. 26, 811-822.
Tanner C. C., Sukias J. P. S. and Upsdell M. P. (1998) Organic matter accumulation during
maturation of gravel-bed constructed wetlands treating farm dairy wastewater. Wat.
Res. 32, 3046-3054.
67
Thurston J. A.(1997) Fate of pathogenic and indicator microorganisms in two subsurface
multi-species constructed welands. Thesis, Department of Soil, Water, and
Environmental Science. University of Arizona, Tucson AZ.
Vinluan E. A. (1996) Survival of microbial indicators in constructed wetlands. Thesis
Department of Soil, Water, and Environmental Science. University of Arizona,
Tucson, AZ., p. 65.
Walsh J. A.and Warren K. S. (1979) Selective primary health care: an interim strategy for
disease control in developing countries. N. Engl. J. Med. 302, 967-974.
Yahya, M. T., Galsomies L., Gerba C. P. and Bales R. C. (1993) Survival of bacteriophages
MS-2 and PRD-I in ground water. Wat. Sci. Tech. 27, 409-412.
68
Table 1. Flow cross-sectional area at different distances from the inlet.
Distance from the inlet (m)
Ac *
20
38
64
85
123
150
168
187
206
240
270
293
315
27.93
25.86
36
72.86
29.74
25.86
66.85
47.70
17.45
21J3
80J7
65.73
21.8
Avg*
41.47
*Ac, cross sectional area (m"); Avg, average cross sectional area (m~)
Table 2. Daily hydraulic loading rate applied on the east polishing basin during the
study.
Hydraulic Load Rate
Inlet
Outlet
Average (mm day"')
88.1
76.4
Standard deviation (mm day'")
5.6
5.2
Variation coefficient (%)
6.4
6.8
69
Table 3. Parameters for mathematical fitting of the convection-dispersion equation
(CDE) to experimental breakthrough curves obtained at the sampling sites 4 and 5.
Input
Output
k
Br"
Ac
M
Outlet
41.5
67
328.88
67
182
Island
D
V
.V
Ac
M
40.4
k
D
V
1600
57.5
1879
62.6
PRDl
Outlet
41.5
1600
57.5
328.88
0.298
7.1 X
10'^
[sland
40.4
7.1 X
1879
62.6
0.298
182
10'^
Ac (m~); M (kg for Br' and pfli's for PRDl); D (m* day"'); V (m day''),
k (day"'); x (m).
Table 4. First moments calculated from experimental and the predicted concentrationtime distribution curves at the sampling sites 4 and 5.
First moment (Days)
Site
Experimental distribution
Predicted distribution
Br-
PRDl
Br
PRDl
Outlet
7.3
5.3
6.8
5.26
Island
3.92
3
3.84
3
70
Table 5.- Mass recovery in tracer studies conducted in treatment wetlands.
Wetland
Surface area
Tracer
500
Infiltration
Source
d
Netter (1997)
(%)
m*
'SSF
Recovery
Lithium
15-100
Uranin
48-50
Eosin
6-7
Bromide
47-99
SSF
100
^AR
86
d
Bowmer (1987)
SSF
19
Bromide
87-94
d
Tanner et al. (1998)
SF and-SSF
300
Lithium
96
d
K.ing etal. (1997)
"SF
1050
Bromide
75 ±11
negligible
Chendorain el al. (1998)
SF
2x 10^
Lithium
106 ±32
negligible
Kadlec (1994)
"SSF, subsurface flow; ''SF, surface flow; "EAR, Eriochrome Acid Red A
""lined with some type of impermeable material.
Settling Basin
Splitter Box
Figure 1. Schematic representation of the east polishing basin showing the injection (I)
and sampling sites at the inlet (1), lysimeter area (2), first (3) and second (4) open water
zones, and outlet (5) of the system.
72
West side
4^r
T
I m
1
r
OJO
r
0.76 m
East side
lo m
3m
Figure 2. Schematic representation of the East and West lysimeter arrays depicting
installation depths (0.3,0.76,1.5, and 3 m) and separation between adjacent lysimeters
(1 m).
73
1.E+00
O
O
D
c
1.E-01 -
PRDl
CO
c
o
u
c
o
u
a>
JO
0)
a:
1.E-02 -i
Br
1.E-03
1.E-04
1.E-05
0
0.85
1.81
2.82
Time since tracer addition (days)
Figure 3. Relative concentration of PRDl and Br'at the sampling site 1 given by the
ratio between actual (C) and input (Co) concentration.
74
Br' (mg I L)
0
5
10
15
20
0
Feb - 25
1
Feb - 26
March - 5
E
SI
a.
Qo
March -2
2
3
PRD1 (pfu/ ml)
0
10
20
30
40
0
Feb - 2 7
March -1
Feb - 2 8
1
E
Q.
O
Q
2
3
Figure 4. Concentration profiles of Br- (A) and PRDl (B) at the East lysimeter array,
sampling site 2.
75
Br- (mg/ L))
10
a.
a>
a
Feb - 2 5
15
20
Feb - 2 7
25
Feb - 2 6
3 -
PRD1 (pfu/ml)
0
10
20
30
40
50
0
Feb - 27
March - 3
1
o
March -4
Figure 5. Concentration profiles of Br- (A) and PRDl (B) at the West Iisymeter array,
sampling site 2.
76
Relative concentration (C/Co)
0.02
0.01
0.03
0.04
0.05
0.06
0
cs
02
I
•
I
0.4
I
£
Q. 0.6.
<U
Q
0.81
V
>
X
I
i=eb-27
Feb-26
F^-25
1.2
Figure 6. Relative concentration of Br- and PRDl expressed by the ratio between
actual (C) and input (Co) concentration observed at three different depths within the
first open water zone, site 3.
CDE
Island
Outlet
D1
0
5
15
10
20
Time (days)
Figure 7. Concentration-time distribution of Br- at sites 4 and 5.
78
2.E+03 CDE
^ 1.E+03 i
a. 5.E+02 -!
O.E+00
0
5
10
15
Time since tracer addition (days)
Figure 8. Concentration-time distribution of PRDl at the sampling site 5.
79
1.00EKX) ^
1.00E-01
Figure 9. Lineal regression of the normalized zero moment and time at the center of
mass of the pulse added into the system and breakthrough curves at sites 4 and 5.
80
6.E+03
O
O
=- 4.E+03
I—•
O Island
O
o
ZE+03 -
O.E+00
2
4
6
8
10
12
14
16
18
Time since tracer addition (days)
Figure 10. Observed and predicted concentration-time distribution of PRDl at the
sampling site 4.
81
APPENDIX B
VIRUS REMOVAL FROM WASTEWATER I N A MULTI-SPECIES
SUBSURFACE-FLOW CONSTRUCTED WETLAND
Juan A. Vidales*', Charles P. Gerba', and Martin M. Karpiscak"
'Depatment of Soil, Water and Envirormiental Science, University of Arizona, Tucson
AZ 85721. "Office of Arid Lands Studies, University of Arizona, Tucson, AZ 85719.
* Corresponding Author
82
Virus Removal from Wastewater in a Multi-Species Subsurface-Flow Constructed
Wetland
Juan A. Vidales', Charles P. Gerba', and Martin M. Karpiscakr
'Department of Soil, Water, and Environmental Science, University of Arizona. "Office
of Arid Lands Studies, University of Arizona.
Abstract.- Virus removal was studied in a multi-species subsurface flow constructed
wetland. Tracer studies and a virus survival test were conducted using Bromide (Br") and
bacteriophage PRDl. These were added simultaneously into the wetland which received
unchlorinated secondary effluent, pre-treated by a duckweed pond. A high degree of plug
flow existed in the hydrodynamic performance of the, 6 year old, gravel bed during the
smdy. Most of the PRDl was recovered during the first four days of sampling; however,
PRDl concentration did not reached the background concentration at the end of the
study. The average bacteriophage removal was 98.8 percent whereas Br' mass recovery
was 75 percent in the spring tracer study. The removal rate of PRDl was estimated to be
-1.17 day"'; in contrast, its inactivation rate in situ for a 12.4 day-period was -0.16 day"'.
Apparently, virus removal is governed by an initial irreversible attachment followed by a
comparatively long inactivation period. The findings of this study suggest that subsurface
flow wetland can remove about 99 percent of the virus load.
83
INTRODUCTION
Water is the principal route for transmission of disease-causing viruses that leads
to significant morbidity and mortality worldwide. Once released into the environment by
infected individuals, viruses can contaminate surface waters and infect, directly or
indirectly, susceptible persons, causing diseases such as diarrhea, fever, poliomyelitis,
gastroenteritis, hepatitis, meningitis, and paralysis (Bitton, 1994; Rusin et al., 2000).
Subsurface flow wetlands represent a low-cost emerging technology that mimics
natural occurring processes to treat wastewater. These constructed ecosystems can
potentially reduce the wastewater virus load by filtration of influent water through
vegetated gravel-filled beds. Gersberg et al. (1987) found that 99 percent of indigenous
F-specific bacteriophage was removed by a 18.5 m x 3.5 m x 0.76 m subsurface flow
wetland operated at a hydraulic loading rate of 5 cm day"'. The authors suggested that
attachment promoted by filtration was the primary removal mechanism. Vinluan (1996)
conducted a removal and survival experiment of indicator microorganisms in a multispecies surface flow wetland. She found that removal of 90 percent of bacteriophages
MS-2, PRDl, and indigenous phage required 3.67, 13.38 and 2.05 days, respectively.
Virus decay showed a slow reduction along the wetland with an increasing removal rate
in the last ten meters of the system. Similarly, a 6 month-study reported by Karpiscak et
84
al. (1995) indicated that pathogenic viruses can be removed significantly from secondary
wastewater effluent in vegetated gravel beds.
Studies of virus removal in subsurface flow wetlands are scarce. Most studies
have analyzed virus removal as the ratio of inlet and outlet concentration associated to
calculated detention time. A tracer test is the best approach to determine the actual
detention time and the amount of mixing within the system. Additionally, it may reveal
the effect of mixing conditions on virus removal. A tracer study conducted
simultaneously with a survival experiment could establish the contribution of virus
inactivation over the removal performance of the system. Gersberg et aL (1987) have
suggested that virus removal is mostly related to a filtration process in a subsurface flow
wetland. This paper reports removal of bacteriophage PRDl and the hydrodynamic
performance of a multi-species subsurface flow wetland, analyzed by tracer studies; as
well as, the results of a survival test in situ conducted to determine the inactivation
behavior of PRD1.
MATERIALS AND METHODS
Bacteriophage PRDl
Bacteriophage PRDl is a (isDNA virion, 62 nm in size (Olsen et
<3/., 1974),
and
isoelectric between pH of 3 and 4 (Loveland et al., 1996). It has a protein capsid that
85
surrounds a lipid-protein membrane similar to the human rotaviruses (Caldentey et
a/.,1990). Several researchers have used it as virus model in soil and aquifer studies
(Powelson and Gerba, 1994; Schijven et aL, 1999; Ryan et a/.,1999) since its size is
similar to rotaviruses and adenoviruses (Rusin et aL, 2000).
Production and Enumeration of PRDl
PRDl stocks were grown by the agar overlay method (Adams, 1959) and
harvested from the agar surface by addition of 6 ml of Tris buffered saline, stock solution
[Trizma base, Sigma Chemical Co., St. Louis MO (63.2 g; NaCl, 163.6 g; KCl, 7.46 g;
and 1.13 g Na-,HP04 anhydrous) dissolved in 1,600 ml of distilled water] diluted 1:10,
per petri dish. After 3 h supernatant was collected in a 250-ml bottle, centrifuged at
22,095 X g for 10 min, and filtrated through a 0.45 |a.m pore size cellulose acetate
membrane filters (Costar, Cambridge, MA).
PRDl was assayed by the double layer agar method described by Adams (1959)
using Trypticase Soy Agar (TSA, Difco, Detroit, MI). A 1-ml aliquot from a Salmonella
typhimurium culture, previously incubated at 37 °C for 24 h in Trypticase Soy Broth
(TSB; Difco, Detroit, MI) was combined with 4 ml of molten agar (TSA, Trypticase Soy
Agar; Difco, Detroit, MI) and 1 ml of sample or sample dilutions added. The agar was
then poured onto a layer of TSA and incubated at 37 °C for 18 h in order to enumerate the
PRDl.
86
Study Site
The study site (Figure 1) was a subsurface-flow wetland at the Constructed
Ecosystem Research Facility (CERF) in Tucson, AZ. The individual plastic lined cells
are trapezoidal in cross-section and measure 8.2 m wide (W) on top, 61 m long (L), and
0.9 m in depth (Karpiscak et ai, 1994). The volume available for water storage before
planting was calculated to be 153 m^ estimating an initial 40 percent porosity (0). On the
bottom of the trench, a layer of 6 to 9 cm of gravel is overlay by 15 to 30 cm of cobbles.
Subsequent gravel additions filled the basin up to 0.9 m of depth (Karpiscak et ai, 1993).
At the influent end, a 4 m-length section, approximately, was filled with 0.9 m of cobbles
over a thin layer of gravel. A pipe with several manually-controlled openings introduced
unchlorinated secondary effluent from the Roger Road Municipal Wastewater Treatment
Facility, previously treated in a duckweed {Lemma gibba) pond at CERF. At the outlet,
water overflowed into a vertical stand pipe (0.19 m in diameter) to leave the system. The
emergent vegetation in the gravel-filter consists of Cottonwood {Populus fremontii), black
willow {Salix nigra), coyote willow {Salix exigua), sycamore {Platanus wrightii), desert
willow {Chilopsis linearis), seep willow {Baccharis glutinosa), bulrush {Scirpus olneyi),
cattail {Typha domingensis), yerba mansa (Anemopsis californica), and giant reed
(Anmdo donax).
87
Tracer Addition and Sampling
Summer 1997
On July 1997, 50 ml of PRDl stock solution at a concentration of 1.8 x 10" pfu
ml"' and 150 g of NaBr were mixed in 20 L of influent wastewater. This mixture was then
added to the wastewater flowing into the wetland for a 2 h-period. Both tracers were
monitored for 10 days by collecting samples every 4 hours at the outlet. Additional
samples were taken from PVC pipes that had holes drilled in their side walls. These pipes
were placed vertically to a depth of 0.45 m at a distance of 20 and 40 m from the inlet. At
the end of the study, during the last three days, 3, 2 and finally one sample per day were
collected at each sampling site. An specific-electrode (ATI Orion Model 9435, Boston,
MA.) was used to determine Br" concentration in the collected samples. At the inlet, flow
rates were monitored continuously by a totalizer flow meter. At the outlet, they were
measured every 4 h. A volume of 2.86 x 10'^ m^ was labeled on the internal well of the
exit pipe and then its bottom neck were blocked with an adjustable sphere-shape rubber.
The filling time of this volume was obtained and the flow rate calculated by the ratio
between the labeled volume and filling time.
88
Spring 2000
On March 26, a second tracer study was performed. Two automatic samplers
were placed at 36 m (Isco, Model 3700; Lincoln, NE) from the inlet and at the outlet of
the wetland (American Sigma, Model 702; New York, NY); later, both samplers were
programed to take a sample every 3 h for 12 days. At 36 m from the inlet, the sampler
collected samples from a previously installed steel pipe with a 0.025 m-diameter and a
length of 0.95 m. Holes were drilled into the pipe wall to permit sampling at a depth of
0.6 m. The tracer suspension consisted of a volume of 20 L of influent water as well as
838 g of NaBr, and 100 ml of bacteriophage PRDl stock suspension at a concentration
of 1.2 X 10" pfu ml"'. It was added at the inlet for a 1.5 h-period. The inflow rate was 0.9
m^ h"' therefore the input concentration (Co) for each tracer was 468 mg L"' of Br and 9.3
X 10^ pfu ml"' of PRDl.
Flow rates were measured as in the summer tracer test of 1997. At the outlet, the
monitoring frequency was changed to three times per day (early morning, afternoon, and
evening). Each day, cleaned and disinfected 1 L-bottles were placed in the sampler and
crushed ice was added to keep the collected samples cooled. The samples were kept in
ice, transported to the laboratory at the University of Arizona, and assayed the same day
of collection. Bromide and bacteriophage were analyzed by using lon-Chromatography
(Dionex 2020i; Sunnyvale, CA) and the double agar layer assay, respectively.
89
Spring 2000: Survival test
Pieces of approximately 10 cm of dialysis tube (Spectra/pro 4, Houston, TX) were
used for incubating PRDl in situ at 36 and 54 m from the inlet. The membrane of this
tube has a molecular weight cut off between 12000-14000 which is defined as the solute
molecular weight that is 90 percent prevented from permeating through the dialysis tube
membrane. The cut off is around 3 orders of magnitude smaller than the molecular
weight of the DNA (24 x 10®; Olsen et ai, 1974) of PRDl; thus this virus is unable to
pass through the membrane. However, it is large enough to allow exchange of different
types of hydrocarbons, alcohols, ketones, esters, oxides, acids, alkalis, and solvents as
well as metal salts.
On same day that the tracers were added, 24 dialysis bags were prepared by
placing 10 cm-pieces of dialysis tube in distilled water to soften them and tying off one
end of each bag. At the same time, a solution of PRDl at a concentration of 2.47 x 10'
pfli ml"' was prepared by diluting one ml of the PRDl stock suspension in influent
wastewater. From this solution, 2 ml-aliquots were poured into the dialysis bags which
were closed by tying their upper end.
A set of 12 bags was placed at both 36 and 54 m from the inlet into a small 0.2
m-deep well. From each site, a dialysis bag was collected daily and the water temperature
measured in the wells, except on March 5. The collected bags were put into 50 cm-
90
centrifuge tubes, placed in ice and taken to the laboratory. The samples were analyzed the
same day.
Parameters Estimation
Mass recovery was calculated by the following expression:
M^ = a Q
dt
where Mo is the tracer mass recovery (Kg), a is a conversion factor (1.44), Q is the
average flow rate at the outlet during the experiment (m^ min"'), t is time (days) and C is
the outlet tracer concentration (mg L"'). Mass recovery can be compared with the added
mass as a criterion of evaluation. Discrepancies between both amounts may be an
indicator of tracer degradation or adsorption, inaccurate measurements of flow rates or
tracer concentrations, or low frequency of sampling. Detention time was determined by
integration of the detention time distribution by the following expression:
^Cit)dt
where
is detention time of the system (days) and t is time (days). The arriving time (t,(,)
was evaluated by the time at 10 percent of Br" mass recovery at the outlet.
91
The variance of the detention tinie distribution with respect to the detention time
can be calculated by:
^
0" =
1 C(t)dt
where cr is the variance of the tracer respond breakthrough curve (days").
Dimensionless variance (c7g") is an indicator of the extension of back-mixing and
it is determined by the ratio of the variance and the square of the tracer detention time:
= a
Relative breakthrough curve is the ratio of the normalized concentration
distributions of two tracers (Harvey and Garabedian, 1991) that can be used to evaluate
relatively removal or transformation of a non-conservative tracer. It is estimated by the
following relationship:
^
RB =
I
[C
L PRD~l ]J/ -dt ^
f [QJ, dt
[QJ.
xlOO
92
where [C pRo-ilto
[C BrJto
PRD-1 and Br pulse concentration during the tracer
injection. Relative attenuation results from subtracting the relative breakthrough curve
from one hundred (RA— 100-RB).
RESULTS AND DISCUSSION
Flow rates and evapotranspiration
Differences in flow rates at both inlet and outlet for the summer and spring tracer
experiments are shown in Table I. During the summer experiment, evapotranspiration
was estimated to be 42 percent based on influent and outflow average flow rates. Some
periods of non-flow at the outlet were commonly observed in the afternoons during
Summer. Forty- two percent is probably an underestimation of the actual value because
the influent water was insufficient to satisfy losses the high evapotranspiration losses
during the non-flow periods. Thus during this part of the day, water losses were more
than 100 percent. However, given the impossibility of measuring the actual losses an
estimation of 100 percent was used.
The spring tracer test showed a small variation of flow rates at both ends of the
wetland, the flow rate was considerably more constant than during the summer
experiment. However, flooded conditions at the outlet, mostly in the mornings, altered
93
the schedule of flow rate monitoring. It was only possible to take the morning, afternoon,
and evening flow rate measurements on five of 12 days. Loses by evapotranspiration
were determined to be 26 percent of average influent flow.
Conservative Tracer (Br)
Summer 1997
A non-typical breakthrough curve at 40 and 56 m from the inlet were observed
during the summer of 1997 (Figure 2). This behavior may be explained by the small Br"
pulse, introduced into the system, that did not increased Br concentrations enough in the
wetland water to diminish the effects of interfering ions, probably, reconcentrated
because of evapotranspirational losses.
At 20 m form the inlet, a sharp peak in concentration was seen after 1.3 days
following introduction of the tracers. Thereafter, a long tailing effect was observed.
Apparently, the first moment of the breakthrough curve up to this point was 2.2 days. At
the outlet, Br' mass recovery was 140 percent and a detention time of 5.6 days was
calculated after integration of the breakthrough curve.
Spring 2000
After extrapolation of the outlet breakthrough curve of Br' to the background
concentration, mass recovery was estimated to be about 75 percent of the initial injected
94
mass. This underestimated mass recovery of Br" may be the result of limited frequency of
monitoring during the flooded periods. Bromide started to arrive at the outlet 1.9 days,
after its addition into the wetland, reaching the background concentration 17.6 days,
later, on the extrapolated breakthrough curve. The actual and nominal detention time (t„)
were estimated to be 5.5 and 8.1 days, respectively. The nominal detention time was
calculated with the average of inlet and outlet flow rate and the effective volume of the
system.
According to the ratio between the actual and nominal detention time, the gravelbed porosity appeared to have decreased about 33 percent from the original effective
volume after 6 years of operation. This is in agreement with findings by Blazejewski and
Murat-Biazejewska (1997) and Tanner et al. (1998) who recently noted that gravel-bed
porosity decreases by accumulation of organic and inorganic matter from wastewater and
vegetation, reducing the hydraulic conductivity and detention time in the wetland.
Sanford et al. (1995) observed a consistent reduction over time of the hydraulic
permeability in four reed-vegetated subsurface flow beds. The high initial hydraulic
conductivity of coarse-gravel and pea-gravel matrix seemed to prevent surface flow.
However, in a sand-and-gravel bed the hydraulic conductivity decreased 87 percent after
one year of leachate addition causing surface flow and a prominent reduction of
treatment efficiency. Tanner et al (1998) reported that the original void space of a 9.5 m
95
X 2 m X 0.4 m vegetated gravel bed operating at an organic loading rate of 5.8 g m'' day"'
was reduced 50 percent after 5 years of operation. Accumulation of organic and inorganic
material primarily take places near the inlet of the wetland system. Kadlec and Watson
(1993) noticed high buildup of solids in the first 100 m of a gravel-bed wetland in
Benton, KY, and Tanner and Sukias (1995) reported a significantly higher accumulation
of organic material in planted wetlands than in unplanted ones. Preferentially, the
accumulation occurred near the inlet and the superficial 10 cm of the gravel bed.
Mixing of Br'
An ideal reactor would function under either plug flow or completely mixed
conditions with a cTg value of zero or one, respectively (Kadlec and Knight, 1996).
Ideally, back mixing becomes infinite in completely-mixed conditions and fractions of
entering water reach the outlet of the system with zero detention time (Teefy, 1996).
Ideal plug flow is opposite in behavior; the solute travels along the filter without backmixing residing in the basin for a period equal to the nominal detention time.
Asymmetry and large dispersion have been common characteristics of the
breakthrough curves from tracer studies conducted in subsurface-flow wetlands.
(Bowmer, 1987; Netter, 1994.; King et al., 1997; Tanner et al, 1998). This
hydrodynamic performance is expected for water flow through a medium with stagnant
zones that promote solute retention as the main cloud moves through the system.
96
Eventually, the solute is released slowly back into the convective zones promoting backmixing and making its breakthrough curve asymmetric.
A negative relationship of
with dimensionless arriving time (tio/tj) and
effective porosity from different experimental sites is illustrated in Table 2. Even though,
the sites are different in dimensions, vegetation, and hydraulic load mixing parameter,
t,o/tj and
cTq
values are comparable. Values of
indicate that subsurface flow
wetlands are more closely related to plug flow than to full-mixed conditions. However,
values such as 0.3 and 0.46 of
indicate a higher degree of full-mixing conditions, an
early arrive of the breakthrough curve (Figure 3), and probably a reduction of the
treatment efficiency. The tendency of hydrodynamic performance of a vegetated gravelfilled bed to plug flow behavior is a result of characteristics such as: a) relatively large
LAV ratio that promote enough time for a uniform distribution of the entering water
across the sectional cross area of the system (Polprasert and Bhattarai, 1985); b)
satisfactory effective
volume, the larger dead
volume the higher dispersion;
consequently, a closer condition to full mixed performance; c) even distribution of
entering water at the inlet to reduce the time for uniform distribution across the water
front (Perrson et aL, 1998), and d) appropriate hydraulic load of operation in contrast
with the hydraulic conductivity of the solid matrix to avoid surface flow.
97
PRDl Removal and Inactivation
The average background concentration of phage infecting PRDl host bacteria in
the influent water was 119 pfu ml"' fluctuating in a range of 104 to 150 pfii ml'' for both
summer and spring experiments. Detention time distribution for Br" and PRDl is
noticeably different (Figure 4). Because bacteriophage removal reduces its original load
through the rooted-gravel filter, the relative virus attenuation was estimated to be 97.6
percent.
Bacteriophage PRDl was mostly recovered during the early stage of the Br'
breakthrough curve. The dashed line in Figure 4 delimits 10 percent of Br' mass recovery
at 3.2 days after the tracer injection; at this time, 65 percent of total recovered PRDl
already had been recovered. Presumably, PRDl traveled through the shortest paths of
the vegetated gravel bed to reach the outlet of the system. The rooted filter design
suggest that the cobble layer is the wetland fraction of shortest detention time where
PRDl was removed inefficiently. Apparently, PRDl is removed by adsorption on the
wetland substratum where the virus had a longer contact period.
Dialysis bag suspension incubated in the gravel bed can be considered to have a
similar thermal, chemical, and biological environment to the wetland aqueous phase to
simulate and evaluate virus survival in the wetland during the tracer experiment (O'Brien
and Newman 1977; Gersberg et al., 1987).
98
Wetland water temperature averaged 16.3 °C. This temperature is in the range
(10-25 °C) where PRDI inactivation rates from -0.01 to -0.18 day"' in secondary effluent
have been reported (Schijven and Hassanizadeh, 2000). Figure 5 shows the inactivation
of PRDI at 36 and 54 m from the inlet. At both sites, PRDI behavior was similar with an
estimated inactivation rate of -0.16 day'' (0.70 R~). This result was comparable to the
inactivation rate reported by Vinluan (1996) in a survival test conducted in a multispecies surface flow wetland under shadow conditions at CERF if the inactivation rate
reported by the author in Log,o day"' is transformed to units per day"' (2.3 x Log,o day"').
On the other hand, the removal rate during the spring experiment was -1.17 day"' (0.99
R~) calculated from the normalized zero moment of injected pulse of PRDI, the
breakthrough curves at 36 and 54 m, and the time at its center of mass. Gersberg et al.
(1987) found a significant difference of MS2 removal in a vegetated gravel bed and its
inactivation in situ by using dialysis bags. Removal rates were estimated to be -0.48 ±
0.19 and -1.14 ± 0.09 day"' under stagnant and flowing conditions, respectively; whereas,
-0.12 day"' was the inactivation rate in the dialysis bags. Somewhat, these results suggest
a similar or longer persistency in subsurface flow wetlands of MS2 than PRDI. Which
contradicts results reported by Yahya et al. (1993) about a longer persistency of PRDI in
groundwater. Apparently, in the current study, virus inactivation and rhizosphere effects
were not the primary removal mechanisms. Studies conducted by Gesberg et al. (1987)
99
suggest that a mechanism responsible of irreversible colloidal collection by solid surfaces
during the filtration enhanced virus adsorption. Additionally, virus attachment on
suspended solids, eventually removed from wastewater by sedimentation or adsorption,
might have played an important role on virus removal from wastewater during the first
4.2 days of the tracer study conducted in CERP.
During the summer experiment of 1999, attenuation of PRDl was characterized
apparently by two processes with very different removal rates (Figure 5). Near the
influent, PRD1 removal was 96.3 percent but after 20 m, a more gradual reduction was
observed, reaching 99.38 percent at the end of the wetland. This apparently non-first
order PRDl removal may result from the sampling pipe location, the heterogeneous
distribution of the tracer between the rooted-gravel bed layers and the high variation of
the flow rates during the study. In contrast, in the spring experiment of 2000, PRDl
reduction showed an apparent first order process; probably, as a consequence of more
even flow rate conditions and a better location of the sampling pipe, intersecting the
cobble layer. The concentration of PRDl in the summer experiment was 36 percent of
the spring test, probably, because of the greater amount added in the former test.
However, the differences in the removal efficiency between both experiments was
negligible, PRDl removal in the second experiment was only 1.12 percent less than in
the summer test.
100
Only few studies have been conducted to determine the ability of constructed
wetlands to remove animal enteroviruses. Results reported by Thurston (1997) showed
that they were reduced by over 83 percent while Karpiscak et al. (1995) found a 98
percent-reduction. Both experiments were conducted in subsurface flow wetlands whit a
five day detention time. Scheuerman, et al. (1997) compared removal rates for
indigenous coliphage and enteroviruses in forested wetlands. The decay rates of
indigenous coliphage were slower than the observed rates for enteroviruses. Similar
results were found by Gersberg et al. (1987) comparing decay rates of human
polioviruses to an indigenous FRNA bacteriophage in vegetated gravel beds receiving
municipal wastewater. The indigenous phage was reduced at slower rates than the human
virus and 1VIS2.
PRDl Flushing
Schijven and Hassanizadeh (2000) indicated that laboratory and field evidence
suggests that virus removal in soils is mainly governed by kinetic processes. Figure 6
shows the long tailing of PRDl in both the sununer of 1997 and the spring of 2000. In
these studies, the outflow PRDl concentration at the end of study was higher than the
background concentration observed in the sampling sites before tracer injection. In soil
column experiments, Dowd et al. (1998) found that low virus concentrations can be
101
observed for extended periods of time after the virus input had ceased. Schijven et al.
(1999) found a long tailing of the virus concentration distribution for long periods of
sampling in soil tracer studies conducted under field conditions. Apparently, this
behavior is produced by virus detachment from the solid phase which is relatively much
slower than attachment. Once retained on the surface of collectors, virus inactivation may
predominantly occur with a small inactivation rate.
CONCLUSIONS
The detention time of the vegetated-gravel bed estimated from the outlet Br'
breakthrough curve observed in the spring 2000 was 5.5 days. The PRDl removal and
inactivation rate was -1.17 and -0.16 day"', respectively. Even though, PRDl recovery
during the summer experiment was 36 percent of the spring one, the total removal was
similar in both experiments. Climatic conditions appear somewhat irrelevant to PRDl
removal, at least for the summer and spring period in a subsurface flow wetland.
Bacteriophage PRD1 was mostly recovered during the early stages of the Br
breakthrough curve suggesting that the recovered firaction of PRD 1 has a low tendency
for attachment or that its removal is strongly affected by preferential flow.
In agreement with the inactivation rate obtained in the survival experiment,
PRDl can survive in the wetland for a significant period of time. Approximately, 13.9
102
days would be required for 90 percent inactivation of the initial amount of bacteriophage.
This is similar to results reported by Vinluan (1996) for 90 percent PRDl reduction in a
survival test conducted in surface flow wetland at CERP under shadded conditions. The
inactivation rate in situ was similar to reported values from others for PRDl (Vinluan,
1996; Schijven and Hassanizadeh, 2000) and comparatively, this decay rate is almost one
order of magnitude less than the removal rate estimated in the spring 2000 tracer test.
Presumably, adsorption was the primary mechanism of virus reduction during the first
4.2 days of PRDl displacement through the rooted-filter. For longer periods, inactivation
may be of greater importance. The long tailing of PRD1 breakthrough curve at the outlet
of the system suggests that virus detachment from wetland substratum can be a long term
source of viruses for water passing through a subsurface flow wetland which appears
capable of reducing PRD 1 by 99 percent of the initial virus load, if the hydrodynamic
conditions of the vegetated gravel bed are close to plug flow behavior.
103
ACKNOWLEDGMENTS
The authors wish to acknowledge technical support from Sue Hopf and Glenn
France from the University of Arizona's Office of Arid Land Studies. Financial support
was provided by Tucson Water, City of Tucson Arizona; Sanitation Districts of Los
Angeles County; United States Environmental Protection Agency; American Water
Works Association Research Foundation; USDA Water Conservation Laboratory; Pima
County Department of Wastewater Management; Water Environment Research
Foundation; City of Phoenix, Arizona; The Subregional Operators Group (Phoenix-area
cities); Water Replenishment District of Southern California; City of Riverside,
California; and City of Los Angeles Department of Water and Power.
DISCLAIMER
The American Water Works Association Research Foundation and the other
agencies listed above not had opportunity to review and comment on this paper,
therefore, none of these agencies necessarily endorse the findings presented here.
104
REFERENCES
Adams M.H. (1959) Bacteriophage. Interscience Publication, New York.
Bitton G. (1994) Wastewater Microbiology. Wiley-Liss, New York.
Blazejewski R. and Murat-BIazejewska S. (1997) Soil clogging phenomena in
constructed wetlands with subsurface flow. Wat. Sci. Tech. 35, 183-188.
Bowmer K. (1987) Nutrient removal from effluents by an artificial wetland: influence of
rhizosphere aeration and preferential flow studied using Bromide and dye tracers.
Wat. Res. 2\, 591-599.
Caldentey, J., Bamford J.K.H. and Bamford D.H. (1990) Structure and assembly of
bacteriophage PRD-1, and Escherichia coli virus with membrane. J. Struct. Biol.
104, 44-51.
Dowd S.C., Pillai S. D., Wang S. and Corapcioglu M. Y. (1998) Delineating the specific
influence of virus isoelectric point and size on virus adsorption and transport
through sandy soils. Appl. Environ. Microbiol. 64, 405-410.
Gersberg R. M., Lyon S. R., Brenner R. and Elkins B. V. (1987) Fate of viruses in
artificial wetlands. Appl. Environ. Microbiol. 53, 731-736.
Harvey R. W. and Garabedian S. P. (1991) Use of colloid filtration theory in modeling
movement of bacteria through a contaminated sandy aquifer. Environ. Sci.
Technol. 25, 178 - 185.
Kadlec R. H. and Knight R. L. (1996) Treatment Wetlands. Lewis Publishers, New York.
Kadlec R. H. and Watson J. T. (1993) Hydraulics and solid accumulation in a gravel bed
treatment wetland. In Constructed Wetlands for Water Quality Improvement.
Lewis, Boca Raton, FL.
Karpiscak, M.M., Foster, K.E. and Hopf, S.B (1993) Pima county constructed ecosystems
research program: annual progress report, phase V. Office of Arid Lands
Studies, The University of Arizona, Tucson AZ.
Karpiscak M. M., Gerba C. P., Watt P. M., Foster K. E. and Falabi J. A. (1995) Multispecies plant systems for wastewater quality improvements and habitat
enhancement. Second International Symposium on Wastewater Reclamation and
Reuse. Iraklio, Greece, pp 37-42.
Karpiscak, M.M., Kennith E. F., Hopf S. B., Bancroft J. M. and Warshall P. J. (1994)
Using water hyacinth to treat municipal wastewater in The Desert Southwest.
Water Resour. Bull. 30, 219-227.
King A.C., Mitchell C. A. and Howes T. (1997) Hydraulic tracer studies in a pilot scale
subsurface flow constructed wetland. Wat. Sci. Tech. 35, 189-196.
Loveland, J.P., Ryan, J.N., Amy, G.L. and Harvey, R.W. (1996) The reversibility of virus
attachment to mineral surfaces. Colloids Surfaces A: Physicochem. Eng. Aspects
107,205-221.
Netter R. (1994) Flow characteristics of planted soil filters. Wat. Sci. Tech. 29, 37-44.
O'Brien R.T. and Newman J. S. (1977) Inactivation of polioviruses and coxsackieviruses
in surface water. Appl. Environ. Microbiol. 33, 334-340.
Olsen R. H., Siak J. and Gray R. H. (1974) Characteristics of PRDl, a plasmid-dependent
broad host range DNA bacteriophage. J. Virol. 14, 689-699.
Perrson J., Somes N. L. G. and Wong T. H. F. (1998) Hydraulic efficiency of constructed
wetlands and ponds. Draft of the Proceedings of the Sixth International
Conference on Wetland Systems for Water Pollution Control. Aguas De Sao
Pedro, Brazil.
Polprasert, C. and Bhattarai K. (1985) Dispersion model for waste stabilization ponds. J.
Environ. Eng. Ill, 45-59.
Powelson, D.K. and Gerba C.P. (1994) Virus removal from sewage effluents during
saturated and unsaturated flow through soil columns. Wat. Res. 28, 2175-2181.
Rusin, P., Enriquez C.E., Johnson D. and Gerba C.P. (2000) Environmentally
Transmitted Pathogens. In Environmental Microbiology ( Edited by Maier R.M.,
Pepper I.L., and Gerba C.P.) pp 447-489. Academic Press, San Diego, CA.
Ryan, J.N., Elimelech M., Ard R.A., Harvey R.W. and Johnson P.R. (1999)
Bacteriophage PRDl and silica colloid transport and recovery in an iron oxidecoated aquifer. Environ. Sci. Technol. 33, 63-73.
Sanford W.E., Steenhuis T.S., Perlange J. Y., Surface J.M. and Peverly J.H. (1995)
Hydraulic conductivity of gravel and sand as substrates in rocked-reed filters.
EcoL Eng. 4,312-336.
Schijven J.F. and Hassanizadeh S. M. (2000) Removal of viruses by soil passage:
overview of modeling, processes, and parameters. Crit. Rev. Environ. Sci.
Technol. 30, 49-127.
Schijven J.K., Hoogenboezem W., Hassanizadeh M. and Peters J. H. (1999) Modeling
removal of bacteriophage MS-2 and PRD-1 by dune recharge at Castricum,
Netherlands. Water Resour. Res. 33, 1101-1111.
Tanner C. C. and Sukias J. P. (1995) Accumulation of organic solids in gravel-bed
constructed wetlands. IVat. Set Tech. 32, 229-239.
Tanner C. C., Sukias J. P. S. and Upsdell M. P. (1998) Organic matter accumulation
during maturation of gravel-bed constructed wetlands treating farm dairy
wastewater. Wat. Res. 32, 3046-3054.
Tanner C. C. (2000) Personal communication.
Teefy S. (1996) Tracer Studies in Water Treatment Facilities: A Protocol and Case
Studies. American Water Works Association, Denver, CO.
108
Thurston, J. (1977) Fate of pathogenic and indicator microorganisms in two subsurface
multi-species
constructed
wetlands.
Master's
thesis.
Soil,
Water,
and
Environmental Science Department. University of Arizona, Tucson AZ.
Vinluan E. A. (1996) Survival of Microbial Indicators in Constructed Wetlands. Master's
thesis. Soil, Water, and Environmental Science Department, University of
Arizona, Tucson AZ.
Yahya, M. T., Galsomies L., Gerba C. P. and Bales R. C. (1993) Survival of
bacteriophages MS-2 and PRD-1 in ground water. Wat. Sci. Tech. 27, 409-412.
109
Table 1.- Flow rates at the inlet and outlet for Summer and Spring tracer
experiments
Parameter
Summer 1997
Inlet
Spring 2000
Outlet
Inlet
Outlet
Mean (m^ h"')
0.78
0.45
0.90
0.66
Standard Deviation (m^ h"')
0.33
0.36
0.11
0.17
Coefficient of variation (%)
42.47
79
11.68
24
Table 2.- Wetland characteristics and Br'breakthrough curve parameters
Sourcc
Dimensions
Vegetation
Hydraulic
Effective
Dimensionless
Dimensionless
(m)
loading rate
porosity
arriving time
variance
LxWxH
(mm day"')
(%)
(t../y
tA X 0
Tanner(1998)
9.5 X 2 X .4
Bulrush"
74
Palla'
A
17.5
0.35
0.3
B
21J
0.53
0.18
C
35.4
0.62
0.11
Bavor et al.
0.19
Schoenoplectus
(1988)'
0.46 = 0.03
Schierup et al.
(1990)'
CERF
61 X 8x0.9
Cited by Kadlec and Knight 1996'*
SchotnopUaus tabernaemontan^
S. vatidus'
•Multi-species
43
26
0.58
0.18
110
Vertical Scctiovs
I 1 Cobbta : 5'toanof itzj:
Emargau nuilti-spcdes vcgdcttioii
61 m
Figure 1. Schematic representation of the vegetated-gravel bed where the tracer
study was conducted. The vertical sections (A-A' and B-B') show gravel and cobble
layers stratification in the wetland. The influent water is distributed along the inlet
by the inlet pipe.
Ill
0.08 -I
0.07 -
S:? 0.06 0.05 0.04 0.03 0.02 0.01 -
0
50
100
150
200
250
300
Time since tracer addition (h)
Figure 2. Relative Concentration distribution of Br' at 20 (•), 40 (•), and 56 m (A)
from the inlet. Summer 1997, expressed by the ratio of the observed (C) and input
concentration (Co).
112
9
8
7
O)
CERF
6
5
4
3
2
1
0
0
0.5
1
1.5
2
2.5
3
3.5
Dimensionless time (t/td)
Figure 3. Detention time distribution from CEEUF and previous reported studies by
Tanner et al. (1998). Curves A, B, and C were normalized on time with information
provided by Tanner (2000).
113
2.00E-02
O
O 1.50E-02
G
c
_o
CO
g 1.00E-02
u
c
o
o
(I)
>
ra 5.00E-03
0)
o:
••
—I—
O.OOE+OO
5
10
15
—I
20
Time since tracer addition (days)
Figure 4. Relative concentration distribution for Br" (•) and PRD1(«) during the
spring tracer experiment expressed by the ratio of observed (C) and input
concentration. The dashed line shows the time at 10 percent Br" mass recovery.
114
10
o
z
z
o
O)
o
0.1
—I
0.01 -
0.001
0
2
4
6
8
10
12
14
Time (days)
Figure 5.- Inactivation and removal performance of PElD-1 in the summer of 1999
(removal • at 20, 40 and 56m) and spring of 2000 (removal • at 36 and 56 m;
inactivation at 36 • and 54 m •) tracer experiment.
115
1.00E-03 -
2 1.00E-04
G
c
o
5 1.00E-05
cou
u
(1>
>
^ 1.00E-06 -
1.00E-07
5
10
15
20
Time since PRD1 addition (days)
Figure 6. Detention time distribution of PRD-1 during the summer (•) and spring
tracer experiment (•). The error bars represent the range of phage background
concentration in the influent water infecting the PRDl host.
116
APPENDIX C
REMOVAL OF CHEMICAL AND MICROBIAL INDICATORS OF POLLUTION
IN A SURFACE FLOW CONSTRUCTED WETLAND
Juan A. Vidales'*, Charles P. Gerba', and Martin M. Karpiscak"
'Department of Soil, Water and Environmental Science, University of Arizona, Tucson
AZ 85721 . "Office of Arid Lands Studies, University of Arizona, Tucson AZ 85719.
'Corresponding Author
117
Removal of Microbial and Chemical Indicators of Pollution in a Surface
Flow Constructed Wetland
Juan A. Vidales', Charles P. Gerba', and Martin M. Karpisaclr
'Department of Soil, Water and Environmental Science, University of Arizona. ^Office of
Arid Lands Studies, University of Arizona.
Abstract- Spatial distribution of physical, chemical, and microbial indicators of pollution
was determined in a constructed wetland operated with chlorinated secondary effluent and
backwash water. Samples were collected before and after vegetation removal. The system
studied consisted of two densely vegetated settling basins (0.35 ha), an artificial stream, and
a 3-ha surface flow wetland, planted with bulrush {Scripus spp.) and cattail {Typha
domingensis). The average inflowofsecondary effluent was 1.9 m^ min"'while the inflow
during backwash water ranged from 0.21 to 0.42 m^ min"'. The system was able to reduce
TSS and BOD5 to tertiary effluent standards, however, monitoring of chloride revealed that
evapotranspiration is concentrating chemical and microbial pollutants. Coliphage removal
during backwash operation was 93 and 46 percent of the influent load at the end of the
system during 1999 and 2000, respectively. During periods when secondary effluent entered
the system, coliphage removal was 65 percent. After vegetation removal, pH and coliphage
density increased significantly (p<0.05) at the outlet of the wetland. The findings of this
study suggest that coliform growth or animal fecal contributions are occurring in the system.
118
INTRODUCTION
Numerous countries are actively using constructed wetlands for wastewater treatment
because they have been effective in reducing biochemical oxygen demand (BOD5), total
suspended solids (TSS) and coliform bacteria (Geller, 1997; Green et al., 1997; Mashauri
et ai, 2000; Perkins and Hunter, 2000). These ecosystems are being used to treat a wide
variety of wastewater in many climatic zones and compared to traditional wastewater
treatment facilities, their operation costs and trained personnel demands are noticeably
lower.
Total and fecal coliforms have been reduced in constructed wetland with efficiencies
over 90 percent. These efficiencies, however, can be decreased significantly depending on
the hydraulic load (Greened a/., 1997; Tanner era/., 1998) and vegetation density (Gersberg
et al., 1989). Moreover, coliform densities may be considerably higher in the outflow of
wetlands if they attract abundant wildlife or receive additional fecal contamination (Kadlec
and Knight, 1996).
Scarce data has been published about removal of viral indicators from wastewater
and particularly from backwash water in constructed wetlands. The available information
reveals that viral removal has been studied under various approaches such as the ratio of
outflow and inflow load of indigenous coliphage (Karpiscak et al, 1995), survival tests
(Vinluan, 1996), and tracer studies (Chendorain et ai, 1998). Although these studies have
119
shown that virus removal may be as high as 99.9 percent of the influent load, higher
concentrations at the outlet have been observed in non-vegetated wetlands (Gersberg et al.,
1989). Other studies have suggested that virus populations decline rapidly with the distance
(Scheuerman er a/., 1989).
This paper reports the findings of a monitoring program for coliphage, and total and
fecal coliforms as well as physical and chemical pollution indicators in a constructed
wetland which was designed to remove suspended solids from backwash water before
aquifer recharge. The wetland received secondary and backwash water at variable flow rates
for 8 months prior to vegetation removal and backwash water for 8 months following
removal.
MATERIALS AND METHODS
The Research Site
Research was conducted at the Sweetwater Wetland and Recharge Facility inTucson,
AZ. The system consists of two polishing subsystems, east and west. These facility has been
in operation since 1997 and were designed to remove suspended solids from wastewater
produced during periodic backwashing of pressure mixed media filters used to treat
secondary effluent at the City of Tucson Reclamation Plant. Residual chlorine in secondary
120
effluent at the pressure mixed media filters is on average I mg L"'. The backwash water
produced during washing does not receive additional chlorine.
Wetland influent water is introduced into a splitter structure that usually diverts the
inflow in equal parts to each of the treatment subsystems. The water then enters a pair of
settling basins with a combined area of 0.35 ha. These basins are densely vegetated with
bulrush (Scirpus spp.) in the East subsystem, and cattail (Typha domingensis) in the West
subsystem. After passing through the settling basins the water enters large surface flow
wetland cells. However, in the East subsystem water flows briefly through a small artificial
stream before entering the wetland ceil. Each of the wetland cells is about 3 ha in area and
contains open water areas (1.2 m deep) and densely vegetated shallow zones (0.3 m deep).
The dominant plants in both the large wetland cells are species of bulrush {Scirpus spp.) and
cattail {Typha domingenses).
Sampling Sites and Analysis
From February to September of 1999 and again in 2000, water samples were
collected in I-L bottles monthly at the backwash splitter box, outlet of the south settling
basin, both ends of the stream, and outlet of the wetland cell in the East subsystem (Figure
1). Concurrently water temperature was taken with a standard mercury thermometer at the
site of sampling. Measurements of biochemical oxygen demand (BOD5), total suspended
121
solids (TSS), sulphate (SO4), chloride (CI"), turbidity, pH, coliphage, and total (TC) and
fecal (FC) coliforms were conducted in the laboratory.
Physical and Chemical Analysis
The 5-day incubation method found in Standard Methods for the Examination of
Water and Wastewater was used to determine BODj. Determination of TSS was made by
filtering a known volume of sample through a precleaned and preweighed glass fiber filter,
drying the filter for at least 24 hours at 100 °C, reweighing the filter, and calculating the
concentration. Sulfate was determined by adding BaClj to a know volume of sample and
measuring the absorbance at 420 nm in a HACH DR/2000 spectrophotometer (Loveland,
CO). If the sample was at all colored, a "blank" sample (without the addition of BaCy was
read and the absorbency subtracted from that reading made in the sample with the BaCl,
added. The absorbency was compared against a standard curve. Chloride was determined
using a chloride-specific electrode method. One ml of Ionic Strength Adjuster (ISA)
Solution (5 M NaNOj) was added to 50 ml of room-temperature sample. The Crspecific
electrode attached to a pH meter (Coming, model M220, New York) is placed in the
solution and the mV of the sample solution read and compared against a standard curve.
Turbidity was measured by using a portable turbidimeter (HACH, model 2 lOOP, Loveland,
CO), pH with a pH meter (model 8005, West Chester, PA) and total and firee chlorine (CI,)
using the DPD method (HACH Spectrophotometer, model DR/2000, Loveland, CO).
122
Coliforms and Coliphages
Coliform bacteria were analyzed within 4 h of sampling by membrane filtration
using mEndo Agar Les and mFC culture media (DIFCO, Detroit, MI) for total coliforms
(TC) and fecal coliforms (FC), respectively. The membrane filters were 47 mm diameter
with a porosity of 0.45 |im (Milipore, Molsheim, France). Sample volumes of 0.1,1, and 10
ml were assayed and incubated at 37 and 44.5 °C for TC and FC, respectively. The colony
formed units (cfii) developed after 24 h of incubation were enumerated.
Coliphage was enumerated by the double layer agar method described by Adams
(1959). A 1-ml aliquot from Escherichia coli ATCC 15597 (ATCC) culture, previously
incubated at 37 °C for 24 h in Trypticase Soy Broth (DIFCO, MI), was combined with 1 ml
of collected sample in a test tube containing molten overlay agar. The agar was poured onto
a layer of Tripticase Soy Agar (DIFCO, MI), and incubated at 37°C for 18 h in order to
enumerate the coliphage plaque forming units (pfu).
Statistical Analysis
The statistical analysis was conducted by using SYSTAT 9 (SPSS Inc., Chicago
ILL). Tests to determine significant difference between observations of BOD5, CI", SO4, T°,
pH, and turbidity, at the sampling sites during backwash operation periods, were conducted
by ANOVA analysis.
123
The high sensibility of the arithmetic mean to extreme values observed for TSS and
microbial indicators during backwash operation suggested that the median may represent
better the central tendency of data sets collected during the study. Therefore, TSS
concentrations and microbial populations were statistically analyzed comparing the median
observations between sites and backwash periods to reduce extreme value effects. The ten
variables studied were compared to each other by Pearson Correlation Analysis.
For the second and third period of sampling, TC, FC, and coliphage concentrations
in the collected samples were represented statistically in box plots. In these figures, the
horizontal line shows the median, ends of the box the 25th and 75th percentiles, error bars
the 10th and 90th percentiles, and extreme values are depicted by circles. A nonparametric
Mann-Whitney U-test was used to compare the median observations between sampling sites
and backwash periods.
RESULTS
Sampling
Based on the type of effluent entering the system and vegetation removal, the
sampling program was divided into three major sampling periods: 1) Secondary effluent
period: from February 12 to March 18 of 1999, the system was receiving chlorinated
secondary effluent at an average rate of 1.84 m^ min"' to facilitate a tracer study. The
124
detention time of the wetland cell determined at the end of the test was 7.3 days (Vidales et
al., 2000). During this time, the influent water was introduced into the splitter box and
flowed into the stream without entering the east pair of settling basins. Periodically, the
basins were flooded with backwash water to keep them wet. On March 18, the system was
returned to normal operation averaging approximately the same flow rate as was used during
the tracer study. The influent water thus flowed from the splitter box passing through both
settling basins as well as into the east polishing wetland cell. A mixture of chlorinated
secondary and backwash water was introduced into the system until March 23 of 1999,
2) Backwash period before vegetation removal: only backwash water was received by
the east subsystem from March 23 to September 21 of 1999. From March 23 to June 30, the
average influent rate was 0.42 m^ min"' with a coefficient of variation of 24 percent. This
influent rate was reduced to 0.21 m^ min"' from the end of Jun through September 21 when
the east subsystem started to be drained for vegetation removal. The theoretical detention
time during this period was in a range of 25 to 50 days.
3) Backwash period after vegetation removal: After vegetation removal during the
Winter of 1999, the system was return to normal operation on February 2000 at an average
inflow rate of 0,32 m^ min"' of backwash water.
These rates were calculated with
information provided by Tucson Water for the period of February 17 to September 10 of
2000. The theoretical detention time was estimated to be between 28 and 38 days.
125
Initial Period of Secondary Operation
BOD5, TSS, and Turbidity
Two samples were collected per sampling site during February and March of 1999.
The influent concentrations of BOD5 and TSS were 29 and 21 mg L"', respectively. Both
concentrations decreased 69 percent after filtration through the settling basin (Figure 2). In
the stream, BOD5 was around 7 mg L"' and remained at this level through the system
reaching 8 mg L"' at the outlet end of the wetland cell for a total removal of some 72 percent.
TSS mean concentration was lowered slightly to 5 mg L"' in the stream and was below the
detection limit (5 mg L"') of the analytical method used in the wetland outflow. Turbidity
was very closely related to BOD5 and TSS distribution in the system. The turbidity of the
water entering the system was 17.9 NTU which fell to 7.8 in the settling basin and 7.6 NTU
in samples from the inlet end of the stream, and then was reduced to 2.8 NTU at the end of
the wetland cell for an average removal efficiency of 84 percent. Pearson correlation
coefficients comparing turbidity to BODj and turbidity to TSS concentrations in effluents
from all sampling sites were 0.93 and 0.97, respectively.
Cr, SO^^', pH, and Temperature
The average concentration of CI" was 126 mg L"' in samples collected from the
splitter box (Figure 3) decreasing to 116.5 mg L"' at the outlet of the wetland. A similar
overall trend was observed for SO4"" with the mean concentration decreasing from 128 mg
126
L"' at the splitter box, to 122.5 mg L"' at the outlet of the wetland. However, the
concentration of SO4 was higher at the outlet of the settling basin, and at the outlet end of
the stream. Pearson correlation analysis revealed a correlation coefficient of 0.76 comparing
average concentrations of SO4to CI".
Water pH was considerably lower at the outflow from the settling basin than at the
splitter box (Table la ). In samples collected from both ends of the stream, water pH
increased from 7.5 to 7.8 reaching 7.6 at the outlet of the wetland which was lower than the
original 7.8 at the splitter box. The mean water temperature at the splitter box was 22.7 °C
decreasing to 20.2 °C at the outlet of the settling basin, 18 °C at the end of the stream and
10.5 °C at the outflow from the wetland (Table I).
On February 19, total and free CU concentrations were 1.19 and 0.14 mg L"' ,
respectively, in water from the splitter box . Both concentrations were below the detection
limit in samples from the other sampling sites. On March 20, CI, was not detected at any
sample point within the east system.
Indicator Microorganisms
The average TC density in water samples from the splitter box was 2.56 logig
increasing to 4.21 logig in the settling basin and decreasing from 4.07 to 3.81 log,o at the
outflow point of the stream (Figure 4). Treatment in the wetland resulted in a reduction of
approximately one order of magnitude. The incoming FC population observed at the splitter
127
box was 1.81 log,o gaining about two log,o in the settling basin. In the stream, the inflow and
outflow concentrations were 3.63 and 3.07 log,o respectively. At the same time, at the
wetland outlet, the observed FC population was 2.35 log,o. Coliphage load in the incoming
water was 3.7 log,o; however, the highest concentration of 3.86 log[o occurred in the settling
basin outflow. At both ends of the stream, coliphage concentrations were 3.9 and 3.78 log^
which decreased to 3.23 logig at the outlet of the wetland for an average removal of 65
percent.
Backwash Operation Before Vegetation Removal
BODg, TSS, Temperature, and Turbidity
Concentrations of TSS, and BOD5 as well as turbidity detected during backwash
operation for the last 6 months of 1999 are reported in Figure 5. ANOVA analysis showed
that the influent turbidity of 155 NTU decreased significantly (p<0.05) to 25 NTU at the
outlet of the settling basin, and 33 NTU at the wetland outflow.
During this backwash period, BOD5 concentrations demonstrated inconsistency
because of the high variability in the water quality. The most critical situation was with
samples from the settling basin where only four samples out of eight gave valid results. The
mean BOD5 concentration in water entering the system was 127 mg L"' which decreased
significantly (p<0.1) to 51 mg L"' in samples collected from the settling basin. In the stream.
128
the BODj concentration went from 37 to 28 mg L"' and then diminished to 14 mg L"' in the
outflow of the wetland.
Similar to BOD5, the mean water temperature also varied along the system. For
example, it was 27.7 °C in the entering water, 25.4 °C in the settling basin, and 19.8 °C in
the wetland outflow (Table lb).
Since the arithmetic mean is highly sensitive to extreme values as shown by the TSS
concentration detected at the outlet of the settling basin on May 21 of 1999 (Figure 5), the
median of the observed concentrations at each sampling site was chosen as a better
representation of their central tendency. The noparametric Maim-Whitney U-test revealed
that the median influent concentration of 159 mg L"' diminished significantly (p<0.05) in
samples from the settling basin, where the estimated median was 8.5 mg L"'. This
concentration was 5 mg L"' in the water from the stream and increased to 6 mg L"' at the end
of the wetland cell. Based on the observed concentrations, the TSS was removed, on
average, more efficiently than BOD5 and turbidity. Finally, the observed removal
efficiencies in the system for these three parameters were about 96, 84, and 76 percent,
respectively.
Cr, SO^^", and pH
From April to September of 1999, ANOVA analysis established that the average
influent level of CI" increased, no significantly (p<0.05), firom 141 to 164 mg L"' in the
129
outflow of the settling basin (Figure 6). This concentratiom declined to 142 and 155 mg L"'
at in and out of the stream; however, at the outlet of the \wetland, an increase of 40 mg L'^
over the former concentration was statistically significatiwe (p<0.05).
Initial concentration levels of sulphate coming into tihe wetland facility averaged 144
mg L"' which was decreased significantly (p<0.05) in^ the settling basin (Figure 7).
Meanwhile, in the stream, concentrations of SO4*" reacheal an approximated level of 125
mg L"' increasing to 146 mg L"' at the end of the polishing wetland. In the same effluents,
water pH ranged from 7.4 to 7.5 at the splitter box, settlimg basin, and stream; whereas, at
the outflow of the wetland, pH decreased, not significativ»ely (p<0.05), to 7.35.
Indicator Microorganisms
Both total and fecal coliforms followed a similar distribution pattern during
backwash operation in 1999. The median concentration of"TC in the influent water was 3.5
Iog[o increasing significantly (p<0.05) lo 4.5 Iog,o in water samples collected from the
settling basin (Figure 8). Practically, no TC removal occurrred in the stream and outlet of the
polishing wetland where a concentration of 4.6 log,o w;.as observed. Similar to TC, FC
populations also increased, significatively (p<0.1), abomt an order of magnitude in the
settling basin (Figure 9). The FC density was stabilized at some 4.0 log,o until the end of the
stream reaching 4.3 log,o in the outflowing water firom the system.
130
The median coliphage load in backwash water at the splitter box was 4.2 log[o
decreasing significatively (p<0.1) to 3.7 logig after passing through the settling basin (Figure
10). The coliphage concentration decreased to 3.4 logig at the end of the stream and 3.1 log,o
in the outflow of the wetland cell, for a total removal of 9 2 percent.
Backwash Operation After Vegetation Removal
BOD5, TSS, Temperature, and Turbidity
The third study started on February 2000, after vegetation removal, with 100 percent
backwash water entering the system. During this period, the average BOD5 load in the
incoming backwash water was 145 mg L"' (Figure 11). The ANOVA test showed that the
38 mg L*'-concentration observed at the outflow from the settling basin was statistically
lower (p<0.05) than the entering BODj load. Little, if any, BODj removal occurred within
the stream since the observed concentrations were 37 and 31 mg L"' at the inlet and outlet,
respectively, decreasing significatively (p<0.05) to 23 mg L"' in the outflow of the wetland
for a total system removal of 84 percent.
The amount of total suspended solids showed a distribution similar to that of BOD5
concentrations and turbidity during this third sampling period. The influent concentration
of 123 mg L'' was lowered significantly (p<0.05) to 7.9 mg L"' in the settling basin effluent.
This BOD5 level increased slightly to 11.6 mg L"' after passing by the stream and wetland
for a removal efficiency of 90 percent.
131
Influent turbidity was reduced by the settling basin from 128 to 57.9 NTU, remaining
without change until the outlet of the stream. At the outlet of the wetland cell, turbidity was
about 33 NTU.
Similar to turbidity, the mean water temperature tended also to decrease from the
inflow to the outflow of the system. Influent water temperature averaged 29.3 °C
decreasing to 25.6 °C at the outlet of the settling basin. Additional decreases were observed
in the stream (25 °C) and again in the wetland outflow (23.4 °C).
After vegetation removal, only at the outlet of the settling basin, turbidity increased
significatively (p<0.05). Similarly, the ANOVA test showed that the mean concentration
for BOD5 observed at the outlet of the wetland was significantly higher than the one
observed
under backwash operation
before vegetation removal. In contrast, the
nonparametric Mann-Whitaey U-test revealed that the median TSS concentration before and
after vegetation removal were statistically (p<0.05) similar.
Cr, 504'-, and pH
The average load of CI' concentration entering the system was 117 mg L"' which was
similar to the outflow of the settling basin (Figure 6). Concentration of CI'was 107 and 115
mg L'' in the inflow and outflow of the stream, respectively, reaching 137 mg L'' at the
outlet of the wetland.
132
The average SO4 concentration in the incoming water was 130 mg L"' which
decreased to 115 mg L"' at the end of the stream (Figure 7). Between sampling sites, the
difference of SO4*' concentrations was not statistically significative (p<0.05). However, the
mean concentration of 148 mg L"' observed at the end of the wetland cell was statistically
(p<0.05) higher than the latter concentration.
The pH readings obtained in water samples from the splitter box to the stream were
about 7.4. Average pH increased to 7.7 in the wetland cell. Statistically, these results were
similar to pH readings obtained during the first period of backwash water except for those
observed at the outlet of tlie wetland. Similarly, there was no statistical evidence to indicate
that vegetation removal from the system caused a significative change (p<0.05) in SO4""
concentrations at the monitoring sites. In contrast, CI" concentrations were significantly
lower (p<0.05) after vegetation removal.
Indicator Microorganisms
The spatial distribution of TC and FC populations along the system are reported in
Figure 8 and 9, respectively. The median TC density in the influent water was 3.8 logiQ.
This concentration increased significatively (p<0.05) to 4.5 logio in the outflow of the
settling basin, stream, and outlet of the wetland. Likewise, FC concentrations increased
significantly (p<0.05) from 3.4 log,o at the splitter box to 4.0 log,o in samples collected from
133
the settling basin. The stream did not reduce FC populations significantly (p<0.05). At the
outflow from the wetland, the median population was 3.7 log,o.
The nonparametric Mann-Whitney U-test revealed that TC and FC densities were
statistically constant (p<0.05) during both backwash operational
periods. Therefore,
vegetation removal did not cause a significant change of coliform distribution in the system.
A statistical representation of coliphage concentrations after vegetation removal can be
appreciated in Figure 10. During this period, an influent coliphage population of 3.8 log,o
was reduced significantly (p<0.05) to 3.5 logm in the settling basin. A similar median
concentration was also observed in the wetland outlet for a removal efficiency of 46 percent.
The nonparametric Mann-Whitney U-test showed that the differences found during the two
backwash periods (Figure 10) were not statistically different for most points in the system.
However, after vegetation removal, a significant increase (p<0.05) of coliphage densities
was observed at the end of the wetland cell.
Correlation Analysis
Correlation analyses were conducted to determine the relationship among the 10
variables measured during the current study. Corresponding Pearson coefficients often
showed no correlation between the paired variables thus only those with correlation
coefficients over 0.5 are reported. The Pearson correlation coefficient comparing TSS to
turbidity in the incoming and stream water was about 0.8. At the other three sites, these
134
coefficients were significantly low suggesting no correlation between both variables. When
coliphage densities were compared to their paired TSS correlation coefficients in the order
of 0.65, 0.68, 0.41, and 0.61 were estimated at the splitter box, settling basin, stream, and
outlet of the wetland. Also, FC populations and TSS concentrations were compared yielding
a negative correlation of -0.6 in samples from the splitter box, -0.8 in the settling basin
effluent, -0.57 and -0.68 in collected samples from both ends of the stream, and -0.44 in the
wetland outflow. Conversely, BOD5 concentrations were positively correlated to TSS levels
in collected samples from the sampling sites. At these locations, the correlation coefficients
were 0.17, 0.94, 0.55, 0.30 and 0.71 from the splitter box to the wetland outflow. When
comparing water pH to CI' concentrations, the correlation coefficients were ranging from
0.57 to 0.77 in the splitter box, settling basin, and stream outflow, respectively. However,
this coefficient was 0.23 at the inlet of the stream and -0.71 at the wetland outlet.
DISCUSSION
Pollution Indicators
A significant increase of turbidit>', BOD5, and TSS concentrations was observed
when the water entering the wetland was switched firom secondary effluent to backwash
water. Observed reductions of turbidity, BOD5, and TSS occurred mostly in the settling
basin where removal of turbidity ranged from 56 to 83 percent, BOD5 firom 60 to 75 percent.
135
and TSS form 62 to 94 percent. A higher removal of TSS and turbidity was observed during
operation with backwash water whereas higher BOD5 removal occurred during the initial
two month period when secondary effluent was entering the system. These parameters
showed a similar evolution in the system; however, when their data sets obtained at each
sampling location were compared to each other a small correlation was often observed.
Apparently, BOD5, TSS, and turbidity levels depend on other factors that are affecting their
interrelationship (Kadlec and Knight, 1996). For example, turbidity may be affected by the
color produced in water by impurities like algae or sediments from reductive environments
( Maier, 2000).
At the wetland outlet, reduction of BODj are comparable to results reported by
Vrhovsek et al. (1996) who found a 89 percent reduction in a subsurface flow wetland
operated at a BOD5 loading rate of 962 mg LRemoval of BODj, and TSS found in the
current study are in general agreement with wetland systems operating across the USA (
Kadlec and Knight, 1996).
The wetland system receiving secondary effluent for the initial two-month period
met on average the Arizona Department of Environmental Quality tertiary standards for
BOD5 and TSS of 10 mg L"'. This level was only slightly exceeded in TSS concentration at
the outlet end of the system for the period of backwash operation following vegetation
removal. In contrast, the average BOD5 outflow concentration was over the accepted value
136
of 10 mg L"' but without exceeding the reconunended level of 30 mg L"' for secondary
effluent, during both backwash operation periods. Chloride is considered highly stable in
most terrestrial environments. In wetlands, its total mass is approximately constant (Kadlec
and Knight, 1996) because its incorporation in plant tissues is negligible (Hayashi et al.,
1998). Consequently, CI' has been used as a conservative tracer to estimate
evapotranspiration in wetland ecosystems (Hayashi et al., 1998). Loses of CI' were not
evident during the initial period of secondary effluent. Probably, the increase of detention
time during backwash water operation promoted the evaporitic enrichment of CI' in the
outflowing water from the polishing wetland. In general, CI" concentrations at the wetland
outlet were lower in 2000 than in 1999 backwash operation; such a condition appears to be
induced by a lower influent concentration. The difference of CI" concentration between
both ends of the polishing wetland was 26 and 19 percent during backwash operation before
and after vegetation removal, respectively.
Sulphate is an essential nutrient for plants; thus it can be retained by plant uptake in
terrestrial environments; however, it is rarely a limiting factor for plant growth in wetlands
(Kadlec and Knight, 1996). The electron acceptor of sulflir-reducing bacteria in anaerobic
environments is SO/'(Maier, 2000) where its presence and a high organic content induces
the production of hydrogen sulfide. Probably, this microbiological process was responsible
for the reduction of SO4
in the settling basin, mostly observed during 100 percent
137
backwash water operation (Figure 7). Similar to CI", a significant increase of S04^' occurred
by an apparent evaporitic enrichment at the outflow of the wetland. Comparing both
sampling periods, vegetation removal from the system did not produce a significant change
of S04^" during backwash operation. However, SO^"' behavior was noticeably impacted by
evaporation as a consequence of the longer detention time during backwash operation
(Figure 3 and 7).
After vegetation removal, water pH was statistically higher at the outflow of the
wetland cell. Probably, vegetation removal may have favored a higher light penetration in
the shallow areas promoting algae proliferation and a higher pH in the outflow of the
wetland cell (Parhad and Rao,
1972; Kadlec and Knight, 1996).
Indicator Microorganisms
Removal efficiency over 90 percent of total and fecal coliforms in surface flow
wetlands have been commonly reported (Gersberg et al., 1989; Kadlec and Knight, 1996;
Karpiscake/a/., 1995; Perkins and Hunter, 2000; Steenera/., 1999; Vrhovseke/a/., 1996).
In this study, the high densities of both total and fecal coliforms observed at the settling
basin and oudet of the east polishing basin are unclear. Growth or recovery of injured
bacteria can occur if the amount of organic matter and temperature are elevated (Gerba,
2000). Studies conducted in soils have found evidences of coliform regrowth associated
138
with inputs of nutrients in the environment coinciding with elevated conditions of
temperature and moisture (Van Donsel et al., 1967). Coliform bacteria such as Klebsiella,
Enterobacter, and Citrobacter have shown the ability to proliferate during wastewater
treatment (Niemi et al. cited by Elmund et al, 1999). Klebsiella was found at high densities
in the outflowing water from a treatment facility receiving municipal wastewater (Elmund
et al., 1999) which was attributed partially to an increase of carbohydrates in wastewater.
In water reservoirs, natural or artificial, animal fecal material have been reported as a
possible source of high densities of total and fecal coliform bacteria (Have, 1973; Kadlec
and Knight, 1996; Moorheade/"^/., 1998). During operation ofthe east polishing subsystem,
the settling basin has shown a high ability to reduce BOD5 and TSS from wastewater. Thus,
it appears that the amount of organic matter introduced into the settling basins are playing
an important role in the multiplication or recovery of total and fecal colifroms.
At the outlet of the system, it is probable that the high densities of both groups of
coliforms are a result of a combination of factors such as: a) total and fecal coliform growth
because organic matter contributions from backwash water, b) increase of total and fecal
coliform densities by evapotranspiration because of the long detention time in the wetland
cell, and c) animal fecal contributions , Sweetwater Wetlands provide habitat for small
mammals, birds and waterfowl. All these factors could be offsetting any total and fecal
coliform removal within the wetland.
139
F-specific RNA bacteriophages have been used as potential indiccator of human
enteroviruses indicator instead of fecal coliforms and fecal streptococci becrause it has been
correlated with human viruses in a wide range of environments (Stetler, 19'84; Havelaar et
ai, 1993). Compared to enteroviruses, the former groups of bacteria hawe shown lower
populations in chlorinated effluents and higher densities in surface water exiposed to animal
fecal contamination (Havelaar et al,1993). In the current study, a moderated correlation of
coliphages compared to TSS was observed during backwash operation. The recovery of
coliphages in samples collected from the wetland outlet revealed a remova^l of 0.43 log for
the initial two-month period, 1.1 log for the second period, and 0.2 log :after vegetation
removal from the wetland. It is important to point out that the higher coliiphage removal
coincided with the highest influent concentration. When comparing removal efficiencies of
coliphage between sampling periods at each sampling location, the rermoval difference
between backwash periods was statistically significative (p<0.05) only at the outlet of the
wetland cell. Frequently, coliphage removal have been found over on«e log reduction
(Gersbergef <2/., 1987, 1989; Chendorainefa/., 1998); However, lower colaphage removals
have been found by Karpiscak et al. (1995) in a duckweed (JLemna spp.') system and by
Gersberg et al. (1989) in nonvegetated wetlands.
140
CONCLUSIONS
Even, after vegetation removal and backwash water operation, the system was able
to decrease BOD5 and TSS to secondary standards required by the Arizona Department of
Environmental Quality. However, little removal was observed in either the stream or the 3
ha surface flow wetland. Based on the results of the current study fecal and total coliform
bacteria are doubtful indicators of human fecal contamination because inputs or recovery
of both groups of coliforms are occurring in the east polishing system. Apparently, coliphage
was a better indicator during the current study although its low level of removal in the
wetland cell suggests additional inputs may be occurring from animal sources.
ACKNOWLEDGMENTS
The authors wish to acknowledge technical support from Sue Hopf and Glenn
France from the University of Arizona's Office of Arid Land Studies. Financial support
was provided by Tucson Water, City of Tucson Arizona; Sanitation Districts of Los
Angeles County; United States Environmental Protection Agency; American Water
Works Association Research Foundation; USDA Water Conservation Laboratory; Pima
County Department of Wastewater Management; Water Environment Research
Foundation; City of Phoenix, Arizona; The Subregional Operators Group (Phoenix-area
141
cities); Water Replenishment District of Southem California; City of Riverside,
California; and City of Los Angeles Department of Water and Power.
DISCLAIMER
The American Water Works Association Research Foundation and the other
agencies listed above not had opportunity to review and comment on this paper,
therefore, none of these agencies necessarily endorse the findings presented here.
142
REFERENCES
Adams, M. H. (1959) Bacteriophage. Interscience Publishers, Inc., New York.
Chendorain M., Yates M. and Villegas F. (1998) The fate and transport of viruses through
surface water constructed wetlands, y. ^/zvzron. OuaL, 27, 1451-1458.
Elmund G. K., Allen M. J. and Rice E. W. (1999) Comparison of Escherichia coli, total
coliform populations as indicators of wastewater treatment efficiency. Water
Environ. Res., 71, 332-339.
Geller G. (1997) Horizontal subsurface flow system in the german speaking countries:
summary of long-term scientific and practical experiences; recommendations. Wat.
Sci. Tech., 35, 157-166.
Gerba C.P. (2000) Indicator Microorganisms. In Environment/il Microbiology (Edited by
Maier R.M., Pepper I.L. and Gerba C.P.). Academic Press, Canada.
Gersberg R.M., Gearheart, R.A. and Ives M. (1989) Pathogen Removal in Constructed
Wetlands In Constructed Wetlands for Wastewater Treatment (Edited by Hammer
D.A.). Lewis Publishers, Michigan.
Gersberg R.M., Lyon S. R., Brenner R. and Elkins B. V. (1987^ Fate of viruses in artificial
wetlands. Appl. Environ. Microbiol., 53, 731-736.
Green M.B., Griffin P., Seabridge J.K. and Dhobie. (1997) Removal of bacteria in
subsurface flow wetlands. Wat. Sci. Tech., 35, 109-116.
143
Have M.R. (1973) Effects of migratory waterfowl on water quality at the Montezume
National Wildlife Refuge, Seneca county. New York. Jour. Research U.S. Geol.
Survey, 1, 725-734.
Havelaar, A. H., Olphen M. and Drost Y. C. (1993) F-specific RNA bacteriophages are
adequate model organisms for enteric viruses in fresh water. Appl. Environ.
Microbiol., 59, 2956-2962.
Hayashi M., Kamp G. and Rudolph D. L. (1998) Water and solute transfer between a prairie
wetland and adjacent uplands, 2. Chloride cycle. J. Hydrol., 207, 56-67.
Kadlec R.H. and Knight R.L. (1996) Treatment Wetlands. Lewis Publishers, New York.
Karpiscak M.M., Gerba C. P., Watt P. M., Foster K. E. and Falabi J. A. (1995) Multi-species
plant systems for wastewater quality improvements and habitat enhancement. In
Second International Symposium on Wastewater Reclamation and Reuse ( Edited by
Angelakis, A., Asano T., Diamadopoulos E., and Techobanoglous G.). lAWQ,
Iraklio, Greece, pp. Z1-A2.
Maier R. M. (2000) Biogeochemical cycling. In Environmental Microbiology (Edited by
Maier R.M., Pepper I.L. and Gerba C.P.). Academic Press, Canada.
Mashauri D.A., Mulungu D. M. M. and Abdulhussein B. S. (2000) Constructed wetland at
the University of Dar Es Salaam. Wat. Res., 34, 1135-1144.
144
Moorhead D.L., Davis W. S. and Wolf C. F. (1998) Coliform densities in urban water of
West Texas. Environmental Health, 14-18.
Parhad N. M. and Rao N. V. (1972) The effect of algal growth on the survival of E coli in
sewage. Indian J. Environ. Hlth., 14, 131-139.
Perkins J., and Hunter C. (2000) Removal of enteric bacteria in a surface flow constructed
wetland in Yorkshire England. Wat. Res., 34, 1941-1947.
Scheuerman P.R., Bitton G. and Farrah S.R. (1989) Fate of Microbial Indicators and
Vinises in a Forested Wetland. Lewis Publishers, Michigan.
Steen P.V., Brenner A., Van Buuren, J. and OronG. (1999) Post-treatment of UASB reactor
effluent in an integrated duckweed and satabilization pond system. Wat. Res., 33,
615-620.
StetlerR. E. (1984) Coliphages as indicators of enteroviruses. Appl. Environ. Microbiol., 48,
668-670.
Tanner C.C., Sukias P.S. and Upsdell M.P. (1998) Relationships between loading rates and
pollutant removal during maturation of gravel-bed constructed wetlands. /. Environ.
Oual, 27, 448-458.
Van Donsel D.J., Geldreich E.E. and Clark, N.A. (1967) Seasonal variations in survival of
indicator bacteria in soil and their contribution to storm-water pollution. Appl.
Microbiol., 15, 1362-1370.
145
Vidales J. A., Gerba C. P. and Karpiscak M. M. (2001) Relative transport of Br- and PRDl
in a surface flow constructed wetland. In process for publication.
Vinluan E.A. (1996) Survival of microbial indicators in constructed wetlands. Soil, Water,
and Environmental Science. University of Arizona, Tucson, AZ., p. 65.
VrhovsekD., Kukanja V. and Bulc T. (1996) Constructed wetland for industrial waste water
treatment. Wat. Res., 30, 2287-2292.
146
Table 1. Average temperature and pH observed in the samples collected from the east
polishing system during the secondary effluent operation ( February and March 1999)
and the 1999 and 2000 backwash periods.
a)Secondary effluent
Parameter
Sampling site
Splitter
Settling
box
basin
Temperature (°C)
22.75
20.25
19
18
10.5
pH
7.8
7.3
7.5
7.8
7.6
Stream out
Outlet
Stream in
Stream out
Wetland
outlet
b)Back\vash water
Sampling site
Parameter
Splitter box
1999
2000
Settling basin
1999
2000
Stream in
1999
2000
1999
2000
199
2000
9
Temperature
27.7
29.3
25.4
25.6
24.6
25.9
24.2
25
19.8
23.4
7.42
7.39
7.5
7.3
7.4
7.4
7.5
7.4
7.3
7.7
CO
pH
147
Innucnt
Settling basin
Stream
Splitter
box
West polishing
subsystem
Stream out
Stream in
Outlet
Figure 1. Schematic representation of the east polishing subsystem and location of the
sampling sites (•).
148
40 -i
D
I
Q
30
2
jBk.
3
I•o
c
TSS (mg L")
BODsCnngL')
Turbidity (NTU)
ra
CO
CO
Q
O
CO
20 T
i
I
'
10 i
Splitter box
Settling basin
Stream in
Stream out
Wetland outlet
Figure 2. Average concentration of TSS, BOD5, and Turbidity in the east polishing
system during secondary effluent operation (February and March 1999).
149
200 -T
•
•
cr
1
160 -1
O)
£
c
_o
120 T
'::r- •
i.'i ^
CO
•
0}
o
c
o
o
80
r"*
40
1
*:? r
•'rv
r» > •_
''•z'.r .
m
'm
%
-•
'.' .-•<
Splitter box
Settling basin
Stream in
Stream out
Wetland outlet
Figure 3. Average concentrations of CI" and SO4 from sampling sites in the east
polishing system during secondary effluent operation (February and March 1999).
150
5 -
I
I Total colifonTE{cfu /100 rri)
KS Fecal colifOTTTE (cfii l^00n^)
I
Splitter box
Settling basin
Stream in
I Coliphage (pfu /100 cri)
Stream out
Outlet
Figure 4. Microbial indicator populations in the east polishing subsystem during
secondary effluent operation (February and March 1999).
151
500
~
IS
3
H
^
400
TSS (mg L-')
• BODjCmgL")
O
Turbidity (MTU)
300
lyj
H
fi 200
O
S3
A
A
100
o <5
0
Splitter
box
Settling
basin
Stream
in
9^
.a ?
Stream
out
Wetland
outlet
Figure 5. Statistical representation of BODj, TSS, and Turbidity observed in tlie east
polishing system during secondary effluent operation (February and March 1999)
period of sampling. The symbol represents the mean concentration, error bars the 25th
and 75th percentile, and the dot (•) the maximum observation.
152
250
^200
r1
CJD
S150
u
100
A
•
i
t
I
I
A
a
iI
I
G
i
50
A
Before vegetation removal
O
After ve{;ctation removal
0
Splitter
box
Settling
basin
Stream
in
Stream
out
Wetland
outlet
Figure 6. Concentration of CI" in the east polishing system during backwasli operation
before (April through September 1999) and after (February through Septemxber 2000)
vegetation removal. The symbol represent the mean concentration, error ! bars the 25
th and 75th percentile, and the dot (•) the maximum observation.
153
200
150
M H ti
•J
Ci}
O
cn
100
50
^
Before vegetalion removal
Q
After vegetaCion removal
0
Splitter
box
Figure 7.
Settling
basin
Stream
in
Stream
out
Wetland
outlet
Statistical representation of SO4 in the east polishing system during
backwash operation before (April through September 1999) and after (February
through September 2000) vegetation removal. The symbol represents the mean
concentration, error bars the 25 th and 75 th percentile, and the dot (•) the maximum
observation.
154
6
a.
Q
Berore vegeiacion removal
D
After vegetation removal
-n
i
ex)
o
3
-
n
I
-
90
2
Splitter
box
Settling
basin
Stream Stream
in
out
Wetland
outlet
Figure 8. Statistical representation of total coliform concentration observed during
backwash operation before (April through September 1999) and after (February
through September 2000) vegetation removal. In the box plot, the horizontal line shows
the median, the box ends indicate the 25"' and 75"* percentile, error bars the lO"* and
90"* percentiles, and the dots (•) extreme values.
155
I I
Before vegetation removal
I i
After vegetation removal
c
o
r
CJ
OCD
Y
5V I
Splitter
box
Settling
basin
I
T
Stream Stream
in
out
Wetland
outlet
Figure 9. Fecal coliform concentrations in the east polishing system observed during
backwash operation before (April through September 1999) and after (February
through September 2000) vegetation removal. In the box plot, the horizontal line shows
the median, the box ends Indicate the 25"* and 75"* percentile, error bars the 10"* and
90"* percentiles, and the dots (•) extreme values.
156
I
[
Before vcgetaiion removal
iI
Arcer vegetation remo\-al
T
C.
acu
OC£
a
AO
1
Splitter
box
Settling
basin
Stream Stream
in
out
Wetland
outlet
Figure 10. Coliphage concentrations in the east polishing system during backwash
operation before (April through September 1999) and after (February through
September 2000) vegetation removal. In the box plot, the horizontal line shows the
median, the box ends indicate the ZS"* and 75"* percentile, error bars the lO"* and 90"*
percentiles, and the dots (•) extreme values.
157
500
A
f 400
1 300
c/f
TSS (mgL-')
•
BOD, (mg L-')
O
Turbidity (NTU)
•
•
§200
•
m
•
100
G?
0
A
Splitter
box
Settling
basin
A
Stream
in
,6
A
Stream
out
^
Wetland
outlet
Figure 11. Statistical representation of BODg, TSS, and Turbidity observed in the east
polishing system during backwash operation before (April through September 1999)
and after (February through September 2000) vegetation removal. The symbol
represents the mean concentration, error bars the 25 th and 75th percentile, and the
dot (•) the maximum observation.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement