United States Office of Research EPA/600/R-06/102
Environmental Protection and Development September, 2006
Agency Washington DC 20460
Performance of
Stormwater Retention
Ponds and Constructed
Wetlands in Reducing
Microbial Concentrations
^PROt*-0
r U.S. Environmental Protection Agency
I Office of Research and Development
^ National Risk Management Research Laboratory
-------
THIS PAGE INTENTIONALLY LEFT BLANK
-------
EPA/600/R-06/102
September, 2006
Performance of Stormwater Retention Ponds
and Constructed Wetlands in Reducing
Microbial Concentrations
By
Scott D. Struck
Ariamalar Selvakumar
Michael Borst
Urban Watershed Management Branch
Water Supply and Water Resources Division
National Risk Management Research Laboratory
Edison, NJ 08837
NATIONAL RISK MANAGEMENT RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OH 45268
-------
Notice
The U.S. Environmental Protection Agency (EPA) through its Office of Research and Development
performed and managed the research described in this report. It has been subjected to the Agency's peer
and administrative review and has been approved for publication as an EPA document. Any opinions
expressed in this report are those of the author and do not, necessarily, reflect the official positions and
policies of the EPA. Any mention of products or trade names does not constitute recommendation for use
by the EPA.
-------
Abstract
Storm water runoff can transport high concentrations of pathogens to receiving waters. Bacteria indicator
organisms, as surrogates for pathogens, are the most often reported cause of receiving water impairments.
Stormwater best management practices (BMPs) are often considered effective tools to mitigate the effects
of stormwater pollutants before they appear in receiving waters. However, BMP performance for pathogen
removal is not well documented. Many questions remain on the transport and fate of indicator bacteria that
enter and exit stormwater BMPs.
The National Risk Management Research Laboratory (NRMRL), part of U.S. EPA's Office of Research and
Development (ORD) investigated the fate of indicator organisms in the stormwater runoff entering and
exiting two commonly used BMPs, constructed wetlands and retention ponds. This research used
controlled-condition, pilot-scale systems that represent larger field-scale systems to determine the dominant
mechanisms that influence the reduction of indicator organism concentrations. The pilot-scale work was
supported by bench-scale laboratory experiments investigating the effects of single parameters such as
temperature, sunlight, and salinity on indicator organism inactivation rates. Presented in this report are the
results of developing techniques for creating bacterially enriched stormwater, bench-scale studies, and the
pilot-scale BMP research. Bench-scale study results show that the temperature and sunlight affect the
inactivation rates significantly. Results from the pilot-scale research suggest that constructed wetlands and
retention ponds lower microbial concentrations in stormwater runoff. Bacteria inactivation generally
followed the first-order, K-C* empirical model that acknowledges an irreducible concentration. Factors
such as sunlight and temperature provide much of the inactivation in indicator bacteria, but other factors
(e.g., predation, sedimentation, filtration, sorption, pH, and BOD) appear to also influence indicator bacteria
concentrations. Future research validating results of the pilot-scale systems to field-scale systems should be
done.
Developing microbial inactivation models to predict effluent concentrations from BMPs will help reduce
the uncertainty and improve the capabilities of surface water quality models. First-order models that do not
consider background concentrations or resuspension, may underestimate actual bacterial concentrations.
in
-------
Foreword
The U.S. Environmental Protection Agency (EPA) is charged by Congress with protecting the Nation's
land, air, and water resources. Under a mandate of national environmental laws, the Agency strives to
formulate and implement actions leading to a compatible balance between human activities and the ability
of natural systems to support and nurture life. To meet this mandate, EPA's research program is providing
data and technical support for solving environmental problems today and building a science knowledge
base necessary to manage our ecological resources wisely, understand how pollutants affect our health, and
prevent or reduce environmental risks in the future.
The National Risk Management Research Laboratory (NRMRL) is the Agency's center for investigation of
technological and management approaches for preventing and reducing risks from pollution that threaten
human health and the environment. The focus of the Laboratory's research program is on methods and
their cost-effectiveness for prevention and control of pollution to air, land, water, and subsurface resources;
protection of water quality in public water systems; remediation of contaminated sites, sediments and
ground water; prevention and control of indoor air pollution; and restoration of ecosystems. NRMRL
collaborates with both public and private sector partners to foster technologies that reduce the cost of
compliance and to anticipate emerging problems. NRMRL's research provides solutions to environmental
problems by: developing and promoting technologies that protect and improve the environment; advancing
scientific and engineering information to support regulatory and policy decisions; and providing the
technical support and information transfer to ensure implementation of environmental regulations and
strategies at the national, state, and community levels.
This publication has been produced as part of the Laboratory's strategic long-term research plan. It is
published and made available by EPA's Office of Research and Development to assist the user community
and to link researchers with their clients.
Sally C. Gutierrez, Director
National Risk Management Research Laboratory
IV
-------
Contents
Notice ii
Abstract iii
Foreword iv
Contents v
List of Figures vii
List of Tables viii
Acronyms and Abbreviations ix
Acknowledgements x
Executive Summary xi
Chapter 1 Introduction 1-1
Urbanization 1-1
Indicator Bacteria 1-2
Stormwater BMPs 1-4
Pollutant Attenuation 1-5
Chapter 2 Factors Affecting Microbial Indicator Concentrations 2-1
Temperature 2-1
Sunlight 2-2
Physical Processes (Sedimentation, Sorption, and Filtration) 2-3
Salinity 2-4
Predation 2-4
Other Potential Factors 2-5
Summary 2-5
Chapter 3 Bench-Scale Research 3-1
Purpose 3-1
Inactivation Rate Models 3-1
Temperature 3-2
Salinity 3-2
Light 3-3
Materials and Methods 3-3
Sample Collection 3-3
Experimental Methods 3-4
Temperature Study 3-5
Salinity Study 3-5
Light Study 3-5
Analysis of Indicator Organisms 3-6
Data Analysis/Reduction 3-7
Results 3-7
Time and Temperature 3-7
Light 3-9
-------
Salinity 3-11
Summary 3-12
Chapter 4 Pilot-Scale Research 4-1
Study Site and Experimental Design 4-1
Creating Bacterially Loaded Stormwater 4-2
Storm Event Simulation 4-3
Water Quality Monitoring, Solids, Light and Bacterial Indicator Sampling 4-3
Sediment Sampling Procedures 4-5
Predation Sampling Procedures 4-5
Data Management 4-6
Statistical Analyses 4-6
Results 4-6
Scaling Consideration 4-16
Summary 4-16
Chapter 5 Discussion of Results and Conclusions 5-1
Effects of Temperature 5-1
Effect of Sunlight/Light Intensity 5-2
Effects of Sedimentation, Sorption, and Filtration 5-2
Effect of Salinity 5-3
Effect of Predation 5-3
Effect of Other Potential Factors 5-4
Inactivation Rates Due to Collective Environmental Factors 5-4
Evaluation of the First-Order Decay Equation 5-5
Conclusions 5-9
Chapter 6 References 6-1
Appendix A Growth of Indicator Bacteria for Pilot-Scale Research 1
Introduction 1
Methods 1
Results and Discussion 2
Conclusion 3
VI
-------
List of Figures
Figure 1-1. Number of reported surface water quality impairments (top 7) since January 1,
1996 1-3
Figure 1-2. Number of approved TMDLs by pollutant (Top 7) since January 1,1996 1-4
Figure 3-1. Drainage area (A) and outfall location (B) of the study in Edison, New Jersey 3-4
Figure 4-1. Pictures of the retention pond (A) constructed wetland (B) treatment systems 4-2
Figure 4-2. Mean turbidity in the retention pond and constructed wetland in all storm events
except October 2005. Whiskers indicate 95% confidence intervals 4-7
Figure 4-3. Effluent concentration of indicator organisms with conductivity and dissolved
oxygen 4-11
Figure 4-4. Effluent concentrations of indicator organisms with ORP and pH 4-12
Figure 4-5. Effluent concentrations of indicator organisms with temperature 4-13
Figure 4-6. Effluent indicator bacteria concentrations with in-situ turbidity 4-13
Figure A-l. Mean indicator bacteria concentrations by day. Vertical bars denote 95%
confidence intervals A-2
vn
-------
List of Tables
Table 1 -1. Range of Concentrations of Indicator Organisms Found in Varying Waterbodies 1-3
Table 2-1. Some Reported Die-off Rates (K) for Indicator Organisms 2-2
Table 3-1. Light Intensities Corresponding to the Height of Light Source above the Water Surface 3-6
Table 3-2. Summary Table of Analytical Procedures 3-6
Table 3-3. Inactivation Rate Constants for Each Indicator Organism at Tested Temperatures in the
Isothermal Experiments 3-8
Table 3-4. Organism-Specific Reference Temperature Rate Constant and Temperature Coefficient
Determined Using Salt-Free Dark Experimental Results 3-9
Table 3-5. Regression-Estimated Values of Inactivation Rate Constants at 26.17°C for Experimental Light
Intensities 3-9
Table 3-6. Regression-Estimated Coefficients from Light Experiment 3-10
Table 3-7. Regression-Estimated Values of Inactivation Rate Constants Determined for Salinity
Concentrations 3-11
Table 3-8. Regression-Estimated Coefficients from Salinity Experiment 3-12
Table 4-1. Target Indicator Organism Densities in Mesocosms after Addition of Enriched Stormwater... 4-2
Table 4-2. Effluent Time of Hand Collected and Programmed Autosample Collection 4-4
Table 4-3. Expected Beginning Densities After Loading and Expected Dilution Factors 4-5
Table 4-4. Average Event in-situ Physical and Chemical Results 4-8
Table 4-5. Inactivation Rates for the Constructed Wetland, Retention Pond, Dark and Light Controls for all
Indicator Bacteria Organisms for each Sampling Event 4-10
Table 4-6. In-situ Indicator Organisms Average Background Concentrations 4-15
Table 4-7. Sediment Bacteria Indicator Organisms Sampled in November of 2004 4-15
Table 4-7. Macroinvertebrate Groups Identified in the Retention Pond and Constructed Wetland 4-16
Table 5-1. Inactivation Rate Coefficients from Batch and Field Studies 5-7
Table 5-2. Retention Pond Overall, Temperature, Sunlight, and Other Factors Rate Coefficients 5-7
Table 5-3. Constructed Wetland Overall, Temperature, Sunlight, and Other Factors Rate Coefficients.... 5-8
Table A-1. Growth Rates of Bacterial Indicators Based on Regression Analyses A-3
Vlll
-------
Acronyms and Abbreviations
ANOVA Analysis of Variance
APHA American Public Health Association
ASCE American Society of Civil Engineers
ASTM American Standard Testing Methods
BHI Brain Heart Infusion
BMP Best Management Practice
BOD Biochemical Oxygen Demand
CPU Coliform Forming Unit
CWA Clean Water Act
DNA Deoxyribonucleic Acid
DO Dissolved Oxygen
DQI Data Quality Indicator
EC Escherichia coli
EMC Event Mean Concentration
EN or ENT Enterococci
EPA US Environmental Protection Agency
FC Fecal Coliforms
FR Federal Register
FS Fecal Streptococci
HOPE High Density Polyethylene
MDE Maryland Department of the Environment
MDL Method Detection Limit
MF Membrane Filtration
MPN Most Probable Number
MS4 Municipal Separate Storm Sewer System
MWLAP Ministry of Water, Land and Air Protection
NPDES National Pollutant Discharge Elimination System
NRMRL National Risk Management Research Laboratory
NSQD National Stormwater Quality Database
NTU Nephelometric Turbidity Unit
ORISE Oak Ridge Institute for Science and Education
ORP Oxygen Reduction Potential
ppt Parts Per Thousand
PVC Poly Vinyl Chloride
RPD Relative Percent Difference
RSD Relative Standard Deviation
SM Standard Methods
SS Suspended Solids
TC Total Coliforms
TMDL Total Maximum Daily Loads
IX
-------
TNTC Too Numerous To Count
TOC Total Organic Carbon
TSS Total Suspended Solids
QA/QC Quality Assurance/Quality Control
USEPA US Environmental Protection Agency
USGS United States Geological Survey
UV Ultraviolet
UWMB Urban Watershed Management Branch
UWRF Urban Watershed Research Facility
WEF Water Environment Federation
WQS Water Quality Standards
YSI Yellow Springs Instruments
-------
Acknowledgements
An undertaking of this type requires the dedication and cooperation of many individuals. The technical
direction and coordination for this project was provided by the technical project team of the Urban
Watershed Management Branch. Many members of the Branch assisted in making this product available to
the public. Special recognition is extended to these members of the Branch:
Anthony N. Tafuri, P.E., Branch Chief
Thomas P. O'Connor, Environmental Engineer
Mary Stinson, Chemical Engineer
Dr. Christopher Nietch, currently with NRMRL, WSWRD, Cincinnati, OH
Carolyn Esposito, Quality Assurance Officer
Thomas P. O' Connor and Mary Stinson performed reviews of this report. Carolyn Esposito reviewed the
quality assurance project plan and this report. Mano Sivaganesan of Water Supply and Water Resources
Division, Cincinnati reviewed the statistical part of this report.
Most monitoring, sampling and laboratory analysis was performed by personnel from US Infrastructure,
Inc., the operating contractor of the Urban Watershed Research Facility under EPA contract # EP-C-04-064.
Most notably, Christa Casciolini, Yogesh Parikh, Toby Swetje, and Clarence Smith should be recognized
for their contribution to the completion of the sampling and analyses.
XI
-------
Executive Summary
USEPA's 305(b) water quality reports consistently show stormwater runoff as a leading cause of water
quality impairment in the United States. Water quality standards (WQS) have been developed through
states and the federal government to improve the condition of our nation's surface and groundwater.
When waters of the United States do not meet the WQS, regulations have been put in place to overcome
impairments. To improve or prevent further degradation of water quality, regulators rely on best
practicable control technologies currently available to reduce the loading of stressors from point sources
(and, at times, non-point sources). Microorganisms are a high priority stressor because of the many
waterbodies that are listed as impaired.
This report documents the efforts to evaluate simple predictive relationships affecting concentrations of
indicator organisms in stormwater runoff based on environmental conditions. The report begins by
describing the breadth of surface water resources affected by bacterial stressors to identify needs for
continued research on understanding the engineering approaches for management of point- and non-point
sources associated with this stressor. In Chapter 2 factors from the scientific literature that influence
bacteria indicator concentrations are reviewed. The subsequent chapters describe bench- and pilot-scale
experiments that attempt to determine the dominant factors that favor a reduction in indicator bacteria
concentrations. Finally, Chapter 5 contains synopses of the results of these experiments and an
assessment of the first-order decay formula's ability to predict bacterial concentrations based on influent
concentrations and other environmental factors. Inactivation rates for each bacteria indicator from these
experiments and coefficients are given.
Combined in this report are bench- and pilot-scale data to assess first-order equations to better predict the
performance of constructed wetland and retention pond best management practices (BMPs). By
measuring varying physical, chemical, and biological parameters that may influence effluent indicator
organism concentrations and characteristics of other stormwater parameters that are often contained in
stormwater runoff factors that most influence inactivation rates were determined. The BMPs used in this
research are small-scale, controlled systems (termed mesocosms) that offer a unique environment for
investigating many parameters that can affect the reduction of indicator organisms.
Detailed in Appendix A is the development of methods to grow and harvest bacteria indicators to provide
an enriched source to increase the concentrations in stormwater for wet weather flow research. The
mesocosm research necessitated this ancillary research as a technique to establish the desired influent
bacteria concentrations. The technique may prove useful to others undertaking similar research.
xn
-------
Chapter 1 Introduction
The Clean Water Act (CWA) of 1972 developed an ambient water quality management program to
measure the condition of a waterbody and determine whether that waterbody meets the criteria associated
with the designated use. By definition, this process depends on setting appropriate water quality
standards (WQSs). Realistic standard setting must balance watershed conditions (hydrologic, ecological,
and land use) against the corresponding need to protect human health, infrastructure, and the
environment.
Where waters of the nation are not meeting established WQSs after implementing best practicable control
technologies currently available to reduce the loading of stressors from point sources or other pollution
control programs, the CWA requires establishing a Total Maximum Daily Load (TMDL) for each
pollutant of concern. As part of the 1987 amendments to the CWA, Congress added Section 402(p) to
cover point-source discharges composed entirely of stormwater. Section 402(p)(2) of the Act requires
permit coverage for discharges associated with industrial activity and discharges from large (over 250,000
people) and medium (between 100,000 and 250,000 people) municipal separate storm sewer systems
(MS4s). These discharges are referred to as Phase IMS4 discharges.
USEPA issued regulations on December 8, 1999 (64 FR 68722), expanding the National Pollutant
Discharge Elimination System (NPDES) stormwater program to include discharges from smaller MS4s
(including all systems within "urbanized areas" and other systems serving populations less than 100,000)
and stormwater discharges from construction sites that disturb one to five acres, with opportunities for
area-specific exclusions. This program expansion is referred to as Phase II.
Urbanization
USEPA used urbanized areas and population to define program boundaries because of the increased risk
to human health in these areas through greater potential for exposure from point and non-point sources
and risks associated with a greater population density. Another component of the selected boundaries is
that urbanization characteristically results in a larger percentage of impervious areas that lead to larger
quantities of stormwater runoff that contribute significant amounts of debris and other pollutants (e.g.,
litter, oils, microorganisms, sediments, nutrients, organic matter, and heavy metals) to receiving waters.
USEPA identified urban stormwater runoff as one of the four leading causes of water quality impairment
related to human activities in lakes and reservoirs (USEPA, 2002). Poor water quality, especially
pathogen contaminated water, can cause illnesses such as gastroenteritis (characterized by vomiting,
diarrhea, abdominal pain or fever) or upper respiratory (ear, nose, and throat) infections to exposed
swimmers. Highly polluted water can occasionally cause serious diseases such as typhoid fever,
1-1
-------
dysentery, hepatitis, and cholera. An epidemiological study conducted in the Santa Monica Bay adjacent
to Los Angeles County, CA found higher risks of a broad range of symptoms, including upper respiratory
and gastrointestinal cases for people swimming closer to storm drains, implicating stormwater runoff as
the source of the illnesses (Haile et al., 1999). Similarly, the Southern California Coastal Water Research
Project showed that more than half of Southern California's shoreline (from Santa Barbara to San Diego)
is unsafe for swimming after rainstorms because of bacteria carried to the ocean by urban runoff (Noble et
al, 2000).
Indicator Bacteria
To protect public health, surface waters are tested for indicators that serve as a proxy for harmful
pathogens. Indicator bacteria are used because it is difficult to measure the pathogens themselves.
Indicator bacteria organisms tested by public health agencies include fecal indicator bacteria such as total
coliforms, fecal coliforms, fecal streptococci, Escherichia coll (E. coll), and enterococci. The
concentrations of these indicators are used to determine the potential for fecal contamination and to
compare to public health-based thresholds. Like the pathogens they represent, fecal indicator bacteria are
found in feces from both human sources (e.g. sewer discharges, and failing septic systems) and non-
human sources (e.g. pets, waterfowl, and farm animals). Historically, total and fecal coliforms with fecal
streptococci have served as the preferred indicators, but there are efforts to substitute enterococci and E.
coll for water quality monitoring because of higher correlation with gastrointestinal illness (Gray, 2000).
E. coll and enterococci are more representative of warm blooded animal fecal contamination in water than
total or fecal coliforms. They both can survive, but generally not grow outside the intestinal tract
(Ashbolt et al, 2001). In 1976, the USEPA recommended that states adopt a bathing WQS of fecal
coliforms not to exceed 200 organisms/100 mL (USEPA, 1976). In 1986, based on the higher correlation,
the USEPA recommended that states revise the recreational water quality microbial criteria to use
enterococci for marine waters and E. coll or enterococci for freshwaters. Suggested criteria are 35
enterococci per 100 mL for marine waters and 33 enterococci per 100 mL and 126 E. coll per 100 mL for
freshwaters (USEPA, 1986). If a single sample exceeds 235 E. coll per 100 mL in freshwater and 104
enterococci per 100 mL in saltwater, the USEPA recommends that a swimming area be closed, or posted
until levels are lower. Several states have established policies that advisories are posted at more
protective levels of indicator bacteria. Although EPA advised the states of the benefits of changing the
indicators, many states continue to use the traditional indicators for a variety of reasons.
Some have questioned the relationship between indicator bacteria and human health risks associated with
pathogen exposure in surface water. Few epidemiological studies have tested the health effects of
exposure to waters receiving direct and recent stormwater runoff. However, Wade et al. (2003)
quantitatively analyzed many studies conducted during the past 50 years. The studies, when analyzed
collectively, provide a weight of evidence showing the regulations suggested by the USEPA for the
enterococci rule (for marine waters) and E. coll rule (for freshwaters) are protective when considering the
risks associated with recreational water contact including swimming.
Stormwater or stormwater-influenced receiving waterbodies can have indicator bacteria concentrations
that greatly exceed WQS. Because elevated fecal indicator bacteria are often associated with stormwater
runoff, some state and local agencies close swimming areas preemptively whenever rainfall exceeds a set
amount based on site-specific studies. In a recent Water Environment Federation (WEF) report (WEF,
2006), the authors summarized 51 studies from around the world that found ranges in concentrations of
total coliforms, fecal coliforms, E. coll, fecal streptococci, and enterococci as reported in Table 1-1.
Sources of these samples included coastal waters, rivers, creeks, drainage canals, and wetlands.
1-2
-------
Table 1-1. Range of Concentrations of Indicator Organisms Found in Varying Waterbodies
Concentration Range
Indicator Organism (per 100 mL)
Total Coliforms
Fecal Coliforms
Eschericia coli (E. coli)
Fecal Streptococci
Enterococci
Source: WEF, 2006; Maestre and Pitt,
IxlO1
IxlO1
IxlO2
IxlO1
IxlO1
2005
-2xl06
- 8 x 106
-2xl06
-2xl04
- 8 x 104
Median
(per 100 mL)
1.2 xlO4
5. IxlO3
1.7 xlO3
1.7 xlO4
When looking at impaired waterbodies in the United States, Maestre and Pitt (2005) using data from the
National Stormwater Quality Database, reported nationwide median concentrations for fecal coliforms,
fecal streptococci, total coliforms and E. coli that are found in Table 1-1. On the US EPA's 303(d) list of
impaired waters, pathogen contamination is the most commonly listed source (7,742 reported
impairments, Figure 1-1) and second for the number of approved TMDL allocations (2,608 TMDLs,
Figure 1-2), indicating the prevalence of this stressor (USEPA, 2004). Of the 2,608 approved TMDL
allocations, more than 84% are for fecal coliforms.
Number of Impairments Reported
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
PATHOGENS
SEDIMENT SILTATIOII
ORGANIC ENRICHMENT LOW DO
FISH CONSUMPTION ADUIS.
PH
Figure 1-1. Number of reported surface water quality impairments (top 7) since January 1,1996.
1-3
-------
Number of TMDLs Approved
METALS
PATHOGENS
NUTRIENTS
SEDIMENT SILTATION
ORGANIC ENRICHMENT LOW DO
UNIONIZED AMMONIA
THERMAL MODIFICATIONS
] 500 1000 1500 2000 2500 3000
i
1
Figure 1-2. Number of approved TMDLs by pollutant (Top 7) since January 1,1996.
Researchers have correlated aquatic microorganism densities with terrestrial watershed factors such as
land use, density of housing, population, development, percent impervious area, and domestic animal
density (Young and Thackston, 1999; Mallin, 1998; Glenne, 1984; Francy et a/., 2000; Selvakumar and
Borst, 2006). Surface runoff samples from more densely populated, sewered areas generally showed
higher bacterial counts than runoff from less developed areas serviced by septic tanks (Young and
Thackston, 1999). Selvakumar and Borst (2006) found microorganism concentrations from high-density
residential areas were higher than those associated with low-density residential and landscaped
commercial areas.
Stormwater BMPs
A stormwater best management practice (BMP) is a technique, measure, or structural control that is used
to manage the quantity and improve the quality of stormwater runoff in the most cost-effective manner.
The USEPA (1999) defines BMPs as "schedules of activities, prohibitions of practices, maintenance
procedures, and other management practices to prevent or reduce the pollution of waters of the United
States." There are two general types of BMPs used to reduce the threat of stormwater runoff pollution
from urbanizing areas: (i) nonstructural or source control BMPs; and (ii) structural or treatment BMPs
(USEPA, 1993).
Nonstructural BMPs refer to those stormwater runoff management techniques that use natural measures to
reduce pollution levels, do not require extensive construction efforts, and either limit the generation of
stormwater runoff, or reduce the amounts of pollutants contained in the runoff. They do not involve
fixed, permanent facilities and they usually work by changing behavior through government regulation
(e.g., planning and environmental laws), persuasion, and economic instruments (Taylor and Wong, 2002).
These BMPs include institutional, educational, or pollution prevention practices.
Structural BMPs are engineered systems and methods designed to provide temporary storage and
treatment of stormwater runoff for the removal of pollutants (MWLAP, 1992; MDE, 2000; Clar et al,
2003). Structural BMPs improve the quality; control the quantity of stormwater runoff or both. The
1-4
-------
USEPA recommends installing stormwater BMPs within the landscape to influence runoff rates and
reduce stressor levels in the stormwater runoff before it reaches the receiving water.
Stormwater runoff is not identical in each location. The character of the drainage area strongly influences
not only the runoff volume and rate from a given rain event but also the stressor concentrations.
Regardless of the landscape where the BMP is installed, the same potential processes occur within the
structure to mitigate the stressors and flow. These processes control the effluent rates and stressor levels
regardless of the designated use of the receiving water. Fundamentally, a constructed wetland or
retention pond of given characteristics attenuates the stressor load runoff regardless of land use or the
receiving waters. The BMP's capabilities are established by their design, construction, and maintenance,
and not whether the device installed is part of the source water protection strategy or is a means to protect
recreational water, e.g., as part of a TMDL strategy.
If it is determined that a BMP approach (including an iterative BMP approach or treatment train) is
appropriate to meet the stormwater component of a TMDL, USEPA recommends that the regulatory
language within the TMDL reflect this. Reductions in concentrations in effluents reaching the
recreational waters depend on BMP performance. To estimate the reduction in stormwater pollutant
concentrations passing through BMPs for developing TMDL allocations, the performance of each BMP
must be well established. Much of the existing information on BMP performance comes from current
literature and the American Society of Civil Engineers (ASCE) International BMP Database
(www.bmpdatabase.org). This database, although one of the largest collections of data on BMP
performance, has a paucity of information to adequately assess the performance of many stormwater
BMPs (Andrews et a/., 2004).
Pollutant Attenuation
BMPs are generally passive tools that use physical, chemical, and biological processes that promote
natural microbiological inactivation to reduce this and other stressor concentrations in the effluent. These
systems are not a means of chemical disinfection as in wastewater treatment. Determining the dominant
mechanisms of stormwater bacterial and pathogen removal by these devices is an important step in
predicting trends in effluent concentrations to meet state and federal WQSs and for developing TMDLs.
Few quantitative studies have been carried out to determine the relative importance of various removal
mechanisms by constructed wetlands and retention ponds for indicator bacteria, consequently the ability
of these BMP treatments to reduce concentrations in stormwater runoff before reaching receiving waters
where WQS must be met, are poorly understood.
Two commonly-used structural BMPs for controlling pollutants in stormwater are constructed wetlands
and retention ponds. Treatment and therefore design within these two systems rely predominately on
slowing water transport time that provides increased settling. Other environmental factors that contribute
to the natural decay process (referred to here as inactivation) in these management practices include
irradiance (sunlight), temperature, turbidity, salinity, toxic substances, and predation. A simple first-order
decay model:
Ct=Coe-K°' (3-1)
Where: Ct = concentration of organism at time t (CFU/100 mL); C0 = concentration of organism at time
zero (CFU/100 mL); K0 = overall inactivation rate constant at the environmental conditions (h"1); and t =
elapsed time since time zero (h); is commonly used to predict the effluent concentrations in these systems.
The literatures report a wide range of values for K, however. Those values are typically based on a single
source making extrapolation to different conditions difficult.
1-5
-------
The primary objectives of this study were to:
(1) determine, using bench-scale studies, the factors most important to evaluate commonly used
inactivation models for indicator bacteria;
(2) document the effects of two types of structural best management practices (retention pond and
constructed wetland) on the removal efficiencies of indicator organisms (total coliforms, fecal
coliforms, E. coli, and enterococci) in stormwater;
(3) evaluate the applicability of first-order decay model;
(4) evaluate the effluent concentrations of indicator organisms in stormwater runoff as they flow
from constructed wetland and retention pond BMPs to determine overall inactivation rates from
these systems for various indicator organisms;
(5) record physical, chemical and biological parameters to determine any correlation with effluent
indicator bacteria concentrations; and
(6) develop relationships to serve as predictors for concentrations of indicator bacteria in the effluent
of selected BMPs.
Investigating the inactivation of indicator organisms from stormwater runoff that passes through retention
ponds and constructed wetlands is a complex undertaking. This project involves the analysis of various
types of environmental and biological factors and multiple laboratory methodologies. A combination of
bench- and pilot-scale studies were selected to take advantage of controls and conditions each scale has to
offer. Bench-scale work was done to identify timing of samples and gain an understanding of the
magnitude select factors would have on bacteria inactivation rates. It was also recognized early that there
was little chance of obtaining constant conditions in a pilot-scale study subject to many environmental
influences. Likewise, the replication of all potential environmental factors and their combinations within
the laboratory was not feasible. Studies of both scales proved to provide the necessary conditions and
controls to complete the project. The following chapter provides, in detail, the primary factors and
supporting literature considered when considering the affect environmental factors have on indicator
organism inactivation.
1-6
-------
Chapter 2 Factors Affecting Microbial Indicator Concentrations
Stormwater runoff often accounts for a large fraction of total microbial loading to many receiving waters
(Jamieson et al, 2004; Crabill et al, 1998; Nix, 1994; Qureshi and Dutka, 1979; Olivieri et al, 1977;
Wanielista, 1977; Geldreich et al, 1968; Weibel et al., 1964) with the potential to adversely impact
drinking water sources, contact recreation areas, and protection and propagation of aquatic life (Sunen
and Sobsey, 1999; Haile et al., 1999). Studies have also identified potential links between stormwater
runoff and waterborne disease outbreaks in human populations (Currieo et al., 2001; Rose etal., 2001).
Once introduced into the environment, microorganisms are affected by various environmental factors.
There are known effects from chemical, physical, and biological sources that influence indicator bacterial
growth, die-off, and inactivation. This chapter reviews some of these factors. While not exhaustive in its
coverage, it covers the factors believed to be the most pertinent to retention ponds and constructed
wetlands.
Temperature
Temperature plays an important role in microorganism die-off and has often been cited as the most
important environmental factor. In general, microorganisms survival is prolonged at lower temperatures
(Ferguson et al, 2003). Experiments conducted by Selvakumar et al. (2004) showed that growth rates of
indicator organisms are greatly reduced at 4°C. Geldreich et al. (1968) noted that organism persistence
remained higher at 10°C than similar samples at 20°C.
In the natural environment, several studies reported different die-off rates for various microorganisms in
surface water (Table 2-1). Medema et al. (1997) found that the die-off of E. coll and enterococci were
approximately ten times faster than die-off of Cryptosporidium parvum oocysts, but die-off rates of
Clostridium perfringens were slower than those of oocysts. They also noted that die-off of these
indicators was faster at 15°C than at 5°C. Dutka and Kwan (1980) reported that E. coll, Streptococcus
faecalis, and Salmonella thompson could survive in 17-18°C waters for at least 28 days and E. coli was
found in greater concentrations than Streptococcus faecalis. Baudisova (1997) reported that the die-off
rate of E. coli is greater than that of total and fecal coliforms. Canteras et al. (1995) noted a clear
negative relationship between die-off and temperature. At 10°C, 36 h was necessary to reduce the
population ofE. coli to 10% of the initial concentration compared to 8.4 h at 42°C. Greater reduction of
the die-off rate was noticed in the range between 10 and 18°C than between 18 and 42°C.
2-1
-------
Table 2-1. Some Reported Die-off Rates (K) for Indicator Organisms
Indicator
Organism
Condition
Reference
Total Coliforms
Total or Fecal
Coliforms
Fecal Coliforms
Enterococci
E. coli
0.042-0.229 Freshwater (20°C)
0.033 Average freshwater (20°C)
0.058 (0.029-0.125) Seawater (20°C)
0.018 River water (12 d)
0-0.1 New York Harbor, salinity
2-18 ppt,
0.104-0.254 dark samples
New York Harbor, salinity
15 ppt and sunlight
1.542-4.583 Seawater and sunlight
0.021 River water (12 d)
0.003 River water (5°C); 42 d
0.009 River water (15°C); 0-14 d
0.003-0.083
Thomann and Mueller (1987)
Baudisova (1997)
Thomann and Mueller (1987)
Thomann and Mueller (1987)
Baudisova (1997)
Thomann and Mueller (1987)
Seawater, salinity 10-30 ppt Thomann and Mueller (1987)
0.022
0.004
0.008
Fecal Streptococci 0.75-2.292
River water (12 d)
River water (5°C); 42 d
River water (15°C); 0-14 d
Seawater and sunlight
Baudisova (1997)
Medemaetal. (1997)
Thomann and Mueller (1987)
Much of the earliest work on bacterial removal assumed that temperature was the most important factor
controlling the removal mechanism, as described by the first-order equation developed by Marais and
Shaw (1961). Studies, such as Klock (1971) and Ferrara and Harleman (1981) also emphasized on first
order concentration reductions with temperature-dependent rate constants.
Recent investigations considered bacterial removal as a more complex mechanism involving interactions
between the physical, chemical and biological systems present in wetlands and retention ponds, although
temperature clearly remains an important parameter. For example, Polprasert et al. (1983), Pearson et al.
(1987a, b), Barzily and Kott (1991), Mara et al. (1992a, b), and Mezrioui et al. (1995a, b) all found that
removal rates of fecal coliforms increased with increasing temperature. No matter which indicator
organism is tested, temperature clearly affects indicator bacteria.
Sunlight
Numerous studies have shown sunlight as an important factor in microorganism die-off though it is
difficult to separate effects from other factors entirely. Sinton et al. (1994) studied inactivation in
sunlight of fecal coliforms and enterococci from sewage and meat works effluent concluding that die-off
rate of fecal coliforms was 2-4 times that of enterococci and inactivation is generally slower at lower light
intensities. Alkan et al. (1995) found that variability of enteric bacteria (i.e., enterococci and E. coli) die-
off due to the effect of sunlight depends on the variability of the intensity of light and other small scale
environmental factors such as turbidity, sewage content, and degree of mixing. Importantly, they further
2-2
-------
reported that the die-off rates of E. coli and enterococci from exposure to light were similar. Canteras et
al. (1995) reported that sunlight was the most important factor affecting die-off of E. coli with 90%
concentration reductions within about 1 h (i.e., inactivation rate of about 0.89 h"1) at 18°C and 8.5% of
salinity when light radiation was greater than 120 W/m2 (12 mW/cm2). Yukselen et al. (2003) studied the
effects of solar radiation and temperature on bacterial die-off rates in Black Sea coastal waters and found
that solar radiation was the most significant factor affecting the mortality of coliform bacteria. No
significant effect of temperature was observed in the presence of solar radiation. However, the effect of
temperature is significant in dark experiments with die-off taking approximately 20 times longer to reach
90% concentration reductions compared to values in the light. Davies-Colley et al. (1999) reported that
sunlight is the main factor causing natural attenuation in waste stabilization ponds, although dissolved
oxygen (DO) and pH can also influence the rate of disinfection. Die-off studies on E. coli and Salmonella
were conducted in two different ecosystems: Morlaix estuary in English Channel and Bay of Toulon on
Mediterranean Sea. In the Morlaix estuary, most of the bacteria were mixed with turbid waters and were
able to survive a long time as light penetration was prevented by suspended matter, lowering the effect of
sunlight. On the contrary, through lack of nutrients and very high sunlight intensity, die-off rates in
Mediterranean waters were high with 90% mortality within 2 h near the water surfaces, and several hours
in deep waters (Pommepuy et al., 1992).
A close relationship was found between the light intensity and the decay rate. Gameson and Gould
(1975) concluded that about half the lethal effect of light is attributable to wavelengths below 370 nm
with an additional quarter of the lethal effect attributable to the 370-400 nm and 400-500 nm bands,
respectively. The effect of longer wavelengths, greater than 500 nm, is negligible.
The exact mechanism whereby microorganisms become non-viable after sunlight exposure is not entirely
clear. Photons can excite exogenous or endogenous sensitizers (e.g., humic acid) present in the water that
damage DNA or other cellular components of the bacteria, directly. Photons can also cause damage
indirectly by promoting the production of free radicals in the presence of dissolved oxygen and organics.
Chamberlin and Mitchell (1978) and Eisenstark (1971) noted that the mechanism of light-induced
bacterial decay depends on the presence of endogeneous sensitizers or chromophores, which adsorb light
energy and cause cell damage directly or by reaction with oxides to form superoxides, which in turn may
cause damage to the cells.
Physical Processes (Sedimentation, Sorption, and Filtration)
Microbes in the water column may associate with particles or remain in the "free" or unassociated phase.
This free phase includes organisms that exist individually or as agglomerated groups (aggregates) held
together by organic and inorganic particles. Microbes associated with particles, particularly denser
inorganic particles, will tend to settle from the water column more quickly than free organisms or those
associated with less dense particles that remain more mobile in the environment. It has also been
observed that microbes associated with particles tend to survive longer in natural waters than free
microbes (Howell et al, 1996; Sherer et al, 1992; Burton et al, 1987; Gerba and Schaiberger, 1975).
These particle associations can affect not only microbial fate and transport, but also the time these
organisms remain a threat to human health.
Gannon et al. (1983) and Davies and Bavor (2000) assessed the performance of stormwater
impoundments and constructed wetlands for microbial concentration reductions by measuring inflow and
outflow, with results suggesting that sedimentation was a primary mechanism of removal. Gannon et al.
(1983) also demonstrated that a significant fraction of fecal coliforms in the water column was retained by
a 5 mm filter, indicating that some bacteria were attached to particles, but making no distinction regarding
the nature of the particles. In subsequent work, Schillinger and Gannon (1985) analyzed the partitioning
2-3
-------
of several bacterial indicators in samples taken from the water column of a stormwater drain under wet
weather conditions. Similar filtration experiments revealed that at least half the bacteria in the water
column passed through a 5 mm filter, but again the nature of the particles was not addressed.
Characklis et al. (2005) found that substantial fractions of five different microbial organisms, including
bacterial, protozoan, and viral indicators, were associated with settleable particles in stormwater. The
results also found partitioning behavior varied by organism and with conditions (e.g., storm vs.
background). This study attempted to correlate the microbe-particle association with specific
environmental factors (e.g., TSS, TOC, particle number) but did not yield strong evidence of a
relationship. However, the results suggest that for some organisms (e.g., fecal coliforms) there may be a
relationship between the fraction that is particle associated and particle concentration, microbial
concentration, or both.
The processes of sedimentation, sorption, and filtration are difficult to separate. In the project discussed
in this report, the physical processes are treated collectively and referred to as sedimentation. Some
consider sedimentation as the main mechanism of pollutant removal in general in constructed wetlands
and retention ponds. With the tendency for a significant portion of some stormwater pollutants to bind to
particulates, sedimentation is often considered as the primary factor when designing treatment practices.
Longer detention periods in retention ponds promote sedimentation of solids in the course and medium
size fractions. Similarly, the presence of extensive vegetation in constructed wetlands can encourage
sedimentation (Pundsack et al., 2001).
Salinity
Osmotic stress can also play a role in the concentrations of concentrations of indicator bacteria. The die-
off rate is generally much faster in marine and estuarine waters than in freshwater (Thoman and Mueller,
1987). Yan et al. (2000) found that both light intensity and salinity have significant effects on the
inactivation of E. coll in wastewater discharged into the ocean through submarine outfall system. Solic
and Krstulovic (1992) found inactivation rate increased as salinity increased. They also noted that higher
salinity and high levels of solar radiation combined produced a synergistic effect, resulting in higher
mortality rates of fecal coliforms. Hanes and Fragala (1967) showed that E. coll, coliforms and
enterococci had greater inactivation rates with increasing salinities, with 6.2, 4.6, and 1.6 times greater
death rate, respectively, at 100% seawater compared to 0% seawater. Similarly, Anderson et al. (1979)
found a decreased survival rate for E. coli with increasing salinity, ranging from 53.5% survival at 10 ppt
after 8 days of exposure, to 2% survival at 30 ppt for the same period. Fuijoka et al. (1981) found that
seawater caused rapid inactivation of fecal streptococci and fecal coliform, whereas the organisms
remained stable for three days in freshwater. Mancini (1978) indicated that components in seawater in
addition to salt may be responsible for inactivation in seawater.
Salinity concentration is an important factor in coastal receiving waters such as estuaries or the coastal
ocean. Although many estuarine and coastal systems receive stormwater runoff, generally stormwater
retention ponds and constructed wetlands are not constructed in areas that are tidally influenced or are in a
continuously saline environment. The project discussed in this report only briefly address bacterial
attenuation from this stressor.
Predation
Wetlands and retention ponds can support a diversity of aquatic animals including micro-crustaceans
(copepods, ostracods, and cladocerans), shrimp, crayfish, insects (dragonfly larvae, water beetles, and
water boatman), pond snails, tadpoles, frogs, and fish. These organisms are a crucial component of
2-4
-------
wetland and shallow open-water ecosystems, providing food-web linkages between plants,
microorganisms and other animals. Predator-prey relationships are important in the control of
mosquitoes (Greenway et al., 2003) and may contribute to the control of bacterial populations (Davies
and Bavor, 2000) in these systems. Green et al. (1997) determined bacterivorous activity was an
important factor in the removal of bacteria in constructed wetlands treating wastewater. Mandi et al.
(1993) determined predation by nematodes and Decamp and Warren (1998) by ciliates and rotifers were
significant factors in determining bacterial densities in constructed wetlands although in the former study
the predation was not quantified. In the latter, Decamp and Warren (1998) determined that ciliates such
as paramecium ingested as many as 13 fluorescently labeled E. coll per cell per minute. Fernandez et al.
(1992a, b) also concluded that predation and competition were extremely important in the removal of
fecal coliforms. As part of a large study to model the removal of fecal coliforms, Troussellier et al.
(1986) investigated the effects of grazing by rotifers and of biological oxygen demand (BOD) loading.
They found rotifers can significantly affect fecal coliform concentrations in aquatic systems.
Other Potential Factors
There are many other factors that can affect the densities of indicator organisms. BOD, pH, and DO have
all previously been mentioned as potential contributing factors to inactivation rates of bacteria. Solic and
Krstuvolic (1992) noted that fecal coliforms thrived at a pH range of 6-7, declining in numbers outside of
this range, with greater rate of mortality in acidic environments. Chemical factors include oxidation,
exposure to biocides excreted by plants, and sorption to organic matter. Additional biological removal
mechanisms may include antimicrobial activity of root exudates (Kickuth and Kaitzis, 1975; Axelrood et
al., 1996), activity of lytic bacteria or viruses (Axelrood et al., 1996), retention in biofilms (Brix, 1997),
and natural die-off (Gersberg et al., 1989a, b).
Few studies have thoroughly investigated the effect of nutrients (including dissolved organic carbon and
trace metals) on the inactivation of microorganisms in stormwater in the environment. Thomas et al.
(1999) found bacterium Campylobacter had low survival rates in nutrient-containing microcosms. In
microcosm studies in fresh and salt waters, Noble et al. (2004) found that nutrient levels had an
insignificant effect on the persistence of fecal indicator bacteria.
Summary
As can be seen, literature values on indicator bacteria inactivation in surface waters are quite variable.
Much of the literature pertains to wastewater treatment and dairy waste studies. There have been few
studies conducted investigating the viability of indicator bacteria using stormwater as a medium. Of the
field studies on surface waters, conditions of the watershed and storm information such as intensity and
duration are not well documented or are incomplete. Therefore, interpreting data and comparing results
are often tenuous at best.
Which of these factors has the greatest influence on indicator bacteria inactivation? Are there
combinations of factors that have the greatest affect? What conditions affect the contribution from each
individual or group of factors? The next three chapters detail both bench- and pilot-scale experiments
conducted for data collection and evaluation of the first-order decay model that uses environmental
factors to determine the inactivation rate, K, to predict effluent concentrations in BMPs.
2-5
-------
Chapter 3 Bench-Scale Research
Purpose
The bench-scale research tested the proposed descriptive relationships between concentration and
identified presumptive controlling variables (time, temperature, light intensity, and salinity) outlined in
the literature and listed below. The work used statistical analysis, specifically nonlinear regression, to
quantify the organism-specific inactivation rate constants for traditional (fecal streptococci, and total and
fecal coliforms) and alternate (enterococci and E. coli) microbial indicators in stormwater. The research
separately assessed the influence of time, temperature, light intensity, and salinity on the inactivation rate
constants to isolate the effects and maintain the analytical load within laboratory capacity. This approach
simplifies the analysis but neglects the potential interactions among the independent variables.
The research exposed stormwater samples to controlled-conditions to measure the change in
microorganism concentrations after known exposure periods. The experimental design selected the
controlled independent variables and their ranges based on the broad, literature-reported influence and the
likelihood of the condition existing in the structural BMPs (e.g., retention ponds and constructed
wetlands). As widely reported in the literature, time, temperature, and light intensity are important
environmental variables which determine the rate of change of indicator organism concentrations.
Salinity was included in this study as it is reported as an environmental factor which influences the
microbial decay and can be potentially important in some BMPs installed in coastal settings or when the
stormwater runoff results from areas where communities apply road salt during winter.
Inactivation Rate Models
The literature reports indicator organism inactivation rates in various water types (Table 2-1). However,
information on inactivation rates for indicator organisms in stormwater and effects of natural factors on
survival rates is limited except for one study by Geldreich et al. (1968).
Most published studies use first-order decay known as Chick's Law to describe indicator organism
inactivation with time. Under this premise, the concentration-time relationship is:
(3-1)
Where: Ct= concentration of organism at time (CFU/100 mL);
C0= concentration of organism at time zero (CFU/100 mL);
3-1
-------
K0= overall inactivation rate constant at the environmental conditions (h"1); and
t = elapsed time since time zero (h).
There are several approaches to estimate the effects of environmental variables on the overall rate
constant. The simplest approach assumes additive effects:
K0=KT+KS+K,+ Kf (3-2)
Where: KT = inactivation rate constant due to temperature (h"1);
Ks = inactivation rate due to salinity (h"1);
KI = inactivation rate constant due to light (h"1); and
Kf = inactivation rate constant due to other factors such as sorption, filtration, and
sedimentation (h"1).
Temperature
The effect of temperature is often approximated by using the Arrhenius-van't Hoff equation (Khatiwada
and Polprasert, 1999):
KT=K2^T(T-20} (3-3)
Where: KT = inactivation rate constant due to temperature at T = T°C (h"1);
K2o = inactivation rate constant due to temperature at T = 20°C (h"1);
T = temperature in °C; and
-------
Light
The effect of light intensity on the inactivation rate constant is normally expressed as:
K, = ®,IZ (3-6)
Where: \ = light proportionality coefficient (cm2/mW-h); and
Iz = light intensity at depth Z below the surface (mW/cm2).
Unlike temperature and salinity which can be reasonably assumed to be uniform throughout the system,
light intensity varies with depth below the water surface. The intensity at a given depth, Iz,, decreases
exponentially with distance (Gameson and Gould, 1975). The value is often estimated as:
/ ^(i_^) (3-7)
rh
Where: I0 = light intensity at the earth surface (mW/cm2);
T = vertical light extinction coefficient (1/m); and
Z = depth of water (m).
The extinction coefficient varies with water proprieties including color and turbidity (Lee and Rast,
1997).
Combining equations, the overall equation using the Canteras etal. (1995) assumption for salinity is:
Ko=K2^T(-T-20^ss+^Iz+Kf (3-8)
Materials and Methods
Sample Collection
Stormwater was collected from an outfall after that drained a 10-acre portion of the Middlesex County
College Campus near the USEPA facility in Edison, New Jersey (Figure 3-1). The drained area was
predominantly campus maintenance buildings and student parking lots. Samples were only collected for
this work when the rain event met the USEPA monitoring guidance (USEPA, 1992). Generally, the
project required at least 3 mm (l/8th in.) total rainfall, preceded by at least 72 h without measurable
precipitation. Automatic samplers (Hach, Loveland, CO) placed in the outfall collected a flow weighted
composite sample when the flow water depth in the outfall initially reached 2.54 cm (1 in.). Area-
velocity flow meters (Hach, Loveland, CO) connected to the automatic samplers triggered the internal
peristaltic pump to add 1-L aliquots to a 20-L, pre-cleaned container when an incremental specified flow
volume was measured. The incremental volume was set based on forecasted total rainfall.
After collection, the samples were transported to the on-site laboratory and allowed to quiescently settle
for 10 to 20 min at room temperature to allow most settling solids to fall to the container bottom. The
water from the settled collection container was transferred leaving about 1 in. in the composite container
bottom to limit the potential effects of settleable particulates on the experiments and avoid interference
with the enumeration process. While continuously stirring the container holding the decanted supernatant
(Stir Pak Mixer), a peristaltic pump transferred aliquots to 250-mL pre-cleaned HDPE bottles. All
subsample bottles were completely filled leaving no headspace.
3-3
-------
Figure 3-1. Drainage area (A) and outfall location (B) of the study in Edison, New Jersey.
Experimental Methods
The experiments were conducted by placing the 250-mL HDPE containers in water baths (Precision, A
Division of Jouan Inc., Winchester, VA) to maintain constant temperature. The temperature of each
water bath was established at least one day before inserting the bottles. Aluminum foil wrapping on the
outside of the bottles prevented light exposure for experiments other than those investigating the effects
of light exposure. The temperature of the water bath and the temperature of an equal volume of deionized
water in separate containers were monitored using a NIST-traceable digital thermometer and recorded at
1-min intervals using logging thermisters (Onset Corp, Bourne, MA). The recorded temperatures
confirmed that the stormwater in the 250-mL containers required from 30 to 340 min to reach the water
bath temperature. The temperature varied less than 1°C during the experiment.
The experiments defined time zero as the time when the sample reached the designated temperature as
described below. Bottles were removed periodically during the experiment for sampling and analysis.
The times when bottles were removed from the water bath were established based on the expected
exponential concentration decline and to collect samples primarily during the normal workday as a cost
control measure. Samples collected from the bottles were analyzed for five indicator organisms following
membrane filtration procedure. A set of four samples was collected for the initial time. Subsequent
sampling collected duplicate samples from the bottle. Time was monitored using commercially-available
(La Crosse Technology, La Crosse, WI) clocks synchronized to the US Naval Observatory atomic clock.
The reported elapsed time for removing the bottle from the temperature bath is believed to be accurate to
within 1 min. DO and pH were monitored daily from independent sample bottles for the duration of the
3-4
-------
experiment.
Temperature Study
The temperature-dependence die-off study targeted temperatures of 10°C, 20°C, and 30°C. Earlier
USEPA research, measuring the diurnal temperature fluctuation in local BMPs showed that summer
temperatures reach about 30°C and fall to less than 5°C in the winter. While extreme water temperatures
in BMPs are likely to span a slightly wider temperature range; the temperature range used in this study
should represent the most often encountered temperatures within the BMP. The mean temperatures
recorded by the data loggers are used in the analysis (usually slightly lower than the targeted
temperatures), however, for simplicity, the descriptive target temperatures were used. Temperature
monitoring was included as part of the light and salinity studies.
Salinity Study
The salinity-dependence die-off study targeted four salinity levels (0, 10, 20, and 30 ppt). The target
temperature was set at 25°C. The salinity range was selected to represent concentrations encountered in
partially diluted seawater. The salinities were established by adding synthetic sea salt (Instant Ocean,
Aquatic Systems, Mentor, OH) directly to stormwater while mixing. When the lowest salinity was
achieved, stormwater was dispensed to the sample bottles. Then, more sea salt was added to achieve the
next highest salinity level. The salinity of the resulting solution was measured using Hach CO 150
Conductivity Meter (Loveland, CO). The analytical results showed salinity concentrations of 0.45, 8.1,
16.1, and 23.5 ppt, while those concentrations for samples other than those in the salinity study were not
measured.
Light Study
The light-dependence study established a target temperature of 25°C. Samples were exposed to light at
four intensities including one dark sample. The respective light intensities for different light conditions
were established using Reptisun 5.0 UVB fluorescent bulbs (Zoo Med Laboratories, Inc., San Luis
Obispo, CA). The manufacturer reports that the bulbs produce light at UVA (320-400 nm) (30%) and
UVB (290-320 nm) (6%) wave lengths. Adjusting the distance between the light source and the surface
of the container controlled light intensities. The distance was maintained at 12.7, 22.9 or 35.6 cm above
the water surface to have light intensities varying between 20 and 100 mW/cm2 (200 to 1000 W/m2).
Average intensity of sunlight is about 120 mW/cm2 (1,200 W/m2). Light intensities at the sample surface
were measured using a light meter (International Light IL 1400A with thermopile detector) daily
throughout the experiments. The measured light intensity varied about 5% during the experiment. Table
3-1 lists the average light intensities measured throughout the experiment.
The container used to hold the sample in the light study was 250-mL thin, flat sided polystyrene flasks
with canted neck and plug seal cap (BD Biosciences, Bedford, MA). This container was selected because
it did not have any effect on the light attenuation and would reduce the depth of water assuring a more
homogenous light dose.
3-5
-------
Table 3-1. Light Intensities Corresponding to the Height of Light Source above the Water
Surface
Distance (cm)
12.7
22.9
35.6
Light Intensity
(mW/cm2)
89.0-97.8
50.2-58.4
19.7-21.9
Average Light Intensity
(mW/cm2)
94.70
55.23
20.86
Analysis of Indicator Organisms
All samples were analyzed for five indicator organisms (total coliforms, fecal coliforms, fecal
streptococci, enterococci, and E. coif) using membrane filtration methods following Standard Methods for
the Examination of Water and Wastewater (APHA et al, 1998) (Table 3-2) and described below.
Table 3-2. Summary Table of Analytical Procedures
Indicator Organism
Total Coliforms
Fecal Coliforms
E. coll
Fecal Streptococci
Enterococci
Method Number
SM 9222B
SM 9222D
SM 9222G
SM 9230C
SM 9230C
Method Title
Membrane Filter Procedure
Membrane Filter Procedure
Membrane Filter Procedure
Membrane Filter Procedure
Membrane Filter Procedure
Total coliforms were determined by incubation on M-Endo agar (24 h at 35°C) and confirmed by gas
formation in lauryl tryptose broth and brilliant green lactose broth. Fecal coliforms were incubated on M-
FC agar (24 h at 44.5°C) and were confirmed by gas formation in lauryl tryptose broth and EC broth. E.
coli levels were measured by transferring the membrane from the Endo-type medium to a nutrient agar
containing 4-methylumbelliferyl-(3-D-glucuronide (NA-MUG) and incubating 4 h at 35°C. Production of
blue fluorescence on the periphery of colonies under long wavelength UV indicated E. coli. Fecal
streptococci were determined by incubation on m-Enterococcus agar (48 h at 35°C). Colonies were
transferred to brain heart infusion (BHI) agar. Transfers were made to BHI broth and incubated at 35°C
for 24 h, with confirmations made by retransfer to bile esculin agar, BHI broth incubated at 45°C, and
BHI with 6.5% NaCl. Growth on bile esculin agar BHI broth verifies that the colony is of the fecal
streptococci group. Growth at 45°C and in BHI with 6.5% NaCl indicates that the colony belongs to the
enterococci group.
Samples were sequentially diluted with sterile buffered water using three dilution factors based on
previous analyses of similar samples. Dilution factors were estimated to obtain the method-recommended
colony count on at least one dilution set. Sequential dilutions usually used at least 10 mL aliquots and
always used at least 5 mL. All results were volume normalized to give concentrations in colony forming
units (CPU) per 100 mL.
Each analytical batch included laboratory blanks and positive controls. Blanks were run before and after
each analytical set. Verification was performed on ten colonies for each organism according to the
procedures listed in Standard Methods (APHA et al., 1998). After incubation, the plates were manually
3-6
-------
enumerated. Positive controls showed the growth of particular indicator organisms.
Data Analysis/Reduction
The data analysis used all incubated plates with colonies in the countable range. The count from each
incubated plate was normalized to the source concentration using the dilution factor and volume filtered.
The uncertainty in each sample was estimated as the propagated error using the methods outlined by
Taylor (1997). The dilution is assumed to be error free. The uncertainty in the filtered volume is
estimated as ±0.4 mL, the tolerance of the ASTM class A graduated cylinders used in this study. The
uncertainty in the number of counted colonies is estimated as 10% of the count with a minimum of 1
colony.
The sample weighted-average concentration and associated uncertainty was calculated using the
uncertainty in the individual estimates as the weighting factor. This approach reduces the multiple (5 to
12) results from the samples and dilutions associated with the original source (bottle) to a single
concentration estimate for the organism representing the experimental result resulting from the
environmental condition. Alternate pooling strategies for combining the multiple analyses can be
developed. For example, the multiple dilutions from each sample could be pooled using the same
weighting strategy to obtain either four or two estimates for each elapsed time.
This data reduction relies on regression analysis of the concentration time series developed under the
established experimental conditions. The approach estimates the effect of the various conditions
(treatments). The treatments identified were based on the breadth of earlier-published research. No
attempt was made to establish why the environmental exposure reduces the measured concentration, e.g.,
cellular wall degradation or DNA damage. Because of the reliance on the statistical techniques, it is
important to prevent artificially increasing the degrees of freedom associated with the analysis that would
suggest greater confidence in the results. The pooling approach described accomplishes the objective of
not artificially increasing the degrees of freedom and placing greater emphasis on results with less
uncertainty. Generally, the concentration time series developed for the individual experiments showed
undetectably low concentrations for the final analyses.
The weighted-average concentration for each organism was regressed on the independent variables using
nonlinear least-squares regression techniques. The nonlinear regression method used the Levenberg-
Marquardt technique (a modified algorithm of the Gauss-Newton least-squares technique) in Statistica
software package (version 7.1, Statsoft, Inc.). All regressions are run at the 95% level of confidence
(oc=0.05). The reported uncertainty in the calculated coefficients is the confidence interval reported by
the Statistica software package. After the regression was complete, an Analysis of Variance (ANOVA)
was run to test the significance of the proposed model.
Results
Time and Temperature
The regression used the proposed time-dependent decay function, known as Chick's Law (equation (3-1).
The concentrations at each elapsed time for a given temperature were used to estimate the inactivation
rate constant for the indicator organism at that temperature. Table 3-3 lists the regression results from the
temperature study with the 25°C results from the light and salinity studies (discussed below). These
experiments were also conducted in salt-free, dark, isothermal conditions and represent data that was
included in the analysis.
3-7
-------
Table 3-3. Inactivation Rate Constants for Each Indicator Organism at Tested
Temperatures in the Isothermal Experiments
Temperature
(°C)
Total
Coliforms
Fecal
Coliforms
K coli
Fecal
Streptococci
Enterococci
Inactivation Rate Constants (h"1)
9.07
19.87
29.32
26.17 l
24.74 2
0.007±0.010
0.017±0.035
0.016±0.045
0.013±0.020
0.0131±0.37*
0.03±0.025*A
0.01±0.140
0.76±0.510*
0.07±0.060*
0.10±0.180
0.027±0.015*
0.085±0.033*
0.136±0.072*
0.019±0.072°
0.029±0.048
0.027±0.022*A
0.076±0.077
0.100±1.300°
0.039±0.055
0.056±0.029*
0.021±0.014
0.095±0.038*
0.150±0.420
0.027±0.015*
0.044±0.038
* Indicates KT value is statistically significant at a = 0.05
A The first concentration is omitted from the analysis as an apparent outlier
0 ANOVA shows regression model is not significant
'Data from light experiment
2Data from salinity experiment
In all but two cases, the post-regression ANOVA confirmed the model significance; however, in about
half the cases the temperature-specific inactivation rate constant was not significant. This is taken to
mean that these results do not provide a reason to reject the first-order decay model but that the data set is
often not numerically sufficient to obtain quantitative estimates of the temperature-specific inactivation
rate constant that excludes zero at the established level of confidence. This result generally occurred
when the time series included few data triplets (time, temperature, concentration) leading to high
uncertainty in the numeric values calculated.
The results listed in Table 3-3 for the portion of the study emphasizing temperature generally demonstrate
an increase in the calculated decay constant with increasing temperature. Completing the same analysis
on the results from the subsequent investigations for the light and salinity studies at roughly 25°C do not
produce rate constants expected by interpolating between the bounding temperatures of 10 and 20°C. The
calculated values (Table 3-1) are not consistent for all organisms although the temperatures were within
about 1.5°C. These later two experiments used stormwater collected at the same outfall but at a different
time.
The experimental design planned to pool the results of the temperature study data across the experimental
temperatures. The weighted-average concentration results were regressed on elapsed time using the same
nonlinear procedures to test the model proposed by equation (3-3). As discussed above, the temperature
study used a common stormwater source for the samples exposed to the selected temperature conditions.
These samples had uniform starting concentrations allowing for the time delays in reaching time zero.
The studies for light and salinity effects used different stormwater samples with dramatically different
initial concentrations as would be expected from samples collected at different times of the year. To pool
the data into a common set, the results were normalized to the estimated initial concentration using the
calculated value of the initial concentration, C0 for the specific source calculated above. After
transforming the data to Ct/C0 the above-described procedures were applied to estimate the reference
temperature decay constant and the temperature coefficient using equation (3-9):
_Q
d
(3-9)
Table 3-4 lists the regression coefficients. In all cases the temperature coefficient (cDT) and the reference
decay constant at 20°C (K2o) are statistically significant. The post-regression ANOVA confirms the
3-8
-------
significance of the regressions and significantly there is consistency of the individual coefficients across
organisms.
Table 3-4. Organism-Specific Reference Temperature Rate Constant and Temperature
Coefficient Determined Using Salt-Free Dark Experimental Results
Indicator Organism
Total Coliforms
Fecal Coliforms
E. coli
Fecal Streptococci
Enterococci
Reference Temperature
Rate Constant (K20) (h'1)
0.016±0.009*
0.042±0.030*
0.036±0.019*
0.047±0.031*
0.042±0.014*
Temperature Coefficient
(0T)
1.057±0.085*
1.090±0.110*
1.023±0.072*
1.044±0.040*
1.057±0.045*
* Coefficient is statistically significant at a=0.05.
The cDT values for all the organisms range between 1.02 and 1.09, which is consistent with the values
reported in the literature. Mancini (1978) and Khatiwada and Polprasert (1999) suggested a value of 1.07
for fecal coliforms which agrees with the calculated value in this work of 1.09±0.11.
The DO monitoring showed a steady decline with time. The rate of decrease increased with temperature.
The DO declined at 0.4, 1.4, and 1.7 mg/L/day at 10, 20, and 30°C, respectively. At 30°C, the DO was
nearly depleted after 60 h. At 20°C, the DO was depleted after 70 h. At 10°C, DO was at 2.75 mg/L after
72 h. Except for total coliforms, the plates produced non-quantitative counts within 23 h at 30°C. This
suggests the die-off is not due to depleted DO, but due to the combined effects of time and temperature.
The pH of the samples varied slightly, but remained within the near- neutral range (6.5 to 7.0) throughout
the experiment. Solic and Krstuvolic (1992) noted that fecal coliforms survived within the pH range of 6
to 7, and declined outside of this range, with greater rate or mortality in acidic environments. The
average TSS in the sample was 41 mg/L. These water quality indicators are within the range reported in
the NSQD (Maestre and Pitt, 2005).
Light
The analysis assumes the light intensity measured at the sample surface is representative of the exposure
throughout the container. The limited water depth in the selected container bottles supports this
assumption. The analysis of the results of the light exposure experiments were examined in a two-step
process. The weighted-average concentration was first used to calculate the overall coefficient under the
established condition of light and temperature (KT=26.17ij) for each indicator at each exposure level using
the model in equation (3-10).
C,=Coe~K'ft (3-10)
Table 3-5 lists the regression-estimated values of KT=26.17jfor each indicator organism at each light level.
The statistical quality of the fecal coliform and E. coli results is generally poor. The increased light
intensity affects the calculated decay constant confirming that light influences the decay process. The
calculated decay constant values increase with increasing light intensity. These results support the
presumptive additive effect of light on the overall rate constant, i.e., KTJ = KT + 0tl.
Table 3-5. Regression-Estimated Values of Inactivation Rate Constants at 26.17°C for
3-9
-------
Experimental Light Intensities
Indicator
Organism
Total Coliforms
Fecal Coliforms
E. coli
Fecal Streptococci
Enterococci
Light Intensity (mW/cm2)
0
0.131±0.037*
0.10±0.18
0.029±0.048
0.056±0.029*
0.044±0.038*
20.86
0.21±0.16*
0.07±0.17«
0.14±0.16
0.205±0.063*
0.202±0.070*
55.23
0.23±0.053*
0.08±0.25«
0.16±0.23
0.254±0.022*
0.19±0.15*
94.7
0.32±0.12*
0.20±0.24
0.26±0.20*
0.469±0.025*
0.96±0.14*
0 Regression result is not statistically significant at a=0.05
The pooled data were then used to estimate the coefficients in the presumptive relationship:
Ct=C0e^KT-26-"+^I}t (3-11)
Table 3-6 lists the estimated values of the constants for each organism. The effect of light on the decay
coefficient varies by a factor of four across the organisms showing a difference in light sensitivity. As
expected, the values of KT=26.17,i generally agree with the values listed in Table 3-5 for the dark
experiments. The values also agree with the expected estimated values using KT = K2o(I)(T~20) evaluated at
26.17°C. The light-free exposure for this data set is included in the previous analysis. The added light
has minimal effects on the decay rates for total and fecal coliforms. There is an order of magnitude
smaller than the effects on the other indicators.
Table 3-6. Regression-Estimated Coefficients from Light Experiment
Indicator Organism
Total Coliforms
Fecal Coliforms
E. coli
Fecal Streptococci
Enterococci
Inactivation Rate
Constant (KT=26.17) (If1)
0.155±0.047*
0.070±0.090
0.040±0.037*
0.046±0.015*
0.034±0.020*
Light Proportionality
Coefficient (0>L)
(cm2/mW-h)
0.0016±0.0012*
0.0130±0.0027
0.0025±0.0019*
0.0057±0.0018*
0.0076±0.0038*
* Coefficient is statistically significant at a=0.05
The intensity of natural sunlight varies during the course of the day. The exposure levels are further
variable when considering the clouds that produce the rainfall and resulting runoff. The clouds will filter
or block incident radiation to differing degrees. This work used artificially generated light to maintain
constant exposure levels. Other researchers reported that sunlight showed the greatest bactericidal effect
on organisms. Most research identifies UVB (290-320 run), UVA (320-400 nm) and blue green visible
light (400-550 nm) as the portion of the solar spectrum responsible for inactivating microorganisms. The
UVB portion of the solar spectrum is believed to be the dominant bactericidal agent causing direct DNA
damage (Sinton et al, 1999). For this reason, UV is used for disinfection in water and wastewater
treatment processes (Ferguson et al., 2003; Giese and Darby, 2000).
The pH of the samples varied between 6.5 and 7.5. Dissolved oxygen content of the samples varied
between 8.2 and 12.3 mg/L. The difference in DO from the temperature study is noteworthy. The decline
3-10
-------
observed in the temperature was not observed in this effort. The light exposure experiments lasted nearly
50 h. Total suspended solids concentration in the sample was measured at 85 mg/L. While about double
the concentration recorded in the temperature study, this is still in the typically reported range.
Salinity
The data analysis to establish the effects of salinity parallels the approach used to evaluate light effects.
The individual time series were analyzed for each organism at each salinity level to obtain a preliminary
assessment of the salinity effect. Table 3-7 lists the values of KT:S determined for each indicator organism
at each salinity level.
The effect of salinity is not consistent across indicator organisms. The rate constant for fecal streptococci
is not significant for any of the salinity levels evaluated. The results for other organisms, e.g., fecal
coliforms and enterococci, are generally significant.
The results listed in Table 3-7 do not show a clear pattern for the salinity effect on the calculated decay
rate. Applying the technique proposed by Canteras et al. (1995) requires fitting the concentrations to the
form:
Ct = C0e-K^< (3-12)
Table 3-7. Regression-Estimated Values of Inactivation Rate Constants Determined for
Salinity Concentrations
Indicator Organism
Total Coliforms
Fecal Colifroms
E. coll
Fecal Streptococci
Enterococci
Salinity (ppt)
0
0.013±0.020
0.065±0.061*
0.020±0.0720
0.039±0.054
0.027±0.015*
8
0.039±0.022*
0.031±0.020*
0.079±0.079*
0.016±0.079°
0.035±0.019*
r, — (\ fK*
16
0.020±0.035
0.058±0.099
0.010±0.110°
0.027±0.098°
0.005±0.011
24
0.044±0.038*
0.417±0.094*
0.012±0.042
0.027±0.089°
0.019±0.008*
0 Regression result is not statistically significant at a=0.05
Table 3-8 lists the regression results using the proposed relationship. Overall, the effect of increased
salinity at the tested concentrations is small. The calculated value of <£s is not generally distinguishable
from unity for organisms other than total and fecal coliforms. This suggests that, for the span of salinity
values studied, the added salt had little effect on the decay rate constant. This supports the results
reported by Canteras et al. (1995) who found largest salinity effect occurs only when the salinity values
were over 35 ppt. Thoman and Mueller (1987) reported that the inactivation of fecal coliforms is
generally much faster in marine and estuarine waters than in freshwater. Mancini (1978) also indicated
that components in seawater in addition to salt may be responsible for greater inactivation effect in
seawater.
The pH of the samples varied between 7.0 and 8.0. Dissolved oxygen content of the samples varied
between 8.3 and 12.2 mg/L. Total suspended solids in the sample were measured at 44 mg/L.
3-11
-------
Table 3-8. Regression-Estimated Coefficients from Salinity Experiment
Indicator Organism
Total Coliforms
Fecal Coliforms
E. coli
Fecal Streptococci
Enterococci
Calculated Coefficients
^(h-1) <£s
O.OlliO.Oll 1.073±0.061*
0.035±0.023* 1.095±0.060*
0.024±0.041 0.990±0.100*
0.025±0.032 0.998±0.086*
0.032±0.016* 0.957±0.036*
* Coefficient is statistically significant at a =0.05
Summary
The results of this experiment demonstrated that the concentration of the tested indicator organisms
decrease exponentially with time. The first-order decay process reasonably models the concentration time
series for the durations tested. The analysis of the weighted-average concentrations enabled developing
organism-specific inactivation rate constants in stormwater assuming time, temperature, light intensity
and salinity are the most significant parameters.
The temperature study indicated that the indicator organisms persisted at higher concentrations at lower
temperatures. The inactivation rates increased with increasing light intensity. The added light has
minimal effects on the inactivation rates for total and fecal coliforms. Different indicator organisms
exhibited different trends with salinity. Taken as a whole, the results indicate that salinity had little or no
effect on inactivation rates for these indicator organisms for the salinities tested.
The difference in temperature results from the temperature study suggests that differences in the
stormwater influence the reference decay rate. The major measured difference in the characterization of
the stormwater was the initial concentration of indicator organisms. This further suggests that the
constants measured in the bench-scale experiments must be viewed as the rate for the specific stormwater
sample evaluated and cannot be extrapolated to all stormwater sources. The variability of the constants
between sources, if any, cannot be estimated from these data.
3-12
-------
Chapter 4 Pilot-Scale Research
This project was designed to determine the factors that influence the rate of microbial inactivation as
urban stormwater passes through retention ponds and constructed wetlands. Research on constructed
wetlands inactivation of fecal indicators in wastewater is well documented (Bavor et al., 1987; Gersberg
et al., 1987; Ottova et al., 1997). Removals of fecal streptococci and coliforms generally exceeded 80%
and 90%, respectively (Kadlec and Knight, 1996). Gersberg et al. (1987) and Garcia and Becares (1997)
concluded that extensively vegetated systems remove indicator bacteria at significantly higher rates from
wastewater than unvegetated systems. However, because of the potentially high indicator bacteria
concentrations in stormwater runoff, the remaining 10 - 20% (assuming a performance of 80-90% is
achieved) in the effluent may increase receiving water concentrations beyond WQS. This is in contrast to
sanitary and combined stormwater and sanitary systems which, other than during sewer overflow,
chemically treat the wastewater routed to treatment plants.
This pilot-scale project intended to build on the bench-scale studies to evaluate the variation of rates of
inactivation using first-order decay with changes in environmental conditions. The mesocosms designed
and constructed for this project offer a unique environment where many parameters concerning
stormwater characterization and flow can be held constant (i.e., characteristics of influent, residence time,
and pollutant loading). By varying testing dates with typical climatic conditions experienced throughout
the year, an assessment of the impact of environmental change on bacterial inactivation rates can be
assessed. A comparison of rates of inactivation with seasonal wet weather events can indicate whether
water quality managers can rely on model predictions to be accurate during periods when loading may be
greatest. More information is needed to determine whether models that use first-order decay functions
when predicting bacteria effluent concentrations from field BMPs (usually as a point source) are
accurately providing effluent concentration predictions and concomitant loads. This can directly impact
the loading allowed when determining TMDLs for meeting WQS.
Study Site and Experimental Design
A pair of rectangular mesocosms of the same size with two different stormwater BMP treatments
(constructed wetland and retention pond) were constructed at the Urban Watershed Research Facility
(UWRF) in Edison, New Jersey (see Figure 4-1). Mesocosm housings were purchased fiberglass
aquaculture tanks. Mesocosms were constructed by placing a perforated PVC under drain with valve,
about 4 cm of pea gravel, 2.5 cm of sand, and 25 cm of topsoil. Local cattail plants were transplanted into
constructed wetland mesocosms. The retention ponds were constructed similarly with like layers and
depths of gravel, sand, and topsoil, but with no plantings. Tanks had a length, width, and depth of 1.78
4-1
-------
meters, 0.74 m, and 0.65 m, respectively with a stormwater volume of approximately 227 L. Both
systems were constructed in August of 2002.
Creating Bacterially Loaded Stormwater
Feed concentrations in the stormwater runoff collected on site were low (lO'-lO3 CFU/100 mL) compared
to many urban watersheds. To achieve a higher loading concentration (104-106 CFU/100 mL) for this
study, a 500 mL aliquot of stormwater runoff was collected from on-site runoff as described in Chapter 3
and placed in growth media to encourage growth of the desired bacteria, consequently producing higher
densities of bacteria in the stormwater. Complete method development is described in Appendix A.
(A)
(B)
Figure 4-1. Pictures of the retention pond (A) constructed wetland (B) treatment systems.
Target indicator bacteria concentrations in mesocosms following addition of the enriched stormwater are
found in Table 4-1. The cultured stormwater was introduced over a 30-45 min time period (in an attempt
to limit thermal shock) to a common supply tank that contained approximately 1000 L of recently
captured stormwater and mixed for 30 min. Constructed wetland and retention pond mesocosms were
filled from this same supply source.
Table 4-1. Target Indicator Organism Densities in Mesocosms after Addition of Enriched
Stormwater
Indicator Organism
Total Coliforms
Fecal Coliforms
E. coli
Enterococci
Target Concentration
(CFU/100 mL)
106- 107
106- 107
104- 105
104- 105
4-2
-------
It is recognized that the relative proportion of indicator bacteria in the bacterially enriched stormwater
will change from the original stormwater and may be less representative of the true indicator bacteria
community. Another recognized difference in the bacterially enriched stormwater is the particle
association of the indicator bacteria is suspected to be different than the source with less bacteria
associated with larger particles and more bacteria associated with finer particles or unattached. Although
this was not measured, it is believed that this property results in a more conservative estimate for the
indicator organisms when using bacterially enriched stormwater for indicator bacterial loading
experiments. The thought is fewer colonies are expected to settle out of the water column through
attachment to larger particles resulting in increased effluent concentrations compared to the stormwater
source.
Storm Event Simulation
Runoff events were simulated by distributing 220 L of collected stormwater runoff from a common
supply tank that was mixed for 30 min using four vortex mixing eductors to each mesocosm. Additions
were made sequentially at an average rate of 0.75 L/s measured using a Seametrics WT-P turbine meter
(SeaMetrics, Kent, WA). At this stormwater discharge rate, it took approximately 5 min to load the BMP
mesocosms. Using the same method and equal volume as in the mesocosms, cultured stormwater was
distributed to a 227 L container placed near the mesocosms serving as a control tank. This tank, based on
proximity, was expected to have the same ambient environmental conditions as the pilot-scale tanks.
Effluent flow rates and hence detention times in the mesocosms were regulated by the effluent pipe
orifice (diameter 2 mm) simulating a riser flow control structure as in field BMPs. Mean effluent flow
rates ranged from 57 - 76 L/hr during the loading event until levels return to their pre-event (static)
levels. Generally, it took between 20-22 h for water levels to return to their original pre-event level.
Between simulated storm events, and after drawdown of stormwater to a static pre-event level during
simulated storm events, mesocosm water levels were maintained through semi-continuous flow from a
nearby water supply regulated by a float valve positioned opposite and slightly above the effluent orifice.
Residence times of the mesocosms were previously determined by Struck et al. (2004) using conservative
dye tracer tests to determine the residence time distribution with Rhodamine WT dye distributed to each
mesocosm. Tracer concentration in effluents were measured with a YSI Rhodamine WT and verified
with a Turner Designs 10 AU field flourometer (Sunnyvale, CA). Exponential curves were fit to tracer
concentration data to calculate the measured mean residence time (Levenspiel, 1999).
Storm events were planned over a two-year period with three simulated storm events each year. No event
was to occur within the same month. The dates selected represented typical climatic conditions during
that time to incorporate changing environmental conditions as a factor in bacterial inactivation.
Water Quality Monitoring, Solids, Light, and Bacterial Indicator Sampling
The constructed wetland, retention pond, and control tank each had a YSI water quality sonde placed on
the sediment surface at a depth of 4 cm in the constructed wetland and 25 cm in the retention pond near
the overflow orifice. These sondes recorded in-situ temperature, DO, pH, conductivity, and turbidity
averaged over 10-minute intervals.
Light intensity was measured using an on-site weather station (Onset Corporation, Bourne, MA). Grab
samples of light were also recorded on six separate occasions between 12:00pm and 3:00pm at the water
surface in the retention pond and constructed wetland using a hand held light photometer (IL1400,
International Light Inc., Newburyport, MA).
4-3
-------
Effluent from the riser pipe of the retention pond and constructed wetland was collected in pre-washed 1-
L HDPE bottles placed at the effluent drainpipe (to collect enough volume for the sample but subject to
continuous replacement). Microbiological and TSS samples were collected from these 1-L containers
using automatic samplers (Hach Company, Love land, CO) and placed into cooled pre-washed 1-L HDPE
bottles within the sampling device at elapsed times found in Table 4-2. Timed samples were also
collected at a depth of 5 cm below the water surface in the control tank and labeled as "light controls". If
grab samples were necessary due to autosampler failure discrete samples using pre-cleaned PVC bottles
were collected to obtain the desired sample.
Six 500-mL bottles were wrapped in aluminum foil filled with 400 mL of inoculated stormwater, sealed,
and placed in the control container to duplicate environmental conditions (other than sunlight) in the
mesocosms and control tank. These samples were collected daily and with the extended times (beyond 90
h) as noted in Table 4-2, as "dark controls" along with the "light" control samples collected from the
control container.
These samples separate light and dark affects on the microbial population in the control groups. It should
be noted that sample #1 in Table 4-2 (time 0) was collected 30 min before stormwater addition for the
constructed wetland and retention pond recording antecedent baseline conditions. Stormwater controls
were then loaded and sampled with sample #1 (time 0) representing the influent concentrations and
conditions of the supply tank and the dark and light controls.
Table 4-2. Effluent Time
Sample #
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
of Hand
h
0
5
10
15
21
27
33
39
45
51
60
69
78
90
114
150
Collected and Programmed Autosample Collection
Sample Collected
influent, light control, in-situ (background)
timed effluent
timed effluent
timed effluent
timed effluent
timed effluent, light control, dark control
timed effluent
timed effluent
timed effluent, light control, dark control
timed effluent
timed effluent, light control, dark control
timed effluent
timed effluent
timed effluent, light control, dark control
timed effluent, light control, dark control
timed effluent, light control, dark control
All samples were transferred to appropriate sized plastic pre-cleaned containers for analyses depending on
the type of analyses and the source of the sample. Samples were transported to the laboratory for
splitting, filtering, and preservation as necessary within specified holding times. If storage was necessary,
samples were stored in a refrigerator at 4°C.
Bacteria indicator samples were analyzed for four indicator organisms (total and fecal coliforms,
enterococci, and E. coli) following the standard procedures listed in Chapter 3 (Table 3-2). Fecal
streptococcus analyses were not done because the increased quantity of samples this indicator added was
beyond laboratory capabilities.
4-4
-------
The expected indicator concentration in the effluent for total and fecal coliform, enterococci, and E. coll
was 104 to 106 CFU/100 mL during the first sampling period. Each timed sample was mixed and split
into three equal volumes of sample for each organism. Three dilutions of each split sample were analyzed
to assure at least one dilution provided concentrations between 5 and 200 colonies per plate. These
ranges were expected to shift as the concentrations of indicator bacteria decrease with time (Table 4-3).
Automatic samplers were programmed to collect two additional 1-L samples 15-20 min following the first
timed sample, allowing enough time for refilling of the effluent collection bottle. These samples served
as additional samples for TSS analysis and secondary indicator bacteria samples in the event of an error in
the first programmed sample collection.
Table 4-3. Expected Beginning Densities After Loading and Expected Dilution Factors
Indicator Organism Expected Density Dilution
(CFU/100 mL)
Retention Pond Mesocosm
Total Coliforms 103 - 105 102,103,104
Fecal Coliforms 103 - 105 102,103,104
E. coli 102- 104 IQ^IO^IO3
Enterococci 102- 104 lO^lO^O3
Constructed Wetland Mesocosm
Total Coliforms
Fecal Coliforms
E. coli
Enterococci
104- 106
104- 106
104- 106
103- 105
103,104,105
103,104,105
103,104,105
102,103,104
Sediment Sampling Procedures
Mesocosms were divided into four spatial areas to randomize sampling location and avoid spatial biases.
Four samples were collected from each section in the retention pond and constructed wetland systems.
Sediments were collected using a clear cylinder (modified syringe) inserted about 5 cm into the
undisturbed sediments and capped at one end. After removing the sediment core from the mesocosm, the
core was transferred to a centrifuge tube. Sediment samples were mixed using a sterile spatula. Samples
were then analyzed using multiple-well fermentation tests for enterococci and E. coli (and total coliforms)
using Enterolert® and Colilert®, respectively, manufactured by IDEXX Laboratories (Westbrook, ME).
This method was selected instead of membrane filtration because of the difficulty in counting colonies
with obstructing sediment and due to the potential of growth inhibition by accumulated sediments on the
filter paper.
Predation Sampling Procedures
Laboratory experiments have suggested that ingestion rates should increase with food concentration and
4-5
-------
that feeding preference should be shown for the most abundant of available foods. Analysis of larger
macroinvertebrates was the most practical method to determine the potential of predation on bacteria
populations, due to the time and cost of other methods. Identification and enumeration of
macroinvertebrates as small as 0.5 mm were done on 5 samples each from the retention pond and
constructed wetland to determine the populations of either bacteriovores (bacteria eating predators) or the
presence of macroinvertebrates. The guides Freshwater Macroinvertebrates of Northeastern North
America (Peckarsky et al., 1990) and An Image-Based Key to the Zooplankton of the Northeast
(University if New Hampshire, 2005) were used for identification of invertebrates. A top-down approach,
looking at the densities of macroinvertebrates including those from a higher trophic level (especially
bacteriovores), was used to qualitatively assess the bacteria concentrations based on the presence and
quantity of bacteria-eating organisms. The assumption is that an increase in the abundance of bacteria
would also lead to an increase in the predators that use bacteria as a major food source. Ideally, when
using this monitoring approach, one would take samples before, during, and after each storm event.
However, because of the enormous analyses requirements to identify and enumerate the
macroinvertebrates, as well as notice of recruitment of macroinvertebrates midway through the study,
macroinvertebrate data are limited to a single sampling event.
Data Management
The volume of data for each simulated storm event was large. Much of the continuous data was placed
into spreadsheets for geometric means and statistical calculations based on time and storm event. For the
purposes of comparing the physical and environmental characteristics data was averaged over the time
between each sample collection as noted in Table 4-2. Bacteria indicator organism concentration data
required some synthesis to calculate sample time averages and inactivation rates as noted in Chapter 3.
Since indicator bacteria samples were split into three samples and then three separate dilutions, it was
possible to have up to 12 indicator concentrations with each organism for each timed sample. This was
usually not the case as often one or more of the three dilutions would result in either values of zero or
TNTC (plates containing over 200 colonies) as designed to bracket the actual culturable colony-forming
units. All plates were enumerated, with the exception of those designated as TNTC. For each split
sample, dilutions with plates containing 1-200 organisms were used for data analysis. Each split sample
was averaged across dilutions. The three split samples for each sample collection time (total of 16 times)
were log transformed and regressed with time. Uncertainty was calculated as in Chapter 3. Each
simulated storm event produced more than 192 data points for indicator bacteria indicators.
Statistical Analyses
Sample sizes were large enough to perform statistical comparisons (ANOVA and correlation analyses)
between constructed wetland, retention pond treatment practices, dark and light control inactivation rates,
and physical and chemical characteristics. Likewise, sample size allowed for regression analyses of the
bacterial indicator concentrations on elapsed time between the constructed wetland, retention pond, and
dark and light controls for the each stormwater event. Slopes of the regressions gave inactivation rate
constants, K, for each event. Chemical parameters were compared between the treatments and controls
using correlation matrices. All statistical analyses that have significant p-values and a = 0.05 were noted.
Results
Physical and Chemical Properties of the Pilot-scale Systems
Physical and chemical parameters measured in the study are listed in Table 4-4. Water temperatures
averaged 2.15°C less in the constructed wetland compared to the retention pond. This difference was
4-6
-------
likely due to shading from the macrophytic vegetation (Typha latifolia, average stem density = 39.3
stems/m2). This temperature difference was more notable in the warmer sampling events (difference of
3.08°C and 1.82°C, respectively) compared to the colder sampling events (difference of 0.46°C).
DO was higher in the retention pond compared to the constructed wetland; the May, July, and September
events had the highest values. The process of decomposition of organic matter in the constructed wetland
can consume some of the DO, causing low concentrations (sometimes near 0 mg/L) during periods of
increased decomposition. Increases in decomposition are temperature-related with a positive correlation
between the two. Also, diurnal fluctuations in DO and temperature were reduced during initial storm
event loading. Values for these parameters did not generally reach pre-event diurnal fluctuations until
after 48 h of detention for most events. Conductivity was nearly the same in the two systems while pH
was circumneutral to alkaline in the retention pond but tended to be acidic in the constructed wetland.
This pattern was noted by Mitch and Gosselink (2000) in constructed wetlands with mineral soils and
lake sediments by Stumm and Morgan (1996). These differences were attributed to the organic matter
build-up in sediments and decomposition which tends to make the pH less than 7. The ORP was much
less (and often negative) in the retention pond compared to the constructed wetland system. The depth of
inundation of the free water in the retention pond was generally three times that of the constructed
wetland. This would substantially increase the potential for more reducing conditions through both
reduced oxygen diffusion with water depth and the lack of photosynthetic oxygen production with
absence of macrophytic vegetation, resulting in lower ORP values in the retention pond. Light intensity
in the constructed wetland was consistently 9-10% of that measured in the retention pond. The difference
in recorded hand-held light intensity for each event was used to adjust the measured irradiance for the
light control and retention pond to calculate a corrected irradiance expected at the surface of the
constructed wetland to make irradiance values between pilot-scale systems comparable.
Most storm events had maximum initial TSS and turbidity values of less than 100 mg/L and 150 NTUs,
respectively, after stormwater loading to the retention pond and constructed wetland. As expected,
turbidity and TSS values decreased with time in each system. Geometric mean turbidity values for
sampling events before October 2005 are shown in Figure 4-2. Turbidity values were averaged for each
time step and then over each sampling event.
Retention Pond
Constructed Wetland
I
60
50
40
30
20
10
0
0 10 20 30 40 50 60 70 80 90 100
Time (h)
60
50
40
30
20
10
10 20 30 40 50 60 70 80 90 100
Time (h)
Figure 4-2. Mean turbidity in the retention pond and constructed wetland in all storm
events except October 2005. Whiskers indicate 95% confidence intervals.
4-7
-------
Table 4-4. Average Event in-situ Physical and Chemical Results
Retention Pond
Valid
Date
Jun-04
Sep-04
Nov-04
May-05
Jul-05
Oct-05
Parameter
Temp (°C)
Cond (mS/cm)
DO (mg/L)
PH
ORP (mV)
Turbidity (NTU)
Irradiance (kJ/m )
Temp (°C)
Cond (mS/cm)
DO (mg/L)
PH
ORP (mV)
Turbidity (NTU)
Irradiance (kJ/m )
Temp (°C)
Cond (mS/cm)
DO (mg/L)
PH
ORP (mV)
Turbidity (NTU)
Irradiance (kJ/m )
Temp (°C)
Cond (mS/cm)
DO (mg/L)
PH
ORP (mV)
Turbidity (NTU)
Irradiance (kJ/m2)
Temp (°C)
Cond (mS/cm)
DO (mg/L)
PH
ORP (mV)
Turbidity (NTU)
Irradiance (kJ/m2)
Temp (°C)
Cond (mS/cm)
DO (mg/L)
PH
ORP (mV)
Turbidity (NTU)
Irradiance (kJ/m2)
N
36
36
36
36
36
36
12
36
36
36
36
36
36
12
42
42
42
42
42
42
14
33
33
33
33
Not
33
12
24
24
24
24
Not
24
8
48
48
48
48
Not
48
16
Mean
27.0
0.316
5.8
8.0
364
11.6
43.9
22.8
0.217
8.5
7.9
-288
11.0
31.6
11.1
0.189
9.5
7.2
-285
7.8
19.0
19.9
0.447
10.5
8.3
Minimum
21.9
0.296
1.9
7.3
200
5.1
<0.1
18.7
0.196
5.8
7.5
-354
7.9
<0.1
5.6
0.159
6.2
7.0
-322
4.0
<0.1
16.6
0.394
8.5
7.3
Maximum
32.2
0.345
11.4
9.1
465
31.1
139.1
27.3
0.237
11.5
8.3
-15
20.7
125.0
16.1
0.216
13.9
7.3
-211
11.5
84.4
27.0
0.555
15.6
9.3
Standard
Deviation
2.9
0.016
2.6
0.6
93
5.7
48.9
2.8
0.011
1.9
0.2
95
3.2
45.1
3.3
0.018
2.0
0.1
29
1.9
26.5
3.2
0.047
1.8
0.5
Standard
Error
0.5
0.00261
0.4
<0.1
15
0.9
14.1
0.5
0.002
0.3
<0.1
16
0.5
12.0
0.5
0.003
0.3
<0.1
4
0.3
7.1
0.6
0.008
0.3
0.1
Constructed Wetland
Valid
N
36
36
36
36
36
36
12
42
42
42
42
42
42
14
42
42
42
42
42
42
14
39
39
39
39
Mean Minimum
25.5 22.2
0.284 0.265
3.6 0.8
6.7 6.6
465 317
10.3 6.5
4.1 <0.1
19.4 16.1
0.204 0.190
4.1 0.7
6.4 6.3
555 524
6.6 3.7
3.4 <0.1
10.1 4.6
0.179 0.130
9.0 2.3
6.3 6.1
403 376
8.4 0.4
1.9 0.0
17.1 15.3
0.680 0.585
1.4 0.1
6.8 6.8
Maximum
29.4
0.316
9.2
7.0
529
23.8
12.8
22.1
0.237
13.4
6.7
591
18.5
11.5
15.9
0.208
14.3
6.8
452
27.6
7.8
22.3
0.771
4.4
7.0
Standard
Deviation
2.2
0.014
2.3
0.1
67
4.8
4.5
1.8
0.013
3.6
0.1
19
3.8
4.7
3.2
0.021
4.1
0.2
19
6.5
2.4
1.9
0.052
1.2
0.1
Standard
Error
0.4
0.002
0.4
<0.1
11
0.8
1.3
0.3
0.002
0.5
<0.1
3
0.6
1.3
0.5
0.003
0.6
<0.1
3
1.0
0.7
0.3
0.008
0.2
<0.1
Recorded
1.8
26.9
26.7
0.253
5.5
7.5
<0.1
<0.1
24.0
0.235
3.0
7.2
11.2
119.0
31.2
0.297
9.5
8.2
2.8
40.2
2.2
0.018
1.9
0.3
0.5
11.1
0.4
0.004
0.4
0.1
39
13
24
24
24
24
4.4 2.5
2.6 <0.1
24.5 23.1
0.259 0.197
1.3 0.1
6.4 6.2
12.0
17.4
27.1
0.371
3.7
6.6
2.4
5.6
1.3
0.054
1.3
0.1
0.4
1.6
0.3
0.011
0.3
<0.1
Recorded
26.0
29.4
14.6
0.156
3.2
7.0
6.5
<0.1
12.6
0.066
<0.1
6.1
92.0
119.0
17.6
0.240
6.3
7.5
20.2
40.3
1.6
0.048
1.6
0.4
4.1
14.3
0.2
0.007
0.2
0.1
24
8
48
48
48
48
61.0 13.5
2.7 <0.1
14.7 11.6
0.184 0.092
2.8 <0.1
6.0 5.8
167.2
10.9
18.4
0.273
9.1
6.5
50.9
3.7
1.7
0.046
3.0
0.1
10.4
1.3
0.3
0.007
0.4
<0.1
Recorded
849.1
18.7
<0.1
<0.1
1236.5
135.3
396.9
39.8
57.3
10.0
48
16
937.4 11.2
1.9 <0.1
2141.4
12.4
782.5
3.7
112.9
0.9
4-8
-------
The October 2005 experimental run had starting TSS values averaging 2,999 mg/L, and turbidity values
averaging 2,173 NTUs, which is near the maximum of the range expected for most storm water runoff.
Active construction in the watershed was evident in the stormwater runoff during this sampling event.
Inclusion of this stormwater runoff greatly increases the variability in solids concentration, overwhelming
the smaller concentrations found in the previous runoff events. Thus some analyses occurred with the
exclusion of this event as noted.
Bacteria Indicator Organisms
Samples were not analyzed for enterococci in the first experiment (June 2004) but resumed for all
subsequent events. Graphs showing the relationship of bacterial indicator organism concentrations for the
constructed wetland and the retention pond with event physical and chemical parameters can be found in
Figures 4-3 through 4-5. Conductivity and DO did not appear to significantly effect bacterial
concentrations over the ranges observed (Figure 4-3). ORP may have moderately affected fecal coliforms
and E. coli concentrations around 200 mV in the retention pond while densities of fecal coliforms, E. coli,
and enterococci decreased between 500 and 600 mV in the constructed wetland, but only three events
were monitored for this parameter. Densities of fecal coliforms and E. coli decreased above a pH of 8.5
in the retention pond but remained unaffected over the range of observed pH values in the constructed
wetland (Figure 4-4). Temperature appears to affect bacterial indicator organism concentrations in the
retention pond. The optimal temperature range that resulted in the greatest number of observed bacteria
colony forming units was between 11°C and 26°C. A similar trend was noticed in the constructed wetland
for a temperature range between 11° C and 23°C (Figure 4-5).
There was a distinct relationship with concentration of all indicator organisms with turbidity in this
experiment. While there was much scatter in concentrations at turbidities less than 20 NTU, at turbidities
greater than 100 NTUs there was a predictable increase in bacteria organism concentrations with
increasing turbidity in both the retention pond and constructed wetland (Figure 4-6). Similar results have
been shown by the United States Geological Survey (USGS) in larger rivers in northern and central
Virginia as well as by the USEPA in smaller streams in northern Virginia (Hyer and Moyer, 2003; Struck
et al, 2006). These solids can potentially affect rates of bacteria attenuation as discussed in Chapter 5.
Overall inactivation rates for all simulated storm events are shown in Table 4-5. Significant differences
were observed between the constructed wetland and retention pond in eight of the bacteria indicators for
the six runoff events. The retention pond had six of these with inactivation rates for: total coliforms in
June and July; E. coli in May; fecal coliforms in July and enterococci in May and November significantly
larger than in the constructed wetland. However, the constructed wetland had a significantly larger
inactivation rate compared to the retention pond for fecal coliforms in June and enterococci in July. Both
treatments had greater inactivation rates compared to light and dark controls in September and November
for total coliforms, E. coli, and fecal coliforms. Retention pond inactivation rates were also greater than
controls for total coliforms in June and July, for E. coli in May and July, and fecal coliforms in May,
June, and July while constructed wetlands rates were greater for E. coli in July, fecal coliforms in May
and June, and enterococci in July. Light controls were greater than dark controls in 9 instances, including
May and June for total coliforms and E. coli, July for fecal coliforms, and May, July, September, and
November for enterococci. This indicates that light does have an impact on bacteria indicator organisms.
Bacteria indicator concentrations decreased with time for the retention pond and constructed wetland
(Figure 4-7). The exponential regression coefficients indicate the inactivation rate. A two step process of
generating an overall inactivation value for each bacterial indicator organism from 0-50 h and from 50-
100 h was used to generate a best fit relationship. This timeframe was determined by maintaining R2
values of regressions greater than 0.70 while varying the time interval between 0 and 100 until the
difference in slope (inactivation rate) was maximized for the majority of the bacteria indicator organisms.
4-9
-------
This method used R2 values (Figure 4-7) of the whole timeframe compared to 0-50 and 50-150 h partial
timeframes. In all instances, the R2 values improved when dividing the duration of the experiment into
the two timeframes, suggesting that inactivation rates vary as a function of time with greater rates of
inactivation during the first 50 h timeframe compared to the second 100 h timeframe.
Table 4-5. Inactivation Rates for the Constructed Wetland, Retention Pond, and Dark and
Light Controls for all Indicator Bacteria Organisms for each Sampling Event
Retention Pond
Month
Total Fecal
Year Coliforms E. coli ColiformsEnterococci
Constructed Wetland
Total Fecal
Coliforms E. coli ColiformsEnterococci
(h1)
June
September
November
May
July
October
2004
2004
2004
2005
2005
2005
0.2419*+ 0.1484
0.144+
0.1653+
0.0949
0.1811*4
0.0437
0.1164+
0.1164+
0.3350*+
0.1957+
0.0524
Dark
0.1814+
0.1192+
0.1485+
0.1417+
0.2610*+
0.0566
Control
0
0
0
0
0
.2030
.1730*
.1717*
.1240
.0512
0.1529
0.1204+
0.1235+
0.1090
0.0733
0.0427
0
0
0
0
0
0
(h1)
.1651
,1204+
,1157+
.0919
,1894+
.0597
Light
0.3277*+
0.1515+
0.1137+
0.1233+
0.1025
0.0536
Control
0.1786
0.1245
0.0852
0.2112*+
0.0594
Month
Total Fecal
Year Coliforms E. coli ColiformsEnterococci
Total Fecal
Coliforms E. coli ColiformsEnterococci
(h1)
June
September
November
May
July
October
* Indicates a
+ Indicates a
* Indicates a
2004
2004
2004
2005
2005
2005
0.0247
0.0700
0.0815
0.0258
0.0637
0.0538
0.0276
0.0563
0.0480
0.0725
0.0509
0.0605
0.0249
0.0527
0.0445
0.0514
0.0619
0.0514
0.0773
0.0711
0.0351
0.0944
0.0194
0.1390*
0.0588
0.0658
0.0679*
0.0720
0.0676
(h1)
0.1502*
0.0789
0.0724
0.1158*
0.0712
0.0862
0
0
0
0
0.
0
.0242
.0760
.0692
.0828
1136*
.0822
0.
0.
0.
0.
0
2027*
1787*
0884*
1681*
.0316
significantly higher value between retention pond and constructed wetland
significantly higher value between retention pond or constructed wetland values and control values
significantly higher value between light
and dark control
values
Fecal coliforms and enterococci in the retention pond were an exception to this generalization. Several of
the inactivation rates during the 50-150 h timeframe had values nearing zero suggesting that these
organisms may have reached or nearly reached background concentrations after 50 h. This is supported
by the average pre-event background concentrations in the retention pond and constructed wetland found
in Table 4-6.
Bacteria Concentrations in Sediment
Results from the sediment bacteria indicator organism concentrations collected and analyzed one day
before and two days after the November 2004 storm event are shown in Table 4-7. Sediment bacteria
increased substantially for total coliforms and E. coli after the storm event. However, concentrations of
enterococci decreased somewhat over the experiment. The bacteria indicator organisms measured before
the experimental run are considered as background concentrations as the previous input of indicator
organisms through stormwater runoff was two months prior to this event.
4-10
-------
Retention Pond
Constructed Wetland
^ 5E+07
-J
S
O 5E+06
O
^H
p 5E+05
to
^ 5E+04
o
"S 5E+03
ra
^3
S 5E+02
o
s
rj 5E+01
O
Ml 5E+00
o
-J
o
0 o
0° 0
t°
o ^ ° o
^8 h r-E° D ° °
O ^ A '"v ^ O
4n 08£° i?
A
o o ° ° o o o
n ^2> ° n ° o A
^[jUtCD A O (3)
0 H A ° °
o S 0 a ^ o A
Bf? r °
cy & ^ n o
B SETA ^ AA a n
o A o ^ FCConc
A EntConc
0.1 0.2 0.3 0.4 0.5 0.6
Conductivity (mS/cm)
^^
"s
I
p
to
U,
e
o
'-^
^
•^
it
u
e
o
o
Ml
O
-J
5E+07
5E+06
5E+05
5E+04
5E+03
5E+02
5E+01
5E+00
o
o o o
A A °
ft A§, A» i °
°fn g^Sp ° * ° o °
n 4 8 o o* n 9
0 A O °
A n 8 80 o •
^V ^A0 ^ A ^ D" A
n A A n
i ^tJi ^ ES A A
XlA A
^o c/
3 n n A
A
DD
D
o TCConc
° ECConc
o FCConc
A EntConc
2 4 6 8 10 12 14 16
Dissolved Oxygen (mg/L)
Figure 4-3. Effluent concentration of indicator organisms with conductivity and dissolved oxygen.
4-11
-------
Retention Pond
Constructed Wetland
^ 5E+07
S
§ 5E+06
g 5E+05
U
^ 5E+04
£ 5E+03
o>
w 5E+02
o
^ 5E+01
S/j
^ 5E+00
-4
^J 5E+07
S
§ 5E+06
i-H
^ 5E+05
U
^ 5E+04
£ 5E+03
u 5E+02
C
O
^ 5E+01
S/j
o 5E+00
O
oo
o
2
A
A °
0°
o a
if
A O
O
n
o o
A
A
0
O 0
0 0
o *•
^ o
0
n
_^25
nu
>
o*
00 -200 0 200 400 6
ORP (mV)
- A
n
C
<3
1
C
5
*
o o
°qj
^A
™^8™
C
o £
"fl* <
n a
° 0
0 <>
§ a
° o
o ag^
QD ^
o
o
n
o
A
) O
1 o o
o
if 8
pa a
Eib 4
^ A 9 ^
O
0 C
3
0
n
o J
0
n ......
n
o
o o
n" "
5E+07
5E+06
5E+05
5E+04
5E+03
5E+02
o TCConc 5E+01
° ECConc
o FCConc 5E+0°
* EntConc
90 -4
5E+07
5E+06
5E+05
5E+04
5E+03
5E+02
o TCConc 5E+01
° ECConc
o FCConc 5E+00
A EntConc
n r
1
n r
^
0
to
o
o
P o
\ n
o
n
t
a n An
4°
^o o
O A
O i.
n
£0 Cf
3 o
0 0
n
a a
^ d
A
A
00
-200 0 200 400
ORP (mV)
~
3
£
A
O
§Q
j
0 0
I
'° 0 1
o
a
o
AB °
I8*i^
f|s|
DA
a
"n""Jn oo ^ " ^
A A °
A
_
A
A J*
n A
u a
a
° TCConc
° ECConc
« FCConc
* EntConc
600
o TCConc
° ECConc
o FCConc
A EntConc
6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5
pH
5.5 6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5
pH
Figure 4-4. Effluent concentrations of indicator organisms with ORP and pH.
4-12
-------
Retention Pond
Constructed Wetland
h-1
s
p
to
o
D
s
o
U
o
h-1
5E+08
5E+06
5E+05
5E+03
5E+02
5E+00
078—n~
B
5E+10
5E+09
5E+08
5E+05
5E+04
TCConc
ECConc 5E+01
FCConc 5E+OQ
EntConc
c
00
o o.
> ra n^
o
5
0
O
n »o
~£AS ^O O
A Q_ Q
-T3—H—o—a-
AID
-Ho_
0
10 15 20 25
Temperature (C°)
30
35
0
10 15 20 25
Temperature (C°)
30
o
o
O
h-1
Figure 4
5E+07
5E+06
5E+05
5E+04
5E+03
5E+02
5E+01
5E+00
TCConc = 1946.3288*exp(0.0046*x)
ECConc = 377.4225*exp(0.0058*x)
FCConc = 873.1594*exp(0.0053*x)
EntConc = 2127.9909*exp(0.0058*x)Q
5. Effluent concentrations of indicator organisms with temperature
5E+07
5E+06
5E+05
5E+04
5E+03
5E+02
5E+01
o
o
CM
O
O
to
o
o
00
o
o
o
o
o
CM
o
o
••fl-
. TCConc
ECConc
. FCConc
EntConc
5E+00
TCConc = 23532.7643*exp(0.0018*x)
ECConc = 2892.7463*exp(0.0026*x)
FCConc = 9072.4396*exp(0.0022*x)
EntConc = 7198.5976*exp(0.003*x)
o TCConc
n ECConc
o FCConc
i EntCon
35
o o
o o o
Turbidity (NTU)
Figure 4-6. Effluent indicator bacteria concentrations with in-situ turbidity.
ooooooooo
ooooooooo
OCM'fl'tOOOOCM'fl1
T-T-T-T-T-CMCMCM
to oo
Turbidity (NTU)
TCConc
ECConc
FCConc
EntConc
4-13
-------
Retention Pond
Constructed Wetland
hJ
S
o
o
to
U
u
o
U
M
O
h-1
»TC
• EC
AFC
oENT
50
100
150
50
100
150
Time (h)
Time (h)
Organism
TC
EC
FC
ENT
JVr 0-50hr
(h1)
-0
-0
-0
-0
.1308
.0534
.0440
.0044
0
0
R2
.895
.628
0.464
0
.765
JVr 50-150hr
(h1)
-0.
-0.
0
-0.
0044
0369
.0547
.0672
0
0
0
0
R2
.937
.876
.956
.937
Organism
TC
EC
FC
ENT
JVr 0-50hr
(h1)
-0.1164
-0.0853
-0.1797
-0.1142
R2
0.959
0.960
0.383
0.824
JVr 50-150hr
(h1)
-0.0270
-0.0361
-0.0298
-0.0180
R2
0.636
0.475
0.664
0.116
Figure 4-7. Indicator organism concentrations with time. Regressions fits are for time = 0-50 h and 50-150 h. Regression
coefficients (k-values) of the exponent (slope) are shown in the tables.
4-14
-------
Table 4-6. In-situ Indicator Organisms Average Background Concentrations
Indicator
Organism
Background Concentration
(CFU/100 mL ± Standard Error)
Retention Pond
Constructed Wetland
Total Coliforms
E. coli
Fecal Coliforms
Enterococci
1.39xl04±3.85xl03
6.42x10° ±7.22x10°
1.02xl04±2.55xl03
3.70x10^8.29x10°
3.37xl04±4.04xl03
2.55x10^6.21x10°
8.09xl03±1.12xl03
2.01x10^ 5.46x10°
Bacteria indicator organisms present in the sediments have the potential to be resuspended in the water
column with turbulent flow or disturbance, and may contribute to maintained or increased effluent
bacteria indicator organism concentrations in the future.
Predation
Table 4-8 shows the groups of organisms identified in the samples. The average number of organisms
enumerated (invertebrate density) in the constructed wetland was 777 and in the retention pond was 442
organisms. However, due to the high variance in the samples, this difference was not significant.
The difference in the number of organisms present (invertebrate taxa richness) was significant between
the constructed wetland and retention pond, averaging 15.75 organisms and 7.5 organisms, respectively.
These groups often had more than one species in a taxonomic group represented in each system as shown
in parenthesis. Ostrocods were not speciated below this subclass.
Table 4-7. Sediment Bacteria Indicator Organisms Sampled in November of 2004
Indicator Organism
Concentrations BEFORE
Stormwater Loading
Indicator Organism
Concentrations AFTER
Stormwater Loading
IIUIKHIUI
Organism
Total Coliforms
E. coli
Enterococci
Total Coliforms
E. coli
Enterococci
Retention Pond
2.25xl04
2.41xl05
>2.41xl05
8.67xl03
4-15
-------
Table 4-7. Macroinvertebrate Groups Identified in the Retention Pond and Constructed
Wetland
Retention Constructed
Taxa
Pond
Wetland
Oligochaetae
Chironomidae
Cladocerans
Coleoptera
Collembola
Copepoda
Ephemeroptera
Hemiptera
Hydracnidia
Ostrocoda
Rotifera
1
0
1479(3)
0
0
22(2)
42
0
0
186
39
231(3)
12
190(2)
9
1591(2)
269 (2)
0
8(2)
9(2)
782
1
*Parentheses indicate the number of taxa identified in that group
Cladocerans were the dominant species present both in richness (3 species) and in density (83%) in the
retention pond. Ostrocods (11%), ephemeroptera (2%), rotifers (2%), copepods (1%), and oligochaetes
(<1%) made up the rest of the composition within the retention pond. The dominant invertebrate species
in the constructed wetlands were collembola, ostrocods, copepods, and oligochaetes composing 51%,
25%, 9%, and 7%, respectively.
Scaling Consideration
Mesocosms have a history of use as a research tool for ecological studies of aquatic and terrestrial
ecosystems (Grice and Reeve, 1982; Odum, 1984; Lalli, 1990; Adey and Loveland, 1991; Beyers and
Odum, 1993; Kangas and Adey, 1996). They have been used in commercial scale applications, such as in
wastewater treatment, food production (Kangas and Adey, 1996), and in ecosystem restoration (Callaway
et a/., 1997). Use of mesocosms, particularly in wetland science, has become more common as a research
tool for use in studies of the fate and effect of pollutants, biogeochemical cycles, and the effects of
nutrients on ecosystem dynamics.
A given condition when considering comparing mesocosms natural ecosystems is that ecological
complexity is to some degree reduced or lost in microcosm or mesocosm studies depending on the size of
the mesocosms being used relative to large ecosystem-scale research. Scale can change nutrient cycling,
the number of trophic levels, number of species within trophic levels, habitat types, and connectivity
between habitats (Beyers and Odum, 1993). Because of this, some caution needs to be used when
extrapolating mesocosm results to larger systems. Once models created using mesocosms are validated in
the field, application of model results at a larger scale can be made.
Summary
The results highlight the varying influence environmental factors have on the inactivation of indicator
bacteria routed into constructed wetland and retention pond systems. The differences between these
treatment systems and light and dark controls are statistically measurable implying some variables may
have a greater influence than others. A detailed discussion of these results and their relation to the bench-
scale study are considered in Chapter 5.
4-16
-------
Chapter 5 Discussion of Results and Conclusions
Microbial contamination from fecal origins in stormwater runoff poses a risk to human health through the
consumption of drinking water and recreational and bathing contact with surface waters. Indicator
bacteria serve as the regulatory meter by which water quality is measured and standards must be met.
This chapter discusses the results of the bench and pilot-scale studies to evaluate the use of the first-order
decay function for predicting indicator bacteria concentrations in BMP effluent. It compares the two
studies to determine similarities and differences in inactivation rate constants, coefficients, and affects of
environmental conditions on bacterial indicators.
A primary factor affecting indicator bacteria, often understated in microorganism studies, is time. Time is
incorporated into every inactivation rate. By definition the inactivation rate is the change in
concentrations of indicator bacteria (presumably decrease) for a designated period. In both the bench-
and pilot-scale experiments, time was always a significant variable when evaluating the parameters
affecting indicator bacteria concentrations. Many types of BMPs increase the storage time and decrease
flow velocities as a primary mechanism of operation.
Results from the bench-scale (Chapter 3) and pilot-scale (Chapter 4) studies show environmental
conditions affect indicator bacteria concentrations in retention ponds and constructed wetlands.
Temperature, sunlight, and salinity, were investigated in these studies. Other environmental factors
typical of constructed wetlands and retention ponds were also considered.
Effects of Temperature
Generally, the results from the bench-scale and pilot-scale experiments agree with the literature (e.g.,
Easton et al., 2005; Ferguson et al., 2003; Geldreich et al., 1968; Medema et al., 1997; and Canteras et
al., 1995) emphasizing temperature generally demonstrates an increase in the calculated inactivation rate
constant with increasing temperature. Similarly, inactivation rates are lower at lower temperatures. This
trend was most notable in the pilot-study during the October 2005 sampling event. Selvakumar et al.
(2004) noted that concentrations of organisms did not change significantly when the samples were stored
at 4°C beyond the standard holding time of 24 h. Geldreich et al. (1968) noted that organism persistence
remained at higher levels at 10°C compared to 20°C. In the pilot-scale experiment the optimal
temperature range for growth (as indicated by overall indicator bacteria concentrations) was similar to
values reported in the literature with indicator concentrations increasing with temperature, reaching a
maximum concentration from 20°-25°C in both the retention pond and constructed wetland. Medema et
al. (1997) found that inactivation was faster at 15°C than at 5°C. Canteras et al. (1995) noted a clear
positive correlation between inactivation and temperature. In their study, when test conditions were at
10°C, 36 h was necessary to reduce the population of E. coli to 10% of the original as opposed to 8.4 h at
5-1
-------
42°C. Greater inactivation was also noticed in the range between 10 and 18°C than between 18 and 42°C.
The pilot-scale study found indicator organism concentrations were much greater over the first 50 h as
compared to the following 100 h. Although over a longer period of time, Easton et al. (2005) reported
that the 0-7 day inactivation rates were much larger than the 7-21 day rates. It may be possible that
during the 0-7 days studied by Easton et al. (2005) there have been varying rates of die-off In the pilot-
scale study, by increasing the temporal resolution of the first 150 h compared to seven days (168 h), the
changes in bacteria inactivation could be more easily observed.
Effect of Sunlight/Light Intensity
Many studies have shown that sunlight is an important factor in bacteria indicator inactivation (Sinton et
al., 1994; Canteras et al., 1995). The bench-scale study supported this by showing that the effect of light
on the overall decay coefficient was substantial, especially for non-coliform bacteria. If, for example,
ambient light levels are 100 mW/cm2, then the light increases the E. coll inactivation rate constant by 0.25
h"1' which is a six-fold increase over the dark value. Similarly, the pilot-scale study shows statistically
lower inactivation rate constants in the dark control compared to the light control for total coliforms and
E. coll for the months of May and June (Table 4-5). According to Table 4-4, June had the greatest
irradiance of any of the dates sampled while May ranked fourth in light intensity (because of cloudy
conditions during the experiment). Enterococci showed the greatest difference in inactivation rate
constants between light and dark controls followed by total coliforms and E .coll. The primary difference
between these controls was the exposure to sunlight. These differences in rate constants, up to 0.12 h"1
for enterococci and E. coll, are substantial and are also supported by the inactivation rates observed in the
bench-scale study for total coliforms, fecal coliforms, fecal streptococci, enterococci, and E. coll indicator
bacteria exposed to the highest light intensity.
Effects of Sedimentation, Sorption, and Filtration
Sedimentation, sorption, and filtration processes are generally accepted as the dominant mechanism for
the removal of solids and other sediment-related stressors such as heavy metals. The settling velocity has
been used as an approximation of the overall rate constant due to these factors in stormwater treatment
systems (Wong and Geiger, 1997). Because the particle settling velocities are related to the grading,
shape, and density of the particles entering the system, settling velocities measured in the laboratory can
only serve as an indicator of the rate constant for sedimentation. Other environmental factors such as
non-ideal flow conditions would be expected to increase solids in the water column through resuspension,
while some posit that higher density vegetation can increase the rate of the settling constant (Wong and
Geiger, 1997).
The pilot-scale experiment compared the effluent concentration of a constructed wetland and a retention
pond. While treatment volume was quite different between the two systems, the major difference
between these systems was the presence of vegetation. Results comparing the TSS and turbidity in the
two systems indicate that the constructed wetland and retention pond showed little difference in turbidity
and effluent TSS. However, settling velocity appears to be greater in the constructed wetland under the
higher (>100 NTUs) sediment loading observed in October 2005.
The difference in indicator bacteria concentrations and the inactivation rate constants between the
constructed wetland and retention pond in our study support settling as a contributing but not primary
factor in bacterial inactivation. With the overall differences in turbidity and TSS between the constructed
wetland and retention pond, relatively small for most of the simulated storm events (Figure 4-2), it is
probable a large portion of the influent may have been unassociated (free) or associated with very fine
particles. Similarly, the effects of settling may be artificially small as an artifact of the manner in which
5-2
-------
the enriched storm water is created (described in Chapter 4). The smaller particles would result in a longer
time necessary to settle in these systems and was likely to occur within the duration of the experiment.
Another possibility is that the increased solids characterized in the influent (especially for October 2005)
may reduce the effect of other environmental variables. Increased particulates may occlude light
penetration or prevent predation by bacteriovores through limiting access and harboring the indicator
bacteria that are agglomerated to these solids.
Wong and Geiger (1997) suggest, when selecting an appropriate AT value for sedimentation, filtration, and
sorption using the settling velocity of the fiftieth percentile sediment grade with adjustments for increased
effectiveness for wetlands having higher vegetation density. However, as experienced in this study, this
may not adequately predict the effluent concentrations of stormwater runoff passing through passive
treatment systems if bacteria are either unassociated with settleable particles or if they are associated with
the fine particle fraction, i.e., less than 2 um in size (Davies and Bavor, 2000).
Effect of Salinity
Salinity was only assessed in the bench-scale study and was not included in the pilot-scale study. The
bench-scale study results indicate that different organisms exhibited different trends at varying salinity
concentrations. Overall, the effect of increased salinity at the tested concentrations was small. The
calculated value of cDs was not generally significant from non-saline controls. This suggests that, for the
span of salinity values studied, the added salt has little effect on the inactivation rate constant and
supports the results reported by Canteras et al. (1995) who found the largest salinity effect occurs when
the salinity values were over 35 ppt. Thoman and Mueller (1987) reported that the inactivation of fecal
coliforms is generally much faster in marine and estuarine waters than in freshwater. Mancini (1978)
indicated that components in seawater in addition to salt may be responsible for inactivation in seawater.
Salinity is less often a factor in most BMPs but it is a consideration when stormwater controls are to be
placed in the coastal and estuarine environments, or when a BMP receives runoff from areas treated with
road salts.
Effect of Predation
Previous research has suggested bacterivory can significantly reduce indicator bacteria organism
concentrations (Green et al., 1997; Mandi et al, 1993; Decamp and Warren, 1998; Pretorius, 1962;
Fernandez et al, 1992a; and Troussellier et al, 1986). There are a variety of invertebrates present in the
constructed wetland and retention ponds in this study. While the retention pond and constructed wetland
had different taxa represented in each of these systems, dominant invertebrates in both systems have been
shown to consume large quantities of indicator bacteria depending on the population size of both
predators and prey. The major difference in species richness between the systems was the retention pond
was dominated by cladocerans while the constructed wetland had populations of oligochaete, collembola,
copepod, and ostrocod invertebrates. The difference in predatory effects of the dominant species on
indicator bacteria concentrations in each system is not known. Characteristics of the constructed wetland
and retention pond do suggest why there may be different invertebrate communities between the different
systems. Constructed wetlands generally had taxa that are associated with greater organic matter (derived
from the macrophytic vegetation) (i.e., oligochaetes, collembola, copepods, and ostrocods) (Peckarsky et
al, 1990). Collembola and ostrocods are reported to feed on detritus algae, fungi, and dead animal matter
with collembolan having special mouthparts for consuming the surface film or underlying bacterial
populations (Peckarsky et al, 1990). Therefore, their numerical importance in the constructed wetlands
was not surprising. Cladocerans and copepods can affect bacterial populations in both wetlands and open
water systems. Both taxa have been shown to consume greater than 25% of the bacterial populations in
near shore areas of lakes (Heath et al, 1999). The association of collembolans with detrital organic
5-3
-------
matter and cladocerans with more open water habitat may explain the relative densities of these
invertebrates in both the retention pond and constructed wetland environments.
The identification and enumeration of the taxa did not provide quantifiable results to determine the
predatory contribution of invertebrates on indicator bacteria concentrations. However it does provide
anecdotal evidence that invertebrates may contribute to the reduction of indicator bacteria in natural
systems. The pilot-scale study, while not directly measuring bacteria indicator inactivation rates due to
predation, included the overall effects of predation by incorporating this effect into the cumulative
inactivation rate as discussed in the section addressing collective environmental factors below.
Effect of Other Potential Factors
The many other factors (i.e., DO, pH, conductivity, oxidation reduction potential) that can contribute to
inactivation rates of indicator bacteria were not directly assessed in this report. The bench-scale study
addressed DO in a peripheral manner while the pilot-scale study did not address other potential factors
individually but grouped their effects into the overall inactivation rates discussed as collective
environmental factors in the next section.
Inactivation Rates Due to Collective Environmental Factors
The data on inactivation rates for microorganisms in stormwater and effects of natural factors on survival
rates are limited except for one study by Geldreich et al. (1968). They reported a decay rate of 0.061 h"1
at 20°C for fecal coliforms. The inactivation rate constant of fecal coliforms at 20°C obtained in the
bench-scale study (0.047±0.031 h"1) was similar to Geldreich et al. (1968) and the 0-50 h and 50-150 h
rates for fecal coliforms observed in the retention pond in the pilot-scale study.
The bench-scale results observed showed a good relationship with the first-order equation. In this portion
of the study, total coliforms had a much slower inactivation rate than other indicator organisms.
Traditional indicators (total coliforms and fecal coliforms) had lower inactivation rates than the alternate
indicators (E. coll and enterococci) suggesting that use of traditional indicators may tend to predict higher
concentrations compared to the alternate indicators. Depending on the stressor(s) for which the BMP is
designed, this could affect the necessary retention time calculated when designing BMPs.
When correlation analysis was done for chemical and physical parameters with overall inactivation rates
in the pilot-scale study, conductivity with E. coll and enterococci with DO were significantly correlated in
the retention pond. The constructed wetland had no significant physical or chemical correlations with
inactivation rates. Conductivity of the in situ water may be a surrogate for total dissolved solids.
However, standard methods suggest the relationship is not constant (APHA et al., 1998). The
relationship between total dissolved solids and conductivity is a function of the type and nature of the
dissolved cations and anions in the water (i.e., the ability of the water to carry a charge). Some total
dissolved solids measuring devices measure the conductivity of the water with the assumption that the
primary dissolved minerals are either a combination of NaCl or KC1. Other anions and cations, such as
sodium sulfate, sodium bicarbonate, or possibly some organic molecules with ionic and cationic charges
can contribute to the conductivity in water samples suggesting total dissolved solids, while not directly
measured in the experiment, may be correlated with E. coli concentrations in the retention pond if other
mineral or organic compounds are present.
The pilot-scale study generally followed the first-order rate equation. A jackknife relationship showing a
different rate constant for the first 50 h compared to longer periods, as in Figure 4-7, was appropriate for
some indicator bacteria. This relationship was also observed by Thomann and Mueller (1987) for bacteria
5-4
-------
distributions in rivers with resistant strains. In addition, with indicator species concentrations (Figure 4-
7) having an average predicted background concentration of lO'-lO4 organisms/100 mL and in situ
background concentrations ranging from 6.42xlO°-3.37xl04 organisms/ 100 mL, there is reasonable
support for changes to the first-order rate equation in wetland and retention pond BMPs. Kadlec and
Knight (1996) suggest that because of residual indicator bacteria populations present in wetlands, bacteria
removal efficiency is a function of the inflow bacteria concentrations. Removal efficiency typically is
higher at high inflow concentrations, but declines to low or negative values when inflow concentrations
are lower than the in situ bacteria production rates. However, during periods when influent flow rates are
turbulent, causing resuspension of the previously settled solids, removal efficiency may not depend on
influent concentrations alone. Because these settled sediments are associated with in situ bacteria
populations, there may be an increase in effluent concentration of indicator bacteria with turbulent or high
flow or when sediments are disturbed by other means (i.e., waterfowl, muskrats, etc.) compared to the
influent concentration. Similarly, sediment resuspension may be more likely to occur in wetland and
retention pond BMPs that are poorly designed, have reached the design life, or are not maintained and
may contribute to lower or negative indicator bacteria inactivation rates (and removal efficiencies).
Evaluation of the First-Order Decay Equation
Recalling that one of the primary objectives of this research was to evaluate the first-order decay equation
for predicting bacteria indicator concentrations affected by environmental conditions, a bench-scale study
to look at selected variables identified in the literature as important to inactivation rates was developed.
With the bench-scale information as the primer, the pilot-scale experiments utilized controlled
mesocosms to further develop the possible effects of typical environmental conditions (similar to
expected field BMP conditions) have on indicator bacteria concentrations in BMP effluent. Both the
pilot-scale study and the bench-scale study demonstrated the first-first order decay function adequately
models indicator bacteria concentrations in the short term. However, during longer periods, the first-
order decay equation may not apply to effluent from these types of BMPs. Literature has reported that the
assumptions for a first-order decay function (i.e., steady flow conditions) may seldom be met in studies
concerning storm water runoff in constructed wetlands and retention ponds (Wong and Geiger, 1997).
Other researchers have suggested using surface area based models for wetlands constructed for the
treatment of wastewater (Kadlec and Knight, 1996). One of these models is known as the K-C* model
which incorporates a concentration term, C*, that represents the background concentration often present
in natural systems. The formula is:
Where: Cout = effluent concentration; Cin = influent concentration; C* = background concentration; K =
rate constant for the water quality parameter being treated based on time of detention.
However, Wong and Geiger (1997) point out that the stochastic nature of stormwater-related systems
introduces significantly different system functions compared to wastewater treatment. These authors
formulate a procedure that incorporates the use of the K-C* model and the interaction between the
requirements for wetland storage for inflow stochasticity and stormwater treatment.
They recommend an adaptation of the Kadlec and Knight's K-C* model with the formula:
Cout=C*+(Cin-C*)e-KAIQ (5-2)
5-5
-------
Where: Cout = effluent water quality target; Cin = influent event mean concentration; C* = background
concentration; K = rate constant for the water quality parameter being treated; A = constructed wetland or
retention pond area; and Q = steady state flow.
It should be noted that equations 5-1 and 5-2 have attempted to incorporate conditions that meet the
assumptions for the first-order decay equation or include environmental realities such as background
concentrations of indicator organisms (C*) to improve prediction of this stressor. The rate constant K,
which governs inactivation rate determinations in the first-order decay equation, is the only means of
incorporating environmental variables to better predict effluent concentrations in surface water models.
The bench and pilot studies discussed in this report have estimated inactivation rate constants for
indicator bacteria. Further, constant coefficients have also been estimated to predict the effect that
environmental factors have on overall indicator bacteria inactivation rates.
Returning to Khatiwada and Polprasert (1999) and Canteras et al. (1995) equations from Chapter 3, the
following formula for overall inactivation rate constant is proposed:
Koveran=K203>TT-20+3>lI + Kf+Kp (5-3)
Where: KoveraU= overall inactivation rate constant; K2o = inactivation rate constant due to temperature at
20°C; cDT = temperature coefficient; \ = light proportionality coefficient; / = light intensity (mW/cm2); Kf
= inactivation rate constant due to other factors such as sorption, filtration, and sedimentation; and Kp =
inactivation rate constant due to predation.
Temperature and light could be quantified through the combination of both studies in this report. Salinity
was found to have little effect on bacteria indicators in the constructed wetland and retention pond
systems used in this study. Due to the inability to separate sorption, sedimentation, predation, and other
environmental factors in the study, substituting the variable Kother instead of Kf and Kp to include these
(and other) processes in one variable is proposed. As a result, the inactivation rate formula from above
can be written as:
K^^K^/^+^I + K^ (5-4)
Where definitions are as above and Kother = inactivation rate constant due to other factors such as sorption,
filtration, sedimentation, predation, pH, DO, conductivity, oxidation reduction potential, etc.
Table 5-1 lists coefficients for light and temperature from the bench-scale study and the light and dark
controls of the field-scale study. Inactivation rates at 20°C (K20) obtained in the field study are larger than
the values obtained from bench-scale experiments except for E. coll. Temperature coefficients (cDT)
obtained from both bench- and pilot-scale studies are not statistically significant different. Light
proportionality coefficients (cDi) obtained from pilot-scale study are much larger than the values obtained
from bench-scale study supporting the conclusion that natural sunlight has a much larger effect on
inactivation rates compared to artificially induced light.
Using the light and dark control inactivation rates, the inactivation rate due to other parameters was
calculated as the measured Kught ^temperature value. Subtracting the Ktemperature value from the Kught ^temperature,
results in a calculated KUght value. The Kother value was then calculated by subtracting Kught and Ktemperature
from the KmeraU value that was measured for the retention pond and constructed wetland. All K values for
the retention pond and constructed wetlands can be found in Tables 5-2 and 5-3, respectively.
5-6
-------
Table 5-1. Inactivation Rate Coefficients from Batch and Field Studies
Indicator u
f~. . J\20\" )
Organism ^ '
TC
FC
EC
ENT
0.016±0.009*
0.042±0.030*
0.036±0.018*
0.042±0.014*
Batch Study
L (cm2/
mW-h)
0.0016
0.0130
0.0025
0.0076
^(h1)
0.066±0.007*
0.053±0.006*
0.057±0.008*
0.054±0.006*
Field Study
L (cm2/
mW-h)
0.0092
0.0047
0.0022
0.0070
Coefficient is statistically significant at a=0.05.
Table 5-2. Retention Pond Overall, Temperature, Sunlight, and Other Factors
Indicator
Organism
TC
FC
Month/Year
June 2004
September 2004
November 2004
May 2005
July 2005
October 2005
Annual Average
June 2004
September 2004
November 2004
May 2005
July 2005
October 2005
Coverall
(measured)
0.242
0.144
0.165*
0.095
0.181
0.044*
0.145
0.181
0.119
0.149*
0.142
0.261
0.057*
if
-"•temp
(measured)
0.025
0.070*
0.082*
0.026
0.064*
0.054*
0.054
0.025
0.053*
0.045*
0.051*
0.062*
0.051*
Klight
(calculated)
(h1)
0.114
-0.011*
-0.016*
0.042*
0.008*
0.014*
0.025
-0.001
0.023*
0.024*
0.033
0.052*
0.031*
Rate Coefficients
Bothers
(calculated)
0.103
0.085
0.010
0.027
0.109
-0.024
0.052
0.157
0.043
0.079
0.059
0.147
-0.025
EC
ENT
Annual Average 0.152
0.048
0.027
Annual Average 0.161
0.053
0.043
Annual Average 0.145
Coefficient is statistically significant at a=0.05
0.059
0.075
0.077
June 2004
September 2004
November 2004
May 2005
July 2005
October 2005
0.148
0.116
0.116*
0.335*
0.196
0.052*
0.028
0.056*
0.048*
0.073
0.051*
0.061*
0.122
0.023*
0.024*
0.043
0.020*
0.025*
-0.002
0.038
0.044
0.219
0.125
-0.034
0.065
September 2004
November 2004
May 2005
July 2005
October 2005
0.203
0.173*
0.172
0.124
0.051
0.077*
0.071*
0.035*
0.094*
0.019*
0.126*
0.108*
0.053*
0.074*
0.013
0.001
-0.006
0.083
-0.044
0.019
0.011
The
temperature
value for the constructed wetland was not directly measured but could be calculated.
To calculate a the Kught +temperature value for the constructed wetland the KUght value from the light control
was multiplied by the weighted average of light intensity expected at the surface of the constructed
5-7
-------
wetland. The weighted average was calculated as 10% of the light that reached the retention pond surface
(based on light meter measurements) for six hours out of 24 hours of effective light exposure, multiplied
by 18 hours out of 24 hours in which the exposure was relatively the same as in the retention pond. This
resulted in a multiplication factor of 0.775 •KUght of the retention pond. All negative calculated values
were assumed to be a propagation of error and are therefore expected to be within the range of error for
the respective inactivation rate constant.
Table 5-3. Constructed Wetland Overall, Temperature, Sunlight, and Other Factors Rate
Coefficients
Indicator
Organism
TC
FC
EC
ENT
Month/Year
June 2004
September 2004
November 2004
May 2005
July 2005
October 2005
Annual Average
June 2004
September 2004
November 2004
May 2005
July 2005
October 2005
Annual Average
June 2004
September 2004
November 2004
May 2005
July 2005
October 2005
Annual Average
September 2004
November 2004
May 2005
July 2005
October 2005
Annual Average
Coverall
(measured)
0.153
0.120
0.124*
0.109
0.073
0.043*
0.104
0.328
0.152
0.114*
0.123
0.103
0.054*
0.146
0.165
0.120*
0.116*
0.092*
0.189*
0.060*
0.124
0.179*
0.125*
0.085*
0.211*
0.059*
0.132
A temp
(measured)
0.025
0.070*
0.083*
0.026
0.064*
0.054*
0.054
0.025
0.053*
0.045*
0.051*
0.062*
0.051*
0.048
0.028
0.056*
0.048*
0.073
0.051*
0.061*
0.053
0.077*
0.071*
0.035*
0.094*
0.019*
0.059
Kught
(calculated)
(h-1)
0.088
-0.009*
-0.013*
0.033*
0.006*
0.011*
0.019
-0.001
0.018*
0.019*
0.025
0.040*
0.024*
0.021
0.095
0.018*
0.019*
0.033
0.016*
0.019*
0.033
0.098*
0.084*
0.041*
0.057*
0.010
0.058
Bothers
(calculated)
0.040
0.059
0.054
0.050
0.003
-0.022
0.031
0.304
0.081
0.050
0.047
0.001
-0.021
0.077
0.042
0.046
0.049
-0.014
0.123
-0.020
0.038
0.004
-0.030
0.009
0.060
0.030
0.015
* Coefficient is statistically significant at a=0.05
Inactivation rate constants vary throughout the year based on different affects of the environmental
factors. In general, the combination of other factors had the greatest effect on inactivation rates in the
retention pond for the indicator bacteria evaluated in this study. Enterococci were an exception to this.
Temperature was the found to be more important than light, however light is still a significant factor and
should be considered when using the first-order equation. In the constructed wetland, temperature had the
greatest effect on inactivation rates for the selected indicator bacteria. Other factors had a greater
influence on inactivation rates for all organisms except for enterococci, where light appeared to be as
5-S
-------
important as temperature. Inactivation rates for fecal coliforms were most affected by other factors
followed by temperature and then light.
Application of the inactivation rate constants found in Tables 5-2 and 5-3 can provide an overall
inactivation rate constant incorporating temperature, light intensity, and a lumped factor for other
environmental variables. The overall inactivation rate constant can be applied to equation 5-2 to
determine the required area necessary to achieve WQSs when designing constructed wetland or retention
ponds. If a background concentration is determined, the overall rate constant can also be applied to
equations 5-1 and 5-2 to predict effluent concentrations. The first-order decay equation is most accurate
when used with the inactivation rates and uncertainties in short-term models to predict stormwater runoff
effluent quality from constructed wetland and retention pond BMPs to improve or prevent further
degradation of water quality. Longer-term modeling would benefit from applying separate inactivation
rates for periods immediately following stormwater runoff and periods unaffected by stormwater runoff.
Conclusions
The studies in this report demonstrated that the concentration of the tested indicator organisms decrease
exponentially with time. The first-order decay process reasonably models the concentration time series
for shorter durations. The first-order decay model is a simple and efficient means of predicting indicator
bacteria concentrations in stormwater runoff effluent from BMPs such as retention ponds and constructed
wetlands. Results from the studies discussed in this report provide new data on inactivation rate constants
coefficients, and uncertainties used in this equation. The factors of light, time, and temperature influence
processes in all retention ponds and wetlands constructed to mitigate the effects of stormwater runoff on
the receiving waters. A combination of other factors (e.g., predation, sedimentation, sorption, filtration,
pH, BOD, pH, and DO) can also contribute to the inactivation of indicator bacteria in constructed wetland
and retention pond BMPs. Reliable rates, coefficients, and the uncertainties expected in the reported
values will add to the accuracy of surface water quality models. Water quality models are a primary tool
for evaluating permit applications (e.g., NPDES) and have an important regulatory role in developing
TMDL allocations. These models should incorporate the affects of BMPs to better model their potential
for improving water quality in the watershed. The incorporation of simple reliable models is an important
step in assuring that the models used in determining bacterial TMDL loading and allocations meet the
state of the science.
BMPs were originally designed to control runoff volumes and rates by attenuating the flow. The
attenuation increases the time between the rainfall-generated runoff and the water reaching the receiving
water. The time lag serves to reduce the concentration of these indicator organisms. Structural BMPs
then can be effective in reducing indicator bacteria concentrations contained in stormwater runoff. Low
inactivation rates may occur in BMPs where inflow bacterial concentrations are lower than the in situ
bacteria productions rates, or turbulent flow through the BMPs causes resuspension of sediments.
Quantitative microbial partitioning estimates can represent critical inputs in areas where sedimentation is
a primary mode of indicator organism inactivation when modeling the location and severity of impaired
waters. The lack of reliable partitioning information currently leads most surface water modeling efforts
to assume that microbes exist in the free phase. The presumption of only free-phase organisms biases
model results to greater dispersion and shorter microbial longevity. However, from the results obtained
from this project, factors such as temperature and light intensity have been shown to be as, or more
important to, indicator bacteria inactivation rates. This would suggest that when attempting to mitigate
bacteria in runoff, watershed managers should construct BMPs to maximize the temperature increase
from solar exposure. Similarly, the added effects of light, even at constant temperature, can increase
inactivation rates, improving BMP performance. The extent to which shading in constructed wetlands,
due to vegetation or the deeper water of retention ponds, attenuates the effect of incident light will vary
5-9
-------
with runoff and in situ water properties (e.g., turbidity, color) in the BMP. It is also important to
recognize that bacteria loading seldom acts as a single environmental stressor of concern. The watershed
manager must consider the effects of the increased effluent temperature on the receiving waters,
particularly when the receiving water is a low-order cold water stream. Also, the results from this study
suggest that the regulatory indicator selected can influence BMP design. The apparent insensitivity of
coliforms to light levels suggests that the shading effects may be reduced when this is selected as the
water quality indicator. When the monitored indicator organism is E. coll or enterococci, the effect of
light would be expected to be greater than for coliforms.
It is accepted that placement of appropriate BMPs in watersheds can lead to improvements in receiving
water quality by reducing the overall load of pollutants to receiving waters. If watershed managers can
reduce microbial loads in waterbodies using the range of possible BMPs, verification of these stormwater
management tools will help MS4 Phase I and Phase II communities reduce microbial loadings and meet
requirements set out by the TMDL process. Long-term microbial load reductions will improve the overall
water quality and could potentially lead to increased consumption of fish and shellfish, increased use of
recreational waters, reduced beach closures, and increased protection of source water used as drinking
water sources.
The limitations of BMP effectiveness in reducing bacterial loading to WQS must be recognized. In most
natural treatment systems there will be an irreducible concentration that is often maintained in system
sediments. Dilution of BMP effluent likely plays a significant role in attaining WQS in receiving water.
However, elimination of bacteria indicators may require chemical treatment. In addition, overall
effectiveness and efficiencies of BMPs hinge on proper design and maintenance of these systems.
5-10
-------
Chapter 6 References
Alkan, U., D.J. Elliott, and L.M. Evison. 1995. Survival of Enteric Bacteria in Relation to Simulated
Solar Radiation and Other Environmental Factors in Marine Waters. Water Research, Vol. 29, No. 9, pp.
2071-2081.
American Public Health Association, American Water Works Association, and Water Environment
Federation. 1998. Standard Methods for the Examination of Water and Wastewater. 20th Edition.
Anderson, I.C., M. Rhodes, and H. Kator. 1979. Sublethal Stress in Escherichia coli: A Function of
Salinity. Applied and Environmental Microbiology, pp. 1147-1152, December.
Ashbolt, N.J., W.O.K. Grabow, and M. Snozzi. 2001. Indicators of Microbial Water Quality, in Water
Quality: Guidelines, Standards, and Health. Risk Assessment and Management for Water-Related
Infectious Disease, Chapter 13, Fewtrell, L. and J. Bartnam, Eds. World Health Organization.
Axelrood, P.E., A.M. Clarke, R.A. Radley, and S.J.V. Zemcov. 1996. Douglas-fir root-associated
microorganisms with inhibitory activity towards fungal plant pathogens and human bacterial pathogens.
Canadian Journal of Microbiology, Vol. 42, pp. 690-700.
Barzily, A. and Y. Kott. 1991. Survival of pathogenic bacteria in an adverse environment. Water
Science and Technology, Vol. 24, No. 2, pp. 395-400.
Baudisova, D. 1997. Evaluation of Escherichia coli as the Main Indicator of Faecal Pollution. Water
Science and Technology, Vol. 35, No. 11-12, pp. 333-336.
Bavor, H.J., D.J. Roser, and S. McKersie. 1987. Nutrient removal using shallow lagoon solid matrix
macrophyte systems. In: Aquatic Plants for Water Treatment and Resource Recovery, Reddy, K.R. and
W.H. Smith, (editors). Magnolia Publishing, Orlando, FL., pp. 227-235.
Brix, H. 1997. Do macrophytes play a role in constructed treatment wetlands? Water Science and
Technology, Vol. 35, No. 5, pp. 11-17.
Brunner, G.W. 2001. HEC- RAS River Analysis System Hydraulic Reference Manual Version 3.0, U.S.
Army Corps of Engineers, Hydrologic Engineering Center.
Burton, G.A. Jr., D. Gunnison, and G.R. Lanza. 1987. Survival of pathogenic bacteria in various
6-1
-------
freshwater sediments. Applied and Environmental Microbiology, Vol. 53, pp. 633-638.
Burton, G.A., Jr. and R.E. Pitt. 2001. Stormwater Effects Handbook: A Toolbox for Watershed
Managers, Scientists, and Engineers. CRC Press, Inc., Boca Raton, FL.
Canteras, J.C., J.A. Juanes, L. Perez, and K.N. Koev. 1995. Modeling the Coliforms Inactivation Rates
in the Cantabrian Sea (Bay of Biscay) from Insitu and Laboratory Determinations of T90. Water Science
and Technology, Vol. 32, No. 2, pp. 37-44.
Chamberlin, C.E. and R. Mitchell. 1978. A decay model for enteric bacteria in natural waters, pp. 325-
348. In R. Mitchell (ed.), Water Pollution Microbiology, Vol. 2. John Wiley & Sons, New York.
Characklis, G.W., M.J. Dilts, O.D. Simmons III, C.A. Likirdopulos, L.H. Krometis, and M.D. Sobsey.
2005. Microbial partitioning to settleable particles in stormwater. Water Research, Vol. 39, pp. 1773-
1782.
Clar, M., B.J. Barfield, and S. Yu. 2003. Considerations in the design of treatment best management
practices (BMPs) to improve water quality. EPA/600/R-03/103. U.S. Environmental Protection Agency,
Office of Research and Development, Cincinnati, OH.
Crabill, C., R. Donald, J. Snelling, R. Foust and G. Southam. 1999. The Impact of Sediment Fecal
Coliform Reservoirs on Seasonal Water Quality in Oak Creek, Arizona. Water Research, Vol. 33, No. 9,
pp. 2163-2171.
Curriero, F.C., J.A. Patz, J.B. Rose and S. Lele. 2001. The Association between Extreme Precipitation
and Waterborne Disease Outbreaks in the United States, 1948-1994. American Journal of Public Health,
Vol. 91, No. 8, pp. 1194-1199.
Davies-Colley, R.J., A.M. Donnison, D.J. Speed and C.M. Ross. 1999. Inactivation of Faecal Indicator
Micro-Organisms in Waste Stabilization Ponds: Interactions of Environmental Factors with Sunlight.
Water Research, Vol. 33, No. 5, pp. 1220-1230.
Davies, C.M. and H.J. Bavor. 2000. The Fate of Stormwater-Associated Bacteria in Constructed
Wetland and Water Pollution Control Pond Systems. Journal of Applied Microbiology, Vol. 89, pp. 349-
360.
Decamp, O. and A. Warren. 1998. Bacterivory in ciliates isolated from constructed wetlands (reef beds)
used for wastewater treatment. Water Research, Vol. 32, pp. 1989-1996.
Dutka, B.J. and K.K. Kwan. 1980. Bacterial Die-off and Stream Transport Studies. Water Research,
Vol. 14, pp. 909-915.
Easton, J.H., J.J. Gauthier, M.M. Lalor, and R.E. Pitt. 2005. Die-off of pathogenic E. coli O157:H7 in
sewage contaminated waters. Journal of the American Water Resources Association, Vol. 41, No. 5, pp
1187-1193.
Eisenstark, A. 1971. Mutagenic and lethal effects of visible and near-ultraviolet light on bacterial cells.
Advances in Genetics, Vol. 16, pp. 167-198.
Ferguson, C., A.M. De Roda Husman, N. Altavilla, D. Deere, and N. Ashbolt. 2003. Fate and Transport
of Surface Water Pathogens in Watersheds. Critical Reviews in Environmental Science and Technology,
6-2
-------
Vol. 33, No. 3, pp. 299-361.
Fernandez, A., C. Tejedor, and A. Chordi. 1992a. Influence of pH on the elimination of faecal coliform
bacteria in waste stabilization ponds. Water, Air and Soil Pollution, Vol. 63, No. 3-4, pp. 317-320.
Fernandez, A., C. Tejedor, and A. Chordi. 1992b. Effect of different factors on the die-off of faecal
bacteria in a stabilization pond purification plant. Water Research, Vol. 26, No. 8, pp. 1093-1098.
Ferrara, R.A. and D.R.F. Harleman. 1981. Dynamic nutrient cycle model for waste stabilization ponds.
Journal of Environmental Engineering, Vol. 107 (EE4), pp. 817-830.
Francy, D.S., D.R. Helsel, and R.A. Nally. 2000. Occurrence and distribution of microbiological
indicators in groundwater and stream water. Water Environment Research, Vol. 72, No. 2, pp. 152-161.
Fujioka, R.S., H.H. Hashimoto, E.B. Siwak, and R.H.F. Young. 1981. Effect of Sunlight on Survival of
Indicator Bacteria in Seawater. Applied and Environmental Microbiology, pp. 690-696.
Gameson, A.L. and D.J. Gould. 1975. Effects of Solar Radiation on the Mortality of Some Terrestrial
Bacteria in Sea Water. In: Discharge of Sewage from Sea Outfalls, Oxford, Pergamen, pp. 209-219.
Gannon, J.J., M.K. Busse, and J.E. Schillinger. 1983. Fecal coliform disappearance in a river
impoundment. Water Research, Vol. 17, No. 11, pp. 1595-1601.
Garcia, M. and E. Becares. 1997. Bacterial removal in three pilot-scale wastewater treatment systems for
rural areas. Water Science and Technology, Vol. 35, No. 5, pp. 197-200.
Geldreich, E.E., L.C. Best, B.A. Kenner, and D.J. Van Donsel. 1968. The Bacteriological Aspects of
Storm water Pollution. Journal of Water Pollution Control Federation, Vol. 40, No. 11, pp. 1861-1872.
Gerba, C.P. and G.E. Schaiberger. 1975. Effects of particles on virus survival in seawater. Journal of
Water Pollution Control Federation, Vol. 47, pp. 93-103.
Gersberg, R.M., R. Brenner, S.R. Lyon, and B.V. Elkins. 1987. Survival of bacteria and viruses in
municipal wastewater applied to artificial wetlands. In: Aquatic Plants for Water Treatment and Resource
Recovery, Reddy, K.R. and W.H. Smith, (editors). Magnolia Publishing, Orlando, FL., pp. 227-235.
Gersberg, R.M., R.A. Gearheart, and M. Ives. 1989a. Pathogen removal in constructed wetlands. In
Constructed Wetlands for Wastewater Treatment. Municipal, Industrial, and Agricultural, ed. D.A.
Hammer, pp. 431-446. Lewis Publishers Inc., Michigan, USA.
Gersberg, R.M., S.R. Lyon, R. Brenner, and B.V. Elkins. 1989b. Integrated wastewater treatment using
artificial wetlands: a gravel marsh case study. In Constructed Wetlands for Wastewater Treatment.
Municipal, Industrial, and Agricultural, ed. D.A. Hammer, pp. 145-152. Lewis Publishers Inc., Michigan,
USA.
Glenne, B. 1984. Simulation of water pollution generation and abatement on suburban watersheds.
Water Resources Bulletin, Vol. 20, No. 2, pp. 211-217.
6-3
-------
Gray, D. 2000. Qualitative Review of Epidemiology Studies, in Regional BEACH Program.
Proceedings: U.S. Environmental Protection Agency, Office of Water, EPA 823-R-00-003, 43-45.
http://www.epa.gov/waterscience/beaches/proceedings/Beach-99.pdf
Green, M.B., P. Griffin, J.K. Seabridge, and D. Dhobie. 1997. Removal of bacteria in subsurface flow
wetlands. Water Science and Technology, Vol. 35, No. 5, pp. 109-116.
Greenway, M., P. Dale and H. Chapman. 2003. An assessment of mosquito breeding and control in four
surface flow wetlands in tropical-subtropical Australia. Water Science and Technology, Vol. 48, pp.
249-256.
Haile, R.W., J.S. Witte, M. Gold. 1999. The Health Effects of Swimming in Ocean Water Contaminated
by Storm Drain Runoff. Epidemiology, Vol. 10, No. 4, pp. 355-363.
Hanes, N.B. and R. Fragala. 1967. Effect of Seawater Concentration on Survival of Indicator Bacteria.
Journal of Water Pollution Control Federation, Vol. 39, pp. 97-104.
Heath, R.T., R. Sturtevant, and P. Enflo. 1999. Modeling the Effects of Nutrient Concentrations on
Ecosystem Stability: Framework for a Great Lakes Model. IAGLR 99 - Modeling Summit.
Howell, J.M., M.S. Coyne, and P.L. Cornelius. 1996. Effect of sediment particle size and temperature on
fecal bacteria mortality rates and fecal coliform/fecal streptococci ratio. Journal of Environmental
Quality, Vol. 25, pp. 1216-1220.
Hyer, K.E., and D.L. Moyer. 2003. Patterns and sources of fecal coliform bacteria in three streams in
Virginia, 1999-2000: U.S. Geological Survey Water-Resources Investigations Report 03-4115, 76 p.
Jamieson, R.C., R.J. Gordon, K.E. Sharpies, G.W. Stratton and A. Madani. 2002. Movement and
Persistence of Fecal Bacteria in Agricultural Soils and Subsurface Drainage Water - A Review. Canadian
Biosystems Engineering.
Kadlec, RH. and R.L. Knight. 1996. Treatment Wetlands. CRC Press, Inc., Boca Raton, FL.
Kenner, B.A. 1978. Fecal Streptococcal Indicators, G. Berg, Ed., Indicators of Viruses in Water and
Food, Ann Arbor Science Pub., Ann Arbor, MI, pp. 147-169.
Khatiwada, N.R. and C. Polprasert. 1999. Kinetics of Fecal Coliform in Constructed Wetlands. Water
Science and Technology, Vol. 40, No. 3, pp. 109-116.
Kickuth, R. and G. Kaitzis. 1975. Mikrobizid wirksame aromaten aus Scirpus lacustris L. Umweltschutz
4-5, 134-135
Klock, J.W. 1971. Survival of coliform bacteria in wastewater lagoons. Journal of Water Pollution
Control Federation, Vol. 50, No. 11, pp. 2971-2083.
Lee, R.W. and W. Rast. 1997. Light Attenuation in a Shallow, Turbid Reservoir, Lake Houston,
Texas. U.S.G.S. Water Resources Investigations Report 97-4064.
Maestre, A. and R. Pitt. 2005. The National Stormwater Quality Databases, Version 1.1- a Compilation
and Analysis of NPDES Stormwater Monitoring Information. Tuscaloosa, AL 3594, The University of
6-4
-------
Alabama.
Mallin, M.A. 1998. Land-use practices and fecal coliform pollution of coastal water. Proceedings of the
14th Annual On-Site Wastewater Treatment Conference, North Carolina State University, Raleigh, NC.
Mallin, M.A., K.E. Williams, B.C. Esham, and R.P. Lowe. 2000. Effect of human development on
bacteriological water quality in coastal watersheds. Ecological Applications, Vol. 10, No. 4, pp. 1047-
1056.
Mandi, L., N. Ouazzani, K. Bouhoum, and A. Boussaid. 1993. Wastewater treatment by stabilization
ponds with and without macrophytes under arid climate. Water Science and Technology, Vol. 28, No. 10,
pp. 177-181.
Mara, D.D. and H.W. Pearson. 1986. Artificial freshwater environments: waste stabilization ponds. In
Biotechnology, ed. W. Schoernborn, pp. 177-206.
Mara, D.D., G.P. Alabaster, H.W. Pearson, and S.W. Mills. 1992a. Waste stabilization ponds: a design
manual for eastern Africa. Lagoon Technology International, Leeds.
Mara, D.D., S.W. Mills, H.W. Pearson, and G.P. Alabaster. 1992b. Waste stabilization ponds: a viable
alternative for small community treatment systems. Journal Institute of Water and Environmental
Management, Vol. 6, No. 1, pp. 72-78.
Marais, G.v.R. and V.A. Shaw. 1961. A rational theory for the design of sewage stabilization ponds in
central and South Africa. Transactions of South African Institute of Civil Engineering, Vol. 13, No. 11,
pp. 205-227.
Maryland Department of the Environment (MDE). 2000. Maryland Stormwater Design Manual Volumes
I & II. Center for Watershed Protection and Maryland Department of the Environment, Baltimore, MD.
http://www.mde.state.md.us.
Medema, G.J., M. Bahar, and P.M. Schets. 1997. Survival of Cryptosporidiumparvum, Escherichia coli,
Faecal Enterococci, and Clostridium perfringens in River Water: Influence of Temperature and
Autochthonous Microorganisms. Water Science and Technology, Vol. 35, No. 11-12, pp. 249-252.
Mezrioui, N., K. Oufdou, and B. Baleux. 1995a. Dynamics of non-01 Vibrio cholerae and fecal
coliforms in experimental stabilization ponds in the arid region of Marrakesh, Morocco, and the effect of
pH, temperature, and sunlight on their experimental survival. Canadian Journal of Microbiology, Vol.
41, No. 6, pp. 489-498.
Mezrioui, N., B. Baleux, and M. Troussellier. 1995b. A microcosm study of the survival of Escherichia
coli and Salmonella typhimurium in brackish water. Water Research, Vol. 29, No. 2, pp. 459-465.
Mitchell, R. and C. Chamberlin. 1978. Survival of Indicator Organisms, G. Berg, Ed., Indicators of
Viruses in Water and Food, Ann Arbor Science Pub., Ann Arbor, MI, pp. 15-37.
Mitsch, W.J. and J.G. Gosselink. 2000. Wetlands. John Wiley & Sons, Inc., New York.
MWLAP. 1992. Best Management Practices to Protect Water Quality, Ministry of Water, Land and Air
Protection, Government of British Columbia.
http://wlapwww.gov.bc.ca/wat/wq/nps/BMP_Compendium/BMP_Introduction/bmphome.htm.
6-5
-------
Nix, P.G., M.M. Daykin, and K.L. Vilkas. 1994. Fecal pollution events reconstructed and sources
identified using a sediment bag grid. Water Environment Research, Vol. 66, No. 6, pp. 814-818.
Noble, R.T., J.H. Dorsey, M. Leecaster, V. Orozco-Borbon, D. Reid, K. Schiff, and S.B. Weisberg. 2000.
A regional survey of the microbiological water quality along the shoreline of the Southern California
Bight. Environmental Monitoring and Assessment, Vol. 64, pp. 435-447.
Oliveri, V.P., C.W. Kruse, K. Kawata, and J.E. Smith. 1977. Microorganisms in Urban Stormwater.
EPA-600/2-77-087. Office of Research and Development, US Environmental Protection Agency,
Cincinnati, OH.
Ottova, V., J. Balcarova, and J. Vymazal. 1997. Microbial characteristics of constructed wetlands.
Water Science and Technology, Vol. 35, No. 5, pp. 117-123.
Pearson, H.W., D.D. Mara, S.W. Mills, and D.J. Smallman. 1987a. Factors determining algal
populations in waste stabilization ponds and the influence of algae on pond performance. Water Science
and Technology, Vol. 19, No. 12, pp. 131-140.
Pearson, H.W., D.D. Mara, S.W. Mills, and D.J. Smallman. 1987b. Physico-chemical parameters
influencing faecal bacterial survival in waste stabilization ponds. Water Science and Technology, Vol.
19, No. 12, pp. 145-152.
Peckarsky, B. L., P. Fraissinet, M. A. Penton, and D. J. Conklin, Jr. 1990. Freshwater macroinvertebrates
of Northeastern North America. Cornell University Press, Ithaca, NY. 442 pp.
Perdek, J.M., R.D. Arnone, M.K. Stinson, and M.E. Tuccillo. 2003. Managing Urban Watershed
Pathogen Contamination. U.S. Environmental Protection Agency, Office of Research and Development,
EPA 600-R-03-111, September.
Pommepuy, M., J.F. Guillaud, E. Dupray, A. Derrien, F. Le Guyader, and M. Cormier. 1992. Enteric
Bacteria Survival Factors. Water Science and Technology, Vol. 25, No. 12, pp. 93-103.
Polprasert, C., M.G. Dissanayake, and N.C. Thanh. 1983. Bacterial die-off kinetics in waste stabilization
ponds. Journal of Water Pollution Control Federation, Vol. 55, No. 3, pp. 285-296.
Pretorius, W.A. 1962. Some observations on the role of coliphage in the number of E. coli in oxidative
ponds. Journal of Hygiene, Vol. 60, pp. 210.
Pundsack, J., R. Axler, R. Hicks, J. Henneck, D. Nordman, and B. McCarthy. 2001. Seasonal pathogen
removal by alternative on-site wastewater treatment systems. Water Environment Research, Vol. 73, No.
2, pp. 204-212.
Qureshi, A.A. and B.J. Dutka. 1979. Microbiological studies on the quality of urban stormwater runoff
in Southern Ontario, Canada. Water Research, Vol. 13, pp. 977-985.
Rose, J.B., P.R Epstein, E.K. Lipp, B.H. Sherman, S.M. Bernard, and J.A. Patz. 2001. Climate
variability and change in the United States: potential impacts of water- and food-borne diseases caused by
microbiologic agents. Environmental Health Perspectives, Vol. 109, No. 2, pp. 211-219.
6-6
-------
Schillinger, J.E., and J.J. Gannon. 1985. Bacterial adsorption and suspended particles in urban
stormwater. Journal of Water Pollution Control Federation, Vol. 57, pp. 384-389.
Selvakumar, A., M. Borst, M. Boner, and P. Mallon. 2004. Effects of Sample Holding Time on
Concentrations of Microorganisms in Water Samples. Water Environment Research, Vol. 76, No. 1, pp.
67-72.
Selvakumar, A., and M. Borst. 2006. Variation of microorganism concentrations in urban stormwater
runoff with land use and seasons. Journal of Water and Health, Vol. 4, No. 1, pp. 109-124.
Sherer, B.M., J.R. Miner, J.A. Moore, and J.C. Buckhouse. 1992. Indicator bacterial survival in stream
sediments. Journal of Environmental Quality, Vol. 21, pp. 591-595.
Sinton, L.W., R.J. Davies-Colley, and R.G. Bell. 1994. Inactivation of Enterococci and Fecal Coliforms
from Sewage and Meatworks Effluents in Seawater Chambers. Applied and Environmental
Microbiology, Vol. 60, No. 6, pp. 2040-2048.
Sinton, L.W., R.K. Finlay, and P. Lynch. 1999. Sunlight inactivation of fecal bacteriophages and
bacteria in sewage-polluted seawater. Applied and Environmental Microbiology, Vol. 65, No. 8, pp.
3605-3613.
Solic, M. and N. Krstulovic. 1992. Separate and Combined Effects of Solar Radiation, Temperature,
Salinity, and pH on the Survival of Faecal Coliforms in Seawater. Marine Pollution Bulletin, Vol. 24, pp.
411-416.
Stumm, W. and J.J. Morgan. 1996. Aquatic Chemistry. John Wiley & Sons, Inc., New York.
Struck, S.D., C.N. Nietch, and M. Borst. 2004. Constructed wetlands vs. retention pond BMPs:
Mesocosm studies for improved pollutant management in urban stormwater treatment. Critical
Transitions in Water and Environmental Resources Management - Proceedings of the 2004 World Water
and Environmental Resources Congress.
Struck, S.D., A. Selvakumar, K. Hyer, and T. O'Connor. 2006. Evaluating the Accotink Creek
Restoration Project for Improving Water Quality, In-Stream Habitat, and Bank Stability. Examining the
Confluence of Environmental and Water Concerns - Proceedings of the 2006 World Water and
Environmental Resources Congress.
Sunen, E., and M.D. Sobsey. 1999. Detection of human enteric viruses in hardshell clams (mercenaria)
by RT-PCR. Journal of VirologicalMethods, Vol. 77, pp. 179-187.
Taylor, J.R. 1997. An Introduction to Error Analysis - The Study of Uncertainties in Physical
Measurements. Second Edition. University of Science Books, Sausalito, California.
Taylor, A. and T. Wong. 2002. Nonstructural Stormwater Quality Best Management Practices - An
Overview of their Use, Cost, and Evaluation. Technical Report 2/11. Cooperative Research Center for
Catchment Hydrology, Victoria, NZ.
Thomann, R.V. and J.A. Mueller. 1987. Principles of Surface Water Quality Modeling and Control.
Harper & Row, Publishers, Inc., New York, NY.
Troussellier, M., P. Legendre, and B. Baleux. 1986. Modeling of the evolution of bacterial densities in
6-7
-------
an eutrophic ecosystem (sewage lagoons). Microbial Ecology, Vol. 12, pp. 355-379.
USEPA. 1974. Analysis and Control of Thermal Pollution. Water Program Operations. EPA-430/I-74-
010.
USEPA. 1976. Quality Criteria for Water, EPA-440/9-76-023, Office of Water, Washington, DC.
USEPA. 1986. Ambient water quality criteria for bacteria- 1986. EPA 440/5-84-002. Office of
Water, Washington, D.C.
USEPA. 1992. NPDES Storm Water Sampling Guidance Document. EPA 833-92-001. Office of Water,
Washington, D.C.
USEPA. 1993. Handbook-Urban Runoff Pollution Prevention and Control Planning. EPA-625-R-93-
004. Washington, D.C.
USEPA. 1999. Preliminary Data Summary on Urban Stormwater Best Management practices. EPA-
821-R-99-012. Office of Water, Washington, D.C.
USEPA. 2002. National Water Quality Inventory 2000 Report. EPA-841-R-02-001. Office of Water,
Washington, D.C.
USEPA. 2004. National Section 303(d) List Fact Sheet.
http ://oaspub .epa.gov/waters/national_rept.control.
USEPA. 1983. Results of the Nationwide Urban Runoff Program. Volume 1 - Final Report. Water
Planning Division, Washington, DC. December NTIS Publication (83): 185552.
Wanielista, M.P. 1977. Nonpoint source effects on water quality. Journal of Water Pollution Control
Federation, Vol. 49, pp. 441-451.
Weibel, S.R., R.J. Anderson, and R.L. Woodward. 1964. Urban land runoff as a factor in stream
pollution. Journal of Water Pollution Control Federation, Vol. 36, pp. 914-924.
Wong, T.H.F. and W.F. Geiger. 1997. Adaptation of wastewater surface flow wetland formulae for
application in constructed Stormwater wetlands. Ecological Engineering, Vol. 9, pp. 187-202.
Yan, L., W. Chang, and M.L. Huang. 2000. Natural Disinfection of Wastewater in Marine Outfall
Fields. Water Research, Vol. 34, No. 3, pp.743-750.
Yukselen, M.A., B. Calli, O. Gokyay, and A. Saatchi. 2003. Inactivation of Coliform Bacteria in Black
Sea Waters Due to Solar Radiation. Environment International, Vol. 29, pp. 45-50.
Young, K.D. and E.L. Thackston. 1999. Housing density and bacterial loading in urban streams.
Journal of Environmental Engineering, Vol. 125, No. 12, pp. 1177-1180.
YSI Inc. 2005. Environmental Monitoring Systems Manual 6-Series.
(http://www.ysi.com/extranet/EPGKL.nsf/)
Software:
6-8
-------
An Image-Based Key To The Zooplankton Of The Northeast (USA) Version 2.0, Copyright0 2003-2005
Center for Freshwater Biology, Department of Zoology, University of New Hampshire, Durham, NH
03824 USA
SigmaStat for Windows Version 2.03, Copyright0 1992-1997 SPSS Inc.
Microsoft® Office Excel 2003, Copyright 1985-2003 Microsoft Corporation.
StatSoft, Inc. 2003. STATIST!CA (data analysis software system), version 6. www.statsoft.com.
6-9
-------
Appendix A Growth of Indicator Bacteria for Pilot-Scale Research
Introduction
To overcome the stochastic nature in stormwater runoff, researchers often employ the use of "synthetic
stormwater", a laboratory-produced stormwater matrix of known pollutant concentrations, in their
research. Literature indicates synthetic stormwater has been used most often for evaluating BMP
performance for the removal of heavy metals and total organic carbon (TOC) (Driscoll et al, 1990; Liu et
al. 2001; Liu et al. 2001) but has also been used in nutrient (nitrogen and phosphorus) (Davis et al., 1998;
Kim et al, 2003), chemical oxygen demand (potassium salt), and total organic carbon (motor oil)
(Passman and Yu, 2001; Hong et al. 2006) studies, as well. However, the authors could not find any wet
weather flow studies that used bacterial indicator-based synthetic stormwater to evaluate the performance
of BMPs. Consequently, the researchers developed a method for creating a stormwater source that has a
known concentration of common indicator organisms (i.e., total coliforms, fecal coliforms, E. coll, and
Enterococci) and contains genotypic representation (assumed, not measured) typical of bacterial
indicators found in urban stormwater for use in evaluating pilot-scale stormwater BMPs.
Methods
Stormwater runoff from the Middlesex County College (9.75 acres) campus was collected in a 11,300 L
tank from an outfall near the Urban Watershed Research Facility (UWRF) in Edison, NJ and stored on
site (Figure 3-1). After thorough mixing, a 60 mL aliquot of stormwater was collected for each sampling
event with 30 mL each placed in a flask containing 1L of tryptic soy (TS) broth (for coliforms and E. coli)
and 1L of brain heart infusion (BHI) broth (for enterococci). A stir bar was added to each flask. The TS
broth flask was then placed on a stir plate in a 37°C incubator while the BHI flask was placed on a stir
plate in a 41°C incubator. Concentrations of bacteria in the inoculated stormwater (broth-stormwater
mixture) were measured daily via membrane filtration using Standard Methods 9222B, 9222D, 9222G,
and 9230C as described in Chapter 3. Methods for positive controls followed manufacturer's
specifications (Becton Dickinson and Company, Sparks, MD).
Trend analysis of the concentration results from five- and seven-day sampling periods determined the
growth curves of indicator bacteria populations. Regression analyses of the growth phases for each
organism were used to determine the rate of growth. Rate of growth, in general, was grouped by days of
positive growth versus days of no or negative growth. Results allowed for future prediction of bacterial
concentrations so that experimental use of indicator bacteria could occur at the time of greatest bacterial
concentration. A measured volume of inoculated stormwater could then be combined with the desired
A-l
-------
volume of original storm water runoff for use in BMP performance studies. The inoculated aliquot and
stormwater was mixed, forming the "synthetic stormwater", and distributed to pilot scale BMPs for
performance evaluation.
Results and Discussion
Average starting concentrations on day zero of total coliforms, fecal coliforms, E. coll, and enterococci
were 2.3 x 105, 1.7 x 104, 5.8 x 103, 3.1 x 103 CFU/100 mL, respectively. All indicator organism
concentrations reached a maximum on day two and then declined (Figure A-2). Enterococci
concentrations declined at a much slower rate with day two and day five concentrations very similar
(Figure A-2; Table A-l). Starting concentrations of fecal coliform were one order of magnitude lower
and E. coll and enterococci starting concentrations were two orders of magnitude lower than the
concentrations of total coliforms. After the second day all indicator bacteria concentrations except for
enterococci were within the same order of magnitude (~109 CFU/100 mL). Enterococci concentrations
remained greater than 109 CFU/100 mL until the seventh (last) day of the experiment.
TC
Day
Figure A-l. Mean indicator bacteria concentrations by day. Vertical bars denote 95%
confidence intervals.
A-2
-------
Table A-l. Growth Rates of Bacterial Indicators Based on Regression Analyses
Growth Rate: Growth Rate:
Indicator Organism Days 0-2 (R2) Days 2-7 (R2)
Total Coliforms 2.565 log (0.922) -0.297 log (0.903)
Fecal Coliforms 3.106 log (0.916) -0.373 log (0.877)
Kcoli 3.356 log (0.913) -0.434 log (0.920)
Enterococci 3.321 log (0.704) -0.087 log (0.823)
Conclusion
The methods for developing indicator bacteria-rich synthetic stormwater in this laboratory experiment can
be useful for applications that specifically require stormwater that is characterized by abundant bacterial
indicator concentrations. Bacterial loading can be closely predicted using growth rate curves. A wide
range of bacterial loading capabilities can be achieved by using this approach to stormwater research.
One must recognize that the relative proportion of indicator bacteria will change from the original
stormwater and may be less representative of the true microbial community structure. Likewise, the
particle association of the indicator bacteria is suspected to be different with less bacteria associated with
larger particles and more bacteria associated with finer particles or unattached as opposed to the initial
stormwater runoff. However, it is believed that this property results in a more conservative estimate for
the indicator organisms when using synthetic stormwater for indicator bacterial loading experiments as
fewer colonies settle out of the water column.
Based on indicator bacteria concentrations from the first sample collection after spiking, there is some
bacterial die-off that occurs when diluting original stormwater with the inoculated stormwater. A
potential cause of this die-off may be thermal shock, although equilibration procedures attempt to
minimize the temperature changes between the collected stormwater and the laboratory grown inoculated
stormwater.
The synthetic stormwater runoff produced in this project has been effectively used in pilot studies for
assessing the performance of small scale controlled stormwater wetland and retention pond BMPs.
Citations
Bavor, H.J., D.J. Roser, and S. McKersie. 1987. Nutrient removal using shallow lagoon solid matrix
macrophyte systems. In: Aquatic Plants for Water Treatment and Resource Recovery, Reddy, K.R. and
W.H. Smith, (editors). Magnolia Publishing, Orlando, FL. pp. 227-235.
Driscoll, E.D., P.E. Shelley, and E.W. Strecker. 1988. Pollutant Loadings and Impacts from Highway
Stormwater Runoff, FHWA-RD-88-006, -008.
Passman, E. A. and S. L. Yu. 2001. "Comparison of Pollutant Removal Performance of Wetland
Vegetation." In American Society of Civil Engineers EWRI World Water and Environmental Resources
Congress in Orlando, FL. Washington, DC.
Gersberg, R.M., R. Brenner, SR. Lyon, and B.V. Elkins. 1987. Survival of bacteria and viruses in
municipal wastewater applied to artificial wetlands. In: Aquatic Plants for Water Treatment and Resource
A-3
-------
Recovery, Reddy, K.R. and W.H. Smith, (editors). Magnolia Publishing, Orlando, FL. pp. 227-235.
Hong, E., E.A. Seagren, and A.P. Davis. 2006. Sustainable Oil and Grease Removal from Synthetic
Stormwater Runoff Using Bench-Scale Bioretention Studies. Water Environment Research, Vol. 78, No.
2. pp. 141-155.
Kim, H., E.A. Seagren, and A.P. Davis. 2003. "Engineered bioretention for removal of nitrate from
Stormwater runoff." Water Environment Research, Vol. 75, No. 4, pp. 355-367.
A-4
------- |