Interactions between a Stormwater Pathogen Indicator, Escherichia coli, and Various Bacteria

Present in an Engineered Infiltration System

Claire Wayner

2

Table of Contents

Preliminary Matters 2-3

Key Words 2

Abbreviations and Acronyms 2

Acknowledgements 2-3

Paper Body 4-19

Introduction 4-6

Materials and Methods 6-9

Results 9-13

Discussion 14-16

Conclusions 16-17

References 17-19

Bibliography 19

Key Words

Engineered infiltration system (EIS), pathogen aquatic pollution, stormwater, biofilm, Escherichia coli

Abbreviations and Acronyms

DNA = deoxyribose nucleic acid

E. coli = Escherichia coli

EIS = engineered infiltration system

P. aeruginosa = Pseudomonas aeruginosa

PCR = polymerase chain reaction

SSS = synthetic stormwater solution

Acknowledgements

I conducted my research project as part of the Preheim Lab Group in the Johns Hopkins

University Department of Environmental Health and Engineering. I’d like to thank my mentor, Dr.

Sarah Preheim, for supporting my research efforts and for being a patient teacher. I’d also like to thank

the other members of the Preheim Lab Group, particularly Andrea Fraser, Yue Zhang, and Dr. Eric

3

Sakowski. The Preheim Lab Group is investigating the biology of EIS in a broader context, but I

designed and conducted my original research project, reported in this paper, independently. Although I

received advice and assistance, all the work displayed in this paper has been a product of my data

collection.

I would also like to thank my high school instructor and mentor, Mrs. Lisa Fridman, for being

such an excellent advisor and my family for supporting my research goals.

4

Introduction

Stormwater runoff has become a salient problem in recent years, as urbanization has increased

the amount of impervious surfaces present, in turn increasing the contamination of the stormwater. As

runoff flows over paved surfaces and enters waterways, it accumulates a variety of pollutants, such as

extra sediment, pesticides, bacterial pathogens, and excess nutrients. These biological pollutants,

primarily pathogens such as Escherichia coli (E. coli) and fecal coliform bacteria, render stormwater

unusable by humans and reduce the safety and quality of urban streams.

Because of the negative impacts

associated with increased stormwater runoff

entering water systems, various solutions have

been engineered to filter runoff and remove

contaminants before contact with the larger

watershed. One such solution that has become

popular in urban areas is an engineered

infiltration system (EIS), a depression filled

with a sediment media, such as sand, that

collects water during storms (see Figure 1),

allowing the water to slowly permeate into the

groundwater supply while removing large

particulate matter and biological pathogens. The

goal of EIS is to improve the quality of

stormwater while minimizing the impact on the hydrology of the larger landscape.

In addition to its ability to reduce the volume of runoff and remove physical contaminants, EIS

has also been shown to be capable of removing multiple varieties of biological contaminants. Pathogenic

bacteria, reported to be the most common contaminant that prevents stormwater from being reused, have

been of particular interest in recent research because of their frequency in stormwater and their

association with human disease, especially E. coli, one of the more common pathogen indicators used in

research. One study revealed that EIS can remove up to 82% of E. coli cells from synthetic stormwater

[1]. Another study explained these high retention rates by purporting that E. coli and other pathogens are

capable of attaching to the sand media while flowing through the EIS [2]. A third study, however, notes

that E. coli cells can potentially remobilize and re-enter stormwater systems after detaching from rapid

5

water flow [3]. This controversy over the true effectivity of EIS at removing pathogens will be explored

in this paper.

Before the effectiveness of EIS in pathogen removal is explored, however, a new factor must be

taken into consideration: the growth of naturally forming bacterial biofilms and their role in the removal

and remobilization of pathogenic bacteria. Biofilms are communities composed of a diverse range of

bacteria species living in a carbohydrate support network, or an extracellular polymeric substance (EPS),

that is attached to some hard surface. Biofilms will undoubtedly naturally proliferate on EIS because of

the high concentration and diversity of bacteria species in urban stormwater. When relating biofilms to

bacterial transport, previous research has detected liquid flow in biofilms, indicating that biofilms are

permeable, allowing bacterial transport through the biofilms [4]. Limited studies have been performed

on the impact of biofilms on pathogenic movement in EIS, however, and none of these studies have

investigated it within the context of natural environmental systems and stormwater runoff. One study

examined the role of Pseudomonas aeruginosa (P. aeruginosa) biofilms grown on glass beads in the

removal of E. coli. These results indicate that E. coli can naturally adhere to and grow within the P.

aeruginosa biofilms grown on EIS and that E. coli are often released upon biofilm detachment, which

can be caused by a variety of environmental factors, such as erosion of the biofilm supporting surface

and human disruption of biofilm habitats [5].

Thus, as summarized above, the true effectiveness of EIS has come into question; it is unclear

whether EIS promotes the permanent removal of pathogens like E. coli or instead harbors thriving

communities of these pathogens, temporarily removing them from the stormwater system but allowing

for eventual remobilization. In addition, no research has investigated the impact of biofilms composed of

bacteria species that naturally occur in urban stormwater on the filtration of pathogens. This study aims

to first characterize the bacteria present in urban stormwater collected from a local Baltimore stream,

Stony Run, and the subset of those bacteria that can grow in a column simulation EIS constructed in the

lab. Not all stormwater bacteria are expected to grow on EIS due to the selectivity of the EIS sediment

media. Biofilms of bacteria present in urban stormwater will be grown on the simulation EIS and will be

identified through sequencing. The relationships between these bacteria and E. coli, the pathogen

indicator used in this research, will be characterized using a growth inhibition assay, which can assess to

what degree the stormwater bacteria able to grow on EIS are able to inhibit E. coli growth. By

developing a bigger picture of the fraction of the bacterial community able to grow on EIS and the

capacity of these bacteria to remove E. coli from stormwater, this study will reveal how biofilms impact

6

E. coli removal in the context of the bacterial community naturally present in the environment and will

attempt to determine the effectiveness of EIS at remediating contaminated stormwater.

Materials and Methods

In order to characterize the bacterial community in urban runoff and the subsection of that

community that is able to grow biofilms on EIS, a column simulation EIS will be constructed in the lab

and inoculated with urban runoff water samples. DNA from the 16S ribosomal region will be extracted

and sequenced from bacteria from the urban runoff to provide a baseline dataset. Bacteria that are

present on the simulation EIS after several weeks will be identified through Sanger sequencing. To then

characterize the relationships between the bacteria that grow on the simulation EIS and E. coli, a growth

inhibition assay will be performed in which bacteria isolates obtained from effluent from the EIS will be

placed on lawns of E. coli and the growth patterns detailed.

1. Runoff Community Characterization

The purpose of this first component of the research is to characterize the bacteria present in

urban runoff as a whole. Water runoff samples were collected from an outfall pipe discharging into a

Baltimore urban stream, Stony Run. 50 mL of each water sample was then filtered using a peristaltic

pump through an in-line Swinnex filter holder onto sterile 0.22 µm filters (Millipore). A 16S ribosomal

library of bacteria species present in water samples was then constructed following the protocol of

Preheim et. al. for DNA extraction and Illumina library setup and sequencing [6].

The first 500 most abundant sequences from the runoff in part 1 were aligned using mother [7]

against the Silva reference alignment [8]. A phylogenetic tree was then generated using FastTree 2 [9]

and visualized using the Interactive Tree of Life [10] for tree visualization. Taxonomic information on

these 500 sequences was obtained using the Ribosomal Database Project Classifier [11].

2. Biofilm-growing Bacteria Community Characterization on EIS

A column simulation EIS was constructed in the lab, and the same runoff samples previously

collected and detailed above were filtered through the EIS, depositing bacteria that were then allowed to

grow up into biofilms.

7

EIS Setup. The simulation EIS (see Figure 2) was constructed out of Econo-Pac 14 cm polypropylene

chromatography columns (Bio-Rad) packed with 6 cm of white quartz sand (SiO2, 50-70 mesh particle

size). Sand and columns were prepared using the method of Siripattanakul et. al. [12] and autoclaved

before usage a total of three times. Columns were connected with sterilized feed tubing to a peristaltic

pump (Cole-Parmer). The column method has been used in other papers as an accepted set-up to imitate

the structure of EIS in the field [1, 2, 3, 5].

Figure 2. A diagram of the column simulation EIS constructed in the lab.

Growth of Bacteria on EIS. For inoculation, 1 L per column of collected stormwater runoff was cycled

through three columns for three hours at the rate of 10 cm/hr. Columns were then gravity drained and

incubated at room temperature for the next 19 days and were fed 1 L of synthetic stormwater solution

(SSS, see end of methods section for contents) on days 5, 12, and 19. SSS was filter sterilized before

usage.

8

3. Growth of Bacteria Isolates from EIS Effluent

The purpose of this section of the research is to isolate particular bacteria species that are able to

grow on the simulation EIS and obtain viable cultures to use in the growth inhibition assay.

Collection and Isolation. On day 19 of the column experiment (detailed above in section 2), effluent

exiting the columns during the first 10 minutes of feeding was collected and grown on plates in 200 µL,

100 µL, and 10 µL concentrations (diluted with SSS to be 200 µL total volume when necessary)

containing 50% purified agar and 50% SSS. Plates were incubated at 22°C for approximately 120 hours

before isolated bacteria were restreaked on additional plates. 64 bacteria were isolated in total.

Sequencing of Isolates. 23 isolated bacteria were sequenced to compare to the larger data set obtained

from sequencing the sand from the EIS (see section 2). Only a portion of the 64 isolates were sequenced

to limit repetition and to provide a representative sampling of what bacteria were expected to grow on

EIS. Identifying the isolates also helped match taxonomic branches with reactions to E. coli, as detailed

below in section 4. After bacteria were purified through rounds of restreaking, colonies were picked and

resuspended in 200 µL of polymerase chain reaction (PCR)-grade H2O. First- and second-step PCR on

the 16S ribosomal region was then performed on extracted DNA with the following conditions: 98°C for

10 minutes to enhance cell lysing; 30 cycles of 98°C for 30 seconds for denaturation, 52°C for 30

seconds for annealing, and 72°C for 60 seconds for extension; and 72°C for 10 minutes to ensure

completion. Forward sequencing on all isolates was performed at the Johns Hopkins School of Medicine

Genetic Resources Core Facility on an Illumina MiSeq with 600 base pairs through the usage of the 27F

primer. Sequenced isolates were then preserved in glycerol cultures at -80°C until the growth inhibition

assay could be performed.

The isolate sequences were then added to the phylogenetic tree generated in part 1 of the

methodology by aligning using mothur [7] against the Silva reference alignment [8] and adding to the

phylogenetic tree generated using FastTree 2 [9] and the Interactive Tree of Life [10] for visualization.

Taxonomic information on these 23 isolates was obtained using the Ribosomal Database Project

Classifier [11].

4. Growth Inhibition Assay

The final component of the research is to characterize relationships between E. coli cells and

isolates through growth inhibition assays performed on SSS and agar plates. The purpose of a growth

inhibition assay is to identify hostile relationships between two bacteria species by placing an isolated

9

culture of one onto a thin lawn culture of another, grown up at the same time. Any hostile or parasitic

reactions are clearly defined by a zone of clearing, indicating that the one bacteria species is able to

outcompete the second species.

The growth inhibition assay was carried out between all 64 isolates and a lawn of E. coli. Saved

glycerol cultures of sequenced isolates were restreaked on plates containing 50% agar and 50% SSS and

incubated at 22°C for 96 hours. Colonies of each isolate were then resuspended in 1 mL of SSS to create

liquid cultures, which are required to perform the growth inhibition assay. After 200 µL of E. coli

culture was spread onto SSS plates as a lawn, 5 µL of liquid culture of isolates was placed on E. coli

cultured plates in defined areas and allowed to dry for 20 minutes onto the lawn of E. coli to ensure that

the droplets remained within their areas. After 48 hours of incubation, isolates were investigated to see if

zones of clearing had formed, indicating an aggressive response toward E. coli. The radius of clearing, if

present, was measured in millimeters around specific isolates from the edge of the isolate colony to the

edge of the clearing zone. Photographs of zones of clearing were also obtained using a standard phone

camera.

SSS Contents:

Ionic composition will consist of 5 mM NaCl, 0.75 mM CaCl2, 0.075 mM MgCl2, 0.30 mM Na2SO4, 1

mM NaHCO3, 0.15 mM NaNO3, 0.07 mM NH4Cl, and 0.02 mM Na2SO4 [2]. To increase carbon

content, 0.0015% (by weight) peptone, 0.0011% meat extract, and 0.0003% urea (16 mg/L BOD) were

added [13]. 3 g/L of yeast extract was also added to increase carbon content. 3 mg/L hexanes were

added to mimic presence of petroleum products naturally found in polluted stormwater [14].

Results

The first phase of this research attempts to identify what bacteria present in runoff are capable of

growing on EIS (referred to hereafter as “isolates”). The sequencing results presented below on both the

runoff and the 23 isolates from EIS do not attempt to identify specific bacterial species. The sequences

obtained were not specific enough to narrow all the isolates down to the genus or species level because

of the low-grade quality of the sequencing used and the similarity among bacteria sequences. Instead,

the family level is used as the most specific measure, and sequences are considered in comparison to

each other instead of independently to draw research conclusions.

10

The sequencing results obtained from the 23 isolates were analyzed using the Ribosomal

Database Project to provide basic phylogenetic information. The results are summarized in Table 1.

Phylum Name Number

of Isolates

Family Name Number

of Isolates

Actinobacteria 5 Microbacteriaceae 2

Micrococcaceae 3

Firmicutes 5 Bacillales_Incertae

Sedis XII

2

Bacillaceae 1 3

Bacteroidetes 7 Flavobacteriaceae 6

Sphingobacteriaceae 1

Proteobacteria 6 Enterobacteriaceae 1

Pseudomonadaceae 3

Comamonadaceae 2

Table 1. Breakdown of the microbial diversity observed within column isolates

obtained using the Ribosomal Database Project Classifier [11].

To compare the 23 isolates to the general runoff community, a phylogenetic tree was

constructed, displayed in Figure 3. A phylogenetic tree helps to visualize how the 23 isolates fit into the

evolutionary patterns of the bacteria present in urban runoff. As stated in the methodology, the first 500

most abundant sequences from the community runoff were used to generate the phylogenetic tree, and

the 23 isolate sequences were added to the tree. All 23 isolates and their labels are highlighted in red for

better visualization, and the number of isolates within particular groupings are identified with red

numbers. The tree is meant to provide a larger visual picture for broader analysis of potential patterns

present in bacteria found on EIS. It is not meant to show specific isolate names or categories.

11

Figure 3. Phylogenetic tree, generated using FastTree 2 [13] and the Interactive Tree of Life [14], with

sequences of bacteria isolates found on EIS overlaid on the diverse range of bacteria found in urban

runoff. Bacteria isolates are highlighted in red. For better visualization, each isolate sequence is

highlighted in red, and the number of isolates within a particular grouping/region is identified with large

red numbering.

12

To explore the interactions between bacteria isolates present on EIS and E. coli, a growth

inhibition assay was then performed after sequencing. The results of the assay can be used to draw

important conclusions about what fraction of the bacteria present on EIS actively antagonize E. coli cells

and, thus, the capacity for E. coli removal or death within a typical EIS installation and whether EIS

remediate stormwater or continually introduce pathogens into the stormwater system.

Radii of clearing in the growth inhibition assays were observed in only 12 out of 64 isolates.

Figures 4 and 5 contain photographs with isolates with radii of clearing and isolates without radii to

show how the distinguishment was made. Figure 6 displays the breakdown in the EIS isolates of how

many behaved in a hostile fashion toward E. coli. The radii were measured at three different points on

the plate and averaged, with the average value across the three isolates being 2.8 mm. Sequencing

results are available for seven out of the 12 isolates that were able to inhibit E. coli growth; out of these

seven, four were members of the genus Chryseobacterium and the other three were members of the

genus Pseudomonas.

Isolate 3z3, member of genus

Chryseobacterium, with a radius of

clearing. Note that the E. coli is present

as a semi-clear film over the entire plate.

Figure 4

13

This pie chart shows the fraction of bacteria isolates tested that showed an adverse reaction to E. coli

(i.e., exhibited zones of clearing).

81%

19%

Isolates' Responses to E. coli

No adverse reaction Inhibit E. coli Growth

Isolate 5z22B*, member of genus

Pseudomonas, with a radius of

clearing.

Figure 5

Figure 6

14

Discussion

1. Comparing the Diversity of the Community Present on EIS to the Community Diversity Present in

Stormwater

By characterizing the bacterial community on EIS compared to the community found in

stormwater runoff, the research aims to identify how EIS is selective. In total, 23 bacteria isolates from

the simulation EIS effluent were sequenced, but a total of 64 different isolates were produced. The

limited sequencing was chosen based on a combination of cost and efficiency; many isolates had similar

phylogenies, and to sequence all of them might result in significant repetition. 23 specific isolates were

selected as representatives of their physiological groups to be sequenced. Yet before any analysis on the

23 sequences obtained may be performed, the accuracy of the limited sequencing performed must be

considered to see what percentage of the total diversity was captured.

The Chao-1 estimator [15] was used to estimate how much of the full diversity of the bacterial

community present on EIS the limited sequencing performed actually captured in order to account for

any error. The estimator formula used is: S*1 = Sobs + (a2/2b), where S*1 is the estimated diversity, Sobs

is the number of taxonomic groups observed through sequencing (i.e., observed diversity), a is the

number of taxonomic groups observed just once, and b is the number of taxonomic groups observed just

twice. For calculating family diversity, substituting 9 for Sobs, 2 for a, and 3 for b, the calculated

estimated family diversity is approximately 10 families. It is important to note that the Chao-1 estimator

is typically used at the species level, which this sequencing did not achieve, so this accuracy may be

limited. However, when using it as a baseline number, the estimator indicates that this sequencing

captured approximately 90% of the family diversity present in EIS at the family level. On the phylum

level, substituting 4 for Sobs, 0 for a, and 0 for b, the calculated estimation is still 4 phyla, indicating that

the sequencing essentially captured the total phyla diversity. Thus, the sequencing of isolates, while

limited to only 23 isolates, essentially captured almost all the diversity present down to the family level

and was fairly accurate when it came to the phylum level.

The sequencing results that were obtained for the 23 isolates can then be compared to the

diversity of the entire runoff community to evaluate how selective EIS is. A preliminary analysis can be

done using the phylogenetic data obtained using the Ribosomal Database Project. Of the 11 phyla and

68 families present in the larger community runoff data (specific phylogenetic data on runoff not

displayed in paper due to length), only four phyla and nine families were observed in the EIS isolates.

This could be due to two main factors, one being the expected selective nature of the lab EIS. The

15

constant temperature of the laboratory environment, coupled with the specific growth media used,

selected for particular organisms best adapted to this simulation EIS. In addition to the selectivity of the

EIS, the plate method used for sequencing, which involved the streaking and restreaking of isolates on

agar plates, was also selective; only certain isolates could grow on an agar media with the given growth

media. If sequencing had been done on the effluent samples collected from the columns instead of on the

isolates grown from the effluent on plates, a more complete and less biased picture might have been

presented. However, the plate isolation technique was necessary to this experiment to save viable isolate

cultures for the growth inhibition assay and future EIS research with these isolates. Yet even when the

selective nature of the EIS and plates is ignored, not all bacteria present in the stream community are

hypothesized to sufficiently grow on EIS. EIS requires bacteria to be able to form or live in robust

biofilms to be able to survive strong liquid flow through sand media. Thus, the diversity of the microbial

community present on EIS is most likely reduced even before taking into account lab conditions. This

could impact E. coli transport by exposing E. coli to a reduced spread of bacteria species that are better

adapted to EIS and thus more aggressively competitive for resources within EIS.

A more in-depth evaluation on how EIS selects for microbial diversity can be performed using

the phylogenetic tree. Overall, the diversity of isolates present on EIS was found to be clustered toward

several distinct sections of the phylogenetic tree instead of evenly distributed. This clustering is not

surprising, given that these isolates are required to have evolved particular beneficial characteristics that

can allow them to survive on EIS. In addition, all but two of the isolates found toward the bottom of the

tree (defined as the groupings with number labels 2, 3, 5, and 1) were Gram positive bacteria, a category

suggesting that these bacteria are primarily soil bacteria. Toward the top of the tree (all other groupings),

all but one of the isolates were Gram negative bacteria. It is at first surprising that Gram negative

bacteria can grow on EIS at all; Gram negative bacteria are typically not found in sediment. This

stratification of isolates by Gram – positive on bottom and negative on top – is an interesting factor that

should be investigated in future research; the Gram could determine whether specific bacteria have

evolved specially for EIS.

The division of isolates into certain areas of the tree suggests that large evolutionary mechanisms

may be at play when it comes to the growth of bacteria on EIS. It is already known that only certain

bacteria are able to grow on EIS because of the specialized environment; this phylogenetic tree confirms

the limited diversity and visualizes it in context with the general evolution of bacteria present in urban

runoff. Instead of an even distribution, isolates are clustered, indicating that only certain evolutionary

16

branches of bacteria are able to grow on EIS. Based on the visual patterns identified in the tree, there is

large potential for future research that targets these groupings of isolates. Different isolates distributed in

different sections of the phylogenetic tree can now be selected for closer examination with E. coli to

assess whether particular evolutionary characteristics of particular groupings of isolates can assist with

E. coli removal. Future research on a simulation EIS like the one constructed previously in the lab would

be necessary to answer these hypotheses, and while this current research does not include higher-scale

analysis of interactions using a simulation EIS, the growth inhibition assay (Section 2, below) provides a

basic description of whether these evolutionary patterns illustrated in the tree translate to interactions

with E. coli.

2. Growth Inhibition Assay and the Limited Capacity for E. coli Hostility in Isolates

As the results indicated, only 19% of the isolates showed a capacity to inhibit E. coli growth,

which is potentially concerning. These low numbers may be impacted by the fact that relationships were

tested on agar plates and not in EIS, where realistically, E. coli and these species would be interacting.

Yet the plates still suggest that only a small fraction of bacteria present on EIS hold potential for

reducing E. coli contamination in stormwater. In most cases, bacteria isolates grew without reacting to

E. coli, suggesting that these species have adapted to E. coli’s presence. If this ratio holds true on EIS,

then EIS could prove to be a host site for E. coli, encouraging its growth and the continual reintroduction

of pathogenic cells into stormwater, putting into question the effectiveness of EIS at remediating

stormwater quality.

These initial results could thus be used to modify the design of EIS to make its stormwater

remediation more effective. Engineers could manipulate the conditions of EIS to be more favorable to

those organisms that antagonize E. coli, thus reducing E. coli concentration in stormwater. Novel

systems utilizing these Chryseobacterium and Pseudomonas isolates to inhibit E. coli growth in captured

polluted stormwater could also be designed. More research in a column simulation EIS is necessary to

see if these results obtained from an assay conducted on agar plates are transferrable to the actual

environment, but the potential is enormous for reducing E. coli populations in urban stormwater.

Conclusions

1) To better understand the microbial communities present in urban stormwater runoff and those

bacteria from runoff able to grow on EIS, a column simulation EIS study was conducted with

runoff samples. Samples from the columns were sequenced and compared to sequencing

17

results from general stormwater samples. Sequencing revealed that only a select portion of

the bacterial diversity present in urban runoff are able to grow on EIS.

2) A phylogenetic tree generated from the sequencing results indicates that there is a limited

spread of diversity of bacteria present on EIS, suggesting that only bacteria specifically

evolved to grow on EIS are capable of doing so. This limited diversity could affect E. coli

pollution and filtration through EIS.

3) An additional growth inhibition assay study indicated that an even smaller subset of the

bacterial community present on EIS have the potential to kill or inhibit the growth of E. coli.

This suggests that the bacterial diversity on EIS, while being incredibly limited, does not

include many species that can reduce the concentration of E. coli in stormwater.

4) The isolates identified through the growth inhibition assay that are able to inhibit E. coli have

promising application in stormwater treatment systems to reduce E. coli pollution.

5) In continuing research, biofilms of various isolates will be grown on a lab simulation EIS and

the removal and remobilization rates of E. coli through these systems measured. These

isolates tested will be selected based on the clusters identified in the phylogenetic tree and

based on the isolates’ ability to inhibit E. coli growth. The rates of E. coli removal can be

used to draw broader conclusions about the roles of underlying physical characteristics of

biofilms and bacterial evolution on the removal of E. coli in EIS and how the design of EIS

could be modified to remove the most E. coli from stormwater, reducing the pollution levels

of stormwater and urban aquatic ecosystems.

References

[1] Zhang, L, Seagren, EA, Davis, AP and Karns, JS (2010). The capture and destruction of Escherichia

coli from simulated urban runoff using conventional bioretention media and iron oxide-coated sand.

Water Environmental Research 82 (8), 701-714.

[2] Mohanty, SK, Torkelson, AA, Dodd, H, Nelson, KL and Boehm, AB (2013). Engineering solutions

to improve the removal of fecal indicator bacteria by bioinfiltration systems during intermittent flow of

stormwater. Environment Science & Technology 47 (19), 10791-10798.

[3] Chandrasena, GI, Deletic, A, Ellerton, J and McCarthy, DT (2012). Evaluating Escherichia coli

removal performance in stormwater biofilters: a laboratory-scale study. Water Science & Technology 66

(5), 1132- 1138.

18

[4] Stoodley, P, Debeer, D and Lewandowski, Z (1994). Liquid flow in biofilm systems. Applied and

Environmental Microbiology 60 (8), 2711-2716.

[5] Liu, Y and Li, J (2008). Role of Pseudomonas aeruginosa biofilm in the initial adhesion, growth and

detachment of Escherichia coli in porous media. Environment Science & Technology 42 (2), 443-449.

[6] Preheim, S, Olesen, S, Spencer, S, Materna, A, Varadharajan, C, Blackburn, M, Friedman, J,

Rodríguez, J, Hemond, H and Alm, E (2016). Surveys, simulation and single-cell assays relate function

and phylogeny in a lake ecosystem. Nature Microbiology 1, 16130.

[7] Schloss, PD, Westcott, SD, Ryabin, T, Hall, JR, Hartmann, M, Hollister, EB, Lesniewski, RB,

Oakley, BB, Parks, DH, Robinson, CJ, Sahl, JW, Stres, B, Thallinger, GG, Van Horn, DJ and

Weber, CF (2009). Introducing mothur: Open-Source, Platform-Independent, Community-Supported

Software for Describing and Comparing Microbial Communities. Applied and Environmental

Microbiology 75 (23), 7537-7541.

[8] Quast, C, Pruesse, E, Yilmaz, P, Gerken, J, Schweer, T, Yarza, P, Peplies, J and Glockner, FO

(2013). The SILVA ribosomal RNA gene database project: improved data processing and web-based

tools. Nucleic Acids Research 41 (D1), D590-D596.

[9] Price, MN, Dehal, PS and Arkin, AP (2010). FastTree 2 – Approximately Maximum-Likelihood

Trees for Large Alignments. PLoS ONE 5 (3), e9490.

[10] Letunic, I and Bork, P (2016). Interactive tree of life (iTOL) v3: an online tool for the display and

annotation of phylogenetic and other trees. Nucleic Acids Research 44, W242-W245.

[11] Wang, Q, Garrity, G, Tiedje, J and Cole, J (2007). Naive Bayesian classifier for rapid assignment of

rRNA sequences into the new bacterial taxonomy. Applied Environmental Microbiology 73 (16), 5261-

5267.

[12] Siripattanakul, S, Wirojanagud, W, McEvoy, J, Casey, F and Khan, E (2009). Atrazine removal in

agricultural infiltrate by bioaugmented polyvinyl alcohol immobilized and free Agrobacterium

radiobacter J14a: A sand column study. Chemosphere 74, 308-313.

[13] Jia, JB, Tang, MY, Chen, X, Li, Q and Dong, SJ (2003). Co-immobilized microbial biosensor for

BOD estimation based on sol-gel derived composite material. Biosensors & Bioelectronics 18 (8), 1023-

1029.

[14] United States Environmental Protection Agency (1999) Preliminary Data Summary of Urban Storm

Water Best Management Practices. Washington, D.C.

19

[15] Chao, A (1984). Non-parametric estimation of the number of classes in a population. Scandinavian

Journal of Statistics 11, 265-270.

Bibliography

[1] United States Environmental Protection Agency (2016). CADDIS Volume 2: Sources, Stressors &

Responses. https://www3.epa.gov/caddis/ssr_urb_is1.html (accessed 12 Dec 2016).

[2] United States Environmental Protection Agency (2016). Combined Sewer Overflows (CSOs):

National Pollutant Discharge Elimination System (NPDES). https://www.epa.gov/npdes/combined-

sewer-overflows-csos (accessed 12 Dec 2016).

[3] Grebel, JE, Mohanty, SK, Torkelson, AA, Boehm, AB, Higgins, CP, Maxwell, RM, Nelson, KL and

Sedlak, DL (2013). Engineered Infiltration Systems for Urban Stormwater Reclamation. Environmental

Engineering Science 30 (8), 437-454.

[4] Donlan, R (2002). Biofilms: Microbial Life on Surfaces. Emerging Infectious Diseases 8 (9), 881-

890.