Ecology and Epidemiology of Antimicrobial Resistance in Urban

ECOLOGY AND EPIDEMIOLOGY OF

ANTIMICROBIAL RESISTANCE IN

URBAN AND PERI-URBAN MESOCARNIVORES

A DISSERTATION

SUBMITTED TO THE FACULTY OF THE

UNIVERSITY OF MINNESOTA

KATHERINE E. L. WORSLEY-TONKS

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS

FOR THE DEGREE OF DOCTOR OF PHYLOSOPHY

MEGGAN E. CRAFT (ADVISOR)

TIMOTHY J. JOHNSON (CO-ADVISOR)

DECEMBER 2020

Acknowledgements

My time at the University of Minnesota has been filled with energy, opportunities, development, and joy. This is because I have been surrounded by brilliant and stimulating mentors, colleagues, and friends. I do not think I could have asked for a better experience. I have been challenged and supported at the same time, which sets up a perfect training environment, and I am extremely grateful for that.

I first would like to express my deepest gratitude to my main advisor Meggan Craft.

Meggan, thank you for taking me on as a student, keeping me motivated throughout my

PhD, and reminding me to always aim high. You have taught me how to be an effective, steady, and rigorous scientist. You have set-up so many opportunities for me, including taking part in other research projects, putting me in touch with your collaborators, and pushing me to really make the most of my PhD by going to conferences, workshops, and even doing an internship at EcoHealth Alliance. I am extremely grateful for everything that you have done for me, and I feel very fortunate to have been your student. I would also like to extend my deepest appreciation to my co-advisor, Tim Johnson, for welcoming me into his research group once I had decided to work on antimicrobial resistance for my PhD. Tim, I could not have done this PhD without your guidance.

Thank you so much for walking me through the world of antimicrobial resistance, for redirecting me when needed, and for giving me the opportunity to learn from you and your research group.

Thank you so much to James Forester, Jeff Bender, and Dominic Travis for serving on my committee, for their rigorous feedback, and for challenging me throughout my PhD.

James, thank you for challenging my methods and for helping troubleshoot approaches,

especially towards the end of my PhD. Thank you, Jeff, for helping me develop my research project and for reminding me to stay motivated and press on. I am especially grateful for your guidance on how to make my research more applied and for helping me identify additional funding when needed. Dom, I cannot thank you enough for helping me break-down my broad research ideas and making them into concrete objectives. I felt lucky to have had a cubicle next to your office because it meant that I could seek your guidance more easily, possibly more often than you had time for. Thank you for always dropping what you were doing to provide guidance. Our conversations about my PhD work, ecosystem health, and my next career step were invaluable. Thank you so much for letting me take up so much of your time.

While not on my committee, I must also give a big thanks to Randy Singer as he has greatly contributed to my training throughout my PhD. Randy, I have learnt so much about the epidemiology of antimicrobial resistance from working with you, and I can’t thank you enough for that. I am so grateful that you helped train me during the second half of my PhD. Elizabeth Miller was also critical for my training. Liz, thank you for trouble-shooting methods with me, for letting me ask you hundreds of questions, and for being a great mentor more generally.

I am indebted to Stan Gehrt, who allowed me to conduct this research as part of his long- term urban mesocarnivore project in Chicago. Thank you, Stan, for being such an amazing mentor and for giving me so much of your time and scientific insight. Your passion for carnivore ecology and conservation is infectious, and your success at developing and maintaining a research project for 25+ years has inspired me to strive for something similar in the future. Chris Anchor was also an important mentor for me in

Chicago. Thank you, Chris, for letting me use your laboratory and for letting me pick your brain on some of my research ideas. I continue to marvel at how much knowledge you have on the history of Chicago both in terms of landscape and wildlife management, and I am so grateful that you shared some of that knowledge with me. Thank you to

Shane McKenzie for helping me take my research ideas and implement them in the field.

Shane is the reason why we were able to collect so much data on raccoons, coyotes, and opossums. Thank you, Shane, for training me on how to do urban mesocarnivore field work and for being so patient with me despite the numerous hiccups along the way. I am beyond grateful to have had the opportunity to interact with the late Donna Alexander, the primary funder of Stan Gehrt’s research. Donna set-up multiple opportunities for me in Chicago. Her devotion to her work and to what she believed in was truly inspiring. I am so sad that she is not able to see the final product of this work but hope that it would have met her standards.

There are several students and research assistants who have played a huge part in this project and/or in my training that I cannot thank enough. These include members of the

Craft lab, including Nick Fountain-Jones, Luis Escobar, Lauren White, Marie Gilbertson,

Janine Mistrick, and Matt Michalska-Smith. Nick and Luis were especially important in my training. I have learnt so much from working with them and I am very grateful that they so willingly taught me countless skills. The help and feedback I got from the

Johnson and Singer labs was also invaluable, and I especially am thankful to Bonnie

Weber and Alison Millis for their help with processing the samples in the lab. Thank you also to members of the Gehrt lab for helping with the field work, especially on days

iii

where we captured 30+ raccoons. Andy Burmesch, Yasmine Hentati, Lauren Ross, and

Steven Winter were especially important to this achievement.

I also would like to thank my colleagues and faculty at the University of Minnesota for their feedback on my work and for teaching me how to be a dedicated scientist. A large part of my training and scientific thinking come from conversations I have had with

Tiffany Wolf, Kim VanderWaal, and Julio Alvarez. Thank you also to the members of the Wildlife Epidemiology Journal Club and Ecosystem Health group for their feedback at different stages of my PhD. Thank you also to the College of Veterinary Medicine more broadly, for funding most of my work.

Finally, and possibly most importantly, I am eternally grateful for the support and encouragement from my family and friends. Thank you to my parents, Pam and Mal, for being the first to encourage me to do this PhD despite probably preferring that I stay closer to home. Thank you for coaching me throughout my PhD, for believing in me, and for reminding me to fight for what I believe in, to stay strong during challenging times, and to have a healthy work-life balance. Thank you to my brothers, David, Andrew, and

Richard, for instilling in me an interest in nature as a child, and thank you especially to

Richard for inspiring me to pursue a PhD. Thank you to my friends at Minnesota, particularly Irene Bueno, Jessica Deere, Kaushi Kanankege, Amy Kinsley, Cata Picasso,

Kim VanderWaal, Liz Miller, Shiv Hayer, George Omondi, Nick Fountain-Jones, Luis

Escobar, for making my time in Minnesota so enjoyable. I will never forget the happy hours, barbecues, hikes, etc. we have had together. Thank you to Claire Aupetit, Floriane

Giraud, Camille Gomez, Suzie Kenny, Camilla Ryan, Sara Heisel, Erin Abernethy, and

Cecilia Sanchez for all the Skype chats that kept me upbeat and sane over the years.

Lastly, thank you to Abdel Lachgar for appearing in my life during the second half of my

PhD. Thank you so much for your love, support, and friendship.

Abstract

Use of antibiotics in human and animal medicine has led to the emergence of many forms of antimicrobial resistant bacteria (ARB). As well as undermining the successful treatment of bacterial infections, intensive use has led to the dissemination of ARB in the community and the environment. While extensive progress has been made in understanding the spread and fate of ARB in the community, the environment has had far less attention. Yet, detection of clinically relevant ARB and associated antimicrobial resistance genes (ARG) in water and soil indicates that the environment could act as an additional exposure pathway for people and domestic animals. ARB have also been isolated from wildlife present in these contaminated environments. Because of this, wildlife are considered to be good candidates for understanding environmental antimicrobial resistance (AMR). While ARB and ARG have been detected in numerous wildlife species across the globe, several questions remain unanswered regarding exposure pathways and what the consequences might be for human and domestic animal health.

The goal of this thesis was to assess the contribution of several anthropogenic sources in shaping the AMR profile of wildlife. To do this, we investigated the ecology of AMR in an urban-suburban context, specifically in the city of Chicago. The wildlife species of focus was the raccoon (Procyon lotor) and the ARB were clinically relevant extended- spectrum cephalosporin resistant (ESC-R) Escherichia coli.

We first investigated the importance of two known anthropogenic sources of AMR in shaping the ESC-R E. coli profile of raccoons: 1) the presence of rivers that were downstream of a wastewater treatment plant and 2) urban context (urban vs. suburban).

We found that the risk of isolating ESC-R E. coli from raccoons was higher when raccoons were sampled at urban sites than at suburban sites. Importantly, we also found that raccoons were more likely to have ESC-R E. coli with transferable ARG when present at sites that were downstream from a WWTP, suggesting that WWTPs may increase the risk for AMR to spread widely in wildlife bacterial communities. We then explored the importance of various landscape factors at predicting isolation of ESC-R E. coli. Landscape factors were examined within estimated raccoon home ranges and we found residential areas and wetlands to increase the risk of isolating ESC-R E. coli from raccoons. This finding is important because most wildlife and environmental AMR research tend to ignore these types of landscape features, and our work highlights that more attention may be warranted.

Next, we explored the interface with domestic animals. For this, we first compared the prevalence and phylogenetic relatedness of ESC-R E. coli of raccoons to that of local domestic dogs (Canis lupus familiaris) and asked whether shared space was important for predicting the sharing of ESC-R E. coli. ESC-R E. coli prevalence in raccoons was three times greater than in dogs, but isolated ESC-R E. coli were phylogenetically similar.

Shared space was important for predicting isolation of ESC-R E. coli from raccoons but not from dogs, and was not important for predicting phylogenetic associations of ESC-R

E. coli. Finally, we compared the AMR profile of raccoons to that of coyotes (Canis latrans), Virginia opossums (Didelphis virginiana), and locally owned and stray dogs.

This final step did not focus on ESC-R E. coli, but rather took a broader look at the AMR profile of the four species by examining the presence of multiple ARG in samples that were pooled by animal species and dog type. The three mesocarnivore pooled samples

vii

had similar numbers and types of ARGs as the stray dog pooled sample, but not the owned dog pooled sample. Collectively, this thesis provides a concrete example for the need to account for the environment when generating surveillance and control strategies for AMR, and identifies several environmental components that could be targeted for surveillance.

viii

Table of Contents

Acknowledgements ...... i Abstract ...... vi List of Tables ...... xii List of Figures ...... xiv List of Abbreviations ...... xvi Chapter 1 - Introduction ...... 1 Chapter 2 – Importance of Anthropogenic Sources at Shaping the Antimicrobial Resistance Profile of a Peri-Urban Mesocarnivore ...... 9

Overview ...... 9

Introduction ...... 10

Methods ...... 14

Study site and design ...... 14 Raccoon handling ...... 16 Phenotypic characterization of ESC-R E. coli ...... 16 Genome assembly and gene content analysis ...... 17 Assessing the plasmid- vs. chromosomal-association of ARG conferring ESC-resistance ... 18 Phylogenetic analysis ...... 19 Statistical analysis ...... 20

Results ...... 22

Prevalence, richness, and characteristics of ESC-R E. coli and associated ARGs isolated from raccoons ...... 22 Importance of anthropogenic sources at influencing the ESC-R E. coli profile of raccoons 27

Discussion ...... 31

Acknowledgements ...... 38

Chapter 3 – Residential Areas and Wetlands are Associated with a Higher Risk of Isolating Antimicrobial Resistant Bacteria from Peri-Urban Wildlife ...... 39

Overview ...... 39

Introduction ...... 40

Methods ...... 43

Study area ...... 43 Raccoon sampling ...... 44 Testing for the presence of ESC-R E. coli ...... 45 Estimating raccoon home ranges ...... 45 Landscape features ...... 46 Statistical analysis ...... 47

Results ...... 48

Discussion ...... 50

Chapter 4 - Antimicrobial Resistant Bacteria at the Wildlife-Domestic Animal Interface – Is Shared Space Important for Microbe Sharing? ...... 54

Overview ...... 54

Introduction ...... 55

Methods ...... 59

Study site and design ...... 59 Raccoon and dog sampling ...... 61 Classification of sites based on raccoon predicted use of residential areas ...... 61 Phenotypic characterization of ESC-R E. coli ...... 62 Sequencing, bioinformatics, and phylogenetic analyses ...... 62 Statistical analysis ...... 64

Results ...... 70

Domestic dog and raccoon characteristics ...... 70 Domestic dogs had a lower prevalence of ESC-R E. coli than raccoons, but dogs and raccoons had phylogenetically similar ESC-R E. coli ...... 71 Presence of domestic dogs at raccoon sites influenced the probability of recovering ESC-R E. coli from raccoons, but this effect was more distinct at suburban than urban sites ...... 74 Presence of raccoons in dog areas did not influence the probability of recovering ESC-R E. coli from domestic dogs ...... 76

None of the four spatial overlap metrics influenced the phylogenetic relatedness of ESC-R E. coli isolated from domestic dogs and raccoons ...... 77

Discussion ...... 77

Acknowledgements ...... 83

Chapter 5 - Characterization of antimicrobial resistance genes in Enterobacteriaceae carried by suburban mesocarnivores and locally owned and stray dogs...... 85

Overview ...... 85

Introduction ...... 86

Methods ...... 88

Mesocarnivore and dog sample collection ...... 88 Sample processing and sequencing...... 89 Sequence processing and analyses ...... 90

Results ...... 91

Discussion ...... 94

Chapter 6 – Conclusion ...... 98 Bibliography ...... 104 Appendix A...... 125 Appendix B...... 130 Appendix C...... 143

List of Tables

Table 2. Model averaging results from binomial generalized linear mixed models of the probability of isolating at least one ESC-R E. coli from raccoons (n = 230)...... 28

Table 3. Richness of ESC-R E. coli sequence types by urban context, presence of a wastewater treatment plant (WWTP) at sampling sites, and collection season ...... 29

Table 5. Top ranked binomial generalized linear mixed models (AICc < 2 from the best fit model) for isolating at least one ESC-R E. coli from raccoons...... 49

Table 6. Model averaging results from binomial generalized linear mixed models of the probability of isolating at least one ESC-R E. coli from raccoons...... 49

Table 7. Description of statistical approaches used...... Error! Bookmark not defined.

Table 8. Generalized linear mixed model results for isolating at least one ESC-R E. coli from raccoons...... 75

Table 10. Number of raccoons for which ESC-R E. coli were isolated on different capture events...... 125

Table 12. Top ranked binomial generalized linear mixed models (AICc < 2 from the best fit model) for isolating at least one ESC-R E. coli from raccoons (n = 211, but 230 with recaptures)...... 128

xii

Table 13. Top ranked Poisson generalized linear mixed models (AICc < 2 from the best fit model) predicting antimicrobial resistance gene richness of ESC E. coli of raccoons (n = 123)...... 129

Table 15. Questions for dog owners...... 130

Table 16. Fecal presence of ESC-R E. coli in domestic dogs in relation to dog age, sex, antibiotic use, and based raccoon spatial overlap...... 130

Table 17. Univariable Poisson generalized linear mixed models core-SNP difference among pairs of ESC-R E. coli isolates recovered from raccoons and domestic dogs. ....131

Table 19. Description of the number of raccoons radio-collared at each site, along with the years of tracking, and maximum and minimum number of relocations obtained per animal...... 139

Table 20. Median maximum distance raccoons travelled outside of each site...... 142

xiii

List of Figures

Figure 1. Sampling sites in urban and suburban Chicago...... 15

Figure 2. Characteristics of ESC-R E. coli and associated ARGs detected in the feces of raccoons sampled in the metropolitan area of Chicago ...... 25

Figure 3. Predictive association of contigs carrying beta-lactam resistance genes with chromosome or plasmid location ...... 26

Figure 4. Raw prevalence of ESC-R E. coli in raccoons based on (A) season; and (B) urban context ...... 28

Figure 5. Raw prevalence of ESC-R E. coli carrying plasmid-associated blaCTX-M or blaCMY ARG based on the presence of WWTP at sampling sites ...... 31

Figure 6. Study sites...... 44

Figure 7. Effect of (A) wetlands and (B) residential areas on the probability of isolating ESC-R E. coli from raccoons ...... 50

Figure 8. Sampling sites in the northwestern portion of the Chicago metropolitan area ..60

Figure 9. Prevalence and phylogenetic associations of ESC-R E. coli isolated from raccoons and domestic dogs...... 73

Figure 10. Mean number of core-SNP differences between pairs of ESC-R E. coli isolates by sequence type (ST) based on whether pairs of isolates were from the same animal species (within) or different animal species (between) ...... 74

Figure 11. Prevalence of ESC-R E. coli in raccoons by urban-suburban context and dog presence (yellow: dogs are present, red = dogs are absent and cannot enter) ...... 76

Figure 13. Movement of antimicrobial resistant bacteria in the community and the environment ...... 103

Figure 14. Map of the Chicago metropolitan area depicting the sites where raccoons were sampled (red and yellow polygons), and the home zip codes of sampled dogs (pink polygons)...... 137

xiv

Figure 15. Sampling sites in the northwestern portion of the Chicago metropolitan area ...... 140

Figure 16. Landcover proportions at each sampling site (n = 7) and surrogate site (n = 1; Dan Ryan) ...... 140

Figure 17. Maps of the eight sites (seven sampling sites and one surrogate site) ...... 141

Figure 18. Relative abundance of the nine most common bacterial genera found in the mesocarnivore, and stray and owned dog samples ...... 143

Figure 19. Correlation between number of individual samples in each pooled sample and antimicrobial resistance gene (ARG) richness ...... 144

List of Abbreviations

AIC Akaike information criterion

AICc Akaike Information Criterion corrected for small sample size

AMR Antimicrobial Resistance

ARB Antimicrobial Resistant Bacteria

ARG Antimicrobial Resistance Gene

DNA Deoxyribonucleic acid

ESC-R E. coli Extended-Spectrum Cephalosporin Resistant Escherichia coli

GLM Generalized Linear Model

GLMM Generalized Linear Mixed Model

GPS Global Positioning System

MCP Minimum Convex Polygon

MGE Mobile Genetic Element

MLST Multilocus Sequence Typing

NCBI National Center for Biotechnology Information

NDM-1 New Delhi Metallo-Beta-Lactamase-1

NLCD National Landcover Database

PERMANOVA Permutational multivariate analysis of variance

xvi

SNP Single-Nucleotide Polymorphism

ST Sequence Type

VHF Very High Frequency

WGS Whole Genome Sequencing

WWTP Wastewater Treatment Plant

xvii

Chapter 1 - Introduction

While antibiotics have undoubtably revolutionized human and animal medicine, today antimicrobial resistance (AMR) is described as one of the greatest medical challenges of our time (Van Der Meer 2013). The World Health Organization has ranked AMR as one of the top ten global public health threats (WHO, 2019), warning that by 2050 antimicrobial-resistant bacteria (ARB) could cause 10 million deaths annually and result in a global financial crisis. For the United States, this amounts to an estimated 35,000 people dying each year from infections caused by ARB (CDC, 2019) and annual healthcare costs to be as high as $20 billion (Hughes 2011). The approach used to overcome this public health crisis has been to restrict and educate on antibiotic use

(Murphy et al. 2017), and in some cases ban the use of antibiotics for livestock growth

(Laxminarayan et al. 2013). While these efforts have resulted in a decline in AMR rates in some countries, global AMR rates continue to rise (WHO, 2019), multi-drug resistant bacteria are becoming more common (Vila 2015), and resistance to last resort antibiotics

(e.g. colistin) has emerged (Liu et al. 2016).

That said, recent advances in intersectoral and global collaborations have greatly enhanced our understanding of AMR. For one, it has become clear that reducing antibiotic consumption alone will not suffice for controlling AMR (Collignon et al.

2018). This is because of the way in which AMR spreads. AMR spreads at two levels: at the bacterial level and at the host level. At the bacterial level, genes that code for AMR can be transferred between bacteria when located on mobile genetic elements such as plasmids. This means that bacteria can acquire resistance genes from related and

unrelated bacteria, and can do so both in hosts and in the environment (Mazodier and

Davies 1991). At the host level, ARB can be transmitted in the community via direct contact with other individuals (within and between species; (Pomba et al. 2017; Pitout et al. 2005), through the food chain (e.g. Börjesson et al. 2016), artificial surfaces, and through environmental pathways such as water or soil (Huijbers et al. 2015). This multi- mode transmission is the reason why ARB that are typically associated with clinical or agricultural settings persist in the community (Holmes et al. 2016), why air travel has resulted in the global spread of novel ARB (e.g. New Delhi Metallo-Beta-Lactamase-1

(NDM-1) -Producing Enterobacteriaceae; Kumarasamy et al. 2010), and why low-income countries have higher AMR rates than high-income countries despite having lower antimicrobial consumption (Collignon et al. 2018).

Of all the potential transmission pathways (e.g. food-borne, direct contact), environmental pathways are frequently the most overlooked (Robinson et al. 2016). Yet, several lines of evidence show that humans and domestic animals can be exposed to ARB through the environment (Huijbers et al. 2015). For example, ARB and antimicrobial resistance genes (ARG) have been detected in tap water in multiple cities across the globe

(e.g. Xi et al. 2009). Water obtained from lakes and rivers used to irrigate plants and livestock is also a non-trivial exposure pathway for humans (Woolhouse et al. 2015).

Importantly, dissemination of ARB and ARG from anthropogenic sources to the environment can result in more widespread occurrence of AMR (Huijbers et al. 2015).

For example, rivers that are downstream from a wastewater treatment plant (WWTP) can transport ARB and ARG long distances and to other ecosystems (e.g. lakes and wetlands;

Martak et al. 2020). Additionally, since plasmid-associated ARG can be transferred to

other bacteria, anthropogenic sources of AMR and the environment can also act as sources of novel ARB (Marti, Variatza, and Balcazar 2014; Rizzo et al. 2013). Thus, the environment plays a role in the transmission and evolution of AMR (Radhouani et al.

2014).

Despite these concerns, our understanding of the spread and fate of ARB and ARG in the environment is limited (Wellington et al. 2013; Arnold et al. 2016). Additionally, while several anthropogenic environmental sources of AMR are predicted to be important , only a handful have been empirically identified as important (Larsson et al. 2018).

Anthropogenic environmental sources come in two forms: point sources (e.g. WWTP;

Rizzo et al. 2013) and non-point sources (e.g. urban runoff, animal defecation, leaky pipes, landfills; Stewart et al., 2007; Surette & Wright, 2017). Most AMR environmental research has focused on the importance of point-sources, particularly WWTPs, probably because effects are easier to quantify in point sources than in non-point sources. Failure to consider and account for other exposure pathways has the potential to threaten human and animal health, and the success of environmental AMR surveillance and control programs. As such, there is a pressing need to better understand the widespread occurrence of AMR in the environment to allow for more targeted surveillance and control.

Wildlife are frequently used as sentinels for understanding the widespread occurrence of infectious agents and contaminants in the environment (Halliday et al. 2007; Smits &

Fernie 2013), and are considered good candidates for understanding the environmental dissemination of AMR (Vittecoq et al. 2016; Furness et al. 2017 but see Swift et al.

2019). In fact, several wildlife species from many different countries have been found to

host ARB and associated ARG (reviewed in Carroll, Wang, Fanning, & Mcmahon, 2015;

Vittecoq et al., 2016), and in most cases proximity to human-dominated areas is associated with a higher risk of detecting ARB and ARG in wildlife (Allen et al. 2010).

Importantly, wildlife carriage of ARB can also have important implications for human and domestic animal health as well as for the spread of AMR in the environment (Carroll et al. 2015). Wildlife have the potential to act as secondary reservoirs of known and/or novel ARB (Greig et al. 2015; Ramey and Ahlstrom 2020), especially if ARG are transferred between bacteria via plasmids (Dolejska and Papagiannitsis 2018).

Additionally, unlike environmental point sources, wild animals are mobile, allowing for the potential widespread dissemination of ARB and/or exposure to humans and domestic animals in areas away from known environmental sources (Arnold, Williams, and

Bennett 2016). Thus, understanding AMR in wildlife is important for human and domestic animal health (Carroll et al. 2015), and for identifying environmental sources of resistance (Dolejska and Literak 2019; Ramey and Ahlstrom 2020; Vittecoq et al. 2016).

So far, wildlife AMR research has focused on exploring the presence of ARB and ARG in various wildlife species and have explored links with human sources. Links with human sources have been quantified based on wildlife proximity to human areas or by comparing ARB and ARG to those found in human and domestic animal populations, or environmental samples. Comparing the AMR profile of wildlife species with distinct characteristics (e.g. aquatic vs. terrestrial, carnivorous vs. herbivorous; Jobbins and

Alexander 2015; Vittecoq et al. 2016) has also been a method used to identify potential environmental exposure pathways. Thus far, the evidence suggests that identifying clinically relevant ARB and ARG in wildlife is more likely when wildlife reside close to

human areas, when wildlife are omnivorous and carnivorous, aquatic, and/or when wildlife use landfills or contaminated rivers (Ahlstrom et al. 2019; Jobbins and Alexander

2015; Vittecoq et al. 2016). There is also evidence for wildlife having similar AMR profiles to local domestic animals, although conflicting results have been found (e.g.

Pesapane, Ponder, and Alexander 2013; Mercat et al. 2016; Viswanathan et al. 2017).

Taken together, these findings highlight that ARB and ARG identified in wildlife are of anthropogenic origin (Allen et al. 2010) and that certain anthropogenic sources could increase exposure risk.

Despite these advances, there are several outstanding questions that need to be answered in order to understand the importance of anthropogenic sources at shaping the AMR profile of wildlife, and what the implications might be for human and domestic animal health (Dolejska and Literak 2019; Ramey and Ahlstrom 2020). One outstanding question is whether known anthropogenic sources of AMR increase the risk for wildlife to act as sources of AMR. Wildlife AMR research thus far has shown that there can be an association between the presence of ARB in wildlife and wildlife exposure to known anthropogenic sources (e.g. Carter et al. 2018; Furness et al. 2017; Swift et al. 2019).

However, it is unclear whether these anthropogenic sources also increase the risk for

AMR to spread in wildlife bacterial communities. Additionally, while previous work has explored links with individual specific anthropogenic sources, environmental context and variation thereof is typically ignored in wildlife AMR research. Yet, environmental AMR research has highlighted that several anthropogenic environmental sources exist (Hooban et al. 2020), the importance of which probably vary with environmental context and time.

Exploring whether certain habitats increase or decrease wildlife exposure risk to ARB

could help identify high-risk areas and the importance of non-point sources of AMR (e.g. urban runoff). Similarly, conflicting results have been found about whether wildlife have similar AMR profiles to domestic animals, which could be associated with the context of the interface. These types of questions can be answered by conducting in-depth, hypothesis-driven epidemiological investigations.

The goal of this thesis was to improve our understanding of the contribution of several anthropogenic sources in shaping the AMR profile of wildlife. Two objectives support this goal: 1) assess the importance of anthropogenic and environmental factors at shaping the AMR profile of wildlife; and 2) compare the AMR profile of wildlife to local domestic animals, and explore whether certain host characteristics and interactions increase AMR profile similarity. To tackle these objectives, we investigated the ecology of AMR in an urban-suburban context, specifically in the city of Chicago. The wildlife species of focus was the raccoon (Procyon lotor) because individuals of this species are abundant, and widely distributed in urban and suburban areas (Gehrt, Riley, and Cypher

2010; Bateman and Fleming 2012), and are known to share several pathogens with domestic animals and humans (e.g. E. coli and Salmonella spp.; Jardine et al. 2011;

Bondo et al. 2016). The ARB of focus were clinically relevant extended-spectrum cephalosporin resistant (ESC-R) Escherichia coli. ESC-R E. coli are an emerging issue in human and veterinary medicine (Bradford 2001; Wieler et al. 2011), and have recently been reported to persist in healthy human (Arpin et al. 2005; Pitout et al. 2005), livestock

(Smet et al. 2010; Ewers et al. 2012), and companion animal populations (Hordijk et al.

2013; Schaufler et al. 2015; Pomba et al. 2017), as well as in the environment (e.g. water sources; Korzeniewska and Harnisz 2013; Bréchet et al. 2014). Wildlife carriage of

ESBL-R E. coli has been reported in over 30 species across the globe (Guenther, Ewers, and Wieler 2011; Wang et al. 2017).

For objective 1, we first investigated the importance of two known anthropogenic sources of AMR in shaping the ESC-R E. coli profile of raccoons (Chapter 2). The two anthropogenic sources of focus were the presence of rivers that were downstream of a

WWTP and urban context (urban vs. suburban). We found that the risk of isolating ESC-

R E. coli from raccoons was higher when raccoons were sampled at urban sites than at suburban sites and when sampled in the spring and summer than winter and fall.

Additionally, we found that raccoons were more likely to have ESC-R E. coli with plasmid-associated ARG when present at sites that were downstream from a WWTP. We then explored the importance of various landscape features present in raccoon home ranges at predicting isolation of ESC-R E. coli (Chapter 3). We found residential areas and wetlands to increase the risk of isolating ESC-R E. coli from raccoons.

For objective 2, since domestic animals can be sources of AMR for wildlife (e.g. Mercat et al. 2016), we compared the prevalence and phylogenetic relatedness of ESC-R E. coli of raccoons to that of local domestic dogs (Canis lupus familiaris) and asked whether shared space was important at predicting the sharing of ESC-R E. coli (Chapter 4). ESC-

R E. coli prevalence in raccoons was three times greater than in dogs, but isolated ESC-R

E. coli were phylogenetically similar. Shared space was important for predicting isolation of ESC-R E. coli from raccoons but not from dogs, and was not important at predicting phylogenetic associations of ESC-R E. coli. Finally, since previous work has suggested that different wildlife species can have different AMR profiles, we compared the AMR profile of raccoons to that of coyotes (Canis latran), and Virginia opossums (Didelphis

virginiana), and to locally owned and stray dogs (Chapter 5). This final chapter did not focus on ESC-R E. coli, but rather took a broader look at the AMR profile of the four species by examining the presence of multiple ARG in samples that were pooled by animal species. We found the raccoon pooled sample to have a similar number and types of ARGs as the coyote and opossum pooled sample, as well as the stray dog pooled sample, but not the owned dog pooled sample. Taken together, these findings suggest that several environmental and anthropogenic factors contribute to wildlife exposure, some of which can be targeted for surveillance.

Chapter 2 – Importance of Anthropogenic Sources at Shaping the Antimicrobial Resistance Profile of a Peri-Urban Mesocarnivore

Overview

Anthropogenically derived antimicrobial-resistant bacteria (ARB) and antimicrobial resistance genes (ARGs) have been detected in wildlife. The likelihood of detecting ARB and ARGs in wildlife increases with wildlife exposure to anthropogenic sources of antimicrobial resistance (AMR). Whether anthropogenic sources also increase the risk for

AMR to spread in bacteria of wildlife is not well understood. The spread of AMR in bacteria of wildlife can be estimated by examining the richness of ARB and ARG, and the prevalence of ARB that have mobilizable ARGs (i.e. ARGs that can be transferred across bacteria via plasmids). Here, we investigated whether raccoons (Procyon lotor) with different exposures to anthropogenic sources, and sampled during different seasons, differed in prevalence and richness of extended-spectrum cephalosporin-resistant (ESC-

R) Escherichia coli, richness of ARGs present in ESC-R E. coli, and prevalence of ESC-

R E. coli with plasmid-associated ARGs. ESC-R E. coli were isolated from over half of the 211 raccoons sampled and were more likely to be isolated from raccoons during the spring and summer than the winter and fall, and from urban than suburban raccoons.

When examining the whole-genome sequences of ESC-R E. coli, 56 sequence types were identified, most of which were associated with the ARGs blaCMY and blaCTX-M. A greater richness of ESC-R E. coli sequence types was found at sites with a wastewater treatment plant (WWTP) than without, but no difference was detected by season or urban context.

ARG richness in ESC-R E. coli did not significantly vary by urban context, presence of a

WWTP, nor season. Importantly, ESC-R E. coli carrying plasmid-associated blaCTX-M

and blaCMY ARGs were more likely to be isolated from raccoons sampled at sites with a

WWTP than without. Our findings indicate that anthropogenic sources shape the AMR profile of wildlife, reinforcing the need to prevent dissemination of AMR into the environment.

Introduction

Use of antibiotics in human and veterinary medicine has led to the emergence of many forms of antimicrobial resistant bacteria (ARB) (WHO, 2014). In addition to undermining the successful treatment of bacterial infections, intensive antimicrobial use has been linked to the widespread dissemination of ARB in the community and the environment (Laxminarayan, Duse, Wattal, Zaidi, et al. 2013; Radhouani et al. 2014).

Hence, what was originally observed only in clinical and agricultural settings is now frequently reported in non-hospitalized people, animals, and the environment. In fact, certain environments can act as important sources of AMR, including rivers connected to wastewater treatment plants (WWTPs) and fecal run-off from agricultural sites

(Kümmerer 2004; Allen et al. 2010; Wellington et al. 2013; Graham et al. 2019). Such environments may directly expose people, domestic animals, and wildlife to ARB

(Berkner, Konradi, and Schönfeld 2014).

Wildlife exposure to ARB has been reported in a number of different species, from different ecosystems and continents (e.g. Cole et al. 2005; Miller, Gammon, and Day

2009; Jobbins and Alexander 2015; Kipkorir et al. 2019). For this reason, there have been concerns regarding the role of wild animals in the dissemination of resistance in the

environment, and whether wildlife could threaten public and domestic animal health

(Vittecoq et al. 2016; Dolejska and Literak 2019). Wildlife may act as important reservoirs, ‘melting pots’, or amplifiers of resistance in the environment (Radhouani et al. 2014; Carroll et al. 2015; Ramey and Ahlstrom 2020; Vittecoq et al. 2016; Arnold,

Williams, and Bennett 2016), potentially facilitating the maintenance and spread of known ARB, and emergence of novel ARB strains (Jones et al. 2008; Karesh et al. 2012;

Ramey and Ahlstrom 2020). However, the likelihood for this to occur depends on whether antimicrobial resistance genes (ARGs) can be transferred between bacteria of wildlife (Allen et al. 2010; Dolejska and Papagiannitsis 2018). ARGs can be transferred between bacteria if they are located on plasmids (a process known as conjugation), but are less likely to transfer if they are located on the chromosome of bacteria, unless they are transferred to plasmids via transposons or integrons. Thus, identifying ARB with plasmid-associated ARGs is of greater interest than identifying ARB with chromosomally associated ARGs because association with transferable plasmids increases the risk for AMR to spread widely in wildlife bacterial communities.

Most wildlife AMR research has focused on investigating the occurrence and similarity of ARB and ARGs in relation to various anthropogenic sources or environments (e.g.

Carter et al. 2018; Swift et al. 2019). However, little is known about whether we should expect ARGs to be plasmid- or chromosomally-associated (Dolejska and Literak 2019).

Environmental AMR research has shown that environments associated with anthropogenic sources of AMR (e.g. rivers connected to WWTPs) are not only more likely to find a greater richness of ARB and ARG (i.e. number of unique ARB and

ARG), but also more plasmid-associated ARGs, than environments that are not, or less

associated with anthropogenic sources of AMR (Berendonk et al. 2015; Rizzo et al.

2013). Whether similar differences hold true for the wildlife in these environments is less well understood, yet it is essential for evaluating the importance of anthropogenic sources at shaping the AMR profile of wildlife. Along those lines, there is evidence to suggest that the prevalence of ARB and ARGs in wildlife can differ when animals are sampled at different times of the year. For example, some rodent and raptor species have been found to have a higher prevalence of ARB and ARG during warmer seasons

(Williams et al. 2011; Miller et al. 2020). Exploring whether the richness of ARB and

ARG and the plasmid-association of ARGs found in wildlife are also likely to differ by season has yet to be done, but could provide insight on whether the same or different

ARB and ARGs contribute to the AMR profile of wildlife over time.

Here, we used whole-genome sequencing and phylogenetic analyses to determine whether wildlife exposed to known anthropogenic sources of AMR were more likely to have a higher prevalence and richness of ARB and ARGs and plasmid-associated ARGs than wildlife that were not exposed. The study took place in the metropolitan area of

Chicago over the course of ten months, and two anthropogenic sources were examined:

1) urban context (urban vs. suburban) (Parker et al. 2016); and 2) presence of a WWTP upstream of sampling sites (Marti, Jofre, and Balcazar 2013; Rizzo et al. 2013;

Wellington et al. 2013). The wildlife species of focus was the raccoon (Procyon lotor) because populations of this species are abundant, widely distributed in urban and suburban areas (Gehrt, Riley, and Cypher 2010; Bateman and Fleming 2012), and raccoons are known to shed ARB and ARGs (Bondo et al. 2016; Bondo et al. 2019;

Worsley‐Tonks et al. 2020). The microorganism of focus was clinically relevant

extended spectrum cephalosporin-resistant (ESC-R) Escherichia coli, which includes both extended-spectrum beta-lactamase (ESBL) and AmpC beta-lactamase producing E. coli. ESC-R E. coli are an emerging issue in human and veterinary medicine (Bradford

2001; Wieler et al. 2011; Woerther et al. 2013), and have been detected in the environment (e.g. Tacão, Correia, and Henriques 2012; Egervärn et al. 2017; Fagerströ et al. 2019) and in the feces of many wildlife species (Guenther, Ewers, and Wieler

2011).

Our objectives were to 1) describe the prevalence, richness, and characteristics of ESC-

R E. coli and associated ARGs in the sampled raccoon population; and 2) determine whether urban context, the presence of a WWTP upstream of capture sites, or season influenced prevalence and richness of ESC-R E. coli sequence types, richness of ARGs in ESC-R E. coli, and prevalence of ESC-R E. coli with plasmid-associated ARGs.

Hundreds of ARGs can confer ESC resistance, but we focused on ARGs from the blaCTX-M and blaCMY families because they are most commonly detected and are of clinical importance (Paterson and Bonomo 2005; Partridge 2015). We predicted that raccoons sampled at urban sites, sites with a WWTP, and in the summer would have a higher prevalence ESC-R E. coli, greater richness of ESC-R E. coli sequence types and associated ARGs, and higher prevalence of ESC-R E. coli with plasmid-associated

ARGs than raccoons sampled at suburban sites, and sites without a WWTP, and in the winter.

Methods

Study site and design

From February-November 2018, raccoons were captured from seven sites in northwestern Chicago (Fig. 1). Sampling took place over the course of four seasons

[winter (mid-December until end of March); spring (beginning of April until end of

June); summer (beginning of July until mid-September); and fall (mid-September until mid-December)]. Three of the capture sites were in urban Chicago (i.e., Damen, DRCA, and Edgebrook; Fig. 1) and four were in suburban Chicago (i.e., Busse, CT, MMWF, and

PC; Fig. 1). Sites were classified as urban if the site and surrounding area were composed of ≥ 80% impervious surface. Otherwise, sites were classified as suburban. Four of the sites had rivers that were downstream from a WWTP (i.e., Busse, Damen, Edgebrook, and MMWF), and three were not downstream from a WWTP (i.e., CT, DRCA, and PC).

Figure 1. Sampling sites in urban and suburban Chicago. The four yellow polygons are sites that had rivers that were downstream from a wastewater treatment plant (WWTP), and the three red polygons are sites not downstream from a WWTP. Damen, DRCA, and Edgebrook are urban sites, and Busse, CT, MMWF, and PC are suburban sites.

Raccoon handling

Raccoons were captured using box traps (Model 108, Tomahawk Live Trap Co.,

Tomahawk, WI, U.S.A.) (as in Prange & Gehrt 2004) and immobilized with an injection of Telazol (Fort Dodge Animal Health, Fort Dodge, Iowa) (Gehrt, Hungerford, and

Hatten 2001). After collecting fecal samples opportunistically from each individual, all captured individuals were aged based on reproductive condition (adult or juvenile) and sexed (male or female). After recovering from immobilization, all animals were released at the capture locations. Fecal samples were stored in brain heart infusion broth and 20% glycerol at -80°C until processing. Captures were approved by the University of

Minnesota’s Institutional Animal Care and Use Committee (protocol ID: 1709-35105A) and by the United States Department of Agriculture (permit number: IDNR W17.0122).

Phenotypic characterization of ESC-R E. coli

To investigate the presence of ESC-R E. coli, we tested E. coli susceptibility to cefotaxime, a 3rd generation cephalosporin. More specifically, samples were enriched in

Lauryl Tryptose Phosphate broth (Difco Laboratories, Detroit, MI, USA) overnight at

37°C and then streaked onto CHROMagar ECC (CHROMagar, Paris, France) containing

2μg/mL of cefotaxime (Albrechtova et al. 2014; Furness et al. 2017). If blue colonies were obtained (indicative of being E. coli), one colony was selected at random, restreaked on CHROMagar ECC containing 2μg/mL of cefotaxime, and incubated overnight at

37°C. Isolates were then grown in 3 ml of LB broth overnight with shaking at 37°C and then stored at -80°C until sequencing.

Genome assembly and gene content analysis

ESC-R E. coli isolates obtained from each sample were subjected to whole genome sequencing (WGS). DNA was extracted from an overnight growth of a single colony for each isolate using the DNeasy Blood and Tissue Kit (Qiagen, Valencia, California) according to the manufacturer’s instructions. WGS of isolates was performed using

NovaSeq Illumina 150bp paired-end sequencing and dual-indexed Nextera XT libraries

(Illumina, USA) at the University of Minnesota Genomics Center (St. Paul, Minnesota,

USA). Raw reads were quality filtered and trimmed using Trimmomatic (version 0.33)

(Bolger, Lohse, and Usadel 2014), which involved removing Illumina Nextera adapters, removing the bases off the start and end of reads if below a threshold quality of 3, having a sliding window of size 4 bp that removed bases if their phred score was < 20, and having the minimum read length be 36 bp.

To identify ARGs, trimmed reads were assembled using the SPAdes assembler (version

3.0) (Bankevich et al. 2012) with default parameters. The quality of assemblies was assessed by examining the N50 score of each isolate (mean = 85,675, range = 32,473-

197,340), which was calculated using QUAST (version 4.3) (Gurevich et al. 2013).

Presence of ARGs on contigs was assessed using the Resfinder database (Zankari et al.

2012), which includes only acquired resistance genes, and not point mutations in chromosomal target genes. Open reading frames were identified using Prokka (version

0.7.17) (Seemann 2014) and were then aligned to the Resfinder database using the NCBI

BLASTn algorithm. ARG presence was based on an identity  90% and a coverage 

80%. When multiple ARG alleles were identified on the same contig and at the same

location (e.g. blaCMY-2 and blaCMY-101), the allele with the highest identity and coverage was selected. In instances when alleles could not be differentiated, the ARG was not classified at the allele level but at the gene family level (e.g. described solely as blaCMY and not blaCMY-2).

Assessing the plasmid- vs. chromosomal-association of ARG conferring ESC- resistance

Given the challenges associated with identifying plasmids from short-read sequencing datasets (Arredondo-Alonso et al. 2017; Orlek et al. 2017), we opted to use two typing programs to classify ARGs conferring ESC-resistance as plasmid or chromosomally- associated: 1) Mlplasmids (Arredondo-Alonso et al. 2018); and 2) MOB-suite (Robertson and Nash 2018). Mlplasmids uses trained machine learning models (specifically Support

Vector Machines), to predict whether contigs are chromosome- or plasmid-associated using pentamer frequencies. Models in this typing program were trained using 583 E. coli genomes (168 chromosomal and 415 with plasmid entities) and have a sensitivity of 71% for detecting the plasmid class (Arredondo-Alonso et al. 2018). The default threshold for classifying contigs as plasmid or chromosome-associated is 50%, but we set the threshold to 70% to reduce the false positive error rate. MOB-suite types and reconstructs plasmids using publicly available Illumina short-read sequencing data. MOB-suite can classify contigs as plasmid-associated with a sensitivity and specificity of 95% and 88%, respectively. In instances where Mlplasmids and MOB-suite predicted different results

(i.e. plasmid in one and chromosomal in the other), we classified contigs as unknown and did not include them in relevant downstream statistical analyses.

Phylogenetic analysis

We explored genetic associations among isolates by subjecting assembled contigs to traditional multilocus sequence typing (MLST) using mlst

(https://github.com/tseemann/mlst) and the in silico E. coli PubMLST typing scheme.

MLST classifies isolates into different sequence types (STs) by exploring the allelic profile of seven housekeeping genes (adk, gyrB, fumC, icd, mdh, purA and recA) unique to E. coli (Wirth et al. 2006). Associations between STs were visualized using minimum spanning trees, which were created in GrapeTree (Zhou et al. 2018).

Deeper phylogenetic associations were explored by performing single-nucleotide polymorphism (SNP)-based phylogenetic analysis from the core genomes of sequenced isolates. A core SNP alignment was created by mapping trimmed reads to the E. coli K-

12 laboratory strain MG1655 genome (Accession number: GCA_000005845.2) using

Snippy version 4.4.0 (https://github.com/tseeman/snippy). Recombinant regions were removed with Gubbins version 2.3.4 (Croucher et al. 2015). A SNP-distance matrix was created using snp-dist version 0.6.3 (https://github. com/tseemann/snp-dists). A maximum likelihood phylogenetic tree was constructed using IQ-TREE version 1.6.12

(Trifinopoulos et al. 2016), where model selection was performed using ModelFinder

(Kalyaanamoorthy et al. 2017), and the tree was validated using 1,000 ultrafast bootstrap repetitions (Hoang et al. 2017). The TVM+F+ASC+R3 model was identified as the best fit based on Bayesian Information Criterion. The resulting phylogenetic tree was visualized and annotated using the iTOL (Interactive Tree of Life) online software

(Letunic & Bork 2016).

Statistical analysis

The importance of urban context, presence of a WWTP at capture sites, or season was examined firstly based on the probability of isolating at least one ESC-R E. coli from raccoons. To do this, we ran a binomial generalized linear mixed model (GLMM) with a logit link function using the ‘lme4’ package (Bates et al. 2014) in R version 4.0.2 (R

Development Core Team 2020). Fixed factors included in the GLMM were urban context

(urban vs. suburban), presence of a WWTP at the site (yes vs. no), and season (fall, winter, spring, summer) (Table 1). Additionally, we included the host factors age

(juvenile vs. adult) and sex (Table 1). Sampling site was included as a random factor because model residuals were significantly spatially autocorrelated (z = 2.83, p = 0.02)

(Table 1) (Dormann et al. 2007), which was tested using a permutation test (999 permutations) for Moran’s I statistic from the ‘spdep’ R package (Bivand et al. 2011).

Spatial standardized weights were calculated using the ‘dnearneigh’ and ‘nb2listw’ functions in the ‘spdep’ package. Because 18 raccoons were captured more than once, we also investigated the need for including ‘animal ID’ as a random factor. To do this, we compared the Akaike information criterion (AIC) values between an intercept model with and without animal ID included as a random factor. There was no significant difference in AIC values between the two models (AIC = 319.49 and 317.9, p = 0.52) indicating that including animal ID as a random factor was not needed (Table 1).

Secondly, we explored whether ESC-R E. coli sequence type richness varied with urban context, presence of a WWTP, and season. Since richness measures can vary greatly with sample size, comparisons were made by sub-sampling the group with the larger sample size 10,000 times to the group with the smaller sample size (as in Mather et al. 2012).

Thirdly, the association between ARG richness in ESC-R E. coli and urban context, presence of a WWTP, and season was assessed by running a GLMM with a Poisson distribution and a log link function using the ‘lme4’package. Capture site was not included as a random factor because model residuals were not significantly spatially autocorrelated (Moran’s I statistic: z = -1.79, p = 0.96). In contrast, including animal ID as a random factor was necessary as it significantly improved model fit (AIC = 629.77 for generalized linear model (GLM) and 604.47 for GLMM, p < 0.0001) and controlled for overdispersion (before including animal ID as a random factor: χ2 = 232.6, p < 0.001; after: χ2 = 73.84, p = 0.99).

Finally, we explored whether urban context, presence of a WWTP, and season influenced the probability of isolating ESC-R E. coli carrying plasmid-associated blaCTX-M or blaCMY from raccoons by running a binomial GLM. Site was not included as a random factor as model residuals were not significantly spatial autocorrelation (Moran’s I statistic: z = -

1.07, p = 0.93). Because of the small sample size for this analysis (n = 62, Table 1), animal ID could not be included as a random factor.

For all models, predictor importance and model fit was performed by starting with a global model and subsequently identifying the most parsimonious model using model selection via the ‘dredge’ function in the ‘MuMIn’ package (Barton 2013). Models were ranked using AIC corrected for small sample size (AICc) (Burnham and Anderson 2002;

Johnson and Omland 2004). If one or more models were within 2 AICc values of the highest-ranking model, model averaging was used to obtain standardized estimates and confidence intervals. Model fit was assessed by calculating the coefficient of determination (r2) (Nakagawa and Schielzeth 2013).

Table 1. Description of the four statistical approaches used to explore the importance of urban context, presence of a WWTP, and season at influencing 1) the prevalence and 2) richness of ESC-R E. coli, 3) the richness of ARG, and 4) prevalence of ESC-R E. coli carrying plasmid-associated ARGs. Outcome variable n Predictor variables Random Analytical factor(s) approach Isolation of at least 230 • urban context • capture Binomial one ESC-R E. coli in (urban/suburban) site GLMM raccoon feces • presence of a WWTP • animal (yes/no) (yes/no) ID* • season (winter, spring, summer, fall) • age (adult/juvenile) • sex (male/female) Richness of 123 • urban context NA Bootstrapping ESC-R E. coli • presence of a WWTP and sequence types • season subsampling Richness of ARGs 123 • urban context • capture Poisson present in • presence of a WWTP site* GLMM ESC-R E. coli • season • animal ID Isolation of at least 62 • urban context NA Binomial one ESC-R E. coli • presence of a WWTP GLM carrying plasmid- • season associated blaCTX-M or blaCMY (yes/no) n = sample size *variable was considered for inclusion as random factor in exploratory analyses but was found to contribute very little to the overall variance (p < 0.05) and was thus excluded from analyses listed here.

Results

Prevalence, richness, and characteristics of ESC-R E. coli and associated ARGs isolated from raccoons

A total of 211 raccoons was sampled between January and November 2018, 17 of which were captured twice and one three times. At least one ESC-R E. coli colony was present

in the fecal samples of 120 raccoons (sample prevalence = 56.9%). Of the 18 raccoons captured more than once, seven had at least one ESC-R E. coli present in feces on both capture events (Table 10). Although ESC-R E. coli were recorded for 128 raccoon fecal samples (including recaptures), WGS was performed on 123 isolates only due to isolation issues and DNA concentration restrictions (isolate NCBI accession numbers can be found in Table 11). MLST analysis revealed that the 123 isolated ESC-R E. coli belonged to 55 known STs and one unknown ST (Fig. 2a). The unknown ST closely resembled ST155

(with variation in the gyrB allele only). The most common STs included ST38, ST68,

ST69, ST162, ST973, and ST1406 (Fig. 2a). The core-SNP based maximum likelihood phylogenetic tree indicated that when raccoons were sampled more than once, isolates did not cluster by raccoon ID (Fig. 2b). Interestingly, while isolates also did not cluster by capture site in general (Fig. 2b), over 50% of isolates obtained from raccoons sampled at DRCA (n = 9) were of the same sequence type (i.e. ST1406; Fig. 2a) and differed by 0

SNPs (Fig. 2b).

Fourteen unique beta-lactam resistance genes (which confer ESC-R) were detected in the isolates, most of which belonged to the blaCTX-M (prevalence = 43.9%), blaCMY

(prevalence = 56.1%), and blaTEM (prevalence = 26.8%) ARG families (Fig. 2c). blaCMY-2 and blaTEM-1B were the most prevalent beta-lactam resistance genes, followed by blaCTX-

M-15, blaCTX-M-55, and blaCTX-M-14 (Fig. 2c). Beta-lactam resistance genes of the blaCTX-M and blaCMY families were distributed throughout the ESC-R E. coli population identified

(Fig. 2b). However, in general, blaCTX-M and blaCMY tended to cluster by ST (Fig. 2a).

The most prevalent non-beta-lactam resistance genes were from the aminoglycoside (e.g.

21.1% for aph(3'')-Ib and 24.4% for aph(6)-Id), tetracycline (e.g. 30.1% for tet(A) and

13% for tet(B)), and sulfonamide classes (e.g. 9.8% for sul1 and 30.1% for sul2), followed by the quinolone class (e.g. 15.4% for qnrS1; Fig. 2c). The median number of

ARGs in an ESC-R E. coli isolate was 4, with over half of isolates having 2-4 ARGs

(53.7%).

Figure 2. Characteristics of ESC-R E. coli and associated ARGs detected in the feces of raccoons sampled in the metropolitan area of Chicago. (A) Minimum spanning tree based on MLST allelic profiles of the 123 ESC-R E. coli isolates recovered from raccoons. Each node represents a unique sequence type (ST) and the size of the node represents the number of isolates classified as each ST. The length of lines connecting nodes represent the number of alleles that are found in common between STs. STs are divided into those that carry blaCTX-M (orange), those that carry blaCMY (blue), and those that carry neither blaCTX-M nor blaCMY but rather blaTEM (grey). (B) core SNP-based maximum likelihood phylogenetic tree of the 123 ESC-R E. coli and heatmap of isolates classified based on ARG families (first 15 bands), capture site (16th band), and raccoon recapture (17th band). The ST of each isolate is also listed on the right side of the heatmap. For ARG families, only those that were detected in more than 10 raccoons are represented on the heatmap. The red color indicates that the ARG family is present. For capture site, each color represents a capture site (purple: Busse, light blue: CT, orange: Damen, red: DRCA, green: Edgebrook, black: MMWF, dark blue: PC). For raccoon recapture, each color represents an individual raccoon). The reference is the laboratory strain E. coli K- 12 MG1655. (C) Percent of all samples for which ARGs were detected in ESC-R E. coli of raccoons. Each color represents an antibiotic class.

When investigating whether contigs with beta-lactam resistance genes were more likely to be chromosomal- or plasmid-associated, we were unable to classify 50% of contigs (77 out of 155 contigs in 52 out of 123 isolates). Mlplasmids predicted chromosomal association when MOB-suite predicted plasmid 11% of the time, and vice versa 39% of the time. In instances when Mlpasmids and MOB-suite predictions were the same, certain beta-lactam resistance genes were more likely to be found on chromosomal- or plasmid- associated contigs. In particular, blaCMY-2, blaTEM-1, blaOXA-1, and blaMOX-4 were typically found on contigs associated with plasmids, whereas blaCTX-M-14, blaCTX-M-15, and blaCMY-27 were generally found on chromosomal contigs (Fig. 3).

Figure 3. Predictive association of contigs carrying beta-lactam resistance genes with chromosome or plasmid location. Predictions were performed using Mlplasmids and MOB-suite. Only contigs that were classified as plasmid- or chromosomally associated by both Mlplasmids and MOB-suite are presented (78 out of 155 contigs. In terms of number of isolates: 71 isolates out of 123 isolates).

Importance of anthropogenic sources at influencing the ESC-R E. coli profile of raccoons

Urban context and season were important predictors of the likelihood of isolating ESC-

R E. coli from raccoons

Season and urban context appeared in all top-ranking models (AICc < 2, Table 12), and were therefore the most important predictors of isolating ESC-R E. coli from raccoons.

Presence of a WWTP and raccoon sex and age each appeared in one of the top four ranking models (Table 12). In general, the top GLMMs explained 39% of the variance for the fixed effects and 55% with site included as a random factor (Table 12). Model averaging revealed that there was a significantly higher probability of isolating ESC-R E. coli from raccoons if they were sampled in the summer and spring than in the winter and fall (Table 2; Fig. 4a) and if raccoons were sampled at urban rather than suburban sites

(Table 2; Fig. 4b). Raccoon age, sex, and the presence of a WWTP did not significantly influence the likelihood of isolating ESC-R E. coli from raccoons (Table 2).

Table 2. Model averaging results from binomial generalized linear mixed models of the probability of isolating at least one ESC-R E. coli from raccoons (n = 230). Predictors were obtained from the top ranking models (AICc < 2; Table 12). Significant terms are depicted in bold. Predictor Mean OR 95% CI Importance Season (spring) 7.68 (2.59 – 22.72) 1.0 Season (summer) 5.2 (2.09 – 12.94) Season (winter) 0.46 (0.19 – 1.11) Urban context (urban) 10.48 (1.45 – 75.93) 1.0 WWTP present (yes) 1.64 (0.3 – 9.16) 0.17 Age (juvenile) 1.34 (0.63 – 2.85) 0.21 Sex (male) 1.18 (0.59 – 2.35) 0.18 Mean OR represents the mean odds ratio, 95% CI the 95% confidence intervals for each mean OR, and Importance the variable importance.

Figure 4. Raw prevalence of ESC-R E. coli in raccoons based on (A) season; and (B) urban context. Whiskers are 95% confidence intervals.

Presence of a WWTP at sampling sites was an important predictor of ESC-R E. coli ST richness, but not ARG richness

Raccoons sampled at sites that were downstream from a WWTP had a greater richness of

ESC-R E. coli sequence types than raccoons sampled at sites that were not downstream from a WWTP (Table 3). This finding remained consistent after sub-sampling the larger group (i.e. WWTP present) to the smaller group (i.e. WWTP not present; Table 3). In

contrast, while raccoons during the summer and spring tended to have a greater richness of ESC-R E. coli STs than raccoons sampled during the fall and winter (Table 3), subsampling the larger groups (i.e. summer and spring) to the smaller group (i.e. fall) revealed that the richness of STs did not vary by season (Table 3). ST richness was also similar between raccoons sampled at urban and suburban sites (Table 3).

Table 3. Richness of ESC-R E. coli sequence types by urban context, presence of a wastewater treatment plant (WWTP) at sampling sites, and collection season. Bootstrap 95% confidence intervals were not presented for predictor levels with the lowest sample sizes. predictor level n ST richness bootstrap 95% CI urban context suburban 57 37 - urban 66 34 (34.01 – 42.30) presence of WWTP no 39 24 - yes 84 44 (30.59 – 39.24) season fall 21 17 - winter 22 22 - spring 27 19 (17.3 – 23.24) summer 53 32 (14.51 – 21.11)

In terms of ARG richness among isolates, model selection revealed that the top two ranking models were the intercept model and a model that only included presence of a

WWTP (Table 13). For the intercept model, animal ID explained 44% of the overall variance (Table 13). In the WWTP model, WWTP explained only 0.3% of the overall variance and thus was not a significant predictor of ARG richness among isolates (p >

0.05).

Presence of a WWTP influenced the probability of isolating ESC-R E. coli carrying plasmid-associated blaCTX-M or blaCMY from raccoons

Presence of a WWTP at sampling sites appeared in the two top ranking models (AICc <

2, Table 14), and was therefore the most important predictor of raccoons shedding ESC-R

E. coli carrying blaCTX-M or blaCMY. Season appeared in one of the two top models, and urban context appeared in none (Table 14). In general, the top GLMs explained a maximum of 22% of the overall variance (Table 14). Model averaging revealed that raccoons had a higher probability of shedding ESC-R E. coli carrying plasmid-associated blaCTX-M or blaCMY if they were sampled at sites with a WWTP than sites without a

WWTP (Table 4; Fig. 5). Season had no significant effect on the probability of raccoons shedding ESC-R E. coli carrying plasmid-associated blaCTX-M or blaCMY (Table 4).

Table 4. Model averaging results from binomial generalized linear models for the probability of isolating at least one ESC-R E. coli carrying plasmid-associated blaCTX-M or blaCMY from raccoons (n = 62). Predictors were obtained from the top ranking models (AICc < 2; Table 14). Significant terms are depicted in bold. Note that the variable ‘urban context’ was dropped during model selection. Predictor Mean OR 95% CI Importance Season (spring) 0.27 (0.03 – 2.82) 0.32 Season (summer) 0.55 (0.05 – 6.4) Season (winter) 0.12 (0.01 – 1.46) WWTP present (yes) 4.1 (1.21 – 13.94) 1.0 Mean OR represents the mean odds ratio, 95% CI the 95% confidence intervals, and Importance the variable importance.

Figure 5. Raw prevalence of ESC-R E. coli carrying plasmid-associated blaCTX-M or blaCMY ARG based on the presence of WWTP at sampling sites. Whiskers are 95% confidence intervals.

Discussion

Our understanding of the importance of anthropogenic sources at shaping the AMR profile of wildlife is in its infancy. Here we show that urban context is an important predictor for the likelihood of isolating ESC-R E. coli from raccoons, with ESC-R E. coli more likely to be identified in raccoons from urban sites than from suburban sites. While the presence of a WWTP at sampling sites did not influence the probability of isolating

ESC-R E. coli, it was an important predictor of both ESC-R E. coli ST richness and the probability of isolating ESC-R E. coli carrying plasmid-associated blaCTX-M or blaCMY.

Season also had an impact on the likelihood of isolating ESC-R E. coli from raccoons, with higher probability in the spring and summer than the fall and winter. Our findings show that both anthropogenic and temporal factors are important at influencing the AMR profile of wildlife.

Detecting ESC-R E. coli in the feces of raccoons is not surprising as ESC-R E. coli have been detected in raccoons and other wildlife species previously (Guenther, Ewers, and

Wieler 2011). However, it is noteworthy that over half of raccoons had ESC-R E. coli and that many were classified as STs of clinical relevance (i.e. commonly identified among infections in humans). In fact, the detection of typically human-associated sequence types such as ST131, ST410, ST10, ST69, and ST23, along with ARGs such as blaCTX-M-15 and blaCTX-M-14, in raccoon feces reinforces the concern that clinically-relevant isolates are present in the environment (Woodford, Turton, and Livermore 2011; Wang et al. 2017) and ARB and ARGs detected in wildlife are of anthropogenic origin (Vittecoq et al. 2016; Wellington et al. 2013). Further, our classifications of beta-lactam ARGs as either chromosomal- or plasmid-associated tended to be similar to those described in human and domestic animal, (Partridge 2015; Hamamoto et al. 2016; 2020; Zurfluh et al.

2015) and wildlife isolates (Guenther et al. 2010; 2017; Tausova et al. 2012; Poirel et al.

2012). Taken together, these findings suggest that the AMR situation unfolding in wildlife might mirror that observed in human and domestic animal communities

(Guenther, Ewers, and Wieler 2011; Wang et al. 2017).

Our result that ESC-R E. coli were more likely to be isolated from raccoons sampled in urban sites than suburban sites is in line with general trends that wildlife that reside close to human-dominated areas are more likely to have a higher prevalence of anthropogenically-derived ARB than wildlife sampled at further distances (Vittecoq et al.

2016; Skurnik 2006; Furness et al. 2017; Dolejska, Cizek, and Literak 2007). What specific factors present in urban areas and absent in suburban areas are driving these differences is unclear and could not be determined in this study. However, several factors are likely to be involved and are probably additive, such as a higher concentration of heavy metals in urban rivers or greater contact with human waste in urban areas

(Almakki et al. 2019; Baker-Austin et al. 2006; Wright and Mason 1999). Combining tools from landscape ecology and wildlife behavior could help tease apart the importance of various environmental factors (Singer, Ward, and Maldonado 2006; Arnold, Williams, and Bennett 2016) and would be an important next step to take.

It was surprising that no association was found between the probability of isolating ESC-

R E. coli and presence of a WWTP at sampling sites, given WWTPs are one primary pathway by which ARB and ARG are disseminated to the environment (Wellington et al.

2013). Further, wildlife that use or reside close to WWTPs are expected to have a higher risk of exposure to ARB and ARGs (Arnold, Williams, and Bennett 2016). In fact, rivers that are downstream of a WWTP are posited to be important AMR exposure pathways for wildlife (Nelson et al. 2008; Radhouani et al. 2014). A lack of association detected in this study could be because sites differed not only by the presence of WWTP, but also in urban context. Since other wildlife research has detected an association with WWTPs

(e.g. Dolejska, Cizek, and Literak 2007; Swift et al. 2019), we expect that WWTPs likely played a role in this study, but that the urban context effect masked the association with

WWTP. However, the presence of a WWTP at sampling sites was a significant predictor of isolating ESC-R E. coli carrying plasmid-associated blaCTX-M or blaCMY. As well as facilitating the dissemination of ARB and ARG to the environment (Wellington et al.

2013), WWTPs can act as hotspots for the horizontal transfer of ARGs among bacteria

(Berendonk et al. 2015; Rizzo et al. 2013). Thus, it is possible that the importance of

WWTPs at facilitating the dissemination of anthropogenically-derived ARB and ARGs to the environment was only apparent when focusing on those ARB that carry plasmid-

associated ARGs. Our results therefore suggest that the association with WWTPs might be more complex than a simple presence-absence association.

Contrary to previous raccoon AMR work (Bondo et al. 2016), we found season to be an important predictor of isolating ESC-R E. coli from raccoons. That said, our finding that isolation of ESC-R E. coli is more likely when raccoons were sampled in the spring and summer is in line with AMR studies in rodents and raptors (Williams et al. 2011; Miller et al. 2020). Seasonal differences could be attributed to ice melt or rainfall increasing the risk of heavy metals and ARB entering river systems, or bacterial growth and proliferation in warmer climates. For example, some ARGs such as tetracycline resistance genes have been found at higher concentrations in rivers during months that experienced highest rainfall (Keen et al. 2018). The month with the highest rainfall in

Chicago is August, a period in which raccoons had a high prevalence of ESC-R E. coli.

Alternatively, seasonal differences could be associated with differences in raccoon diet, which is thought to be influential for AMR in other wildlife systems (e.g. Jobbins and

Alexander 2015). Seasonal differences could also have been caused by differences in raccoon population dynamics. The spring and summer are seasons in which raccoons breed and rear their young, while the fall and winter are seasons during which raccoons disperse and den, respectively (Rosatte et al. 2010). A higher prevalence of ESC-R E. coli in the spring and summer could be associated with a higher contact rate among raccoons.

Further, while we did not detect a significant difference by age, it is possible that younger raccoons were more prone to shedding than older raccoons because of having less stable gut flora, a phenomenon that has been observed in other systems (e.g. Williams et al.

2011; Miller et al. 2018). Stratifying raccoons into a larger number of age classes could have made age differences more distinct.

Another potential reason for seasonal differences could be a function of seasonal differences in antibiotic prescription in the human population. Caucci et al. (2016) found that a rise in ARGs in WWTP outflow coincided with a rise in outpatient antibiotic prescription, both of which occurred during the fall and winter (Caucci et al. 2016). It is possible that an increase in the prevalence of ESC-R E. coli in raccoons in the spring and summer could be a delayed effect of antibiotic pressure occurring in the human population months beforehand. Teasing apart these possibilities requires comparing the

AMR profile of raccoons to those of local humans. An indirect way to make these comparisons would be to examine the interaction effect of anthropogenic factors and season, and we advocate that this be explored in future work.

While ESC-R E. coli ST richness did not differ by season we expect that seasonal differences exist because previous research has found seasonal variation in E. coli subtypes in other settings (e.g. Rödiger et al. 2015). One potential reason for not detecting differences by season could be that exploring richness by groups of animals is less informative than exploring richness within animals. Examining ST richness at the sample level was not done in this study, because only one isolate per sample was sequenced. We suspect that there were host level differences in ST richness by season and anthropogenic sources and we advocate that more research be done in this area.

The finding that raccoons sampled at sites with a WWTP had a greater richness of ESC-R

E. coli STs than raccoons sampled at sites without a WWTP is in line with previous

environmental AMR research (e.g. Akiyama and Savin 2010) and suggests that raccoons at WWTP sites are more likely to display an anthropogenic source profile of AMR.

While several factors could be influencing this finding, it highlights a need to better understand the association between AMR in wildlife and their exposure to waters derived from WWTPs. Freshwater studies have shown that the prevalence, richness, and abundance of ARB and ARGs in river systems tend to be higher downstream than upstream of WWTPs (e.g. Marti, Jofre, and Balcazar 2013; Bueno et al. 2020). Exploring whether similar associations hold true for wildlife sampled up and downstream from a

WWTP would help tease apart the role of WWTP in shaping the AMR profile of wildlife.

A potential limitation of our work is that only two anthropogenic sources of AMR were explored, but there are others that could be important at influencing the AMR profile of raccoons (e.g. urban runoff; Almakki et al. 2019). Indeed, most of our statistical models explained < 40% of the overall variance. It is possible that a large portion of the unexplained variance was attributed to other anthropogenic factors not accounted for in this study. Alternatively, it could be because we did not explore the interface with domestic animals. In urban settings, pets can be important reservoirs and sources of AMR for humans (Guardabassi, Schwarz, and Lloyd 2004), and may also be reservoirs for wildlife. Another important limitation worth attention is the lack of comparison with the anthropogenic source sites themselves. The assumption in this study was that the AMR profile of raccoons was associated with urban context and the presence of a WWTP.

However, no soil or water samples were collected from sites. A final limitation worth mentioning is the chromosomal versus plasmid classification. Predictions made should be taken with caution because the classification programs Mlpasmids and MOB-suite only

agreed 50% of the time. Nevertheless, the fact that several of the predicted plasmid- and chromosomal- associations have been reported in other research, suggests that when

Mlpasmids and MOB-suite agreed, the plasmid/chromosomal classification was likely to be robust.

In conclusion, associations detected with urban context and the presence of a WWTP lends support to the hypothesis that some AMR in wildlife is derived from anthropogenic sources (Vittecoq et al. 2016). However, differences in the importance of each factor at different biological levels of resistance (i.e. prevalence of ARB for urban context versus prevalence of ARB with plasmid-associated ARGs for WWTP) highlights the complex ways in which anthropogenic sources may influence the AMR profile of wildlife.

Importantly, several wildlife AMR studies stress that isolation of ARB and detection of

ARGs differs by wildlife species, and thus different species may play different roles in the dissemination of AMR in the environment (Williams et al. 2011; Jobbins and

Alexander 2015; Vittecoq et al. 2016; Torres et al. 2020). Here we show that there can even be great variation among individuals of the same species. Whether raccoons act as reservoirs or sentinels of AMR remains unknown, but given their association with both terrestrial and aquatic systems (Gehrt and Fritzell 1998; Henner et al. 2004), raccoons may act as important conduits for the introduction of ARB disseminated via rivers into terrestrial systems. More generally, our results lend support for the hypothesis that wildlife exposed to anthropogenic sources of AMR (in our case urban sites and sites with a WWTP) are more likely to have an anthropogenic source profile of AMR. While several investigations have shown that wildlife sampled at sites with anthropogenic sources of AMR tend to have a higher prevalence of ARB, few have explored differences

based on plasmid vs. chromosomal association of ARGs. Comparing the chromosomal vs. plasmid association of ARG using Mlpasmids and MOB-suite was insightful, and we recommend this approach be used in future wildlife and environmental AMR research.

Acknowledgements

Funding was provided by Donna Alexander from the Cook County Animal and Rabies

Control, the Max McGraw Wildlife Foundation, the Forest Preserve District of Cook

County, the National Science Foundation (DEB-1413925 and 1654609), and CVM

Research Office UMN Ag Experiment Station General Ag Research Funds (MIN-62-

098). The authors extend many thanks to the Gehrt lab for field and technical assistance, particularly Gretchen Anchor, Andy Burmesch, Yasmine Hentati, Lauren Ross, Katie

Robertson, Missy Stallard, Sean Sullivan, Steven Winter, and Ashley Wurth. The authors also thank members of the Johnson lab, particularly Bonnie Weber, Alison Millis, and

Emily Clarke for laboratory assistance. Finally, many thanks to the Minnesota

Supercomputing Institute for bioinformatic support.

Chapter 3 – Residential Areas and Wetlands are Associated with a Higher Risk of Isolating Antimicrobial Resistant Bacteria from Peri-Urban Wildlife

Overview

Antimicrobial resistance is a public and animal health issue, and the fate of antimicrobial resistant bacteria (ARB) and antimicrobial resistance genes (ARGs) to the environment remains largely unknown. While ARB and ARG can enter the environment via several pathways, only a handful of pathways have been examined. Here, we use wildlife as sentinels to explore the importance of various landscape features (e.g. residential areas, urban parks) at influencing exposure risk to ARB. Specifically, we asked whether the risk of isolating extended-spectrum cephalosporin resistant Escherichia coli (ESC-R E. coli) from raccoons increased when certain landscape features were present in raccoon home ranges (e.g. residential areas). Six landscape features were examined, including the proportion of residential areas, commercial areas, and wetlands. Of 211 raccoons sampled, ESC-R E. coli were isolated from over half of individuals (56.7%). Proportion of residential areas and wetlands in raccoon home ranges were the most important predictors for isolating ESC-R E. coli from raccoons, along with season. Raccoons with more residential area and wetlands in their home range were more likely to have ESC-R

E. coli. Presence of a river in raccoon home ranges along with proportion of highly developed land, crops, and lawns were not important at predicting isolation of ESC-R E. coli from raccoons. Our findings highlight that environments other than rivers that are downstream from wastewater treatment could be sources of AMR and require further attention.

Introduction

Antimicrobial resistance (AMR) is a threat to public and animal health. While antibiotic consumption is the primary cause of AMR, exposure to antimicrobial resistant bacteria

(ARB) via contact with other individuals or the environment (e.g. water, soil; Huijbers et al. 2015) is also important (Furuya & Lowy 2006; Huijbers et al. 2015). In fact, ARB can be disseminated to the environment from human sources (e.g. wastewater treatment plants (WWTPs) (Wellington et al. 2013; Berendonk et al. 2015). This is concerning for two main reasons. First, little is known about the spread and fate of ARB in the environment, but environments receiving ARB could act as additional pathways of exposure for people and animals (Wellington et al. 2013; Arnold et al. 2016). Second, antimicrobial resistance genes (ARGs) present in anthropogenically-derived ARB could be transferred to bacteria in the environment via plasmids (a process referred to as horizontal gene transfer; Summers 2006), causing widespread dissemination of resistance in environmental bacterial communities.

Most efforts to control the dissemination of AMR to the environment have focused on

WWTPs (reviewed by Pruden et al. 2013). However, in many cases ARB and ARG can remain in the sewage outflow (Korzeniewska, and Harnisz 2013) and be released to rivers and lakes (Huijbers et al. 2015; Bréchet et al. 2014). Further, several other pathways can be involved in the dissemination of ARB and ARG to the environment. For example, urban runoff, leaky septic tanks, storm water discharge, domestic animal waste (Stewart et al. 2007; Surette and Wright 2017; Almakki et al. 2019), and to a lesser extent wildlife

(Graham et al. 2019) can all contribute to ARB contamination of the environment. While

it has become clear that rivers, especially those that are downstream from a WWTPs can be an important exposure pathways (Marti, Variatza, and Balcazar 2014), the importance of other environments is less well understood. Yet, recent work has shown that other environmental factors may also contribute to exposure. For example, urbanization can be an important predictor of environmental AMR (Almakki et al. 2019). Similarly, ARB and

ARG associated with clinics have been detected in wetlands and urban parks (Yan et al.

2019; Adelowo et al. 2020). These findings indicate that a broader look at environmental

AMR may be needed to understand the extent to which the environment may pose a threat to human and animal health.

One way to take a broader look at environmental AMR is to examine AMR in wildlife, and explore whether the AMR profile of wildlife is associated with specific environmental factors. ARB have been isolated from numerous wildlife species (Vittecoq et al. 2016; Ramey and Ahlstrom 2020), and because exposure risk tends to increase with proximity to human-dominated areas (Skurnik et al. 2006; Dolejska, Cizek, and Literak

2007; Furness et al. 2017), wildlife are considered good sentinels for AMR (Vittecoq et al. 2016; Furness et al. 2017; Torres et al. 2020; although see Swift et al. 2019). Thus far, most wildlife AMR research has explored links to human sources by comparing the AMR profile of wildlife that are exposed and not exposed to anthropogenic pressures. For example, comparisons have been made between urban and non-urban wildlife (Carter et al. 2018; Atterby, Ramey, Gustafsson Hall, et al. 2016) and between wildlife exposed and not exposed to livestock (Kozak et al. 2009). Whilst differences in ARB prevalence between wildlife exposed and not exposed to anthropogenic pressures have been detected, an important next step is to determine which environmental factors at

anthropogenic sites increase exposure risk (Arnold, Williams, and Bennett 2016; Ramey and Ahlstrom 2020).

One approach to explore environmental sources at a finer scale is to investigate whether certain landscape features at sampling locations (e.g. residential areas, wetlands) are associated with a higher risk of isolating ARB from wild animals. This type of approach is frequently used for identifying pathways by which wildlife are exposed to pathogens

(e.g. Carver et al. 2016). In many cases, landscape features are quantified in the area where wild animals are sampled (e.g. Gras et al. 2018), or within the estimated home range of radio-collared wild animals (e.g. Sevila et al. 2014). Taking a similar approach for ARB could help identify environmental sources and tease apart the importance of various anthropogenic pressures (e.g. Ahlstrom et al. 2019).

Here, we investigated the importance of various landscape features at influencing the occurrence of ARB in raccoons (Procyon lotor) that were sampled in the metropolitan area of Chicago, USA. Raccoons are ideal for exploring associations between ARB occurrence and landscape features because they utilize numerous habitats (e.g. anthropogenic, natural, and aquatic; Bateman and Fleming 2012) and have relatively small home ranges (~1-3 km2; Šálek, Drahníková, and Tkadlec 2015), making it possible to relate ARB occurrence with the environment where raccoons were sampled. The ARB of focus were extended-spectrum cephalosporin resistant (ESC-R) Escherichia coli. ESC-

R E. coli are typically associated with clinical settings (Bradford 2001) but there is increasing evidence of onward transmission outside of clinics (Arpin et al. 2005; Pitout et al. 2005), and ESC-R E. coli have been identified in rivers (e.g. Lenart-boroń, Kulik, and

Jelonkiewicz 2020) and in wildlife (Guenther et al. 2011).

Our objective was to determine how the probability of isolating ESC-R E. coli associates with landscape features present in the predicted home range of each raccoon. Several landscape features were explored, such as the presence of rivers and streams and proportion of residential areas and wetlands in raccoon home ranges. Since rivers are recognized as important exposure AMR pathways, and ESC-R E. coli have been detected in some cases (e.g. Lenart-boroń, Kulik, and Jelonkiewicz 2020), we expected the presence of a river or stream in raccoon home ranges to be an important predictor of isolating ESC-R E. coli from raccoons. Further, based on previous work (Allen et al.

2011), we expected wetlands and residential areas to be important (Adelowo et al. 2020).

Methods

Study area

In January - November 2018, we captured raccoons in six core areas in the Northwestern portion of the Chicago metropolitan area. The six sites differed in public access, rivers connected to WWTPs, and landcover type (Fig. 6a). Three sites were open for public access (i.e. Busse, Edgebrook, and PC) and three sites were not (i.e. Crabtree, DRCA, and MMWF; Fig. 6a). Three sites had rivers that were downstream from a WWTP (i.e.

Busse, Edgebrook, and MMWF) and three sites did not (i.e. Crabtree, DRCA, and PC;

Fig. 6a). Sites furthest away from the core of Chicago (i.e. Busse, Crabtree, PC, and

MMWF) were mostly composed of low intensity developed land (i.e. single housing units), open developed space (e.g. parks and golf courses), and natural and cropland areas

(Fig. 6b), while sites closest to the core of Chicago (i.e. Edgebrook and DRCA) were

mostly composed of low, medium, and high developed land (i.e. residential areas, apartment complexes, and industrial areas) with some open developed space (Fig. 6b).

Figure 6. Study sites. (A) Map of the Chicago metropolitan area. The six study sites are represented as polygons. Yellow polygons are sites that have rivers that are downstream from a wastewater treatment plant (WWTP) and red polygons are sites that have rivers that are not downstream from a WWTP. (B) Landcover proportions at each sampling site. Landcover types were obtained from the 2011 National Landcover Database (NLCD).

Raccoon sampling

Raccoons were captured using box traps (Model 108, Tomahawk Live Trap Co.,

Tomahawk, WI, U.S.A.) (as in (Prange and Gehrt 2004), and were transported to a research laboratory. At the research laboratory, raccoons were immobilized with an injection of Telazol (Fort Dodge Animal Health, Fort Dodge, Iowa). Faecal samples were collected opportunistically from the rectum of immobilized raccoons and were stored in

20% glycerol and brain heart infusion broth at -80°C. After recovering from immobilization, all raccoons were released at the capture locations. Raccoon captures were approved by the University of Minnesota (IACUC ID: 2013A00000012-R1) and the

United States Department of Agriculture (permit: IDNR W17.0122).

Testing for the presence of ESC-R E. coli

To investigate the presence of ESC-R E. coli, we tested E. coli susceptibility to cefotaxime. We enriched the samples in Lauryl tryptose phosphate broth (Difco

Laboratories, Detroit, MI, USA) overnight at 37⁰C, then streaked them onto

CHROMagar ECC containing 2μg/mL of cefotaxime (Albrechtova et al. 2014; Furness et al. 2017). If blue colonies were obtained (representative of E. coli), one blue colony was selected at random, restreaked on CHROMagar ECC containing 2μg/mL of cefotaxime, and incubated overnight at 37⁰C.

Estimating raccoon home ranges

Landcover proportions were calculated within raccoon estimated home ranges. Typically, animal home ranges are estimated by fitting animals with very high frequency (VHF) or global positioning system (GPS) collars and collecting locations from each animal at set time points. In this study, none of the raccoons were collared. However, 124 raccoons were fitted with VHF collars as part of previous work in Chicago (range of years: 1995-

2017) and can be used to predict the mean home range of raccoons sampled in this study

(see Supplementary materials for details on telemetry data gathered on raccoons tracked in previous years). Each home range was estimated using a circular buffer, with the

centre of the buffer being the capture location. The diameter of the buffer represented the mean home range size for raccoons sampled at each site. The mean home range size was estimated by calculating and plotting 95% minimum convex polygon (MCP) home range estimates for each VHF-monitored raccoon using the ‘adehabitatHR’ package (Calenge

2015) in the statistical software R (R Development Core Team 2020) and subsequently calculated a mean home range for raccoons sampled at each one of the six sites.

Landscape features

We considered six landscape features for inclusion in statistical models. One landscape feature was the presence of a river or stream in raccoon home ranges. River and stream shapefiles were obtained from the Cook County Maps and Geospatial Data website

(https://www.cookcountyil.gov/CookCentral). All other landscape features were proportions of one or several landcovers present in raccoon home ranges, and were obtained from the 2011 National Landcover Database (2011 NLCD)

(https://www.mrlc.gov) (spatial resolution: 30 meters). Since we expected landcovers associated with anthropogenic activity and wetland areas to be important predictors for isolating ESC-R E. coli from raccoons, and natural green spaces to be less important, we grouped landcover data into six variables: 1) residential areas (representing ‘developed, low intensity’ and ‘developed, medium intensity’ in NLCD), 2) highly developed areas that included apartment complexes and industrial areas (‘developed, high intensity’ in

NLCD); 3) wetlands (‘woody wetlands’ and ‘emergent herbaceous wetlands’ in NLCD) ;

4) lawn areas such as urban parks or golf courses (i.e. ‘developed, open space’ in NLCD);

5) pasture or cropland areas (‘pasture/hay’ and ‘cultivated crops’ in NLCD); and 6)

natural vegetation (representing all forest, shrubland, and herbaceous landcovers in

NLCD). Natural vegetation was excluded from further analyses because of being highly correlated with residential areas (r > 0.6).

Statistical analysis

The association between probability of isolating ESC-R E. coli and landscape features was explored using a binomial generalized linear mixed models (GLMM) using the

‘lme4’ R package (Bates et al. 2014). As well as exploring associations with the six landscape features, season was included as a fixed factor since it was found to be important at predicting ESC-R E. coli occurrence in raccoons in previous work (Worsley-

Tonks et al., Chapter 2). Study site was included as a random intercept. To identify the most parsimonious model, we used an information theory approach, comparing GLMMs with different variable combinations, and used Akaike Information Criterion corrected for small sample size (AICc) to rank models (Burnham and Anderson 2002; Johnson and

Omland 2004) using the ‘MuMIn’ package in R (Barton 2013). Models within < 2

AICc values of the top ranking model were combined to obtain mean effect sizes and

95% confidence intervals (a method often referred to as ‘model averaging’(Burnham and

Anderson 2002). All continuous predictors were centred and standardized to facilitate interpretation of main effects and to perform model averaging (Grueber et al. 2011).

Multicollinearity among predictors was assessed by calculating the variance inflation factor using the ‘car’ package. Model fit was assessed by calculating the marginal and

2 2 conditional coefficients of determination (rm and rc , respectively) (Nakagawa and

2 2 Schielzeth 2013). rm is the variance explained by the fixed predictors, and rc the variance explained by the fixed and random predictors (Nakagawa and Schielzeth 2013).

Results

Of the 215 raccoons tested for ESC-R E. coli, 112 shed at least one ESC- R E. coli isolate

(56.7%). Raccoons sampled at Edgebrook, DRCA, and PC had the highest prevalence

(70%, 64.3%, and 58% respectively), followed by raccoons sampled at Busse and

MMWF (43.8% and 40.5% respectively) and subsequently Crabtree (13.6%).

Model selection revealed that season and the proportion of residential areas and wetlands present in raccoon home ranges were the most important predictors of isolating ESC-R E. coli from raccoons because all three variables appeared in all top ranking models (Table

5). As shown in previous work (Worsley-Tonks et al. Chapter 2), the likelihood of isolating ESC-R E. coli from raccoons was higher when raccoons were sampled in the spring and summer than winter and fall (Table 6). Additionally, there was a positive association between isolation of ESC-R E. coli and proportion of both residential areas and wetlands in raccoon home ranges (Table 6; Fig 7). Proportion of crops, lawn, and high developed land were not significant predictors of isolating ESC-R E. coli as all three predictors appeared in only three of the seven top ranking models (Table 5). Additionally, presence of a river or stream in raccoon home ranges was the least important predictor and appeared in none of the top-ranking models.

Table 5. Top ranked binomial generalized linear mixed models (AICc < 2 from the best fit model) for isolating at least one ESC-R E. coli from raccoons. 2 2 Model Predictors included in each model k AICc wi r m r c 1 season + residential areas + wetlands 7 0.00 0.11 0.35 0.43 2 season + residential areas + wetlands + 8 0.06 0.11 0.36 0.48 crops 3 season + residential areas + wetlands + 8 0.35 0.09 0.34 0.44 lawn 4 season + residential areas + wetlands + 8 1.10 0.06 0.35 0.42 high developed 5 season + residential areas + wetlands + 9 1.28 0.06 0.34 0.43 lawn + high developed 6 season + residential areas + wetlands + 9 1.36 0.04 0.34 0.47 crops + lawn 7 season + residential areas + wetlands + 9 1.94 0.04 0.36 0.47 crops + developed high

Table 6. Model averaging results from binomial generalized linear mixed models of the probability of isolating at least one ESC-R E. coli from raccoons. Predictors were obtained from the top ranking models (Table 1). Significant terms are depicted in bold. Predictor Mean OR 95% CI Importance Season (spring) 10.51 (3.2 – 34.55) 1.0 Season (summer) 7.17 (2.72 – 18.89) Season (winter) 0.57 (0.22 – 1.46) Residential areas 1.87 (1.12 – 3.13) 1.0 Wetlands 2.46 (1.32 – 4.59) 1.0 Crops 1.31 (0.85 – 2.03) 0.39 Lawns 0.76 (0.48 – 1.19) 0.39 Developed high 0.8 (0.5 – 1.29) 0.31 Mean OR represents the mean odds ratio, 95% CI the 95% confidence intervals for each mean OR, and Importance the variable importance.

Figure 7. Effect of (A) wetlands and (B) residential areas on the probability of isolating ESC-R E. coli from raccoons. Black lines represent smooth functions and grey area the 95% confidence intervals.

Discussion

The fact that ESC-R E. coli were isolated from over half of the 215 raccoons sampled emphases two main points. First, ARB typically associated with clinical settings are common in synanthropic wildlife (Vittecoq et al. 2016; Ramey and Ahlstrom 2020), which has important implications for human and domestic health. Second, finding clinically-relevant ARB such as ESC-R E. coli in raccoons suggests that environments used or resources exploited by these wild animals are sources of AMR (e.g. human waste) or are receiving clinically-relevant ARB and ARG from anthropogenic sources

(e.g. aquatic environments). Taken together these points highlight a need to identify environmental factors that may contribute to wildlife exposure. By estimating the home ranges of raccoons, we identified landscape features within urbanized landscapes that could contribute to raccoon exposure to ESC-R E. coli.

One landscape feature that appeared as a top predictor in all statistical models was the proportion of residential areas present in raccoon home ranges. The more residential area present in raccoon home ranges the more ESC-R E. coli were likely to be isolated from

raccoon feces. This finding is in line with previous work showing that wildlife present in urban areas, or that are close to urban areas, have a higher risk of shedding ARB than wildlife present in non-urban areas, or further away from urban areas (Skurnik et al.

2006; Dolejska, Cizek, and Literak 2007; Furness et al. 2017). However, identifying this trend at the home range level provides more insight on the potential pathways of exposure. Given that raccoons exploit human waste, it is possible that raccoons are exposed when foraging in residential areas. Alternatively, since ESC-R E. coli have mostly been detected in aquatic environments (Guenther, Ewers, and Wieler 2011;

Hooban et al. 2020), water bodies present in residential areas may be sources of ESC-R

E. coli. This is especially likely to be the case if households have septic tanks rather than sewage systems. Several lines of evidence suggest that septic tanks and leaky pipes can be important sources of ARB, and groundwater can facilitate widespread dissemination in the local environment. An important next step would be to determine whether higher risk of isolating ESC-R E. coli in wildlife is due to use of residential areas or to the fact that wildlife reside in parks surrounded by residential areas. Comparing the AMR profile between two wildlife species that differ in their use of residential areas could help tease apart this relationship.

Proportion of wetland areas in raccoon home ranges was also positively correlated with the isolation of ESC-R E. coli from raccoons. This is an important finding because most wildlife AMR research has focused on exploring the importance of rivers as potential exposure pathways for wildlife and little attention has been given to wetlands. Thus, this finding emphasizes that multiple environmental factors should be considered when exploring exposure pathways for wildlife. Wetland areas, in particular, may be important

additional environmental sources of AMR because of being at the receiving end of rivers, streams, and urban runoff. While no water or soil samples were collected from wetlands as part of this study, given that previous environmental AMR research has detected ARB and ARG from wetlands (e.g. Adelowo et al. 2020) and raccoons forage in and around water bodies more generally (Stuewer 1943), wetlands could be an important AMR exposure pathway and require further attention.

While the proportion of wetlands in raccoon home ranges was important for predicting the isolation of ESC-R E. coli from raccoons, the presence of a river or stream in raccoon home ranges was not significant. In fact, it was the least important predictor for isolating

ESC-R E. coli. This was surprising given that rivers, especially those that are downstream from a WWTP can be reservoirs of AMR in the environment (Marti, Variatza, and

Balcazar 2014). One reason for not detecting an association could be that rivers present in raccoon home ranges did not have ESC-R E. coli, and that raccoons were exposed through other environmental pathways. That said, given that aquatic wildlife species, especially aquatic avian species are considered to be more prone to exposure to ARB, we suspect that rivers played a part in shaping the AMR profile of raccoons.

It is possible that a lack of association detected with rivers was because we sampled raccoons across multiple seasons, yet dissemination of ARB and ARG to aquatic systems can vary by season (Keen et al. 2018; Garner et al. 2017). Season was an important predictor for isolating ESC-R E. coli in the raccoons, where raccoons sampled during the spring and summer were more likely to have ESC-R E. coli than raccoons sampled during the fall and winter (a discussion of the importance of season at shaping the AMR profile of raccoons can be found in Worsley-Tonks et al. Chapter 2). Another potential

reason for not detecting an association with rivers could be due to the way in which raccoon home ranges were estimated. A circular buffer around capture locations may not adequately estimate the home range of raccoons, especially if the capture location is on the fringe of the ‘true’ home range of animals. Thus, the importance of rivers at shaping the AMR profile of raccoons requires further investigation.

In conclusion, by exploring the importance of several landscape features at influencing the occurrence of ESC-R E. coli in raccoons, we show that residential areas and wetlands are important predictors for isolating ESC-R E. coli from raccoons. These findings are timely given that recent environmental AMR research has stressed the need to take a broader look at AMR in the environment (Ramey and Ahlstrom 2020; Hooban et al.

2020) Further, this study reinforces the need to explore the importance of non-point sources of AMR at influencing the dissemination of ARB and ARG to the environment

(Yan et al. 2019; Hooban et al. 2020), and suggests that examining factors associated with residential areas and wetlands could be a useful starting point.

Chapter 4 - Antimicrobial Resistant Bacteria at the Wildlife- Domestic Animal Interface – Is Shared Space Important for Microbe Sharing?

Overview

Wildlife can be exposed to anthropogenically-derived antimicrobial resistant bacteria

(ARB) via many pathways. Spatial overlap with domestic animals is ranked as a prominent pathway. However, in some cases the interface with domestic animals appears to not be important. These differences could in part be associated with the way in which wildlife-domestic animal interfaces are quantified. Here, we investigated the prevalence and phylogenetic relatedness of extended-spectrum cephalosporin resistant (ESC-R)

Escherichia coli in raccoons (Procyon lotor) and domestic dogs (Canis lupus familiaris).

We asked whether the presence of one species influenced the prevalence of ESC-R E. coli in the other species, and whether different types of spatial overlap influenced phylogenetic associations of ESC-R E. coli isolated from dogs and raccoons. We sampled

211 raccoons and 176 dogs in seven sites in northwestern Chicago, USA. Fecal samples were cultured and tested for the presence of ESC-R E. coli to investigate prevalence, and whole-genome sequencing was used to explore phylogenetic associations. Four metrics were used to quantify the spatial overlap between raccoons and dogs: (1) presence of dogs at raccoon sites; (2) raccoon predicted use of residential areas; (3) frequency at which dogs visit raccoon sites; and (4) dog home mailing zip code overlap with raccoon sites. Raccoons had a significantly higher prevalence of ESC-R E. coli (55.2%) than dogs

(16.5%). However, phylogenetic analyses revealed that ESC-R E. coli identified in raccoons and dogs were not genetically distinct. Dog spatial overlap with raccoons was

not important for isolating ESC-R E. coli from dogs. In contrast, raccoon overlap with dogs was important for the isolation of ESC-R E. coli from raccoons, but only at suburban, not urban sites. In terms of phylogenetic relatedness of ESC-R E. coli, none of the four spatial overlap metrics examined were important at influencing the phylogenetic relatedness of ESC-R E. coli isolated from raccoons and domestic dogs. A higher prevalence of ESC-R E. coli in raccoons than dogs underscores a need to incorporate wildlife into antimicrobial resistance surveillance efforts. Further, the varying importance of different types of spatial overlap at influencing the prevalence and phylogenetic associations of ESC-R E. coli among dogs and raccoons highlights the complexity of wildlife-domestic animal interfaces. Combining bacterial genomic information with animal behavior data can help elucidate which interfaces should be targeted for management and surveillance.

Introduction

Human encroachment into natural habitats and urbanization increase the extent to which humans and domestic animals interface with wildlife. Greater contact between humans, domestic animals, and wildlife can increase the risk of infectious agent spillover (Daszak,

Cunningham, and Hyatt 2000; Plowright et al. 2017). Our understanding of this phenomenon has mostly been driven by pathogen spillover from wildlife into human or domestic animal populations (Karesh et al. 2012; Plowright et al. 2017). However, infectious agents can also spillover from human sources into wild animal populations through the environment, which can threaten public and domestic animal health if

wildlife cause further spread and spill-back into the human and/or domestic animal populations (Daszak, Cunningham, and Hyatt 2000).

A quintessential example of this is the dissemination of anthropogenically-derived antimicrobial resistant bacteria (ARB) (Carroll et al. 2015; Radhouani et al. 2014;

Dolejska and Literak 2019). Wildlife shedding of ARB has been recorded from many different species, sampled in different ecosystems, and identified across continents

(Vittecoq et al. 2016; Dolejska and Literak 2019). In general, wild animals are more likely to shed anthropogenically-derived ARB if they are closer to human dominated areas (Österblad et al. 2001; Allen et al. 2010). Wildlife can be exposed via numerous pathways (e.g. rivers, landfills, sewage runoff; reviewed in Allen et al., (2010) and

Vittecoq et al., (2016)), and the order of importance of each pathway is a topic of intense debate (Ramey & Ahlstrom 2020).

One pathway that has received attention is the interface with domestic animals. In general, wildlife that co-exist with domestic animals tend to host bacteria with similar levels of resistance to various antimicrobials (e.g. Kozak et al. 2009; Literak et al. 2009), and in some cases have phylogenetically similar ARB to those from local domestic animals (e.g. Rwego et al. 2008; Pesapane et al. 2013; Mercat et al. 2016). However, wildlife that co-exist with domestic animals have also been found to shed bacteria that are susceptible to available antimicrobials, or that are resistant but phylogenetically distinct from those shed by proximate domestic animals (e.g. Benavides et al. 2012; Viswanathan et al. 2016; Bonardi et al. 2019). These differences could in part be associated with the type of bacteria used to trace transmission or sharing of ARB (e.g. Enterococcus,

Escherichia coli; Tormoehlen et al., 2019), the types of host species involved (e.g.

Jobbins and Alexander 2015), and to some extent the isolation technique and resolution of molecular tools being used (Boolchandani, Souza and Dantas, 2019).

Another potential reason for these differences could be associated with the degree to which wildlife and domestic animals overlap in space and time, and the ecological context in which interfaces occur. Investigations of antimicrobial resistance (AMR) at the wildlife-domestic animal interface have mostly been based solely on phylogenetic relatedness of ARB, and on whether or not wildlife and domestic animals are present in the same environment. However, the type of spatial overlap quantified will likely influence the extent to which domestic animals and wildlife share microbes (Rwego et al.

2008; Vanderwaal et al. 2014). Combining bacterial genetic information with animal behavior data could help identify important links, and this approach has already proven useful for understanding other potential pathways by which wildlife could be exposed.

For example, combining bacterial genetics with animal movement and inter-specific interaction data has helped understand wildlife exposure to AMR via landfills, and the potential for onward transmission within wild animal populations (e.g. Ahlstrom et al.

2019; Miller et al. 2019). Taking a similar approach for the wildlife-domestic animal interface could provide insight on whether domestic animals are an important pathway of exposure for wildlife, or whether other anthropogenic pressures are more influential.

Here, we investigated the sharing of ARB between wildlife and domestic animals and how extent of sharing varies with the type of metric used to quantify wildlife and domestic animals spatial overlap. Most wildlife AMR research exploring the importance of the domestic animal interface, thus far, has focused on livestock (Kozak et al. 2009;

Mercat et al. 2016; Woolhouse et al. 2015; Hassell et al. 2019). Little is known regarding

the companion animal – wildlife interface, which is surprising given the potential role of companion animals in the transmission cycle of ARB (Ewers et al. 2012; Guardabassi,

Schwarz, and Lloyd 2004; Damborg et al. 2016; Pomba et al. 2017), their close contact with humans, and frequent overlap with wildlife, particularly urban-adapted wildlife (e.g. mice, raccoons; Gehrt, Riley and Cypher, 2010; Mackenstedt, Jenkins and Romig, 2015).

Raccoons (Procyon lotor), in particular, are an ideal wild animal to investigate the sharing of ARB between wildlife and domestic animals as they are urban exploiters

(Bateman and Fleming 2012; Gehrt, Riley, and Cypher 2010), share several infectious agents with domestic dogs (e.g. canine distemper virus, Leptospira and Salmonella spp.;

Rentería-Solís et al., 2014; Very et al., 2016; Straub et al., 2020), and are known to shed

ARB and ARGs (Bondo et al. 2016; Worsley‐Tonks et al. 2020). The microorganism of focus here was extended-spectrum cephalosporin resistant Escherichia coli (ESC-R E. coli). ESC-R E. coli are of increasing concern for public and animal health (Pitout et al.

2005; Mathers, Peirano, and Pitout 2015), are increasingly reported in companion animals (Wieler et al. 2011; Cummings, Aprea, and Altier 2015; Liu, Thungrat, and

Boothe 2016), and have been reported in over thirty wildlife species (Guenther et al.

2011).

Our objectives were to 1) explore the extent to which raccoons and domestic dogs have similar ESC-R E. coli profiles in terms of prevalence and phylogenetic relatedness; 2) determine whether spatial overlap with dogs influences the probability of isolating ESC-

R E. coli in raccoons, and vice versa for domestic dogs; and 3) investigate whether spatial overlap between raccoons and dogs influences the phylogenetic association of ESC-R E. coli isolated from raccoons and domestic dogs. Spatial overlap was quantified using four

metrics: 1) presence of dogs at raccoon sites; 2) raccoon predicted use of residential areas; 3) frequency at which dogs visit raccoon sites; and 4) dog home mailing zip code overlap with raccoon sites. We hypothesized that domestic dogs would have a higher prevalence of ESC-R E. coli than raccoons because of antimicrobial use in dogs, and that the presence of dogs would increase the risk that raccoons having ESC-R E. coli. Further, we expected dogs and raccoons to have similar ESC-R E. coli. However, we expected the strength of genetic associations among isolates recovered from dogs and raccoons to be more pronounced when raccoons were predicted to use residential areas and when dogs regularly visited raccoon sites.

Methods

Study site and design

In February-November 2018, raccoons were captured from seven sites in northwestern

Chicago, of which four were suburban and three were urban (Fig. 8). Sites were classified as urban if the site and surrounding area (i.e. ~1.5-2 km buffer around each site) were composed of ≥ 80% impervious surface. Otherwise, sites were classified as suburban.

Out of the seven sites, three were considered to be ‘dog use’ sites because they were accessible to people and domestic dogs (i.e., Busse, Edgebrook, and PC), and four were

‘non-dog use’ sites as they were not accessible to people and domestic dogs (i.e., CT,

Damen, DRCA, and MMWF). Domestic dogs were sampled at each of the three ‘dog use’ sites where wildlife were captured (i.e. Busse, Edgebrook, and PC). Domestic dogs were also sampled at dog parks (a park in which dogs mingle off leash) that were closest to three of the dog use sites (i.e. Busse, Edgerbook, and PC; Fig. 8).

Figure 8. Sampling sites in the northwestern portion of the Chicago metropolitan area. Polygons depict raccoon sites. The four red polygons are sites where domestic dogs were not allowed to enter (i.e. CT, Damen, DRCA, and MMWF). The three yellow polygons are sites where dogs were allowed to enter (i.e. Busse, Edgebrook, and PC). Blue stars represent dog parks.

Raccoon and dog sampling

Raccoons were captured using box traps (Model 108, Tomahawk Live Trap Co.,

Tomahawk, WI, U.S.A.) (as in Prange & Gehrt, 2004) and immobilized with an injection of Telazol (Fort Dodge Animal Health, Fort Dodge, Iowa). Fecal samples were collected opportunistically from the rectum of each immobilized raccoon. After recovering from immobilization, all raccoons were released at the capture locations. Captures were approved by the University of Minnesota’s Institutional Animal Care and Use Committee

(protocol ID: 1709-35105A) and by the Illinois Department of Natural Resources (permit number: IDNR W17.0122).

For every dog sampled, a standardized survey (Table 15) was given to dog owners detailing the age and sex of each dog, as well as history of antibiotic use in the past year.

Information on dog breed was collected but not included in analyses as most dogs were of mixed breeds. Dog owners were also asked about the frequency at which they visited the park where the dogs were sampled (e.g. 5 times a week), as well as their home zip code. Dog fecal samples were collected by dog owners using their own dog waste bags or bags provided by investigators. All dog and raccoon fecal samples were stored in brain heart infusion broth and 20% glycerol at -80⁰C until further analyses.

Classification of sites based on raccoon predicted use of residential areas

Classifying sites as ones in which raccoons are predicted to use or not use residential areas was done by using very high frequency (VHF) telemetry data collected on raccoons captured and radiotracked in previous years (1990-2017) (i.e. raccoons were not radiocollared during this study). Methods and results are explained in full in the

supplementary materials. From the results in the supplementary materials, CT, DRCA,

Edgebrook, and PC were classified as sites where raccoons are predicted to use residential areas, and Busse, Damen, and MMWF were classified as sites where raccoons are predicted to not use residential areas.

Phenotypic characterization of ESC-R E. coli

To investigate the presence of ESC-R E. coli, we tested E. coli susceptibility to cefotaxime. We enriched the samples in Lauryl tryptose phosphate broth (Difco

Laboratories, Detroit, MI, USA) overnight at 37⁰C, then streaked them onto

Sequencing, bioinformatics, and phylogenetic analyses

ESC-R E. coli isolates obtained from each sample were subjected to whole genome sequencing (WGS). DNA was extracted using the DNeasy Blood and Tissue Kit (Qiagen,

Valencia, California) according to the manufacturer’s instructions. WGS of isolates was performed using NovaSeq Illumina 150 bp paired-end sequencing with dual-indexed

Nextera XT libraries (Illumina, USA) at the University of Minnesota Genomics Center

(St. Paul, Minnesota, USA). Raw reads were quality checked and trimmed using

Trimmomatic version 0.33 (Bolger, Lohse, and Usadel 2014), which involved removing

Illumina Nextera adapters, removing the bases off the start and end of reads if below a threshold quality of 3, having a sliding window of size 4 bp that removed bases if their phred score was < 20, and having the minimum read length be 36 bp.

Genetic associations among isolates were explored by determining the multi-locus sequence type (MLST) of each isolate. To do this, trimmed reads were assembled using

SPAdes assembler (version 3.0) (Bankevich et al. 2012) with default parameters. The quality of assemblies was assessed by examining the N50 score of each isolate (mean =

85,675, range = 32,473-197,340) which we calculated using QUAST (version 4.3)

(Gurevich et al. 2013). Isolates were then classified into different sequence types (STs) using mlst (https://github.com/tseemann/mlst) and the in silico E. coli PubMLST typing scheme. Associations between STs were visualized using minimum spanning trees, which were created in GrapeTree (Zhou et al. 2018).

12 laboratory strain MG1655 genome (Accession number: GCA_000005845.2) using

(Trifinopoulos et al. 2016), where model selection was performed using ModelFinder

(Kalyaanamoorthy et al. 2017), and the tree was validated using 1,000 ultrafast bootstrap repetitions (Hoang et al. 2017). The TVM+F+ASC+R3 model was identified as the best

fit based on Bayesian Information Criterion. The resulting phylogenetic tree was visualized and annotated using the iTOL (Interactive Tree of Life) online software

(Letunic and Bork 2016).

Statistical analysis

Objective 1: Similarity of ESC-R E. coli isolated from raccoons and dogs based on prevalence, richness, and phylogenetic relatedness

The sample prevalence of ESC-R E. coli and 95% confidence intervals for raccoons and dogs were calculated using the ‘prevalence’ package in R version 4.0.2 (R Development

Core Team 2020). Comparisons of the prevalence of ESC-R E. coli by species were performed using the Fischer’s exact test (Table 7). Using a similar approach to Mather et al. (2012), the richness of ESC-R E. coli STs (number of unique STs found in raccoons and dogs. STs were characterized using multilocus sequence typing (MLST)) were compared between raccoons and domestic dog populations by bootstrapping the raccoon sample (n = 124) to the size of the dog sample (n = 29) using 1,000 replicates with replacement. Deeper phylogenetic associations between ESC-R E. coli isolated from domestic dogs and raccoons were explored by quantifying the pairwise genetic distance between isolates using the ‘cophenetic.phylo’ function in the ‘ape’ R package (Paradis,

Claude, and Strimmer 2004). Phylogenetic clustering by animal species (dog vs. raccoon) was assessed by performing permutational multivariate analysis of variance

(PERMANOVA) using the ‘adonis2’ function in the ‘vegan’ package (Oksanen et al.

2013) with the number of permutations set to 999. PERMANOVA can be used on any type of pairwise matrix (Anderson et al. 2001) and can be used to identify factors shaping

microbe phylogenetic associations (Fountain-Jones et al. 2018). The assumption of homogeneity of variance was validated using the ‘betadisper’ function in ‘vegan’.

Objective 2: Influence of raccoon spatial overlap with dogs on the probability of isolating ESC-R E. coli in raccoons, and vice versa for domestic dogs.

The outcome variable for this analysis was presence of at least one ESC-R E. coli isolate in the feces of domestic dogs and raccoons (yes vs. no) (Table 7). For raccoons, the interface with dogs was quantified using two metrics: 1) whether raccoons were sampled at a site where dogs could enter (dog presence: yes vs. no); and 2) whether raccoons were sampled at a site where raccoons are predicted to use residential areas (residential use: yes vs. no; see methods section 2.3 and supplementary materials for more information). Associations were explored using a binomial generalized linear mixed model (GLMM) with a logit link function using the ‘lme4’ package (Bates et al. 2014).

Other predictors included season and urban-suburban context because both can influence the likelihood of isolating ESC-R E. coli from raccoons (Worsley-Tonks et al.

Chapter 2). We did not include raccoon age or sex as fixed effects because neither were expected to be important (Worsley-Tonks et al. Chapter 2). The interaction between dog presence and urban-suburban context was also explored, but not the interaction between residential use and urban-suburban context nor between residential use and dog presence because significant multicollinearity was detected in both cases (tested by calculating the variance inflation factor using the ‘car’ R package). Because 18 raccoons were captured more than once, we investigated the need for including ‘animal ID’ as a random factor.

To do this, we compared the Akaike information criterion (AIC) values between an intercept model with and without animal ID included as a random factor. There was no significant difference in AIC values between the two models (AIC = 319.49 and 317.9, p

= 0.52) indicating that including animal ID as a random factor was not needed (Table 7).

While there was no spatial autocorrelation in model residuals (Moran’s I statistic: z = 1.2, p = 0.12), capture site was included as a random factor since it improved model fit (AIC

= 317.9 when capture site was not included in intercept model and 300 when included, p

< 0.0001).

For dogs, the interface with raccoons was quantified using two metrics: 1) whether dog home zip code overlapped with at least one of the seven raccoon sites (zip code overlap: yes vs. no); and 2) whether dogs visited one of the raccoon sites more than three times per week (raccoon site visit: yes vs. no). Dog age, sex, and antibiotic use, as well as whether dogs were sampled at a forest preserve vs. dog park are expected to be important at influencing whether dogs shed antimicrobial-resistant E. coli (Procter et al.

2014; Leonard et al. 2015; Wedley et al. 2017) and were therefore included as fixed effects in our analyses. Given only a small number of dogs had ESC-R E. coli in their feces (i.e. 32 dogs, see Results section 3.2), to avoid overfitting the model, the two raccoon interface predictors were analyzed in separate models. The importance of predictors associated with dog characteristics (i.e. age, sex, antibiotic use, and sampled at forest preserve vs. dog park) were evaluated in univariable models and were included in the three raccoon interface models if they were significant in univariable models

(significance level was set at p < 0.1). Generalized linear models (GLMs) with a binomial family and a logit link function were used for all three models using the ‘lme4’ package.

Objective 3: Importance of the type of spatial overlap at influencing genetic relatedness of ESC-R E. coli isolated from raccoons and dogs

The outcome variable in this analysis was pairwise SNP distance of ESC-R E. coli. Only pairwise SNP differences for raccoon – dog ESC-R E. coli isolates were examined (dog – dog and raccoon – raccoon pairwise SNP differences were not examined). We used four metrics to evaluate the importance of spatial overlap between raccoons and domestic dogs: 1) dog presence; 2) zip code overlap; 3) raccoon site visit; and 4) residential use

(Table 7). The importance of each of the four metrics at influencing the phylogenetic clustering of ESC-R E. coli was assessed by running univariable GLMMs with a Poisson distribution using the ‘lme4’ package. To run this analysis, we converted the distance matrix to a pairwise distance list and set raccoon and dog IDs as random factors (Table

7).

Table 7. Description of statistical approaches used.

Analytical n Outcome variable Predictor variable Random approach factor Fischer’s exact 406 Contingency table • species (dog/raccoon) NA test of ESC-R E. coli (presence/absence) Bootstrapping 152 ST richness • species NA and subsampling Univariable 152 Pairwise SNP • species NA PERMANOVA distance of ESC-R E. coli Multivariable 230 ESC-R E. coli • dog presence (dogs capture site binomial presence (yes/no) present at raccoon raccoon GLMM for site: yes/no) ID* raccoons • residential use (site where raccoons are predicted to use residential areas: yes/no) • season (fall, winter, spring, summer) • urban-suburban context (urban/suburban) • urban context*dog presence Univariable and 176 ESC-R E. coli • dog age (in years)** NA multivariable presence (yes/no) • dog sex binomial (male/female)** GLMMs for • forest preserve vs. dogs dog park** • dog antibiotic use (on antibiotics in last 12 months: yes/no)** • raccoon site visit (dog visits raccoon site > 3 times or ≤ 3 times a week) • zip code overlap (dog home zip code overlapped with raccoon site: yes/no)

Univariable 152 Pairwise SNP • dog presence raccoon ID Poisson distance of ESC-R • residential use dog ID GLMMs E. coli (pairwise • raccoon site visit distance list) • zip code overlap *variable was considered for inclusion as random factor in exploratory analyses but was found to contribute little to the overall variance (p < 0.05) and was thus excluded from analyses listed here. **variable was examined in univariable analysis before being included in multivariable analysis if p < 0.1. If p ≥ 0.1, the variable was excluded from multivariable analysis.

Results

Domestic dog and raccoon characteristics

Of the 176 domestic dogs sampled, 22 were sampled from the same household as at least one other sampled dog. Based on survey results, a total of 64 dogs were ≤ 2 years of age,

75 were between 2 and 7, 34 were older than 7, and 3 had no age data. Stratified by sex,

99 dogs were males, 66 were females, and 11 had no data. In terms of antibiotic use,

30.1% of sampled dogs were on some form of antibiotic in the 12-months prior to sampling, 53.4% were not, 11.9% owners were unsure, and 4.5% owners did not respond.

Further, dogs differed based on where they were sampled; 48.3% of dogs were sampled at wildlife sites and 51.7% at local dog parks. In terms of frequency of park use, 35.8% of dog owners stated that they took their dog to the park where sampling took place more than 3 times per week, 28.4% reported visiting 1-3 times per week, 30.1% reported visiting once every two weeks or less, and 5.7% did not respond. Most sampled dogs lived in the northwestern portion of the Chicago area (based on home zip code; Fig.

S4.1), and 64% of dogs had their home zip code that overlapped with at least one raccoon site (Fig. S4.1).

Of the 211 raccoons sampled (17 of which were captured twice and one three times),

67.3% were sampled in suburban areas and 41.7% in urban areas. Stratified by dog use site, 63.5% were sampled in areas where dogs could enter and 36.5% in areas where dogs were prohibited.

Domestic dogs had a lower prevalence of ESC-R E. coli than raccoons, but dogs and raccoons had phylogenetically similar ESC-R E. coli

With a sample prevalence of 55.2% (95% CI = 49%-62%) and 16.5% (95% CI = 11.7%-

22.7%) for raccoons and domestic dogs, respectively (Fig. 9a), there was a significantly higher odds of recovering at least one ESC-R E. coli isolate from raccoons than domestic dogs (Fisher’s exact test: OR = 3.37, 95% CI = 2.12-5.48; p < 0.0001). Of the 152 ESC-R

E. coli isolates recovered from raccoons and domestic dogs (123 from raccoons and 29 from domestic dogs), raccoons had a total of 55 unique STs and one unknown (the unknown ST closely resembled ST155, with variation in the gyrB allele only) and dogs had 20 unique STs and two unknown (one of the unknowns closely resembled ST58, with variation in the parA allele only, and the other was dissimilar to all STs) (Fig. 9b).

Accounting for differences in samples sizes, bootstrapping the raccoon sample size to the dog sample size (i.e. from n = 123 to n = 29) revealed that the raccoon and dog populations likely shed a similar richness of STs (95% CI for raccoons = 16.1-23.8 using

1,000 bootstrap replicates). Of the STs detected, ST38 was most commonly detected in raccoon samples (8.8%), followed by ST973 (7.26%), and both ST68 and ST162 (4.8%)

(Fig. 9b). For dogs, ST68 was most common (13.8%), followed by ST297 (10.3%) (Fig.

9b).

In terms of phylogenetic similarity, raccoons and dogs had 12 STs in common, including

ST10, ST38, ST68, and ST131 (Fig. 9b). Core SNP-based phylogenetic analyses revealed that within-species average core SNP differences were similar to between-species average core SNP differences (raccoon – raccoon: 455.8 mean core SNP difference; dog

– dog: 489.2, raccoon – dog: 480). Further, the maximum likelihood phylogenetic tree

showed no clustering by species, with dog and racoon samples randomly interspersed throughout the tree (Fig. 9c), which was supported by a lack of significant difference in the clustering of ESC-R E. coli isolates by animal species (PERMANOVA: F1, 151 = 0.53, p = 0.74). Focusing on isolates that belonged to one of the 12 STs shared between raccoons and dogs, pairs of isolates differed by less than 20 SNPs in all cases and were similar both within and between animal species (Fig. 10).

Figure 9. Prevalence and phylogenetic associations of ESC-R E. coli isolated from raccoons and domestic dogs. (A) Prevalence of ESC-R E. coli. Whiskers represent 95% confidence intervals and numbers above 95% confidence intervals are sample sizes. (B) minimum spanning tree of ESC-R E. coli sequence types (STs) detected in raccoons (blue) and domestic dogs (orange). The size of nodes represents the number of isolates. The length of lines connecting nodes represents the number of allele differences. (C) core SNP-based maximum likelihood phylogenetic tree of the 152 ESC-R E. coli and heatmap of isolates classified based on host species (i.e. raccoon: orange and dog: blue). The reference is E. coli K-12 strain MG1655.

Presence of domestic dogs at raccoon sites influenced the probability of recovering

ESC-R E. coli from raccoons, but this effect was more distinct at suburban than urban sites

The odds of isolating ESC-R E. coli from raccoons varied significantly by season, urban- suburban context, and presence of dogs (Table 8). The odds of isolating ESC-R E. coli from raccoons was higher in the spring and summer than fall, but not winter, and in urban areas than in suburban areas (Table 8). Importantly, while the presence of dogs was important, it was only significant when examining the interaction with urban-suburban context (Table 8). The odds of isolating ESC-R E. coli from raccoons was higher at sites where dogs were present than at sites where dogs were absent, but only at suburban sites

(Fig. 11). At urban sites, there was no significant difference in the odds of isolating ESC-

R E. coli from raccoons based on dog presence (Fig. 11). Finally, the odds of isolating

ESC-R E. coli from raccoons did not significantly vary based on raccoon predicted use of residential areas (Table 8).

Table 8. Generalized linear mixed model results for isolating at least one ESC-R E. coli from raccoons. Predictor variable OR 95% CI p season (spring) 8.56 (2.86 – 25.58) < 0.001 season (summer) 5.24 (2.1 – 13.08) < 0.001 season (winter) 0.48 (0.2 – 1.16) 0.1 urban-suburban context (urban) 38.56 (6.84 – < 0.001 217.39) residential use (yes) 0.33 (0.09 – 1.14) 0.08 dog presence (yes) 6.21 (1.8 – 21.45) 0.004 urban context (urban)*dog presence 0.11 (0.01 – 0.84) 0.03 (yes) OR represents the odds ratio for each predictor, 95% CI the 95% confidence intervals, and p the p-value. Significant terms are depicted in bold (with 95% CI not overlapping with 1 and p < 0.05).

Figure 11. Prevalence of ESC-R E. coli in raccoons by urban-suburban context and dog presence (yellow: dogs are present, red = dogs are absent and cannot enter). Whiskers are 95% confidence intervals.

Presence of raccoons in dog areas did not influence the probability of recovering ESC-

R E. coli from domestic dogs

Univariable analyses based on dog characteristics revealed that the odds of isolating

ESC-R E. coli did not significantly vary based on dog sex, antibiotic use, nor based on whether they were sampled at dog parks (p > 0.1; Table 16). However, dog age had a marginal effect (OR = 0.88, 95% CI = 0.76-1.02, p = 0.09) and was therefore included in multivariable models. Multivariable GLMs revealed that none of the two metrics used to evaluate the importance of the interface with raccoons were significant predictors for dogs having ESC-R E. coli (i.e. zip code overlap and raccoon site visit; Table 16).

None of the four spatial overlap metrics influenced the phylogenetic relatedness of

ESC-R E. coli isolated from domestic dogs and raccoons

There was no significant differences in the number of core-SNP differences among pairs of ESC-R E. coli isolates recovered from domestic dogs and raccoons based on any of the four spatial overlap metrics examined (p ≥ 0.05; Table 17).

Discussion

Wildlife and domestic animals can have phylogenetically similar ARB when they overlap in space (e.g. Subbiah et al., 2020), but what types of spatial overlap are more likely to facilitate sharing is not well understood. Here, we show that raccoons and domestic dogs sampled in the northwestern metropolitan area of Chicago have phylogenetically similar

ESC-R E. coli, but these resistant bacteria are more likely to be recovered from raccoons than domestic dogs. That said, the likelihood that raccoons have ESC-R E. coli increased when they were sampled at sites where dogs were present, with this association being only apparent in suburban areas. In term of genetic relatedness of ESC-R E. coli, we found that none of the four metrics used to explore spatial overlap between raccoons and dogs influenced the phylogenetic relatedness of ESC-R E. coli isolated from raccoons and dogs.

The finding that raccoons had a higher prevalence of ESC-R E. coli than domestic dogs was unexpected given domestic dogs are considered reservoirs for AMR (Guardabassi et al. 2004). That said, wildlife could have a higher prevalence due to exposure to ARB and

ARG through pathways that domestic dogs were less frequently, or not exposed to. For

example, lakes and rivers are known to be important pathways for the dissemination of anthropogenically-derived ARB into the environment (Zhang, Zhang, and Fang 2009;

Wellington et al. 2013; Surette and Wright 2017), and water-associated wildlife species are especially likely to be exposed (Veldman et al. 2013; Jobbins and Alexander 2015;

Nelson et al. 2008). Raccoons tend to select habitats with water bodies (Gehrt and Fritzell

1998; Henner et al. 2004) because a large proportion of their foods are in or along rivers and lakes (Stuewer 1943). Thus, it is possible that raccoons had a higher prevalence of

ESC-R E. coli than domestic dogs because they were exposed to ARB via contaminated water sources. This is, however, speculative as no environmental samples were collected as part of this study. While previous work has suggested that differences in the prevalence of certain ARB between animals species could be attributed to differences in the host gut hospitability to certain bacteria (Gordon and Cowling 2003; Guenther,

Ewers, and Wieler 2011; Radhouani et al. 2014), it is unlikely to be of importance here because ESC-R E. coli have previously been isolated from domestic dogs in both clinical and community settings (Mathys et al. 2017; Schaufler et al. 2015). As such, differences in exposure risk is likely a more plausible explanation for the prevalence differences detected between domestic dogs and raccoons. One approach to identify possible differences associated with host physiological characteristics or environmental factors would be to make comparisons between domestic dogs and stray dogs (Worsley‐Tonks et al. 2020). Comparisons of the ARG profile of samples from raccoon, stray dog, domestic dog, and other wildlife species revealed that the pooled stray dog sample was more similar to the pooled raccoon sample than the pooled domestic dog sample (Worsley-

Tonks et al. 2020). This study lends support for environmental factors being more important than host physiological characteristics.

While raccoons tended to have a higher sample prevalence of ESC-R E. coli than domestic dogs, the presence of domestic dogs at raccoon sites was an important predictor of recovering ESC-R E. coli isolates from raccoons, but only at suburban sites. Previous work has shown the presence of domestic animals to be an important determinant of isolating ARB from wildlife (e.g. Kozak et al., 2009). However, since the domestic dog population tended to have a low prevalence of ESC-R E. coli (16.5%), the presence of domestic dogs themselves is unlikely to be the main factor associated with the differences detected at suburban sites. Instead, the difference in the presence of a large and diverse range of people (with and without dogs) and the anthropogenic waste left at parks was potentially more influential. While water bodies are predicted to be the primary pathway of wildlife exposure to ARB, anthropogenic waste is also thought to be important

(Dolejska and Literak 2019). For example, wildlife species (e.g. particular gulls) will often have a higher prevalence of ARB and ARG when using landfills (e.g. Ahlstrom et al., 2019), and phylogenetically similar ARB to those detected in landfills (e.g. Nelson et al., 2008) or other wildlife taxa sampled at landfills (Ahlstrom et al. 2018). Raccoons are generalist and opportunistic feeders (Lotze and Anderson 1979), and in urban and suburban areas they will feed on anthropogenic waste present in parks, either on the ground or in trash cans (Hoffmann and Gottschan 1977). Thus, raccoons sampled at suburban sites where dogs and people were allowed to enter may have had a higher prevalence than raccoons sampled at suburban sites where dogs and people were not allowed to enter because of the higher exposure to people and anthropogenic waste. A

lack of difference detected at urban sites could be due to the fact that raccoons at all urban sites were equally likely to be exposed to anthropogenic waste.

The fact that interface with raccoons was not a significant predictor of recovering ESC-R

E. coli from domestic dogs, whether it be based on dog home zip code overlap with raccoon sites or frequency at which dogs visit raccoon sites, tends to suggest that the interface with raccoons did not influence the AMR profile of domestic dogs. This could be because domestic dogs have a low risk of coming into contact with raccoon feces at parks, or with environmental factors that could increase exposure (e.g. contaminated waters), because of having to be leashed by law. However, raccoons have the potential to pose a risk to dogs if raccoons visit residential backyards, particularly if raccoon densities are high. Exploring whether the risk of isolating ARB from dogs increases when dogs reside in areas where raccoons occur at high densities and have a high prevalence of ARB would be a useful next step to take.

The finding that dog age, sex, antibiotic use, and dog park vs. forest preserve sampling were not important predictors of isolating ESC-R E. coli from domestic dogs was surprising given that in previous work ARB were more likely to be isolated from older, female dogs, on antibiotics in previous months, and/or using dog parks (Wedley et al.

2017; Leonard et al. 2015; Procter et al. 2014). For antibiotic use, a lack of association could be due to not knowing the specific types of antibiotics or the timing of when antimicrobials were prescribed to dogs, and we suspect that in most cases dogs were not given second and third generation cephalosporins (Pomba et al. 2017). For dog parks, we suspect that differences were not detected in this study because domestic dogs sampled at non-dog park sites could also have been taken to dog parks outside of the time at which

sampling took place. More broadly, the limited information available on dog movement and locations (e.g. movement between parks and specific home address) could be a reason for why no associations were detected with regards to dog overlap with, and proximity to raccoon sites.

While domestic dog and raccoon populations differed in ESC-R E. coli prevalence, ESC-

R E. coli isolated from the two animal species were phylogenetically similar. This was especially true for isolates that belonged to the same ST, where they differed by less than

10 core-SNPs in most cases. Such a high degree of similarity could reflect transmission among domestic dogs and raccoons sampled. However, as well as having STs in common with domestic dogs, raccoons also had several STs that were not detected in dogs and are typically associated with human sources, such as ST23, ST224, ST410, ST167 (Wieler et al. 2011). Further, ARB identified in wildlife have previously been attributed to human sources (Pesapane, Ponder, and Alexander 2013; Bonnedahl et al. 2009), especially in urban areas (Atterby, Ramey, Hall, et al. 2016; Schaufler et al. 2018). This conforms with the general consensus that people tend to play a more important role in the circulation of

ARB and ARG in the community and the environment than companion animals (Smet et al. 2010). Hence, raccoons may have acquired ESC-R E. coli through exposure to human- derived sources of AMR rather than through contact with feces of domestic dogs.

Nevertheless, other human-associated STs, such as ST131 and ST10 (Wieler et al. 2011), were found in both raccoons and domestic dogs. Companion animals and people can have several ESC-R E. coli in common (Ewers et al. 2010), either because of direct transmission or parallel microevolution (Ewers et al. 2012). Thus, it is possible that

domestic dogs and raccoons had similar ESC-R E. coli because individuals of both species were exposed to human-associated AMR via different pathways.

The absence of an association between any of the four spatial overlap metrics and phylogenetic relatedness of ESC-R E. coli isolated from raccoons and dogs suggests that domestic dogs are unlikely to be an important pathway of exposure for raccoons.

However, this finding should be taken with caution for several reasons. Firstly, few ESC-

R E. coli isolates were recovered from domestic dogs (n = 29), meaning that this analysis may have suffered from a lack of statistical power. Secondly, some of the spatial overlap metrics we used have weaknesses, such as the ‘dog home zip code spatial overlap’ variable. Having location data for dogs at the zip code level may not have provided enough resolution to explore spatial overlap with raccoon sites. Similarly, the variable created to classify raccoon use of residential areas was generated based on radio- telemetry data collected from raccoons tracked in previous years. It is possible that raccoons tested for ESC-R E. coli did not conform to our residential use classification, particularly for sites where a relatively small number of raccoons were radiocollared

(e.g. n = 12 for the surrogate of Edgebrook and DRCA). Future work should prioritize collecting both host behavior and bacterial genomics information on the same individuals

(Arnold, Williams, and Bennett 2016; Ahlstrom et al. 2019).

In terms of public health management of AMR, the most important finding of this study was the difference in prevalence of ESC-R E. coli between domestic dogs and raccoons.

We were over three times more likely to recover ESC-R E. coli from raccoons than from dogs. Environmental and wildlife AMR research has been grossly overlooked in understanding the epidemiology of ARB (Dolejska and Literak 2019; Perez and Villegas

2015), and our study highlights the need for research on wildlife AMR. In this sense, it is apparent that a “One Health” approach is needed to tackle AMR (Perez and Villegas

2015). To date, much wildlife AMR research has advocated to target avian species (in particular gulls) as sentinels for AMR in the environment. We argue that mammalian species that reside in close proximity to humans, such as raccoons, could also be important targets. The fact that raccoons spend a large proportion of time in residential areas and along rivers and lakes (Gehrt and Fritzell 1998; Prange and Gehrt 2004) makes them especially useful for understanding the spread and maintenance of ARB in urban and suburban environments. Since raccoons in many regions across the United States are annually tested for pathogens such as rabies virus, testing for the presence of clinically relevant ARB and storing isolates for future genomics work would be a productive surveillance measure to initiate. While this study was limited by the inability to collect behavioral information on sampled raccoons, we suggest wildlife AMR research take advantage of pre-existing knowledge of the behavior of wildlife involved (in terms of datasets and literature) to make inference about animal habitat use.

Acknowledgements

Funding was provided by Donna Alexander from the Cook County Animal and Rabies

Control, the Max McGraw Wildlife Foundation, the Forest Preserve District of Cook

County, the National Science Foundation (DEB-1413925 and 1654609), and CVM

Research Office UMN Ag Experiment Station General Ag Research Funds (MIN-62-

098). We extend many thanks to the Gehrt lab for field and technical assistance, particularly Gretchen Anchor, Andy Burmesch, Yasmine Hentati, Lauren Ross, Katie

Robertson, Missy Stallard, Steven Winter, and Ashley Wurth. We also thank members of

the Johnson lab, particularly Bonnie Weber, Alison Millis, and Emily Clarke for laboratory assistance. Finally, many thanks to the Minnesota Supercomputing Institute for bioinformatic support.

Chapter 5 - Characterization of antimicrobial resistance genes in Enterobacteriaceae carried by suburban mesocarnivores and locally owned and stray dogs

Chapter published as: Worsley‐Tonks, K. E., Miller, E. A., Gehrt, S. D., McKenzie, S. C.,

Travis, D. A., Johnson, T. J., & Craft, M. E. (2020). Characterization of antimicrobial resistance genes in Enterobacteriaceae carried by suburban mesocarnivores and locally owned and stray dogs. Zoonoses and Public Health, 67(4), 460-466.

Overview

The role of wildlife in the dissemination of antimicrobial resistant bacteria (ARB) and antimicrobial resistance genes (ARGs) in the environment is of increasing concern. We investigated the occurrence, richness, and transmissibility potential of ARGs detected in the feces of three mesocarnivore species: the coyote (Canis latrans), raccoon (Procyon lotor), and Virginia opossum (Didelphis virginiana), and of stray and owned dogs in suburban Chicago, Illinois USA. Rectal swabs were collected from live-captured coyotes

(n = 32), raccoons (n = 31), and Virginia opossums (n = 22). Fresh fecal samples were collected from locally owned (n = 13) and stray dogs (n = 18) and from the live-captured mesocarnivores, when available. Fecal samples and rectal swabs were enriched to select for Enterobacteriaceae and pooled by mesocarnivore species and dog type (owned or stray). Pooled enriched samples were then analyzed for the presence of ARGs using shotgun sequencing. The three mesocarnivore and stray dog samples had twice as many unique ARGs compared to the owned dog sample, which was partly driven by a greater richness of beta-lactamase genes (genes conferring resistance to penicillins and

cephalosporins). Raccoon and stray dog samples had the most ARGs in common, suggesting possible exposure to similar environmental sources of ARGs. In addition to identifying clinically relevant ARGs (e.g. blaCMY and qnrB), some ARGs were linked to the class 1 integrase gene, intI1, which may indicate anthropogenic origin. Findings from this pilot investigation suggest that the microbial communities of suburban mesocarnivores and stray dogs can host ARGs that can confer resistance to several antimicrobials used in human and veterinary medicine.

Introduction

Antimicrobial resistant bacteria (ARB) have been detected in numerous wildlife species across the globe (Vittecoq et al. 2016), particularly in animals that reside in human- dominated environments (Guenther, Ewers, and Wieler 2011; Wang et al. 2017; Arnold,

Williams, and Bennett 2016; Dolejska and Literak 2019). If ARB become established in wildlife bacterial communities, wild animals could act as sources of known and/or novel resistant strains and facilitate widespread dissemination in the environment (Radhouani et al. 2014; Carroll et al. 2015). However, the potential for this to occur depends on the occurrence, richness, and transmissibility of antimicrobial resistance genes (ARGs) both at the microbial and host community level (Martiny et al. 2011; Pal et al. 2016). In particular, ARGs that can be transferred between related and unrelated bacterial species

(via mobile genetic elements [MGEs]) have the potential to become widespread within microbial communities in wildlife. Further, wildlife can acquire ARB and ARGs via multiple routes (e.g. through contact with contaminated water, food sources, and other animals (Vittecoq et al. 2016)). The exposure of wildlife to ARGs in the environment can

vary with wildlife species due to species characteristics such as diet and habitat use

(Swift et al., 2019; Vittecoq et al., 2016). Hence, the distribution of ARGs should be examined simultaneously in the context of the microbe and ecology of the host.

While a variety of ARGs have been detected in wildlife, most investigations have identified genes by testing for the presence of key ARGs found in specific bacterial species. Because ARGs can be transferred between different bacterial species horizontally via MGEs, focusing on specific bacterial species may underestimate the richness of ARGs present in bacterial communities of wildlife (Allen et al. 2010).

Shotgun metagenomic sequencing overcomes this issue by simultaneously sequencing all genetic material, including ARGs, present in a microbial community (Pal et al. 2016).

However, metagenomic sequencing often requires significant sequencing depth to detect

ARGs in bacterial species that occur in low abundance (Zaheer et al. 2018). Targeted enrichment of a specific group of bacterial taxa, combined with shotgun sequencing, allows for ARG profiling of multiple strains within the selected group while avoiding the need for significant sequencing depth (Peto et al. 2019).

Here, we used an enriched shotgun sequencing approach to: 1) investigate the occurrence and richness of ARGs in enterobacterial microbial communities of suburban mesocarnivores; 2) compare and contrast mesocarnivore ARGs to those of locally owned and stray domestic dogs; and 3) explore the transmissibility potential of ARGs in the mesocarnivore and dog bacterial communities by investigating their linkage to genes associated with MGEs. We focused on three mesocarnivore species, including the coyote

(Canis latrans), raccoon (Procyon lotor), and Virginia opossum (Didelphis virginiana), because they are known to share several infectious agents with domestic animals and

humans (e.g. antimicrobial resistant Escherichia coli and Salmonella spp.; Bondo,

Boerlin, et al., 2016; Bondo et al., 2019; Jardine et al., 2012; Jay-Russell, Hake, Bengson,

Thiptara, & Nguyen, 2014). In addition, all of these mesocarnivore species are abundant in urban settings throughout the USA (Gehrt, Riley, and Cypher 2010). Given that raccoons and opossums frequently exploit anthropogenic food sources (Bateman and

Fleming 2012), we predicted that their ARG profiles would be more similar to domestic dogs than to coyotes. Further, since stray dogs are free-roaming animals, we expected them to have the greatest richness of ARGs, having ARGs in common with both mesocarnivores and owned dogs.

Methods

Mesocarnivore and dog sample collection

We opportunistically captured 32 coyotes, 22 opossums, and 31 raccoons in different parts of the northwestern suburbs of the Chicago Metropolitan Area, IL, USA from

January-March 2017. Trapping methods for coyotes and raccoons are discussed in Gehrt,

Anchor, & White (2009) and Prange & Gehrt (2004), respectively. Briefly, all animals were live-captured. Coyotes were captured with padded foothold traps and cable restraint devices, and opossums and raccoons with Tomahawk box traps. Coyotes and raccoons were immobilized with an injection of Telazol. Opossums were not immobilized and were handled with care using safety gloves. Sterile flocked swabs (Puritan Medical

Products, Guilford, Maine, USA) were collected from all captured mesocarnivores and fresh fecal samples were collected opportunistically. Coyote captures were approved by

Ohio State University and the University of Minnesota (IACUC IDs: 2013A00000012-

R1 and 1703-34694A, respectively), and opossum and raccoon captures by the United

States Department of Agriculture (permit: IDNR W17.0122).

Domestic dogs were subdivided into two groups: owned dogs and stray dogs. Fecal samples for owned dogs were collected opportunistically from owners that visited forest preserves and dog parks near the mesocarnivore capture locations (n = 5) and from veterinarian dog owners in the local area (n = 8). Fecal samples from stray dogs were collected within 24 hours of dogs arriving in a local dog shelter (n = 18). All mesocarnivore fecal samples and rectal swabs, and dog fecal samples were stored in brain heart infusion broth with 20% glycerol at -80°C until processing.

Sample processing and sequencing

Once thawed, 1 ml of each sample was enriched in 9 ml brain heart infusion broth at

37°C for 3 hours. Ten ml of Lauryl tryptose phosphate broth was added and cultured overnight to select for the growth of Enterobacteriaceae. Deoxyribonucleic acid (DNA) was extracted from all samples using the PowerSoil DNA Isolation Kit (Qiagen, Hilden,

Germany) and samples were subsequently pooled by mesocarnivore species and dog type

(stray and owned dog) (see Table 21 for sample sizes of each pool). For the mesocarnivores, fecal samples were included in pooled samples when available, otherwise rectal swabs were used. Library construction and Illumina HiSeq2500 2x150 bp sequencing were performed at the University of Minnesota Genomics Center.

Sequencing depth ranged from 64-69 million reads per pooled sample (see Table 21 for

NCBI accession numbers), which increased the probability of detecting rare ARGs and

MGEs (Jonsson et al. 2016).

Sequence processing and analyses

Raw reads were quality checked and trimmed using Trimmomatic (version 0.33) (Bolger,

Lohse, and Usadel 2014). To remove host sequences, reads were mapped to host genomes using BWA-MEM (version 0.7.17) (Li and Durbin 2009). To the best of our knowledge, no coyote, opossum, or raccoon genomes are accessible online. We therefore used the genomes of closely related species: for the coyote, we used the domestic dog

(Canis lupus familiaris) genome (accession number: GCA_000002285.2), for the opossum we used the gray short-tailed opossum (Monodelphis domestica, accession number: GCF_000002295.2), and for the raccoon we used the lesser panda (Ailurus fulgens, accession number: GCA_002007465.1). Non-host reads were then taxonomically classified using Kraken (version 1.1.1) (Wood and Salzberg 2014) and aggregated to the genus level.

For ARG and MGE identification, non-host reads were de novo assembled using metaSPAdes with default parameters (version 3.10) (Nurk et al. 2017). The resulting contigs were quality assessed using MetaQUAST (version 4.3) (Mikheenko, Saveliev, and Gurevich 2016) and annotated using Prokka (version 0.7.17) (Seemann 2014). ARGs were then identified by aligning coding sequences to the ResFinder database (accessed on

01 Aug 2018) (Zankari et al. 2012) using the NCBI BLASTn algorithm. To investigate the transmissibility potential of ARGs, we searched for genetic features associated with

MGEs (Siguier, Gourbeyre, and Chandler 2014). Plasmid replicons were identified using the PlasmidFinder database (Carattoli et al. 2014), while integrons were identified using the INTEGRALL database (Moura et al. 2009). All BLAST hits were filtered based on

their sequence identity ( 90%), sequence coverage ( 80%), query coverage ( 80%), E- value ( 1E-10), and bitscore ( 400).

All downstream analyses were conducted in R (version 3.4.0) (R Development Core

Team 2020). We used Pearson correlation to determine whether there was an association between the number of ARGs detected and the number of individual samples in each pooled sample type. Sample dissimilarities based on ARG profiles were calculated using the Jaccard distance from the ‘proxy’ package (Meyer and Buchta 2015).

Results

Aminoglycoside, beta-lactam, macrolide, quinolone, and tetracycline resistance genes were among the most prevalent ARGs found in the mesocarnivore and dog samples (Fig.

12a). The coyote, opossum, raccoon, and stray dog samples had a greater richness of

ARGs compared to the owned dog sample (Fig. 12a), a pattern that was also detected when exploring the richness of the most abundant bacterial genera (Fig. S5.1). Although there were fewer individual samples pooled for the owned dog sample than the other four pooled samples, the correlation between ARG richness and the number of individual samples in each pool was not statistically significant (Pearson’s r = 0.77; p = 0.13; Fig.

S5.2). The owned dog sample had half the number of unique ARGs compared to other samples, which was partly driven by a lower richness in beta-lactam resistance genes

(Fig. 12a-b). Specifically, the only beta-lactam gene detected in owned dogs was blaTEM whereas twelve other beta-lactam genes were detected in the mesocarnivore and/or stray dog samples (e.g. blaSHV, blaACT and blaCMY; Fig 12b). Other clinically relevant ARGs detected in the mesocarnivore samples, that were not identified in the owned dog sample,

included those conferring resistance to sulfonamides (sul1 and sul2) and aminoglycosides

(e.g. aadA) (Fig. 12b). Additional genes found only in the mesocarnivore and stray dog samples conferred resistance to aminoglycosides (aph(3”)-Ib), fosfomycin (fosA), and tetracyclines (tet(A)).

By calculating the proportion of ARGs in common between pairs of pooled samples for mesocarnivores, stray dogs, and owned dogs, we found the owned dog sample to be the most dissimilar from all samples (Jaccard distance for: owned dog-stray dog = 0.74, owned dog-coyote = 0.76, owned dog-opossum = 0.77, owned dog-raccoon = 0.78; Fig.

12c). The stray dog and raccoon samples had the most similar ARG profiles compared to all other sample pairs (Jaccard distance = 0.47; Fig. 12c). The coyote-opossum and coyote-raccoon samples had the next most similar ARG profiles (Jaccard distance = 0.57 and 0.58, respectively; Fig. 12c).

Figure 12. Patterns of antimicrobial resistance genes (ARGs) detected in enriched pooled fecal samples and rectal swabs of mesocarnivores, and stray and owned dogs in suburban Chicago, Illinois, USA. a) Richness of ARGs by antibiotic class and b) types of ARGs identified in the pooled samples. c) Dissimilarity (Jaccard distance) between pairs of samples based on the proportion of ARGs in common. Greater Jaccard distance values (lighter blue) indicate a higher degree of dissimilarity between samples while lower Jaccard distance values (darker blue) indicate less dissimilarity between samples.

When investigating the transmissibility potential of ARGs, we found no plasmid replicons on any assembled contig also containing at least one ARG. However, 11

contigs were found to contain both ARGs and integrons (Table 9). For example, the pooled coyote sample had two ARGs on the same contigs as class 1 integron elements: dfrA with the class 1 integrase gene, intI1, and sul1 with the qacE1, which is found in the 3’CS region of class 1 integrons (Table 9).

Table 9. Summary of antimicrobial resistance genes (ARGs) detected in enriched pooled fecal samples and rectal swabs of mesocarnivores, and stray and owned dogs in suburban Chicago, Illinois, USA. ARGs listed are only those that were in close proximity to genetic features associated with mobile genetic elements (MGEs). Information on the distance between ARGs and associated MGE genes on assembly contigs is also provided. Contig Genetic Pooled Contig ID length ARG MGE distance sample (bp) (bp) coyote NODE_2674 5051 tet(G) orf1(class 1 integron) 90 NODE_6368 2173 dfrA intI1 157 NODE_8413 1604 sul1 qacE1 6 opossum NODE_2269 5620 dfrA intI1 145 NODE_3152 4271 blaCARB ISCR1 1248 NODE_3675 3778 qnrB orf2 (class 1 integron) 42 NODE_7767 1895 aadA qacE1 164 raccoon NODE_1351 5534 blaTEM tnpA (IS26) 242 stray 183 dog NODE_2225 4950 blaTEM tnpR (class 1 integron) owned NODE_83 31036 blaTEM tnpR (class 1 integron) 183 dog NODE_903 6317 dfrA intI1 171

Discussion

By using an enriched shotgun metagenomic sequencing approach, we identified a wide range of ARGs in the enterobacterial communities of suburban mesocarnivores, some of which were also found in other mesocarnivore studies that targeted specific bacterial species (e.g. E. coli and Salmonella sp.; Bondo, Pearl, et al., 2016; Bondo et al., 2019). Several of the ARGs detected in the coyote, raccoon, and opossum samples

were of clinical relevance (e.g. blaSHV, blaCMY, qnrB, aadA, sul1, and sul2), suggesting an anthropogenic origin and the potential for these suburban mesocarnivores to carry drug-resistant bacteria. For example, blaCMY genes encode AmpC beta-lactamases, which confer resistance to many clinically important beta-lactam antibiotics, including extended spectrum cephalosporins (Philippon, Arlet, and Jacoby 2002). These genes are frequently found on plasmids in a variety of pathogenic bacteria, including

Klebsiella pneumoniae, Enterobacter aerogenes, and multiple Salmonella serovars

(Philippon, Arlet, and Jacoby 2002). Additionally, intI1, an integrase gene considered to be an indicator of anthropogenic pollution (Gillings et al. 2015; Ma et al. 2016), was detected in the opossum and coyote samples, which further suggests an anthropogenic origin.

In terms of similarities with owned and stray dogs, we found mesocarnivores to have similar ARG profiles to stray dogs, a pattern that was also detected in a stray dog and coyote study that targeted antimicrobial-resistant E. coli and Salmonella (Jay-Russell et al. 2014). That said, it was surprizing that the mesocarnivore and stray dog samples had a greater richness of beta-lactam genes and other ARGs than the owned dog sample given that owned dogs are considered reservoirs of ARB (Guardabassi, Schwarz, and

Lloyd 2004). This difference could be because coyotes, raccoons, opossums, and stray dogs were exposed to ARB and ARGs through environmental sources that owned dogs were not exposed. For example, water bodies that are connected to wastewater treatment plants have previously been proposed to act as important exposure routes of antimicrobial resistance for wildlife (Cole et al. 2005; Dolejska, Cizek, and Literak 2007) and may be responsible for some of the resistance patterns observed in the mesocarnivore and stray

dogs sampled in this study. Likewise, wildlife species that have similar diets are expected to have similar antimicrobial resistance profiles because they have similar exposure routes (Jobbins and Alexander 2015; Vittecoq et al. 2016). While dietary information on sampled mesocarnivores and stray dogs could not be obtained in this study, it is possible that the stray dogs, unlike the owned dogs, foraged in similar locations and/or on similar resources as the mesocarnivores (particularly the raccoons) and may therefore have been exposed to similar environmental sources of ARB and ARGs.

Alternatively, given that some of the ARGs detected in the three mesocarnivore and stray dog samples have been found in owned dogs sampled in other investigations (e.g. blaSHV, blaCMY, qnrB; (Costa et al. 2007; Aly et al. 2012; Rocha-Gracia et al. 2014), it is possible that the differences found between owned and stray dogs, and mesocarnivores were from sampling bias. The characteristics and lifestyle of an owned dog, including size, eating a raw diet, or visiting a dog park, can increase the risk of carrying ARB and ARGs (Procter et al. 2014; Leonard et al. 2015). The attribute information of the owned dogs was not gathered in this study, so it is unknown if the owned dogs had characteristics or lifestyles that reduced their risk of shedding ARB and ARGs. Although not statistically significant, we expect that the smaller number of samples pooled for owned dogs compared to coyotes, opossums, raccoons, and stray dogs likely influenced differences observed in

ARG richness. The non-significant association observed between ARG richness and number of individual samples pooled for the five animal groups was likely due to low statistical power. Hence, additional research is needed to determine if there is a biological significance in ARG richness between owned dogs and both stray dogs and mesocarnivores.

When investigating the transmissibility potential of ARGs, none were linked to plasmids, which facilitate the transfer of ARGs between bacterial species (Bennett 2009). Hence, there was no evidence for horizontal gene transfer in the mesocarnivore, stray dog, and owned dog samples, which could in part be due to the reduced detection of plasmids from shotgun metagenomic data (Quince et al. 2017). Indeed, many of the identified ARGs are commonly found on plasmids (Partridge et al. 2018). However, several ARGs in both mesocarnivore and dog samples were linked to genetic features associated with class 1 integrons, which can be involved in the movement of ARGs across bacterial genomes, and therefore to plasmids when present (Mazel 2006).

In conclusion, we detected several clinically relevant ARGs and anthropogenically associated MGEs in mesocarnivores and stray dogs sampled in urbanized landscapes.

These results suggest potential exposure through anthropogenic sources and lends support for using wildlife (Bondo et al. 2016; Furness et al. 2017), and possibly stray dogs, as sentinels of antimicrobial resistance in the environment. More broadly, our findings highlight that human and animal exposure to ARB and associated ARGs may not be solely associated with antimicrobial prescription and clinical visits, but also with exposure through the community and via environmental sources of resistance (e.g. via rivers receiving effluent from municipal or hospital treatment facilities; Laxminarayan et al., 2013; Wellington et al., 2013). Future antimicrobial resistance research performed in wildlife should consider using shotgun sequencing approaches as it allows for a wide variety of antimicrobial resistance genes to be detected.

Chapter 6 – Conclusion

While substantial advances have been made in understanding antimicrobial resistance

(AMR) in the environment, several questions remain unanswered regarding sources and potential exposure pathways. Further, recent work has stressed that to effectively surveil

AMR in the environment, and mitigate further spread, requires taking a broader look at different environmental compartments (Wellington et al. 2013; Marti, Variatza, and

Balcazar 2014). Wildlife are one of those compartments. Understanding AMR in wildlife is beneficial both in terms of identifying environmental sources of AMR and understanding the role of wildlife in the transmission and maintenance of AMR.

So far, wildlife AMR research has shown that: 1) wildlife can carry ARB and ARG that are typically associated with clinical and agricultural settings (Wang et al. 2017; Vittecoq et al. 2016); 2) some ARG found in wildlife can be associated with plasmids, and thus have the potential to be transferred widely in wildlife bacterial communities (Dolejska and Papagiannitsis 2018); 3) wildlife are more likely to have ARB and ARG the closer they are to human dominated areas (e.g. Bonnedahl et al. 2009); 4) wildlife can have similar ARB and ARG to local domestic animals (e.g. Mercat et al. 2016; Pesapane,

Ponder, and Alexander 2013); and 5) the prevalence of ARB and ARG can differ by wildlife species (Jobbins and Alexander 2015; Vittecoq et al. 2016). Since wildlife can be exposed to ARB and ARG via numerous pathways (Vittecoq et al. 2016; Guenther,

Ewers, and Wieler 2011), an important next step is to gain better insight on the types of anthropogenic sources that are most likely to be sources of AMR for wildlife and determine to what extent these sources shape the AMR profile of wildlife.

The goal of this thesis was to assess the role of several anthropogenic sources at shaping the AMR profile of wildlife. In Chapter 2, we focused on two known environmental anthropogenic sources of resistance and investigated not only whether these sources increase the risk of isolating ARB from wildlife but also whether they increase the risk for resistance to be widespread in wildlife bacterial communities. The two known environmental anthropogenic sources of focus were rivers that were downstream from a wastewater treatment plant (WWTP) and urban context (urban vs. suburban). We found that the risk of isolating Extended-spectrum cephalosporin-resistant (ESC-R) Escherichia coli from raccoons was higher when raccoons were sampled at urban sites than at suburban sites, which is consistent with previous work for proximity to human-dominated areas more generally (Skurnik et al. 2006; Dolejska, Cizek, and Literak 2007; Furness et al. 2017). Importantly, we also found that raccoons were more likely to have ESC-R E. coli with transferable ARG when present at sites that were downstream from a WWTP.

This finding is an important contribution to wildlife AMR research because it suggests that wildlife residing near, or utilizing, anthropogenic environmental sources of AMR have the potential to display a ‘source-type’ profile of AMR. This is concerning because it raises the question of whether efforts to control AMR in the environment could be hampered if AMR is maintained in wildlife bacterial communities. An important next step is to explore how the AMR profile of wildlife varies when AMR is controlled at the anthropogenic environmental sources.

Chapter 3 aimed to take a broader look at pathways by which wildlife could be exposed to ARB through the environment. This work stemmed from the concern that wildlife

AMR research takes too narrow of an approach at exploring exposure pathways for

wildlife (Arnold, Williams, and Bennett 2016), focusing mainly on the importance of

WWTPs for example. By examining the environment at the home range level, in this chapter we explore the importance of several landscape features at influencing the isolation of ESC-R E. coli from raccoons. Landscape features that appeared as most important were proportion of residential areas in raccoon home ranges as well as wetlands. ESC-R E. coli were more likely to be isolated from raccoons when more residential area and wetlands were present in their home ranges. This finding is an important contribution to both wildlife and environmental AMR research because it raises awareness that environments that are not typically associated with detection of clinically relevant ARB could be important exposure pathways.

Chapter 4 and 5 explored the interface with domestic animals. Specifically, we compared the AMR profile of wildlife to local domestic dogs (Canis lupus familiaris), and explored whether certain interactions (Chapter 4) and host characteristics (Chapter 5) increased

AMR profile similarity. Chapter 4 compared the ESC-R E. coli of raccoons to those of local domestic dogs, and asked whether shared space was an important predictor for both isolating ESC-R E. coli and having phylogenetically similar ESC-R E. coli.

Counterintuitively, the prevalence of ESC-R E. coli in raccoons was three times higher than that of domestic dogs. Interestingly, the presence of domestic dogs at raccoon sites was an important predictor for isolating ESC-R E. coli from raccoons, but only at suburban sites. Given that raccoons have a higher prevalence of ESC-R E. coli than domestic dogs, we suspect that domestic dogs themselves are likely not the main factor increasing the risk of isolating ESC-R E. coli from raccoons sampled at these sites.

Rather, we presume that the high human density at these sites and exposure to various

100

anthropogenic food sources and waste could provide a better explanation for the pattern detected. Another important finding from Chapter 4 was that in many cases ESC-R E. coli isolated from raccoons and domestic dogs were phylogenetically similar, and in many cases differed by less than 10 single nucleotide polymorphisms (SNP). This finding is noteworthy because shared space was not important at predicting the phylogenetic associations of ESC-R E. coli isolated from raccoons and dogs. While no firm conclusion can be made because approaches used to quantify shared space had several limitations, we speculate that phylogenetically similar ESC-R E. coli detected in raccoons and domestic dogs likely originated from humans, and that raccoons and domestic dogs were exposed via different pathways (e.g. direct contact with humans for dogs, exposure to

WWTP or hospital effluent for raccoons).

Chapter 5 took a broader look at the AMR profile of raccoons and domestic dogs by exploring similarities in antimicrobial resistance genes (ARGs) in samples that were pooled by animal species. Importantly, as well as examining ARG in raccoons, we explored the ARG profile of coyotes (Canis latrans) and Virginia opossums (Didelphis virginiana) that were sampled in the same area. Further, domestic dogs were subdivided into owned and stray dogs. The raccoon, coyote, and opossum pooled samples had similar ARG profiles. Interestingly, the stray dog pooled sample had a similar ARG profile to the mesocarnivore pooled samples, but the owned dog pooled sample had fewer

ARG in common with both the mesocarnivore pooled samples and stray dog pooled sample. This was partly because the owned dog pooled sample had fewer beta-lactam

ARGs (some of which can confer ESC-R). While conclusions can only be speculative because epidemiological characteristics of mesocarnivores and dogs sampled for this

101

chapter were unknown, detecting a similar ARG profile between the mesocarnivore and stray dog sample may lend support for the theory that wildlife are exposed to ARB and

ARG through sources that owned dogs are not, or less, exposed to.

Collectively, this thesis identifies several anthropogenic and environmental factors contributing to AMR of wildlife (Fig. 13). More specifically, this thesis finds that known anthropogenic environmental sources of AMR can not only increase wildlife exposure to

ARB but can also increase the risk for AMR to spread widely in wildlife bacterial communities. This thesis also raises further awareness that environments not typically associated with AMR (e.g. wetlands) may act as important exposure pathways and require further attention. Finally, comparing the AMR profile of raccoons to that of local domestic dogs was particularly insightful. The counter intuitive finding that prevalence of

ESC-R E. coli was three times higher in raccoons than domestic dogs, in itself, highlights that control of AMR can only be attained if AMR is examined in both humans, domestic animals, wildlife, and the environment. To conclude, this thesis provides a concrete example for the need to account for the environment when generating surveillance and control strategies for AMR, and identifies several environmental components that could be targeted for surveillance.

102

Figure 13. Movement of antimicrobial resistant bacteria in the community and the environment. Grey arrows represent the movement of antimicrobial resistant bacteria between compartments. Movement can be directional or mutual. Pink arrows represent movements that were explored as part of this thesis.

103

Bibliography

Adelowo, Olawale Olufemi, Odion Osebhahiemen Ikhimiukor, Camila Knecht, John Vollmers, Mudit Bhatia, Anne Kirstin Kaster, and Jochen A. Müller. 2020. “A Survey of Extended-Spectrum Betalactamase-Producing Enterobacteriaceae in Urban Wetlands in Southwestern Nigeria as a Step towards Generating Prevalence Maps of Antimicrobial Resistance.” PLoS ONE 15 (3): 1–19. https://doi.org/10.1371/journal.pone.0229451. Ahlstrom, Christina A., Jonas Bonnedahl, Hanna Woksepp, Jorge Hernandez, Björn Olsen, and Andrew M. Ramey. 2018. “Acquisition and Dissemination of Cephalosporin-Resistant Escherichia coli in Migratory Birds Sampled at an Alaska Landfill as Inferred through Genomic Analysis.” Scientific Reports 8 (1): 7361. https://doi.org/10.1038/s41598-018-25474-w. Ahlstrom, Christina A., Jonas Bonnedahl, Hanna Woksepp, Jorge Hernandez, John A. Reed, Lee Tibbitts, Björn Olsen, David C. Douglas, and Andrew M. Ramey. 2019. “Satellite Tracking of Gulls and Genomic Characterization of Fecal Bacteria Reveals Environmentally Mediated Acquisition and Dispersal of Antimicrobial Resistant Escherichia coli on the Kenai Peninsula, Alaska.” Molecular Ecology 28 (10): 2531–45. https://doi.org/10.1111/mec.15101. Akiyama, Tatsuya, and Mary C. Savin. 2010. “Populations of Antibiotic-Resistant Coliform Bacteria Change Rapidly in a Wastewater Effluent Dominated Stream.” Science of the Total Environment 408 (24): 6192–6201. https://doi.org/10.1016/j.scitotenv.2010.08.055. Albrechtova, Katerina, Ivo Papousek, Helene De Nys, Maude Pauly, Etile Anoh, Arsene Mossoun, Monika Dolejska, et al. 2014. “Low Rates of Antimicrobial-Resistant Enterobacteriaceae in Wildlife in Tai National Park, Côte d’Ivoire, Surrounded by Villages with High Prevalence of Multiresistant ESBL-Producing Escherichia Coli in People and Domestic Animals.” PLoS ONE 9 (12): e113548. https://doi.org/10.1371/journal.pone.0113548. Allen, Heather K., Justin Donato, Helena Huimi Wang, Karen A. Cloud-Hansen, Julian Davies, and Jo Handelsman. 2010. “Call of the Wild: Antibiotic Resistance Genes in Natural Environments.” Nature Reviews Microbiology 8 (4): 251–59. https://doi.org/10.1038/nrmicro2312. Allen, Samantha E., Patrick Boerlin, Nicol Janecko, John S. Lumsden, Ian K. Barker, David L. Pear, Richard J. Reid-Smith, and Claire Jardine. 2011. “Antimicrobial Resistance in Generic Escherichia coli Isolates from Wild Small Mammals Living in Swine Farm, Residential, Landfill, and Natural Environments in Southern Ontario, Canada.” Applied and Environmental Microbiology 77 (3): 882–88. https://doi.org/10.1128/AEM.01111-10. Almakki, Ayad, Estelle Jumas-Bilak, Hélène Marchandin, and Patricia Licznar-Fajardo. 2019. “Antibiotic Resistance in Urban Runoff.” Science of the Total Environment

104

667: 64–76. https://doi.org/10.1016/j.scitotenv.2019.02.183. Aly, Sherine A., Nipattra Debavalya, Sang-Jin Suh, Omar A. Oryazabal, and Dawn M. Boothe. 2012. “Molecular Mechanisms of Antimicrobial Resistance in Fecal Escherichia Coli of Healthy Dogs after Enrofloxacin or Amoxicillin Administration.” Canadian Journal of Microbiology 58 (11): 1288–94. https://doi.org/10.1139/w2012-105. Arnold, Kathryn E, Nicola J Williams, and Malcolm Bennett. 2016. “Disperse Abroad in the Land’: The Role of Wildlife in the Dissemination of Antimicrobial Resistance.” Biology Letters 12 (8). https://doi.org/10.1098/rsbl.2016.0137. Arpin, Corinne, Véronique Dubois, Jeanne Maugein, Jacqueline Jullin, Brigitte Dutilh, Jean-Philippe Brochet, Gilberte Larribet, Isabelle Fischer, and Claudine Quentin. 2005. “Clinical and Molecular Analysis of Extended-Spectrum-Lactamase- Producing Enterobacteria in the Community Setting.” Journal of Clinical Microbiology 43 (10): 30. https://doi.org/10.1128/JCM.43.10.5048-5054.2005. Arredondo-Alonso, Sergio, Malbert R.C. Rogers, Johanna C. Braat, Tess D. Verschuuren, Janetta Top, Jukka Corander, Rob J.L. Willems, and Anita C. Schürch. 2018. “Mlplasmids: A User-Friendly Tool to Predict Plasmid- and Chromosome- Derived Sequences for Single Species.” Microbial Genomics 4 (11). https://doi.org/10.1099/mgen.0.000224. Arredondo-Alonso, Sergio, Rob J Willems, Willem Van Schaik, and Anita C Schürch. 2017. “On the (Im)Possibility of Reconstructing Plasmids from Whole-Genome Short-Read Sequencing Data.” Microbial Genomics 3 (10). https://doi.org/10.1099/mgen.0.000128. Atterby, Clara, Andrew M. Ramey, Gabriel Gustafsson Hall, Josef Järhult, Stefan Börjesson, and Jonas Bonnedahl. 2016. “Increased Prevalence of Antibiotic- Resistant E. Coli in Gulls Sampled in Southcentral Alaska Is Associated with Urban Environments.” Infection Ecology & Epidemiology 6 (1): 32334. https://doi.org/10.3402/iee.v6.32334. Baker-Austin, Craig, Meredith S. Wright, Ramunas Stepanauskas, and J. V. McArthur. 2006. “Co-Selection of Antibiotic and Metal Resistance.” Trends in Microbiology. Elsevier Current Trends. https://doi.org/10.1016/j.tim.2006.02.006. Bankevich, Anton, Sergey Nurk, Dmitry Antipov, Alexey A Gurevich, Mikhail Dvorkin, Alexander S Kulikov, Valery M Lesin, et al. 2012. “Original Articles SPAdes: A New Genome Assembly Algorithm and Its Applications to Single-Cell Sequencing.” Journal of Computational Biology 19 (5): 455–77. https://doi.org/10.1089/cmb.2012.0021. Barton, Kamil. 2013. “Package ‘ MuMIn ’. R Package Version 1.” Bateman, P. W., and P. A. Fleming. 2012. “Big City Life: Carnivores in Urban Environments.” Journal of Zoology 287 (1): 1–23. Bates, Douglas, Martin Mächler, Ben Bolker, and Steve Walker. 2014. “Fitting Linear

105

Mixed-Effects Models Using Lme4.” Bennett, P M. 2009. “Plasmid Encoded Antibiotic Resistance: Acquisition and Transfer of Antibiotic Resistance Genes in Bacteria.” British Journal of Pharmacology 153 (S1): S347–57. https://doi.org/10.1038/sj.bjp.0707607. Berendonk, Thomas U, Célia M Manaia, Christophe Merlin, Despo Fatta Kassinos, Eddie Cytryn, Fiona Walsh, Helmut Bürgmann, et al. 2015. “Tackling Antibiotic Resistance: The Environmental Framework.” Nature Reviews Microbiology 13 (5): 310–17. Berkner, Silvia, Sabine Konradi, and Jens Schönfeld. 2014. “Antibiotic Resistance and the Environment-There and Back Again Science & Society Series on Science and Drugs.” EMBO Reports 15 (7): 740–44. https://doi.org/10.15252/embr.201438978. Bivand, R., M. Altman, L. Anselin, R. Assunção, O. Berke, G. A. Bernat, and W. Müller. 2011. “Spdep: Spatial Dependence: Weighting Schemes, Statistics and Models.” OpenScience. Bolger, Anthony M, Marc Lohse, and Bjoern Usadel. 2014. “Genome Analysis Trimmomatic: A Flexible Trimmer for Illumina Sequence Data.” Bioinformatics 30 (15): 2114–20. https://doi.org/10.1093/bioinformatics/btu170. Bonardi, Silvia, Luca Bolzoni, Renato Giulio Zanoni, Marina Morganti, Margherita Corradi, Stefano Gilioli, and Stefano Pongolini. n.d. “Limited Exchange of Salmonella Among Domestic Pigs and Wild Boars in Italy.” EcoHealth 16. Accessed May 11, 2020. https://doi.org/10.1007/s10393-019-01418-2. Bondo, K J, D L Pearl, N Janecko, P Boerlin, R J Reid‐Smith, J Parmley, and C M Jardine. 2016. “Epidemiology of Antimicrobial Resistance in Escherichia coli Isolates from Raccoons (Procyon Lotor) and the Environment on Swine Farms and Conservation Areas in Southern Ontario.” PLoS ONE 11 (11): 165303. https://doi.org/10.1371/journal.pone.0165303. Bondo, Kristin, David L. Pearl, Nicol Janecko, Richard J. Reid‐Smith, E. Jane Parmley, J. Scott Weese, Joyce Rousseau, Eduardo Taboada, Steven Mutschall, and Claire M. Jardine. 2019. “Salmonella, Campylobacter, Clostridium Difficile, and Anti‐ microbial Resistant Escherichia coli in the Faeces of Sympatric Meso‐mammals in Southern Ontario, Canada.” Zoonoses and Public Health 66 (4): 406–16. https://doi.org/10.1111/zph.12576. Bonnedahl, J, M Drobni, M Gauthier-Clerc, J Hernandez, and S Granholm. 2009. “Dissemination of Escherichia coli with CTX-M Type ESBL between Humans and Yellow-Legged Gulls in the South of France Dissemination of Escherichia coli with CTX-M Type ESBL between Humans and Yellow-Legged Gulls in The.” PLoS ONE 4 (6): 5958. https://doi.org/10.1371/journal.pone.0005958. Boolchandani, Manish, Alaric W DSouza, and Gautam Dantas. 2019. “Sequencing-Based Methods and Resources to Study Antimicrobial Resistance.” Nature Reviews Genetics. https://doi.org/10.1038/s41576-019-0108-4.

106

Börjesson, Stefan, Sofia Ny, Maria Egervärn, Jakob Bergström, Åsa Rosengren, Stina Englund, Sonja Löfmark, and Sara Byfors. 2016. “Limited Dissemination of Extended-Spectrum β-Lactamase–and Plasmid-Encoded AmpC–Producing Escherichia coli from Food and Farm Animals, Sweden.” Emerging Infectious Diseases 22 (4): 634–40. https://doi.org/10.3201/eid2204.151142. Bradford, Patricia A. 2001. “Extended-Spectrum-Lactamases in the 21st Century: Characterization, Epidemiology, and Detection of This Important Resistance Threat” 14 (4): 933–51. https://doi.org/10.1128/CMR.14.4.933-951.2001. Bréchet, Caroline, Julie Plantin, Marlène Sauget, Michelle Thouverez, Daniel Talon, Pascal Cholley, Christophe Guyeux, Didier Hocquet, and Xavier Bertrand. 2014. “Wastewater Treatment Plants Release Large Amounts of Extended-Spectrum β - Lactamase – Producing Escherichia coli Into the Environment” 58: 1658–65. https://doi.org/10.1093/cid/ciu190. Bueno, Irene, Claudio Verdugo, Omar Jimenez-Lopez, Pedro Pablo Alvarez, Gerardo Gonzalez-Rocha, Celia A. Lima, Dominic A. Travis, et al. 2020. “Role of Wastewater Treatment Plants on Environmental Abundance of Antimicrobial Resistance Genes in Chilean Rivers.” International Journal of Hygiene and Environmental Health 223 (1): 56–64. https://doi.org/10.1016/j.ijheh.2019.10.006. Burnham, Kenneth P., and David Raymond Anderson. 2002. Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach. Springer. Calenge, Clement. 2015. “Home Range Estimation in R: The AdehabitatHR Package.” Carattoli, Alessandra, Ea Zankari, Aurora García-Fernández, Mette Voldby Larsen, Ole Lund, Laura Villa, Frank Møller Aarestrup, and Henrik Hasman. 2014. “In Silico Detection and Typing of Plasmids Using PlasmidFinder and Plasmid Multilocus Sequence Typing.” Antimicrobial Agents and Chemotherapy 58 (7): 3895–3903. https://doi.org/10.1128/AAC.02412-14. Carroll, D., J. Wang, S. Fanning, and B. J. Mcmahon. 2015. “Antimicrobial Resistance in Wildlife: Implications for Public Health.” Zoonoses and Public Health 62 (7): 534– 42. https://doi.org/10.1111/zph.12182. Carter, Diana L., Kathryn M. Docherty, Sharon A. Gill, Kathleen Baker, Jordan Teachout, and Maarten J. Vonhof. 2018. “Antibiotic Resistant Bacteria Are Widespread in Songbirds across Rural and Urban Environments.” Science of the Total Environment. https://doi.org/10.1016/j.scitotenv.2018.01.343. Carver, S, S N Bevins, M R Lappin, E E Boydston, L M Lyren, M Alldredge, K A Logan, et al. 2016. “Pathogen Exposure Varies Widely among Sympatric Populations of Wild and Domestic Felids across the United States.” Ecological Applications 26 (2): 367–81. https://doi.org/10.1890/15-0445. Caucci, Serena, Antti Karkman, Damiano Cacace, Marcus Rybicki, Patrick Timpel, Veiko Voolaid, Robert Gurke, Marko Virta, and Thomas U. Berendonk. 2016. “Seasonality of Antibiotic Prescriptions for Outpatients and Resistance Genes in Sewers and Wastewater Treatment Plant Outflow.” FEMS Microbiology Ecology 92

107

(5): 1–10. https://doi.org/10.1093/femsec/fiw060. Cole, Dana, David J V Drum, David E Stalknecht, David G White, Margie D Lee, Sherry Ayers, Mark Sobsey, and John J Maurer. 2005. “Free-Living Canada Geese and Antimicrobial Resistance.” Emerging Infectious Diseases 11 (6): 935–38. https://doi.org/10.3201/eid1106.040717. Collignon, Peter, John J. Beggs, Timothy R. Walsh, Sumanth Gandra, and Ramanan Laxminarayan. 2018. “Anthropological and Socioeconomic Factors Contributing to Global Antimicrobial Resistance: A Univariate and Multivariable Analysis.” The Lancet Planetary Health 2 (9): e398–405. https://doi.org/10.1016/S2542- 5196(18)30186-4. Costa, Daniela, Patricia Poeta, Yolanda Sáenz, Ana Cláudia Coelho, Manuela Matos, Laura Vinué, Jorge Rodrigues, and Carmen Torres. 2007. “Prevalence of Antimicrobial Resistance and Resistance Genes in Faecal Isolates Recovered from Healthy Pets.” Veterinary Microbiology 127 (2). https://doi.org/10.1016/j.vetmic.2007.08.004. Croucher, Nicholas J, Andrew J Page, Thomas R Connor, Aidan J Delaney, Jacqueline A Keane, Stephen D Bentley, Julian Parkhill, and Simon R Harris. 2015. “Rapid Phylogenetic Analysis of Large Samples of Recombinant Bacterial Whole Genome Sequences Using Gubbins.” Nucleic Acids Research 43 (3): 15. https://doi.org/10.1093/nar/gku1196. Cummings, Kevin J, Victor A Aprea, and Craig Altier. 2015. “Antimicrobial Resistance Trends among Canine Escherichia coli Isolates Obtained from Clinical Samples in the Northeastern USA, 2004–2011.” The Canadian Veterinary Journal. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4357913/pdf/cvj_04_393.pdf. Damborg, P., E. M. Broens, B. B. Chomel, S. Guenther, F. Pasmans, J. A. Wagenaar, J. S. Weese, et al. 2016. “Bacterial Zoonoses Transmitted by Household Pets: State-of- the-Art and Future Perspectives for Targeted Research and Policy Actions.” Journal of Comparative Pathology 155 (1): S27–40. https://doi.org/10.1016/j.jcpa.2015.03.004. Daszak, Peter, Andrew A Cunningham, and Alex D Hyatt. 2000. “Emerging Infectious Diseases of Wildlife-Threats to Biodiversity and Human Health.” Science 287 (5452): 443–49. Dolejska, M., A. Cizek, and I. Literak. 2007. “High Prevalence of Antimicrobial- Resistant Genes and Integrons in Escherichia coli Isolates from Black-Headed Gulls in the Czech Republic.” Journal of Applied Microbiology 103 (1): 11–19. https://doi.org/10.1111/j.1365-2672.2006.03241.x. Dolejska, Monika, and Ivan Literak. 2019. “Wildlife Is Overlooked in the Epidemiology of Medically Important Antimicrobial Resistant Bacteria.” Antimicrobial Agents and Chemotherapy 63 (8): Dolejska, Monika, and Ivan Literak. "Wildlife Is O. https://doi.org/10.1128/AAC.01167-19. Dolejska, Monika, and Costas C Papagiannitsis. 2018. “Plasmid-Mediated Resistance Is

108

Going Wild.” Plasmid 99: 99–111. https://doi.org/10.1016/j.plasmid.2018.09.010. Dormann, Carsten F, Jana M Mcpherson, Miguel B Araújo, Roger Bivand, Janine Bolliger, Gudrun Carl, Richard G Davies, et al. 2007. “Methods to Account for Spatial Autocorrelation in the Analysis of Species Distributional Data: A Review.” Ecography 30 (5): 609–28. https://doi.org/10.1111/j.2007.0906-7590.05171.x. Egervärn, Maria, Stina Englund, Marianne Ljunge, Christer Wiberg, Maria Finn, Mats Lindblad, and Stefan Börjesson. 2017. “Unexpected Common Occurrence of Transferable Extended Spectrum Cephalosporinase-Producing Escherichia coli in Swedish Surface Waters Used for Drinking Water Supply.” Science of the Total Environment 587–588 (June): 466–72. https://doi.org/10.1016/j.scitotenv.2017.02.157. Ewers, C., A. Bethe, T. Semmler, S. Guenther, and L.H. Wieler. 2012. “Extended- Spectrum β-Lactamase-Producing and AmpC-Producing Escherichia coli from Livestock and Companion Animals, and Their Putative Impact on Public Health: A Global Perspective.” Clinical Microbiology and Infection 18 (7): 646–55. https://doi.org/10.1111/j.1469-0691.2012.03850.x. Ewers, C, A Bethe, T Semmler, S Guenther, and L H Wieler. 2012. “Extended-Spectrum b-Lactamase-Producing and AmpC-Producing Escherichia coli from Livestock and Companion Animals, and Their Putative Impact on Public Health: A Global Perspective.” Clinical Microbiology and Infection 18 (7): 646–55. https://doi.org/10.1111/j.1469-0691.2012.03850.x. Ewers, Christa, Mirjam Grobbel, Ivonne Stamm, Peter A Kopp, Ines Diehl, Torsten Semmler, Angelika Fruth, et al. 2010. “Emergence of Human Pandemic O25:H4- ST131 CTX-M-15 Extended-Spectrum-b-Lactamase-Producing Escherichia coli among Companion Animals.” Journal of Antimicrobial Chemotherapy 65 (4): 651– 60. https://doi.org/10.1093/jac/dkq004. Fagerströ, Anna ID, Paula Mö lling, Faisal Ahmad Khan, Martin Sundqvist, Jana Jass, and Bo Sö derquist. 2019. “Comparative Distribution of Extended-Spectrum Beta- Lactamase-Producing Escherichia coli from Urine Infections and Environmental Waters.” https://doi.org/10.1371/journal.pone.0224861. Fountain-Jones, Nicholas M., William D. Pearse, Luis E. Escobar, Ana Alba-Casals, Scott Carver, T. Jonathan Davies, Simona Kraberger, et al. 2018. “Towards an Eco- Phylogenetic Framework for Infectious Disease Ecology.” Biological Reviews 93 (2): 950–70. https://doi.org/10.1111/brv.12380. Furness, Lauren E., Amy Campbell, Lihong Zhang, William H. Gaze, and Robbie A. McDonald. 2017. “Wild Small Mammals as Sentinels for the Environmental Transmission of Antimicrobial Resistance.” Environmental Research 154: 28–34. https://doi.org/10.1016/j.envres.2016.12.014. Garner, Emily, Romina Benitez, Emily von Wagoner, Richard Sawyer, Erin Schaberg, W. Cully Hession, Leigh Anne H. Krometis, Brian D. Badgley, and Amy Pruden. 2017. “Stormwater Loadings of Antibiotic Resistance Genes in an Urban Stream.”

109

Water Research 123: 144–52. https://doi.org/10.1016/j.watres.2017.06.046. Gehrt, S D, L L Hungerford, and S Hatten. 2001. “Drug Effects on Recaptures of Raccoons.” Wildlife Society Bulletin, 833–37. Gehrt, Stanley D., Chris Anchor, and Lynsey A. White. 2009. “Home Range and Landscape Use of Coyotes in a Metropolitan Landscape: Conflict or Coexistence?” Journal of Mammalogy 90 (5): 1045–57. Gehrt, Stanley D, and Erik K Fritzell. 1998. “Resource Distribution, Female Home Range Dispersion and Male Spatial Interactions: Group Structure in a Solitary Carnivore.” Animal Behaviour 55 (5): 1211–27. Gehrt, Stanley D, Seth P Riley, and Brian L Cypher. 2010. Urban Carnivores: Ecology, Conflict, and Conservation. Johns Hopkins University Press. Gillings, Michael R, William H Gaze, Amy Pruden, Kornelia Smalla, James M Tiedje, and Yong-Guan Zhu. 2015. “Using the Class 1 Integron-Integrase Gene as a Proxy for Anthropogenic Pollution.” The ISME Journal 9 (6): 1269–79. https://doi.org/10.1038/ismej.2014.226. Gordon, D. M., and Ann Cowling. 2003. “The Distribution and Genetic Structure of Escherichia coli in Australian Vertebrates: Host and Geographic Effects.” Microbiology 149 (12): 3575–86. https://doi.org/10.1099/mic.0.26486-0. Graham, David W, Gilles Bergeron, Megan W Bourassa, James Dickson, Filomena Gomes, Adina Howe, Laura H Kahn, et al. 2019. “Complexities in Understanding Antimicrobial Resistance across Domesticated Animal, Human, and Environmental Systems.” Ann. N.Y. Acad. Sci 17. https://doi.org/10.1111/nyas.14036. Gras, P, S Knuth, K Börner, L Marescot, S Benhaiem, A Aue, U Wittstatt, B Kleinschmit, and S Kramer-Schadt. 2018. “Landscape Structures Affect Risk of Canine Distemper in Urban Wildlife.” Frontiers in Ecology and Evolution 6 (SEP). https://doi.org/10.3389/fevo.2018.00136. Greig, J., A. Rajić, I. Young, M. Mascarenhas, L. Waddell, and J. LeJeune. 2015. “A Scoping Review of the Role of Wildlife in the Transmission of Bacterial Pathogens and Antimicrobial Resistance to the Food Chain.” Zoonoses and Public Health 62 (4): 269–84. https://doi.org/10.1111/zph.12147. Grueber, C. E., S. Nakagawa, R. J. Laws, and I. G. Jamieson. 2011. “Multimodel Inference in Ecology and Evolution: Challenges and Solutions.” Journal of Evolutionary Biology 24 (4): 699–711. Guardabassi, L., Stefan Schwarz, and David H. Lloyd. 2004. “Pet Animals as Reservoirs of Antimicrobial-Resistant Bacteria: Review.” Journal of Antimicrobial Chemotherapy 54 (2): 321–32. https://doi.org/10.1093/jac/dkh332. Guenther, Sebastian, Christa Ewers, and Lothar H. Wieler. 2011. “Extended-Spectrum Beta-Lactamases Producing E. Coli in Wildlife, yet Another Form of Environmental Pollution?” Frontiers in Microbiology 2 (DEC): 1–13. https://doi.org/10.3389/fmicb.2011.00246.

110

Guenther, Sebastian, Mirjam Grobbel, Janine Beutlich, Astrid Bethe, Nicole D Friedrich, Andreas Goedecke, Antina Lübke-Becker, Beatriz Guerra, Lothar H Wieler, and Christa Ewers. 2010. “CTX-M-15-Type Extended-Spectrum Beta-Lactamases- Producing Escherichia coli from Wild Birds in Germanye Mi4_148 641..645.” Environmental Microbiology Reports 2 (5): 641–45. https://doi.org/10.1111/j.1758- 2229.2010.00148.x. Guenther, Sebastian, Torsten Semmler, Annegret Stubbe, Michael Stubbe, Lothar H Wieler, and Katharina Schaufler. 2017. “Chromosomally Encoded ESBL Genes in Escherichia coli of ST38 from Mongolian Wild Birds.” Journal of Antimicrobial Chemotherapy 72 (5): 1310–13. https://doi.org/10.1093/jac/dkx006. Gurevich, Alexey, Vladislav Saveliev, Nikolay Vyahhi, and Glenn Tesler. 2013. “Genome Analysis QUAST: Quality Assessment Tool for Genome Assemblies” 29 (8): 1072–75. https://doi.org/10.1093/bioinformatics/btt086. Hamamoto, Kouta, Toshiro Tokunaga, Nobuyoshi Yagi, and Itaru Hirai. 2020. “Characterization of blaCTX-M-14 Transposition from Plasmid to Chromosome in Escherichia coli Experimental Strain.” International Journal of Medical Microbiology 310 (2): 151395. https://doi.org/10.1016/j.ijmm.2020.151395. Hamamoto, Kouta, Shuhei Ueda, Takehiko Toyosato, Yoshimasa Yamamoto, and Itaru Hirai. 2016. “High Prevalence of Chromosomal blaCTX-M-14 in Escherichia coli Isolates Possessing blaCTX-M-14.” https://doi.org/10.1128/AAC.00108-16. Hassell, James M, Melissa J Ward, Dishon Muloi, Judy M Bettridge, Timothy P Robinson, Sam Kariuki, Allan Ogendo, et al. 2019. “Clinically Relevant Antimicrobial Resistance at the Wildlife–Livestock–Human Interface in Nairobi: An Epidemiological Study.” The Lancet Planetary Health 3 (6): e259–69. https://doi.org/10.1016/S2542-5196(19)30083-X. Henner, Chrissie M, Michael J Chamberlain, Bruce D Leopold, and L Wes Burger. 2004. “A Multi-Resolution Assessment of Raccoon Den Selection.” The Journal of Wildlife Management 68 (1): 179–87. Hoffmann, Cliff O, and Jack L Gottschan. 1977. “Numbers, Distribution, and Movements of a Raccoon Population in a Suburban Residential Community.” Journal of Mammalogy 58 (4): 623–36. Holmes, Alison H., Luke S.P. Moore, Arnfinn Sundsfjord, Martin Steinbakk, Sadie Regmi, Abhilasha Karkey, Philippe J. Guerin, and Laura J.V. Piddock. 2016. “Understanding the Mechanisms and Drivers of Antimicrobial Resistance.” The Lancet 387 (10014): 176–87. https://doi.org/10.1016/S0140-6736(15)00473-0. Hooban, Brigid, Aoife Joyce, Kelly Fitzhenry, Carlos Chique, and Dearbháile Morris. 2020. “The Role of the Natural Aquatic Environment in the Dissemination of Extended Spectrum Beta-Lactamase and Carbapenemase Encoding Genes: A Scoping Review.” Water Research 180: 1–12. https://doi.org/10.1016/j.watres.2020.115880. Hordijk, Joost, Anky Schoormans, Mandy Kwakernaak, Birgitta Duim, Els Broens,

111

Cindy Dierikx, Dik Mevius, and Jaap A Wagenaar. 2013. “High Prevalence of Fecal Carriage of Extended Spectrum β-Lactamase/AmpC-Producing Enterobacteriaceae in Cats and Dogs.” Frontiers in Microbiology 4: 242. https://doi.org/10.3389/fmicb.2013.00242. Hughes, James M. 2011. “Preserving the Lifesaving Power of Antimicrobial Agents.” JAMA - Journal of the American Medical Association 305 (10): 1027–28. https://doi.org/10.1001/jama.2011.279. Huijbers, Patricia M. C., Hetty Blaak, Mart C. M. de Jong, Elisabeth A. M. Graat, Christina M. J. E. Vandenbroucke-Grauls, and Ana Maria de Roda Husman. 2015. “Role of the Environment in the Transmission of Antimicrobial Resistance to Humans: A Review.” Environmental Science & Technology 49 (20): 11993–4. https://doi.org/10.1021/acs.est.5b02566. Jardine, C, R J Reid-Smith, N Janecko, M Allan, and S A McEwen. 2011. “Salmonella in Raccoons (Procyon Lotor) in Southern Ontario, Canada.” Journal of Wildlife Diseases 47 (2): 344–51. https://doi.org/10.7589/0090-3558-47.2.344. Jardine, Claire M., Nicol Janecko, Mike Allan, Patrick Boerlin, Gabhan Chalmers, Gosia Kozak, Scott A. McEwen, and Richard J. Reid-Smith. 2012. “Antimicrobial Resistance in Escherichia coli Isolates from Raccoons (Procyon Lotor) in Southern Ontario, Canada.” Applied and Environmental Microbiology 78 (11): 3873–79. https://doi.org/10.1128/AEM.00705-12. Jay-Russell, Michele T., Alexis F. Hake, Yingjia Bengson, Anyarat Thiptara, and Tran Nguyen. 2014. “Prevalence and Characterization of Escherichia coli and Salmonella Strains Isolated from Stray Dog and Coyote Feces in a Major Leafy Greens Production Region at the United States-Mexico Border.” PLoS ONE 9 (11). https://doi.org/10.1371/journal.pone.0113433. Jobbins, S E, and K A Alexander. 2015a. “From Whence They Came - Antibiotic- Resistant Escherichia coli in African Wildlife.” Journal of Wildlife Diseases 51 (4): 811–20. https://doi.org/10.7589/2014-11-257. Johnson, Jerald B, and Kristian S Omland. 2004. “Model Selection in Ecology and Evolution.” Trends in Ecology & Evolution 19 (2): 101–8. Jones, Kate E, Nikkita G Patel, Marc A Levy, Adam Storeygard, Deborah Balk, John L Gittleman, and Peter Daszak. 2008. “Global Trends in Emerging Infectious Diseases” 451 (2): 990–94. https://doi.org/10.1038/nature06536. Jonsson, Viktor, Tobias Österlund, Olle Nerman, and Erik Kristiansson. 2016. “Statistical Evaluation of Methods for Identification of Differentially Abundant Genes in Comparative Metagenomics.” BMC Genomics 17 (1): 78. https://doi.org/10.1186/s12864-016-2386-y. Kalyaanamoorthy, Subha, Bui Quang Minh, Thomas K.F. Wong, Arndt Von Haeseler, and Lars S. Jermiin. 2017. “ModelFinder: Fast Model Selection for Accurate Phylogenetic Estimates.” Nature Methods 14 (6): 587–89. https://doi.org/10.1038/nmeth.4285.

112

Karesh, William B, Andy Dobson, James O Lloyd-smith, Juan Lubroth, and Matthew A Dixon. 2012. “Ecology of Zoonoses : Natural and Unnatural Histories.” Keen, Patricia L, Charles W Knapp, Kenneth J Hall, and David W Graham. 2018. “Seasonal Dynamics of Tetracycline Resistance Gene Transport in the Sumas River Agricultural Watershed of British Columbia, Canada.” Science of the Total Environment 628–629: 490–98. https://doi.org/10.1016/j.scitotenv.2018.01.278. Kipkorir, Kebenei C, Paul O Ang’ienda, David M Onyango, and Patrick O Onyango. 2019. “Antibiotic Resistance of Escherichia coli from Humans and Black Rhinoceroses in Kenya.” EcoHealth, 1–11. https://doi.org/10.1007/s10393-019- 01461-z. Korzeniewska, Ewa, and Monika Harnisz. 2013. “Extended-Spectrum Beta-Lactamase (ESBL)-Positive Enterobacteriaceae in Municipal Sewage and Their Emission to the Environment.” Journal of Environmental Management 128: 904–11. https://doi.org/10.1016/j.jenvman.2013.06.051. Korzeniewska, Ewa, Anna Korzeniewska, and Monika Harnisz. 2013. “Antibiotic Resistant Escherichia coli in Hospital and Municipal Sewage and Their Emission to the Environment.” Ecotoxicology and Environmental Safety 91: 96–102. https://doi.org/10.1016/j.ecoenv.2013.01.014. Kozak, Gosia K., Patrick Boerlin, Nicol Janecko, Richard J. Reid-Smith, and Claire Jardine. 2009. “Antimicrobial Resistance in Escherichia coli Isolates from Swine and Wild Small Mammals in the Proximity of Swine Farms and in Natural Environments in Ontario, Canada.” Applied and Environmental Microbiology 75 (3): 559–66. https://doi.org/10.1128/AEM.01821-08. Kumarasamy, Karthikeyan K., Mark A. Toleman, Timothy R. Walsh, Jay Bagaria, Fafhana Butt, Ravikumar Balakrishnan, Uma Chaudhary, et al. 2010. “Emergence of a New Antibiotic Resistance Mechanism in India, Pakistan, and the UK: A Molecular, Biological, and Epidemiological Study.” The Lancet Infectious Diseases 10 (9): 597–602. https://doi.org/10.1016/S1473-3099(10)70143-2. Kümmerer, Klaus. 2004. “Resistance in the Environment.” Journal of Antimicrobial Chemotherapy 54 (2): 311–20. https://doi.org/10.1093/jac/dkh325. Larsson, D. G.Joakim, Antoine Andremont, Johan Bengtsson-Palme, Kristian Koefoed Brandt, Ana Maria de Roda Husman, Patriq Fagerstedt, Jerker Fick, et al. 2018. “Critical Knowledge Gaps and Research Needs Related to the Environmental Dimensions of Antibiotic Resistance.” Environment International 117 (January): 132–38. https://doi.org/10.1016/j.envint.2018.04.041. Laxminarayan, Ramanan, Adriano Duse, Chand Wattal, Anita K M Zaidi, Heiman F L Wertheim, Nithima Sumpradit, Erika Vlieghe, et al. 2013. “Antibiotic Resistance— the Need for Global Solutions.” The Lancet Infectious Diseases 13 (12): 1057–98. https://doi.org/10.1016/S1473-3099(13)70318-9. Lenart-boroń, Anna M, Klaudia Kulik, and Ewelina Jelonkiewicz. 2020. “Toxic / Hazardous Substances and Environmental Engineering Isolated in Proximity to a

113

Sewage Treatment Plant.” Journal of Environmental Science and Health, Part A 0 (0): 1–10. https://doi.org/10.1080/10934529.2020.1826774. Leonard, E. K., D. L. Pearl, N. Janecko, R. L. Finley, R. J. Reid-Smith, J. S. Weese, and A. S. Peregrine. 2015. “Risk Factors for Carriage of Antimicrobial-Resistant Salmonella Spp and Escherichia coli in Pet Dogs from Volunteer Households in Ontario, Canada, in 2005 and 2006.” American Journal of Veterinary Research 76 (11): 959–68. Letunic, Ivica, and Peer Bork. 2016. “Interactive Tree of Life (ITOL) v3: An Online Tool for the Display and Annotation of Phylogenetic and Other Trees.” Nucleic Acids Research 44. https://doi.org/10.1093/nar/gkw290. Li, H., and R. Durbin. 2009. “Fast and Accurate Short Read Alignment with Burrows- Wheeler Transform.” Bioinformatics 25 (14): 1754–60. https://doi.org/10.1093/bioinformatics/btp324. Liu, Xiaoqiang, Kamoltip Thungrat, and Dawn M. Boothe. 2016. “Occurrence of OXA- 48 Carbapenemase and Other β-Lactamase Genes in ESBL-Producing Multidrug Resistant Escherichia coli from Dogs and Cats in the United States, 2009–2013.” Frontiers in Microbiology 7 (July): 1057. https://doi.org/10.3389/fmicb.2016.01057. Liu, Yi Yun, Yang Wang, Timothy R. Walsh, Ling Xian Yi, Rong Zhang, James Spencer, Yohei Doi, et al. 2016. “Emergence of Plasmid-Mediated Colistin Resistance Mechanism MCR-1 in Animals and Human Beings in China: A Microbiological and Molecular Biological Study.” The Lancet Infectious Diseases 16 (2): 161–68. https://doi.org/10.1016/S1473-3099(15)00424-7. Lotze, Joerg-Henner, and Sydney Anderson. 1979. “Procyon Lotor.” Mammalian Species, no. 119 (June): 1. https://doi.org/10.2307/3503959. Ma, Liping, Yu Xia, Bing Li, Ying Yang, Li Guan Li, James M. Tiedje, and Tong Zhang. 2016. “Metagenomic Assembly Reveals Hosts of Antibiotic Resistance Genes and the Shared Resistome in Pig, Chicken, and Human Feces.” Environmental Science and Technology 50 (1): 420–27. https://doi.org/10.1021/acs.est.5b03522. Mackenstedt, U, D Jenkins, and T Romig. 2015. “The Role of Wildlife in the Transmission of Parasitic Zoonoses in Peri-Urban and Urban Areas.” International Journal for Parasitology: Parasites and Wildlife 4 (1): 71–79. Martak, Daniel, Charles P Henriot, Marion Broussier, Charlotte Couchoud, Benoit Valot, Marion Richard, Julie Couchot, et al. 2020. “High Prevalence of Human-Associated Escherichia coli in Wetlands Located in Eastern France” 11 (September): 1–7. https://doi.org/10.3389/fmicb.2020.552566. Marti, E, J Jofre, and J L Balcazar. 2013. “Prevalence of Antibiotic Resistance Genes and Bacterial Community Composition in a River Influenced by a Wastewater Treatment Plant.” PLoS ONE 8 (10): 78906. https://doi.org/10.1371/journal.pone.0078906. Marti, Elisabet, Eleni Variatza, and Jose Luis Balcazar. 2014. “The Role of Aquatic

114

Ecosystems as Reservoirs of Antibiotic Resistance.” Trends in Microbiology 22 (1): 36–41. https://doi.org/10.1016/j.tim.2013.11.001. Martiny, Adam C., Jennifer B. H. Martiny, Claudia Weihe, Andrew Field, and Julie C. Ellis. 2011. “Functional Metagenomics Reveals Previously Unrecognized Diversity of Antibiotic Resistance Genes in Gulls.” Frontiers in Microbiology 2 (November): 238. https://doi.org/10.3389/fmicb.2011.00238. Mather, Alison E, Louise Matthews, Dominic J Mellor, Richard Reeve, Matthew J Denwood, Patrick Boerlin, Richard J Reid-Smith, et al. 2012. “An Ecological Approach to Assessing the Epidemiology of Antimicrobial Resistance in Animal and Human Populations.” Proceedings of the Royal Society B: Biological Sciences 279 (1733): 1630–39. https://doi.org/10.1098/rspb.2011.1975. Mathers, Amy J, Gisele Peirano, and Johann D D Pitout. 2015. “The Role of Epidemic Resistance Plasmids and International High-Risk Clones in the Spread of Multidrug- Resistant Enterobacteriaceae.” Clinical Microbiology Reviews 28 (3): 565–91. https://doi.org/10.1128/CMR.00116-14. Mathys, D. A., D. F. Mollenkopf, C. A. Bremer, J. B. Daniels, and T. E. Wittum. 2017. “Prevalence of AmpC- and Extended-Spectrum β-Lactamase-Harbouring Enterobacteriaceae in Faecal Flora of a Healthy Domestic Canine Population.” Zoonoses and Public Health 64 (7): 554–60. https://doi.org/10.1111/zph.12341. Mazel, Didier. 2006. “Integrons: Agents of Bacterial Evolution.” Nature Reviews Microbiology 4 (8): 608–20. https://doi.org/10.1038/nrmicro1462. Mazodier, Philippe, and Julian Davies. 1991. “Gene Transfer between Distantly Related Bacteria.” Annual Review of Genetics 25 (1): 147–71. Meer, J. W. Van Der. 2013. “The Infectious Disease Challenges of Our Time.” Frontiers in Public Health 1 (7): 1–2. https://doi.org/10.1038/ncomms2343. Mercat, Mathilde, Olivier Clermont, Méril Massot, Etienne Ruppe, Michel De Garine- Wichatitsky, Eve Miguel, Hugo Valls Fox, et al. 2016. “Escherichia coli Population Structure and Antibiotic Resistance at a Buffalo/Cattle Interface in Southern Africa.” Applied and Environmental Microbiology 82 (5): 1459–67. https://doi.org/10.1128/AEM.03771-15. Meyer, D, and C Buchta. 2015. “Proxy: Distance and Similarity Measures. R Package Version 0.4–15.” Mikheenko, Alla, Vladislav Saveliev, and Alexey Gurevich. 2016. “MetaQUAST: Evaluation of Metagenome Assemblies.” Bioinformatics 32.7: 1088–90. https://doi.org/10.1093/bioinformatics/btv697. Miller, Elizabeth A, Julia B Ponder, Michelle Willette, and Timothy J Johnson. 2020. “Merging Metagenomics and Spatial Epidemiology to Understand the Distribution of Antimicrobial Resistance Genes from Enterobacteriaceae in Wild Owls .” Applied and Environmental Microbiology. https://doi.org/10.1128/AEM.00571-20. Miller, R V, K Gammon, and M J Day. 2009. “Antibiotic Resistance among Bacteria

115

Isolated from Seawater and Penguin Fecal Samples Collected near Palmer Station, Antarctica 1.” Canadian Journal of Microbiology 55 (1): 37–45. https://doi.org/10.1139/W08-119. Moura, A., M. Soares, C. Pereira, N. Leitao, I. Henriques, and A. Correia. 2009. “INTEGRALL: A Database and Search Engine for Integrons, Integrases and Gene Cassettes.” Bioinformatics 25 (8): 1096–98. https://doi.org/10.1093/bioinformatics/btp105. Murphy, David, Antonia Ricci, Zanda Auce, J. Gabriel Beechinor, Hanne Bergendahl, Rory Breathnach, Jiří Bureš, et al. 2017. “EMA and EFSA Joint Scientific Opinion on Measures to Reduce the Need to Use Antimicrobial Agents in Animal Husbandry in the European Union, and the Resulting Impacts on Food Safety (RONAFA).” EFSA Journal 15 (1). Nakagawa, Shinichi, and Holger Schielzeth. 2013. “A General and Simple Method for Obtaining R 2 from Generalized Linear Mixed-Effects Models.” Methods in Ecology and Evolution 4 (2): 133–42. Nelson, M, SH Jones, C Edwards, and JC Ellis. 2008. “Characterization of Escherichia coli Populations from Gulls, Landfill Trash, and Wastewater Using Ribotyping.” Diseases of Aquatic Organisms 81 (1): 53–63. https://doi.org/10.3354/dao01937. Nurk, Sergey, Dmitry Meleshko, Anton Korobeynikov, and Pavel A Pevzner. 2017. “MetaSPAdes: A New Versatile Metagenomic Assembler.” Genome Research 27 (5): 824–34. https://doi.org/10.1101/gr.213959.116. Oksanen, Jari, F Guillaume Blanchet, Michael Friendly, Roeland Kindt, Pierre Legendre, Dan Mcglinn, Peter R Minchin, et al. 2013. “Package ‘Vegan.’” Community Ecology Package, Version 2. Orlek, Alex, Nicole Stoesser, Muna F. Anjum, Michel Doumith, Matthew J. Ellington, Tim Peto, Derrick Crook, et al. 2017. “Plasmid Classification in an Era of Whole- Genome Sequencing: Application in Studies of Antibiotic Resistance Epidemiology.” Frontiers in Microbiology. Frontiers Research Foundation. https://doi.org/10.3389/fmicb.2017.00182. Österblad, Monica, Kai Norrdahl, Erkki Korpimäki, and Pentti Huovinen. 2001. “Antibiotic Resistance. How Wild Are Wild Mammals?” Nature 409 (6816): 37–38. https://doi.org/10.1038/35051173. Pal, Chandan, Johan Bengtsson-Palme, Erik Kristiansson, and D G Joakim Larsson. 2016. “The Structure and Diversity of Human, Animal and Environmental Resistomes.” Microbiome 4 (1): 54. https://doi.org/10.1186/s40168-016-0199-5. Paradis, Emmanuel, Julien Claude, and Korbinian Strimmer. 2004. “APE: Analyses of Phylogenetics and Evolution in R Language.” Bioinformatics Applications Note 20 (2): 289–90. https://doi.org/10.1093/bioinformatics/btg412. Parker, D., M.K. Sniatynski, D. Mandrusiak, and J.E. Rubin. 2016. “Extended-Spectrum β-Lactamase Producing Escherichia coli Isolated from Wild Birds in Saskatoon,

116

Canada.” Letters in Applied Microbiology 63 (1): 11–15. https://doi.org/10.1111/lam.12589. Partridge, Sally R. 2015. “Resistance Mechanisms in Enterobacteriaceae.” Pathology 47 (3): 276–84. http://www.lahey.org/Studies/,. Partridge, Sally R, Stephen M Kwong, Neville Firth, and Slade O Jensen. 2018. “Mobile Genetic Elements Associated with Antimicrobial Resistance.” Clinical Microbiology Reviews 31 (4): e00088-17. Paterson, David L, and Robert A Bonomo. 2005. “Extended-Spectrum-Lactamases: A Clinical Update.” Clinical Microbiology Reviews 18 (4): 657–86. https://doi.org/10.1128/CMR.18.4.657-686.2005. Perez, Federico, and Maria Virginia Villegas. 2015. “The Role of Surveillance Systems in Confronting the Global Crisis of Antibiotic-Resistant Bacteria.” Current Opinion in Infectious Diseases 28 (4): 375. https://doi.org/10.1097/QCO.0000000000000182. Pesapane, R, M Ponder, and K A Alexander. 2013. “Tracking Pathogen Transmission at the Human-Wildlife Interface: Banded Mongoose and Escherichia coli.” EcoHealth 10 (2): 115–28. https://doi.org/10.1007/s10393-013-0838-2. Peto, Leon, Nicola J Fawcett, Derrick W Crook, Tim EA Peto, Martin J Llewelyn, and A Sarah Walker. 2019. “Selective Culture Enrichment and Sequencing of Faeces to Enhance Detection of Antimicrobial Resistance Genes.” BioRxiv, February, 560292. https://doi.org/10.1101/560292. Philippon, Alain, Guillaume Arlet, and George A Jacoby. 2002. “Plasmid-Determined AmpC-Type-Lactamases” 46 (1): 1–11. https://doi.org/10.1128/AAC.46.1.1- 11.2002. Pitout, Johann D D, Patrice Nordmann, Kevin B Laupland, and Laurent Poirel. 2005. “Emergence of Enterobacteriaceae Producing Extended-Spectrum b-Lactamases (ESBLs) in the Community.” Journal of Antimicrobial Chemotherapy 56: 52–59. https://doi.org/10.1093/jac/dki166. Plowright, Raina K., Colin R. Parrish, Hamish McCallum, Peter J. Hudson, Albert I. Ko, Andrea L. Graham, and James O. Lloyd-Smith. 2017. “Pathways to Zoonotic Spillover.” Nature Reviews Microbiology 15 (8): 502–10. https://doi.org/10.1038/nrmicro.2017.45. Poirel, Laurent, Anaïs Potron, Carolina De, La Cuesta, Timothy Cleary, Patrice Nordmann, and L Silvia Munoz-Price. 2012. “Wild Coastline Birds as Reservoirs of Broad-Spectrum-Lactamase-Producing Enterobacteriaceae in Miami Beach, Florida.” https://doi.org/10.1128/AAC.05982-11. Pomba, Constança, Merja Rantala, Christina Greko, Keith Edward Baptiste, Boudewijn Catry, Engeline van Duijkeren, Ana Mateus, et al. 2017. “Public Health Risk of Antimicrobial Resistance Transfer from Companion Animals.” Journal of Antimicrobial Chemotherapy 72 (4): 957–68. https://doi.org/10.1093/jac/dkw481.

117

Prange, Suzanne, and Stanley Gehrt. 2004. “Changes in Mesopredator-Community Structure in Response to Urbanization.” Canadian Journal of Zoology 82 (11): 1804–17. https://doi.org/10.1139/z04-179. Procter, T. D., D. L. Pearl, R. L. Finley, E. K. Leonard, N. Janecko, R. J. Reid-Smith, J. S. Weese, A. S. Peregrine, and J. M. Sargeant. 2014. “A Cross-Sectional Study Examining the Prevalence and Risk Factors for Anti-Microbial-Resistant Generic Escherichia coli in Domestic Dogs That Frequent Dog Parks in Three Cities in South-Western Ontario, Canada.” Zoonoses and Public Health 61 (4): 250–59. https://doi.org/10.1111/zph.12064. Pruden, Amy, D. G. Joakim Larsson, Alejandro Amézquita, Peter Collignon, Kristian K. Brandt, David W. Graham, James M. Lazorchak, et al. 2013. “Management Options for Reducing the Release of Antibiotics and Antibiotic Resistance Genes to the Environment.” Environmental Health Perspectives 121 (8): 878–85. https://doi.org/10.1289/ehp.1206446. Quince, Christopher, Alan W Walker, Jared T Simpson, Nicholas J Loman, and Nicola Segata. 2017. “Shotgun Metagenomics, from Sampling to Sequencing and Analysis.” Nature Biotechnology 35 (9): 833. R Development Core Team. 2020. “R: A Language and Environment for Statistical Computing.” Viana, Austria. Radhouani, Hajer, Nuno Silva, Patr??cia Poeta, Carmen Torres, Susana Correia, and Gilberto Igrejas. 2014. “Potential Impact of Antimicrobial Resistance in Wildlife, Environment, and Human Health.” Frontiers in Microbiology 5 (FEB): 1–12. https://doi.org/10.3389/fmicb.2014.00023. Ramey, Andrew M., and Christina A. Ahlstrom. 2020. “Antibiotic Resistant Bacteria in Wildlife: Perspectives on Trends, Acquisition and Dissemination, Data Gaps, and Future Directions.” Journal of Wildlife Diseases 56 (1): 1–15. https://doi.org/10.7589/2019-04-099. Rentería-Solís, Z, C Förster, A Aue, U Wittstatt, G Wibbelt, and M König. 2014. “Canine Distemper Outbreak in Raccoons Suggests Pathogen Interspecies Transmission amongst Alien and Native Carnivores in Urban Areas from Germany.” Veterinary Microbiology 174 (1–2): 50–59. https://doi.org/10.1016/j.vetmic.2014.08.034. Rizzo, L., C. Manaia, C. Merlin, T. Schwartz, C. Dagot, M.C. Ploy, I. Michael, and D. Fatta-Kassinos. 2013. “Urban Wastewater Treatment Plants as Hotspots for Antibiotic Resistant Bacteria and Genes Spread into the Environment: A Review.” Science of The Total Environment 447 (March): 345–60. https://doi.org/10.1016/J.SCITOTENV.2013.01.032. Robertson, James, and John H E Nash. 2018. “MOB-Suite: Software Tools for Clustering, Reconstruction and Typing of Plasmids from Draft Assemblies.” Microbial Genomics 4 (8). https://doi.org/10.6084/m9.figshare.6177188. Rocha-Gracia, R C, G Cortés-Cortés, P Lozano-Zarain, F Bello, Y Martínez-Laguna, and C Torres. 2014. “Faecal Escherichia coli Isolates from Healthy Dogs Harbour CTX-

118

M-15 and CMY-2 β-Lactamases.” The Veterinary Journal 203: 315–19. https://doi.org/10.1016/j.tvjl.2014.12.026. Rödiger, Stefan, Toni Kramer, Ulrike Frömmel, Jörg Weinreich, Dirk Roggenbuck, Sebastian Guenther, Katharina Schaufler, Christian Schröder, and Peter Schierack. 2015. “Intestinal Escherichia coli Colonization in a Mallard Duck Population over Four Consecutive Winter Seasons.” Environmental Microbiology 17 (9): 3352–61. https://doi.org/10.1111/1462-2920.12807. Rosatte, Rick, Mark Ryckman, Karen Ing, Sarah Proceviat, Mike Allan, Laura Bruce, Dennis Donovan, and J. Chris Davies. 2010. “Density, Movements, and Survival of Raccoons in Ontario, Canada: Implications for Disease Spread and Management.” Journal of Mammalogy 91 (1): 122–35. https://doi.org/10.1644/08-mamm-a- 201r2.1. Rwego, Innocent B., Gilbert Isabirye-Basuta, Thomas R. Gillespie, and Tony L. Goldberg. 2008. “Gastrointestinal Bacterial Transmission among Humans, Mountain Gorillas, and Livestock in Bwindi Impenetrable National Park, Uganda.” Conservation Biology 22 (6): 1600–1607. https://doi.org/10.1111/j. Šálek, Martin, Lucie Drahníková, and Emil Tkadlec. 2015. “Changes in Home Range Sizes and Population Densities of Carnivore Species along the Natural to Urban.” Mammal Review 45 (1): 1–14. Schaufler, K, K Nowak, A Düx, T Semmler, L Villa, L Kourouma, K Bangoura, L H Wieler, F H Leendertz, and S Guenther. 2018. “Clinically Relevant ESBL- Producing K. Pneumoniae ST307 and E. Coli ST38 in an Urban West African Rat Population.” Frontiers in Microbiology 9 (FEB). https://doi.org/10.3389/fmicb.2018.00150. Schaufler, Katharina, Astrid Bethe, Antina Lübke-Becker, Christa Ewers, Barbara Kohn, Lothar H Wieler, and Sebastian Guenther. 2015. “Putative Connection between Zoonotic Multiresistant Extended-Spectrum Beta-Lactamase (ESBL)-Producing Escherichia coli in Dog Feces from a Veterinary Campus and Clinical Isolates from Dogs.” Infection Ecology & Epidemiology 5 (1): 25334. https://doi.org/10.3402/iee.v5.25334. Seemann, Torsten. 2014. “Genome Analysis Prokka: Rapid Prokaryotic Genome Annotation.” Bioinformatics 30 (14): 2068–69. https://doi.org/10.1093/bioinformatics/btu153. Sevila, Julie, Céline Richomme, Hervé Hoste, Mónica G. Candela, Emmanuelle Gilot- Fromont, Annie Rodolakis, Nicolas Cebe, Denis Picot, Joël Merlet, and Hélène Verheyden. 2014. “Does Land Use within the Home Range Drive the Exposure of Roe Deer (Capreolus Capreolus) to Two Abortive Pathogens in a Rural Agro- Ecosystem?” Acta Theriologica 59 (4): 571–81. https://doi.org/10.1007/s13364-014- 0197-6. Siguier, Patricia, Edith Gourbeyre, and Mick Chandler. 2014. “Bacterial Insertion Sequences: Their Genomic Impact and Diversity.” FEMS Microbiology Reviews 38

119

(5): 865–91. https://doi.org/10.1111/1574-6976.12067. Singer, Randall S., Michael P. Ward, and George Maldonado. 2006. “Can Landscape Ecology Untangle the Complexity of Antibiotic Resistance?” Nature Reviews Microbiology 4 (12): 943–52. https://doi.org/10.1038/nrmicro1553. Skurnik. 2006. “Effect of Human Vicinity on Antimicrobial Resistance and Integrons in Animal Faecal Escherichia Coli.” Journal of Antimicrobial Chemotherapy. Smet, Annemieke, An Martel, Davy Persoons, Jeroen Dewulf, Marc Heyndrickx, Lieve Herman, Freddy Haesebrouck, and Patrick Butaye. 2010. “Broad-Spectrum β- Lactamases among Enterobacteriaceae of Animal Origin: Molecular Aspects, Mobility and Impact on Public Health.” FEMS Microbiology Reviews. John Wiley & Sons, Ltd. https://doi.org/10.1111/j.1574-6976.2009.00198.x. Stewart, Jill R, Rebecca J Gast, Roger S Fujioka, Helena M Solo-Gabriele, J Scott Meschke, Linda A Amaral-Zettler, Erika Del Castillo, et al. 2007. “Environmental Health The Coastal Environment and Human Health: Microbial Indicators, Pathogens, Sentinels and Reservoirs” 7 (2): 3. https://doi.org/10.1186/1476-069X-7- S2-S3. Straub, Mary H, Molly Church, Elle Glueckert, and Janet E Foley. 2020. “Raccoons (Procyon lotor) and Striped Skunks (Mephitis mephitis) as Potential Reservoirs of Leptospira Spp. in California.” Vector-Borne and Zoonotic Diseases 20 (6): 418–26. https://doi.org/10.1089/vbz.2019.2528. Stuewer, Frederick W. 1943. “Raccoons: Their Habits and Management in Michigan.” Ecological Monographs 13 (2): 203–57. https://doi.org/10.2307/1943528. Subbiah, Murugan, Mark A Caudell, Colette Mair, Margaret A Davis, Louise Matthews, Robert J Quinlan, Marsha B Quinlan, et al. 2020. “Antimicrobial Resistant Enteric Bacteria Are Widely Distributed amongst People, Animals and the Environment in Tanzania.” Nature Communications 11 (1): 1–12. https://doi.org/10.1038/s41467- 019-13995-5. Summers, Anne O. 2006. “Genetic Linkage and Horizontal Gene Transfer , the Roots of the Antibiotic Multi-Resistance Problem.” Animal Biotechnology 17 (2): 125–35. Surette, Matthew D, and Gerard D Wright. 2017a. “Lessons from the Environmental Antibiotic Resistome.” Annual Review of Microbiology 71: 309–29. https://doi.org/10.1146/annurev-micro-090816. Swift, Benjamin M C, Malcolm Bennett, Katie Waller, Christine Dodd, Annie Murray, Rachel L Gomes, Bethan Humphreys, et al. 2019. “Anthropogenic Environmental Drivers of Antimicrobial Resistance in Wildlife.” Science of the Total Environment 649: 12–20. https://doi.org/10.1016/j.scitotenv.2018.08.180. Tacão, Marta, António Correia, and Isabel Henriques. 2012. “Resistance to Broad- Spectrum Antibiotics in Aquatic Systems: Anthropogenic Activities Modulate the Dissemination of blaCTX-M-Like Genes.” https://doi.org/10.1128/AEM.00359-12. Tausova, D., M. Dolejska, A. Cizek, L. Hanusova, J. Hrusakova, O. Svoboda, G. Camlik,

120

and I. Literak. 2012. “Escherichia coli with Extended-Spectrum-Lactamase and Plasmid-Mediated Quinolone Resistance Genes in Great Cormorants and Mallards in Central Europe.” Journal of Antimicrobial Chemotherapy 67 (5): 1103–7. https://doi.org/10.1093/jac/dks017. Thi Hoang, Diep, Olga Chernomor, Arndt von Haeseler, Bui Quang Minh, Le Sy Vinh, and Michael S Rosenberg. 2017. “UFBoot2: Improving the Ultrafast Bootstrap Approximation.” Mol. Biol. Evol 35 (2): 518–22. https://doi.org/10.5281/zenodo.854445. Tormoehlen, Kristin, Yvette J Johnson-Walker, Emily W Lankau, Maung San Myint, and John A Herrmann. 2019. “Considerations for Studying Transmission of Antimicrobial Resistant Enteric Bacteria between Wild Birds and the Environment on Intensive Dairy and Beef Cattle Operations.” PeerJ 7: e6460. https://doi.org/10.7717/peerj.6460. Torres, Rita Tinoco, Joana Fernandes, João Carvalho, Mónica V. Cunha, Tânia Caetano, Sónia Mendo, Emmanuel Serrano, and Carlos Fonseca. 2020. “Wild Boar as a Reservoir of Antimicrobial Resistance.” Science of the Total Environment. Elsevier B.V. https://doi.org/10.1016/j.scitotenv.2019.135001. Trifinopoulos, Jana, Lam-Tung Nguyen, Arndt von Haeseler, and Bui Quang Minh. 2016. “W-IQ-TREE: A Fast Online Phylogenetic Tool for Maximum Likelihood Analysis.” Nucleic Acids Research 44. https://doi.org/10.1093/nar/gkw256. Vanderwaal, Kimberly L, Edward R Atwill, Lynne A Isbell, and Brenda Mccowan. 2014. “Quantifying Microbe Transmission Networks for Wild and Domestic Ungulates in Kenya.” Biological Conservation 169: 136–46. https://doi.org/10.1016/j.biocon.2013.11.008. Veldman, Kees, Peter Van Tulden, Arie Kant, Joop Testerink, and Dik Mevius. 2013. “Characteristics of Cefotaxime-Resistant Escherichia coli from Wild Birds in The Netherlands.” https://doi.org/10.1128/AEM.01880-13. Very, K. J., M. K. Kirchner, N. Shariat, W. Cottrell, C. H. Sandt, E. G. Dudley, S. Kariyawasam, and B. M. Jayarao. 2016. “Prevalence and Spatial Distribution of Salmonella Infections in the Pennsylvania Raccoon (Procyon lotor).” Zoonoses and Public Health 63 (3): 223–33. https://doi.org/10.1111/zph.12222. Vila, Jordi. 2015. “Multidrug-Resistant Bacteria Without Borders: Role of International Trips in the Spread of Multidrug-Resistant Bacteria.” Journal of Travel Medicine 22 (5): 289–91. https://doi.org/10.1111/jtm.12231. Viswanathan, M., D. L. Pearl, E. N. Taboada, E. J. Parmley, S. Mutschall, and C. M. Jardine. 2017. “Molecular and Statistical Analysis of Campylobacter Spp. and Antimicrobial-Resistant Campylobacter Carriage in Wildlife and Livestock from Ontario Farms.” Zoonoses and Public Health 64 (3): 194–203. https://doi.org/10.1111/zph.12295. Vittecoq, Marion, Sylvain Godreuil, Franck Prugnolle, Patrick Durand, Lionel Brazier, Nicolas Renaud, Audrey Arnal, et al. 2016. “Antimicrobial Resistance in Wildlife.”

121

Journal of Applied Ecology 53 (2): 519–29. https://doi.org/10.1111/1365- 2664.12596. Wang, Jing, Zhen-Bao Ma, Zhen-Ling Zeng, Xue-Wen Yang, Ying Huang, and Jian-Hua Liu. 2017. “The Role of Wildlife (Wild Birds) in the Global Transmission of Antimicrobial Resistance Genes.” Science Press Zoological Research 38 (2): 55–80. https://doi.org/10.24272/j.issn.2095-8137.2017.003. Wedley, Amy L., Susan Dawson, Thomas W. Maddox, Karen P. Coyne, Gina L. Pinchbeck, Peter Clegg, Tim Nuttall, Miranda Kirchner, and Nicola J. Williams. 2017. “Carriage of Antimicrobial Resistant Escherichia coli in Dogs: Prevalence, Associated Risk Factors and Molecular Characteristics.” Veterinary Microbiology 199: 23–30. https://doi.org/10.1016/j.vetmic.2016.11.017. Wellington, Elizabeth M H, Alistair B A Boxall, Paul Cross, Edward J. Feil, William H. Gaze, Peter M. Hawkey, Ashley S. Johnson-Rollings, et al. 2013. “The Role of the Natural Environment in the Emergence of Antibiotic Resistance in Gram-Negative Bacteria.” The Lancet Infectious Diseases 13 (2): 155–65. https://doi.org/10.1016/S1473-3099(12)70317-1. Wieler, Lothar H., Christa Ewers, Sebastian Guenther, Birgit Walther, and Antina Lübke- Becker. 2011. “Methicillin-Resistant Staphylococci (MRS) and Extended-Spectrum Beta-Lactamases (ESBL)-Producing Enterobacteriaceae in Companion Animals: Nosocomial Infections as One Reason for the Rising Prevalence of These Potential Zoonotic Pathogens in Clinical Samples.” International Journal of Medical Microbiology 301 (8): 635–41. https://doi.org/10.1016/j.ijmm.2011.09.009. Williams, N.J., C. Sherlock, T.R. Jones, H.E. Clough, S.E. Telfer, M. Begon, N. French, C.A. Hart, and M. Bennett. 2011. “The Prevalence of Antimicrobial-Resistant Escherichia coli in Sympatric Wild Rodents Varies by Season and Host.” Journal of Applied Microbiology 110 (4): 962–70. https://doi.org/10.1111/j.1365- 2672.2011.04952.x. Wirth, Thierry, Daniel Falush, Ruiting Lan, Frances Colles, Patience Mensa, Lothar H. Wieler, Helge Karch, et al. 2006. “Sex and Virulence in Escherichia coli: An Evolutionary Perspective.” Molecular Microbiology 60 (5): 1136–51. https://doi.org/10.1111/j.1365-2958.2006.05172.x. Woerther, Paul-Louis, Charles Burdet, Elisabeth Chachaty, and Antoine Andremont. 2013. “Trends in Human Fecal Carriage of Extended-Spectrum-Lactamases in the Community: Toward the Globalization of CTX-M.” https://doi.org/10.1128/CMR.00023-13. Wood, Derrick E, and Steven L Salzberg. 2014. “Kraken: Ultrafast Metagenomic Sequence Classification Using Exact Alignments.” Genome Biology 15 (3): R46. https://doi.org/10.1186/gb-2014-15-3-r46. Woodford, Neil, Jane F. Turton, and David M. Livermore. 2011. “Multiresistant Gram- Negative Bacteria: The Role of High-Risk Clones in the Dissemination of Antibiotic Resistance.” FEMS Microbiology Reviews. Oxford Academic.

122

https://doi.org/10.1111/j.1574-6976.2011.00268.x. Woolhouse, Mark, Melissa Ward, Bram Van Bunnik, Jeremy Farrar, M Ward, and Mark Woolhouse. 2015. “Antimicrobial Resistance in Humans , Livestock and the Wider Environment.” Philosophical Transactions of the Royal Society B: Biological Sciences 370 (1670): 20140083. Worsley‐Tonks, Katherine E. L., Elizabeth A. Miller, Stanley D. Gehrt, Shane C. McKenzie, Dominic A. Travis, Timothy J. Johnson, and Meggan E. Craft. 2020. “Characterization of Antimicrobial Resistance Genes in Enterobacteriaceae Carried by Suburban Mesocarnivores and Locally Owned and Stray Dogs.” Zoonoses and Public Health 67 (4): 460–66. https://doi.org/10.1111/zph.12691. Wright, P., and C. F. Mason. 1999. “Spatial and Seasonal Variation in Heavy Metals in the Sediments and Biota of Two Adjacent Estuaries, the Orwell and the Stour, in Eastern England.” Science of the Total Environment 226 (2–3): 139–56. https://doi.org/10.1016/S0048-9697(98)00383-0. Xi, Chuanwu, Yongli Zhang, Carl F. Marrs, Wen Ye, Carl Simon, Betsy Foxman, and Jerome Nriagu. 2009. “Prevalence of Antibiotic Resistance in Drinking Water Treatment and Distribution Systems.” Applied and Environmental Microbiology 75 (17): 5714–18. https://doi.org/10.1128/AEM.00382-09. Yan, Zhen Zhen, Qing Lin Chen, Yu Jing Zhang, Ji Zheng He, and Hang Wei Hu. 2019. “Antibiotic Resistance in Urban Green Spaces Mirrors the Pattern of Industrial Distribution.” Environment International 132 (June): 105106. https://doi.org/10.1016/j.envint.2019.105106. Zaheer, Rahat, Noelle Noyes, Rodrigo Ortega Polo, Shaun R. Cook, Eric Marinier, Gary Van Domselaar, Keith E. Belk, Paul S. Morley, and Tim A. McAllister. 2018. “Impact of Sequencing Depth on the Characterization of the Microbiome and Resistome.” Scientific Reports 8 (1): 5890. https://doi.org/10.1038/s41598-018- 24280-8. Zankari, Ea, Henrik Hasman, Salvatore Cosentino, Martin Vestergaard, Simon Rasmussen, Ole Lund, Frank M Aarestrup, and Mette Voldby Larsen. 2012. “Identification of Acquired Antimicrobial Resistance Genes.” Journal of Antimicrobial Chemotherapy 67: 2640–44. https://doi.org/10.1093/jac/dks261. Zhang, Xu-Xiang, Tong Zhang, and Herbert H P Fang. 2009. “Antibiotic Resistant Genes in Water Environment.” Article in Applied Microbiology and Biotechnology 82 (3): 397–414. https://doi.org/10.1007/s00253-008-1829-z. Zhou, Zhemin, Nabil Fareed Alikhan, Martin J. Sergeant, Nina Luhmann, Cátia Vaz, Alexandre P. Francisco, João André Carriço, and Mark Achtman. 2018. “Grapetree: Visualization of Core Genomic Relationships among 100,000 Bacterial Pathogens.” Genome Research 28 (9): 1395–1404. https://doi.org/10.1101/gr.232397.117. Zurfluh, Katrin, Melinda Glier, Herbert Hächler, and Roger Stephan. 2015. “Replicon Typing of Plasmids Carrying blaCTX-M-15 among Enterobacteriaceae Isolated at the Environment, Livestock and Human Interface.” Science of the Total Environment

123

521–522 (July): 75–78. https://doi.org/10.1016/j.scitotenv.2015.03.079.

124

Appendix A.

Table 10. Number of raccoons for which ESC-R E. coli were isolated on different capture events. Eighteen raccoons were captured more than once, 17 were captured twice and one three times. For the raccoon sampled three times, at least one ESC-R E. coli was isolated on all three occasions. Of the six raccoons that shed at least once ESC E. coli on both occasions, only one shed the same ST both times (ST162). ESC-R E. coli isolated ESC-R E. coli isolated Number of on 1st capture event on 2nd capture event raccoons - - 5 + + 7 + - 5 - + 1 + indicates that ESC-R E. coli were isolated and – indicates that ESC-R E. coli were not isolated.

Table 11. National Center for Biotechnology Information (NCBI) accession number of each ESC-R E. coli isolate collected from the feces of raccoons sampled in suburban and urban Chicago, Illinois, USA. Isolate ID Accession number 17 SAMN16077032 21 SAMN16077033 25 SAMN16077034 29 SAMN16077035 33 SAMN16077036 37 SAMN16077037 41 SAMN16077038 45 SAMN16077039 49 SAMN16077040 53 SAMN16077041 57 SAMN16077042 61 SAMN16077043 73 SAMN16077044 77 SAMN16077045 81 SAMN16077046 85 SAMN16077047 89 SAMN16077048 93 SAMN16077049

125

97 SAMN16077050 101 SAMN16077051 105 SAMN16077052 109 SAMN16077053 113 SAMN16077054 117 SAMN16077055 121 SAMN16077056 125 SAMN16077057 129 SAMN16077058 133 SAMN16077059 137 SAMN16077060 141 SAMN16077061 145 SAMN16077062 196 SAMN16077063 200 SAMN16077064 204 SAMN16077065 208 SAMN16077066 216 SAMN16077067 220 SAMN16077068 224 SAMN16077069 228 SAMN16077070 232 SAMN16077071 236 SAMN16077072 240 SAMN16077073 244 SAMN16077074 248 SAMN16077075 252 SAMN16077076 256 SAMN16077077 260 SAMN16077078 265 SAMN16077079 273 SAMN16077080 277 SAMN16077081 281 SAMN16077082 289 SAMN16077083 293 SAMN16077084 297 SAMN16077085 301 SAMN16077086 309 SAMN16077087 313 SAMN16077088 317 SAMN16077089 321 SAMN16077090

126

325 SAMN16077091 329 SAMN16077092 337 SAMN16077093 341 SAMN16077094 345 SAMN16077095 353 SAMN16077096 357 SAMN16077097 361 SAMN16077098 365 SAMN16077099 369 SAMN16077100 377 SAMN16077101 381 SAMN16077102 405 SAMN16077103 409 SAMN16077104 413 SAMN16077105 417 SAMN16077106 421 SAMN16077107 425 SAMN16077108 429 SAMN16077109 433 SAMN16077110 437 SAMN16077111 443 SAMN16077112 451 SAMN16077113 455 SAMN16077114 459 SAMN16077115 463 SAMN16077116 467 SAMN16077117 471 SAMN16077118 475 SAMN16077119 479 SAMN16077120 483 SAMN16077121 487 SAMN16077122 492 SAMN16077123 512 SAMN16077124 519 SAMN16077125 523 SAMN16077126 527 SAMN16077127 531 SAMN16077128 535 SAMN16077129 539 SAMN16077130 543 SAMN16077131

127

547 SAMN16077132 551 SAMN16077133 555 SAMN16077134 556 SAMN16077135 560 SAMN16077136 564 SAMN16077137 568 SAMN16077138 572 SAMN16077139 576 SAMN16077140 580 SAMN16077141 584 SAMN16077142 588 SAMN16077143 592 SAMN16077144 649 SAMN16077145 653 SAMN16077146 657 SAMN16077147 661 SAMN16077148 664 SAMN16077149 665 SAMN16077150 670 SAMN16077151 674 SAMN16077152 678 SAMN16077153 682 SAMN16077154

Table 12. Top ranked binomial generalized linear mixed models (AICc < 2 from the best fit model) for isolating at least one ESC-R E. coli from raccoons (n = 211, but 230 with recaptures). 2 2 Model Predictors included in each model k AICc wi r m r c 1 season + urban context 6 0.00 0.37 0.37 0.55 2 season + urban context + age 7 1.54 0.17 0.37 0.55 3 season + urban context + WWTP 7 1.83 0.15 0.39 0.54 4 season + urban context + sex 7 1.91 0.14 0.37 0.54 k represents the number of estimated coefficients in each model, ΔAICc the difference in 2 AICc between a given model and the top model, wi the renormalized Akaike weights, r m 2 and r m the marginal and conditional coefficients of determination.

128

Table 13. Top ranked Poisson generalized linear mixed models (AICc < 2 from the best fit model) predicting antimicrobial resistance gene richness of ESC E. coli of raccoons (n = 123). 2 2 Model Predictors included in each model k AICc wi r m r c 1 intercept 2 0.00 0.47 0 0.44 2 WWTP 3 1.71 0.2 0.003 0.44 k represents the number of estimated coefficients in each model, ΔAICc the difference in 2 AICc between a given model and the top model, wi the renormalized Akaike weights, r the coefficients of determination.

Table 14. Top ranked binomial generalized linear models (ΔAICc < 2 from the best fit model) predicting isolation of at least one ESC E. coli carrying plasmid-associated blaCTX-M or blaCMY from raccoons (n = 62). 2 Model Predictors included in each model k AICc wi r 1 WWTP 2 0.00 0.44 0.11.5 2 WWTP + season 5 1.47 0.21 0.23 k represents the number of estimated coefficients in each model, ΔAICc the difference in 2 AICc between a given model and the top model, wi the renormalized Akaike weights, r the coefficients of determination

129

Appendix B.

Table 15. Questions for dog owners. Questions for dog owners What is the age, sex, and breed of your dog? Has your dog been on antibiotics in the last 12-months? How frequently do you take your dog to this park? What is your home zip code?

Table 16. Fecal presence of ESC-R E. coli in domestic dogs in relation to dog age, sex, antibiotic use, and based raccoon spatial overlap. Analysis Model Predictor OR 95% CI p univariable 1 sex (male) 2.13 0.85 – 5.38 0.11 2 forest preserve vs. dog 1.98 0.86 – 4.54 0.11 park (dog park) 3 dog antibiotics use (yes) 2.0 0.83 – 4.84 0.12 multivariable 1 dog age 0.88 0.76 – 1.02 0.09 zip code overlap (yes) 0.8 0.34 – 1.87 0.61 2 dog age 0.89 0.76 – 1.03 0.11 raccoon site visit ( > 3 1.18 0.5 – 2.78 0.71 times a week) OR represents the odds ratio, 95% CI the 95% confidence intervals, and p the p-value. Significant terms are depicted in bold (with 95% CI not overlapping with 1 and p < 0.05).

130

Table 17. Univariable Poisson generalized linear mixed models core-SNP difference among pairs of ESC-R E. coli isolates recovered from raccoons and domestic dogs. Model Predictor variable Estimate SE Z p 1 raccoon residential use (yes) 0.05 0.04 1.27 0.2 2 dog home zip code overlap -0.008 0.11 -0.07 0.94 (yes) 3 dog presence (yes) 0.01 0.04 0.27 0.79 4 raccoon site visit (>3 times a -0.02 0.12 -0.17 0.03 week) Estimate represents the log-mean count, SE the standard errors, Z the z-value, and p the p-value. Significant terms are depicted in bold (p < 0.05).

Table 18. National Center for Biotechnology Information (NCBI) accession number of each ESC-R E. coli isolate collected from the feces of raccoons sampled in suburban and urban Chicago, Illinois, USA. Isolate ID Accession number Animal species 17 SAMN16077032 Raccoon 21 SAMN16077033 Raccoon 25 SAMN16077034 Raccoon 29 SAMN16077035 Raccoon 33 SAMN16077036 Raccoon 37 SAMN16077037 Raccoon 41 SAMN16077038 Raccoon 45 SAMN16077039 Raccoon 49 SAMN16077040 Raccoon 53 SAMN16077041 Raccoon 57 SAMN16077042 Raccoon 61 SAMN16077043 Raccoon 73 SAMN16077044 Raccoon 77 SAMN16077045 Raccoon 81 SAMN16077046 Raccoon 85 SAMN16077047 Raccoon

131

89 SAMN16077048 Raccoon 93 SAMN16077049 Raccoon 97 SAMN16077050 Raccoon 101 SAMN16077051 Raccoon 105 SAMN16077052 Raccoon 109 SAMN16077053 Raccoon 113 SAMN16077054 Raccoon 117 SAMN16077055 Raccoon 121 SAMN16077056 Raccoon 125 SAMN16077057 Raccoon 129 SAMN16077058 Raccoon 133 SAMN16077059 Raccoon 137 SAMN16077060 Raccoon 141 SAMN16077061 Raccoon 145 SAMN16077062 Raccoon 196 SAMN16077063 Raccoon 200 SAMN16077064 Raccoon 204 SAMN16077065 Raccoon 208 SAMN16077066 Raccoon 216 SAMN16077067 Raccoon 220 SAMN16077068 Raccoon 224 SAMN16077069 Raccoon 228 SAMN16077070 Raccoon 232 SAMN16077071 Raccoon 236 SAMN16077072 Raccoon 240 SAMN16077073 Raccoon 244 SAMN16077074 Raccoon 248 SAMN16077075 Raccoon 252 SAMN16077076 Raccoon

132

256 SAMN16077077 Raccoon 260 SAMN16077078 Raccoon 265 SAMN16077079 Raccoon 273 SAMN16077080 Raccoon 277 SAMN16077081 Raccoon 281 SAMN16077082 Raccoon 289 SAMN16077083 Raccoon 293 SAMN16077084 Raccoon 297 SAMN16077085 Raccoon 301 SAMN16077086 Raccoon 309 SAMN16077087 Raccoon 313 SAMN16077088 Raccoon 317 SAMN16077089 Raccoon 321 SAMN16077090 Raccoon 325 SAMN16077091 Raccoon 329 SAMN16077092 Raccoon 337 SAMN16077093 Raccoon 341 SAMN16077094 Raccoon 345 SAMN16077095 Raccoon 353 SAMN16077096 Raccoon 357 SAMN16077097 Raccoon 361 SAMN16077098 Raccoon 365 SAMN16077099 Raccoon 369 SAMN16077100 Raccoon 377 SAMN16077101 Raccoon 381 SAMN16077102 Raccoon 405 SAMN16077103 Raccoon 409 SAMN16077104 Raccoon 413 SAMN16077105 Raccoon

133

417 SAMN16077106 Raccoon 421 SAMN16077107 Raccoon 425 SAMN16077108 Raccoon 429 SAMN16077109 Raccoon 433 SAMN16077110 Raccoon 437 SAMN16077111 Raccoon 443 SAMN16077112 Raccoon 451 SAMN16077113 Raccoon 455 SAMN16077114 Raccoon 459 SAMN16077115 Raccoon 463 SAMN16077116 Raccoon 467 SAMN16077117 Raccoon 471 SAMN16077118 Raccoon 475 SAMN16077119 Raccoon 479 SAMN16077120 Raccoon 483 SAMN16077121 Raccoon 487 SAMN16077122 Raccoon 492 SAMN16077123 Raccoon 512 SAMN16077124 Raccoon 519 SAMN16077125 Raccoon 523 SAMN16077126 Raccoon 527 SAMN16077127 Raccoon 531 SAMN16077128 Raccoon 535 SAMN16077129 Raccoon 539 SAMN16077130 Raccoon 543 SAMN16077131 Raccoon 547 SAMN16077132 Raccoon 551 SAMN16077133 Raccoon 555 SAMN16077134 Raccoon

134

556 SAMN16077135 Raccoon 560 SAMN16077136 Raccoon 564 SAMN16077137 Raccoon 568 SAMN16077138 Raccoon 572 SAMN16077139 Raccoon 576 SAMN16077140 Raccoon 580 SAMN16077141 Raccoon 584 SAMN16077142 Raccoon 588 SAMN16077143 Raccoon 592 SAMN16077144 Raccoon 649 SAMN16077145 Raccoon 653 SAMN16077146 Raccoon 657 SAMN16077147 Raccoon 661 SAMN16077148 Raccoon 664 SAMN16077149 Raccoon 665 SAMN16077150 Raccoon 670 SAMN16077151 Raccoon 674 SAMN16077152 Raccoon 678 SAMN16077153 Raccoon 682 SAMN16077154 Raccoon 1 SAMN16533493 Dog

13 SAMN16533494 Dog 151 SAMN16533495 Dog 155 SAMN16533496 Dog 159 SAMN16533497 Dog 163 SAMN16533498 Dog 167 SAMN16533499 Dog 171 SAMN16533500 Dog 175 SAMN16533501 Dog

135

179 SAMN16533502 Dog 183 SAMN16533503 Dog 187 SAMN16533504 Dog 191 SAMN16533505 Dog 264 SAMN16533506 Dog 269 SAMN16533507 Dog 373 SAMN16533508 Dog 385 SAMN16533509 Dog 389 SAMN16533510 Dog 393 SAMN16533511 Dog 496 SAMN16533512 Dog 5 SAMN16533513 Dog 504 SAMN16533514 Dog 508 SAMN16533515 Dog 596 SAMN16533516 Dog 690 SAMN16533517 Dog 693 SAMN16533518 Dog 696 SAMN16533519 Dog 699 SAMN16533520 Dog 9 SAMN16533521 Dog

136

Figure 14. Map of the Chicago metropolitan area depicting the sites where raccoons were sampled (red and yellow polygons), and the home zip codes of sampled dogs (pink polygons). For the raccoon sites, red polygons represent areas where dogs were not allowed to enter and yellow polygons sites where dogs were allowed to enter.

Supplementary Methods and Results Classifying sites as ones in which raccoons were predicted to use or not use residential areas

1) Methods

Because radio-telemetry data were not collected for raccoons sampled during this study, we used information from raccoons tracked in previous years to classify sample sites as ones in which raccoons were predicted to use or not use residential areas. While radio- telemetry information on raccoons sampled in previous years were gathered for most sites

(specifically Busse, CT, MMWF, PC; Table SM1), three of the seven sites did not have

137

historical telemetry data (i.e. Damen, DRCA, and Edgebrook). However, for two of the sites that had no telemetry data (i.e. Edgebrook and DRCA), a surrogate site was used

(i.e. Dan Ryan). While not in the same location (Fig. SM1), Dan Ryan was ecologically similar to Edgebrook and DRCA (Fig. SM2, SM3). The seventh site (i.e. Damen) had no telemetry data and no surrogate site, but was classified as a site in which raccoons were not predicted to use residential areas because the site and surrounding area were mostly composed of industrial buildings and highways (i.e. an area with no domestic dogs present).

For sites with telemetry data, relocations were obtained from each raccoon twice a week.

Only raccoons that were tracked for more than 6 months and had a minimum of 30 relocations were included (Table SM1). Raccoons with radio-telemetry data were classified as using or not using residential areas based on maximum distance traveled outside of the park. If maximum distance was equal to zero, the raccoon was classified as not using residential areas, and if maximum distance was greater than zero, the raccoon was classified as using residential areas. To test whether our classification of raccoon use residential areas (yes/no) varied by season, we ran logistic regressions for each site. The outcome variable was raccoon use of residential areas (yes/no) and the predictor variable was season (fall, spring, summer, winter). Raccoon use of residential areas did not significantly vary by season for any of the sites (p > 0.05).

To make inference for raccoons sampled as part of this study (and for which no telemetry data were collected), we obtained the median for all radio-collared raccoons at each site.

If median maximum distance was equal to zero, the site was classified as one in which raccoons are predicted to use residential areas. If median maximum distance was greater

138

than zero, the site was classified as one in which raccoons are predicted to not use residential areas.

Table 19. Description of the number of raccoons radio-collared at each site, along with the years of tracking, and maximum and minimum number of relocations obtained per animal. Site n surrogate n at Year of Range of included site surrogate tracking relocations from site site obtained per animal Busse 32 - 1995-2003 48-462 Crabtree 21 - 2009-2014 37-360 Damen 0 none 0 - - DRCA 0 Dan Ryan 12 2014-2017 32-226 Edgebrook 0 Dan Ryan 12 2014-2017 32-226 MMWF 30 - 1995-2003 33-499 PC 31 - 2009-2014 32-561

139

Figure 15. Sampling sites in the northwestern portion of the Chicago metropolitan area. Polygons and points depict sites. The three red polygons are sites where no dogs were present or could enter (i.e. CT, Damen, and DRCA). The four yellow polygons are sites where dogs were present or could enter (i.e. Busse, Edgebrook, MMWF, and PC). The blue point represents the surrogate site (i.e. Dan Ryan for Edgebrook and DRCA).

Figure 16. Landcover proportions at each sampling site (n = 7) and surrogate site (n = 1; Dan Ryan). Whether dogs were present or could enter each park is also represented.

140

Figure 17. Maps of the eight sites (seven sampling sites and one surrogate site). Landcovers represented included urban developed high (cyan), urban developed medium (grey), urban developed low (green), and all other landcovers (white).

Results

Median maximum distance results revealed that, on average, the furthest raccoons traveled outside of parks ranged from 0 km (in the case of Busse and MMWF; Table

SM1) and 0.28 km (in the case of Dan Ryan; Table SM1). Given these findings, Busse and MMWF (and Damen) were classified as sites in which raccoons do not use residential areas and CT, Edgebrook (Dan Ryan was the surrogate), DRCA (Dan Ryan was the surrogate), and PC as sites in which raccoons use residential areas.

141

Table 20. Median maximum distance raccoons travelled outside of each site. Site Median max distance traveled outside of site (km) Busse 0 Crabtree 0.78 Damen - SVNC (surrogate for DRCA) 0.18 Dan Ryan (surrogate for Edgebrook) 0.28 MMWF 0 PC 0.19

142

Appendix C.

Table 21. Summary of shotgun metagenomic sequencing data obtained from enriched pooled fecal samples and rectal swabs of mesocarnivores, and stray and owned dogs in suburban Chicago, Illinois, USA. Pooled Number of Accession Number of Number of Number of sample samples number raw paired trimmed trimmed pooled reads paired paired (total reads reads post (fecal, host swab)) filtering coyote 32 (18, 14) SRR10098752 69,693,772 58,683,355 54,470,428 opossum 22 (6, 16) SRR10098750 64,720,391 54,380,236 51,678,049 raccoon 31 (23, 8) SRR10098751 66,611,670 55,336,184 51,810,242 stray 18 (18, 0) SRR10098748 66,712,494 56,573,712 55,534,998 dog owned 13 (13, 0) SRR10098749 66,370,970 56,771,796 55,389,127 dog

Figure 18. Relative abundance of the nine most common bacterial genera found in the mesocarnivore, and stray and owned dog samples.

143

Figure 19. Correlation between number of individual samples in each pooled sample and antimicrobial resistance gene (ARG) richness. The grey shaded area represents 95% confidence limits around the regression line

144