Host Disease and Environmental Factors Associated with Zoonotic Pathogens in Urban Norway Rats (Rattus Norvegicus)

Home , Pasteurella dagmatis

Host Disease and Environmental Factors Associated with Zoonotic Pathogens in Urban Norway Rats (Rattus norvegicus)

by Jamie Lee Rothenburger

A Thesis presented to The University of Guelph

In partial fulfilment of requirements for the degree of Doctor of Philosophy in Pathobiology

Guelph, Ontario, Canada

ABSTRACT

Host Disease and Environmental Factors Associated with Zoonotic Pathogens in Urban Norway Rats (Rattus norvegicus)

Jamie Lee Rothenburger Advisor: University of Guelph, 2017 Dr. C. Jardine

The purpose of this research was to investigate the role of environmental and intra-host factors in the epidemiology and ecology of zoonotic pathogen carriage by urban Norway rats (Rattus norvegicus). Rats are the hosts of many zoonotic pathogens, including Yersinia pestis and Leptospira interrogans, the causative agents of plague and leptospirosis, respectively. Knowledge of the ecology and epidemiology of zoonotic pathogens in their rat hosts is important for understanding and mitigating the risk to people. Most studies of rat-associated zoonotic pathogens investigate rat demographic characteristics. But many factors across hierarchical levels of biological organization can influence pathogens in hosts. These include environmental and intra-host factors such as co-infections and disease.

Using samples and data collected during a year-long trap and removal study of rats set in Vancouver, British Columbia, I first assessed microenvironmental features, time-lagged weather variables and rat abundance for associations with four potentially zoonotic pathogens carried by rats (Bartonella tribocorum, Clostridium difficile, antimicrobial resistant Escherichia coli and methicillin-resistant Staphylococcus aureus). Significant factors included temperature, precipitation, specific land use and pavement condition. No pathogens were associated with rat abundance. These results may inform predictive modeling, targeted surveillance activities and specific interventions.

Next, I used pathological analyses to document the spectrum of macroscopic and microscopic disease found in these rats. The most severe and frequent lesions were infectious and

inflammatory. Finally, I assessed the most common lesions and parasite infections for associations with three zoonotic pathogens (B. tribocorum, C. difficile and L. interrogans). Parasite infections were associated with B. tribocorum, while C. difficile and L. interrogans were associated with specific lesions, and rats were rarely co-infected with multiple zoonotic pathogens.

The impact of the environment, weather, lesions and parasitic infections varied depending on the zoonotic pathogen. Collectively, these results suggest a possible dynamic interplay among these factors, adding to the growing knowledge of zoonotic pathogen ecology in rats. The disease ecology methods and concepts developed by this research are broadly applicable to the study of the epidemiology and ecology of zoonoses in other hosts and ecosystems.

iii DEDICATION

In memory of my mom, Valerie Rothenburger.

ACKNOWLEDGEMENTS First and foremost, I would like to thank my advisory committee for their help, support and guidance during these years. Special thanks to my advisor, Dr. Claire Jardine. I’m so grateful she was willing to take me on as a graduate student. She was a continual source of wisdom, support, intellectual challenge and mentorship. I also value learning from her leadership and management styles. It was such a pleasure to work with her these last three years and I will take many invaluable lessons she’s imparted to me into my future career. I am incredibly grateful to Dr. Chelsea Himsworth for starting me down this wild rat journey and inviting me to be a part of the Vancouver Rat Project. I am thankful for the opportunity to study under Dr. David Pearl, who challenged me to push the boundaries of my epidemiological knowledge. Finally, many thanks to Dr. Nicole Nemeth for her unwavering support and enthusiasm for the project, as well as her pathology boards solidarity.

My gratitude goes to the many people who made important contributions to this research endeavour. Many thanks to my fellow graduate students for their advice, wisdom, encouragement, statistical help and writing feedback: Christine, Katie, Diana, Kathryn, Jon, Nadine, Shannon and Kristen. It was great fun and I will always remember the important lesson of teamwork. Kirbee Parsons and Alice Feng were the “field crew,” Dr. Victoria Chang was the hard-working summer student and Dr. Heather Anhold volunteered during the rat sampling portion of the Vancouver Rat Project. Debra Rempel welcomed me into her home in Abbotsford for months at a time and generously assisted with labeling thousands of sample bags. There also were many individuals at the British Columbia Ministry of Agriculture Animal Health Centre who made these studies possible. Special thanks to Erin Zabek for bacterial culture and Sandra Etheridge and Joanne Taylor for preparing over 1000 histology slides. Also, much appreciation goes to Erin Zabek, Julie Bidulka, Michael Kosoy and Scott Weese for pathogen testing. The Canadian Institutes of Health Research funded the Vancouver Rat Project (MOP– 119530).

I would also like to gratefully acknowledge the financial support that allowed me to continue my studies: Natural Sciences and Engineering Research Council Alexander Graham Bell Canada Graduate Scholarship-Doctoral, Canadian Federation of University Women Dr. Margaret

McWilliams Pre-Doctoral Fellowship, Imperial Order Daughters of the Empire War Memorial Scholarship, Ontario Veterinary College Graduate Student Fellowship and the University of Guelph Dean’s Tri-Council Scholarship.

There are many people that have inspired my career and generously offered their wisdom and mentorship for which I am eternally grateful. First of all, thank-you to Dr. William Karesh for writing Appointment at the Ends of the World—this book was one of the main reasons I went to veterinary school and it certainly started me down the wildlife track. Dr. Nick Nation, the first veterinary pathologist I ever met, encouraged and inspired me throughout my graduate studies. I thank Dr. Piper Treuting for her collaborations, contagious enthusiasm and prompt answers to random rat pathology questions. Finally, many thanks to Dr. Ted Leighton for supervising my Master’s program, encouraging me to further my studies, reminding me of the big picture and prompting me to aspire for scientific greatness.

My family has been an incredible source of support and balance during these challenging years. Thank-you so much for everything you have done to may this all possible. Special credit goes to my husband, Todd, for his willingness to move to Guelph. His love, patience, steadiness in the face of adversity, voice of reason, Excel skills and task mastering were all invaluable to me during my studies. I’m especially grateful for my dad, Tex, for his long-term enthusiasm and encouragement during my continuing educational efforts and quest for knowledge. He also deserves credit for sparking my love of nature and wildlife. My sister, Teri, has always been a tremendous friend, pragmatic advisor and source of extracurricular fun. I also thank the large, boisterous extended Rothenburger family for all that they have done for me. Since our first year together at the University of Alberta, Allison Raher and Jessica Thiessen have been my constant cheerleaders, sources of encouragement and grounding presences. Thank-you all a million times over.

“The complexity of things – the things within things – just seems to be endless. I mean nothing is easy, nothing is simple” –Alice Munro

STATEMENT OF WORK

Jamie L. Rothenburger was solely responsible for the preparation of this thesis.

Dr. Chelsea Himsworth conceived of and directed the Vancouver Rat Project. Kirbee Parsons and Alice Feng collected rats and environmental data during sampling portion of the Vancouver Rat Project. Dr. Victoria Chang did many rat autopsies, managed image storage and entered data. Dr. Heather Anholt volunteered with the project and assisted with rat autopsies and sample collection. At the British Columbia Ministry of Agriculture Animal Health Centre, Erin Zabek performed bacterial culture, and Sandra Etheridge and Joanne Taylor prepared histology slides. In addition, Erin Zabek, Julie Bidulka, Dr. Michael Kosoy (Bartonella & Rodent-Borne Diseases Laboratory, Centers for Disease Control and Prevention, Fort Collins, CO) and Dr. Scott Weese (Department of Pathobiology, Ontario Veterinary College) did zoonotic pathogen testing. Dr. Piper Treuting (University of Washington) confirmed the presence of poxvirus in two cutaneous lesions using electron microscopy. Dr. Treuting and Dr. Frederick (Ted) Leighton (University of Saskatchewan) consulted on the pathology analyses. Dr. Krista La Perle (The Ohio State University) consulted on the analysis of thyroid lesions and performed immunohistochemistry on samples of thyroid tissue. Kaylee Byers (University of British Columbia) identified ectoparasites. Jessica Thiessen (Ink Bird Studios, Edmonton, Alberta) illustrated Figure 1.1. Discussions with Kate Bishop-Williams informed the creation and analyses of time-lagged weather data.

vii TABLE OF CONTENTS

ACKNOWLEDGEMENTS ...... v

STATEMENT OF WORK ...... vii

TABLE OF CONTENTS ...... viii

LIST OF TABLES ...... xiii

LIST OF FIGURES ...... xv

LIST OF ABBREVIATIONS ...... xvii

1. INTRODUCTION & RESEARCH OBJECTIVES ...... 1 1.1 WILD ANIMALS AS A SOURCE OF HUMAN INFECTIOUS DISEASES ...... 1 1.2 BACKGROUND ON NORWAY RATS ...... 1 1.3 RAT-ASSOCIATED ZOONOTIC AND POTENTIALLY-ZOONOTIC PATHOGENS ...... 3 1.4 BACKGROUND ON THE VANCOUVER RAT PROJECT ...... 5 1.5 TOWARDS A BROADER UNDERSTANDING OF FACTORS INFLUENCING THE ECOLOGY OF RAT-ASSOCIATED ZOONOTIC PATHOGENS ...... 6 1.6 RESEARCH OBJECTIVES ...... 7 1.7 FIGURES ...... 8 1.8 REFERENCES ...... 9

2. ENVIRONMENTAL FACTORS AND ZOONOTIC PATHOGEN ECOLOGY IN URBAN EXPLOITER SPECIES* ...... 14 2.1 ABSTRACT ...... 14 2.2 INTRODUCTION ...... 15 2.3 MATERIALS AND METHODS ...... 16 2.4 PATHOGEN ECOLOGY VARIES AMONG LOCATIONS ...... 17 2.4.1 Varying Pathogen Prevalence By Location ...... 17 2.4.2 Varying Pathogen Genetic And Phenotypic Diversity By Location ...... 18 2.4.3 Strengths And Limitations Of Locational Studies ...... 19 2.5 PATHOGEN ECOLOGY VARIES AMONG HABITAT TYPES ...... 19 2.5.1 Comparing Urban Habitats to Other Habitat Types ...... 19 2.5.2 Comparing Different Habitat Types within Cities ...... 20

viii

2.5.3 Strengths and Limitations of Habitat Studies ...... 21 2.6 DISADVANTAGED URBAN AREAS ARE ASSOCIATED WITH INCREASED PATHOGEN PREVALENCE ...... 22 2.6.1 Strengths and Limitations of Studies in Disadvantaged Urban Areas ...... 23 2.7 SEASONALITY AND WEATHER EFFECTS ON PATHOGEN CHARACTERISTICS IN WILDLIFE HOSTS ...... 23 2.7.1 Strengths and Limitations of Seasonal and Weather Studies ...... 25 2.8 OTHER ENVIRONMENTAL CHARACTERISTICS ...... 25 2.9 DIRECTIONS FOR FUTURE RESEARCH ...... 26 2.9.1 Study Quality ...... 26 2.9.2 Impact of the Urban Environment on Pathogen Ecology ...... 27 2.10 CONCLUSIONS ...... 27 2.11 TABLES ...... 29 2.12 FIGURE ...... 31

3. BEYOND ABUNDANCE: HOW MICROENVIRONMENTAL FEATURES AND WEATHER INFLUENCE BARTONELLA TRIBOCORUM INFECTION IN WILD NORWAY RATS (RATTUS NORVEGICUS)* ...... 39 3.1 ABSTRACT ...... 39 3.2 INTRODUCTION ...... 41 3.3 MATERIALS AND METHODS ...... 43 3.3.1 Study Design ...... 43 3.3.2 Environmental and Weather Characteristics ...... 43 3.3.3 Statistical Analysis ...... 44 3.3.4 Variable Assessment and Construction ...... 45 3.3.5 Statistical Modeling ...... 45 3.4 RESULTS ...... 46 3.4.1 Environmental Variables ...... 46 3.4.2 Weather and Season Variables ...... 46 3.4.3 Rat Abundance ...... 47 3.5 DISCUSSION ...... 48 3.6 TABLES ...... 53 3.7 FIGURES ...... 59 3.8 REFERENCES ...... 62

4. ENVIRONMENTAL FACTORS ASSOCIATED WITH THE CARRIAGE OF BACTERIAL PATHOGENS IN NORWAY RATS* ...... 65 4.2 INTRODUCTION ...... 66 4.3 METHODS ...... 67 4.3.1 Study Design ...... 67 4.3.2 Microenvironmental and Weather Characteristics ...... 68 4.3.3 Statistical Analyses ...... 68 4.4 Results ...... 71 4.4.1 Clostridium difficile ...... 71 4.4.2 Antimicrobial Resistant Escherichia coli ...... 71 4.4.3 Methicillin-resistant Staphylococcus aureus ...... 72 4.4.4 Model Diagnostics ...... 72 4.5 DISCUSSION ...... 73 4.5.1 Clostridium difficile ...... 73 4.5.2 Antimicrobial resistant Escherichia coli ...... 74 4.5.3 Methicillin-resistant Staphylococcus aureus ...... 74 4.5.4 Conclusions ...... 75 4.6 TABLES ...... 78 4.7 REFERENCES ...... 86

5. PATHOLOGY IN WILD NORWAY RATS (RATTUS NORVEGICUS) ...... 90 5.1 ABSTRACT ...... 90 5.2 INTRODUCTION ...... 90 5.3 MATERIALS AND METHODS ...... 92 5.3.1 Rat Collection ...... 92 5.3.2 Autopsy and Sample Collection ...... 92 5.3.3 Pathology Analyses ...... 92 5.3.4 Lesion Categorization ...... 93 5.4 RESULTS ...... 93 5.4.1 Cardiovascular Lesions ...... 94 5.4.2 Digestive Tract Lesions ...... 94 5.4.3 Endocrine Lesions ...... 95 5.4.4 Hemolymphatic Lesions ...... 95 5.4.5 Integumentary Lesions ...... 95

5.4.6 Musculoskeletal, Adipose and Connective Tissue Lesions ...... 96 5.4.7 Respiratory Lesions ...... 96 4.8 Reproductive Lesions ...... 97 5.4.9 Urinary Lesions ...... 97 5.4.10 Other Lesions and General Conditions ...... 98 5.5 DISCUSSION ...... 99 5.5.1 Lesions Associated with Bacterial Infections ...... 100 5.5.2 Parasitic Lesions ...... 101 5.5.3 Idiopathic Inflammatory Lesions ...... 103 5.5.4 Traumatic and Anthropogenic Lesions ...... 104 5.5.5 Degenerative Conditions ...... 106 5.5.6 Neoplastic, Proliferative and Congenital Lesions ...... 107 5.6 CONCLUSIONS ...... 108 5.7 TABLES ...... 112 5.8 FIGURES ...... 124 5.9 SUPPLEMENTAL FIGURES ...... 127 5.9 REFERENCES ...... 131

6. THE DEVIL IS IN THE DETAILS—HOST DISEASE AND CO-INFECTIONS ASSOCIATED WITH ZOONOTIC PATHOGEN CARRIAGE IN NORWAY RATS (RATTUS NORVEGICUS) ...... 140 6.1 ABSTRACT ...... 140 6.2 INTRODUCTION ...... 142 6.2 METHODS ...... 144 6.2.1 Study Design ...... 144 6.2.2 Autopsy, Pathology and Zoonotic Pathogen Analyses ...... 144 6.2.3 Statistical Analyses ...... 145 6.3 RESULTS ...... 147 6.3.1 Bartonella tribocorum ...... 147 6.3.2 Clostridium difficile ...... 147 6.3.3 Leptospira interrogans ...... 148 6.3.4 Co-infections Among Zoonotic Pathogens ...... 148 6.3.5 Model Diagnostics ...... 149 6.4 DISCUSSION ...... 149

6.4.1 Bartonella tribocorum ...... 149 6.4.2 Clostridium difficile ...... 150 6.4.3 Leptospira interrogans ...... 151 6.4.4 Co-infections ...... 152 6.4.5 Implications ...... 153 6.5 CONCLUSIONS ...... 155 6.6 REFERENCES ...... 156 6.7 TABLES ...... 162 6.8 FIGURE ...... 174

7. GENERAL DISCUSSION ...... 175 7.1 SUMMARY OF MAIN RESULTS ...... 175 7.2 LIMITATIONS OF STUDY DESIGN ...... 176 7.3 RESEARCH IMPLICATIONS AND SIGNIFICANCE ...... 178 7.4 FUTURE RESEARCH DIRECTIONS ...... 179 7.5 FINAL THOUGHTS ...... 180 7.6 REFERENCES ...... 182

APPENDIX A ...... 183

APPENDIX B ...... 186

APPENDIX C ...... 196

xii

LIST OF TABLES

Table 2.1 Summary of emergent themes ...... 29 Table 2.2 Summary of future research directions ...... 30 Table 3.1 Characteristics and univariable associations of significant microenvironmental features among Norway rats (Rattus norvegicus) infected with Bartonella tribocorum in Vancouver, Canada...... 53 Table 3.2 Characteristics and univariable associations of significant season, abundance and weather variables among Norway rats (Rattus norvegicus) infected with Bartonella tribocorum in Vancouver, Canada...... 54 Table 3.3 Results from two-variable models assessing confounding based on causal diagrams among Norway rats (Rattus norvegicus) infected with Bartonella tribocorum in Vancouver, Canada. Adjusted values in bold indicated a confounding relationship (>30% change in coefficients when confounder was added to the model)...... 55 Table 4.1 Characteristics and results of univariable associations of microenvironmental, weather and abundance variables considered for a multivariable model among Norway rats (Rattus norvegicus) carrying Clostridium difficile in Vancouver, Canada...... 78 Table 4.2 Characteristics and univariable associations of microenvironmental, weather and abundance variables considered for a multivariable model among Norway rats (Rattus norvegicus) carrying antimicrobial-resistant Escherichia coli in Vancouver, Canada...... 80 Table 4.3 Characteristics and results of univariable associations of microenvironmental, weather and abundance variables considered for a multivariable model among Norway rats (Rattus norvegicus) carrying methicillin-resistant Staphylococcus aureus in Vancouver, Canada. . 82 Table 4.4 Results of multivariable mixed logistic regression models with random effect for block assessing associations among environmental and weather variables with Clostridium difficile, methicillin-resistant Staphylococcus aureus and antimicrobial resistant Escherichia coli carriage in Norway rats (Rattus norvegicus) captured in Vancouver Canada...... 84 Table 5.1 Description and prevalence of major histological lesions in a sample of wild urban Norway rats (Rattus norvegicus) trapped in Vancouver, Canada...... 112 Table 5.2 Neoplastic and proliferative lesions in a sample of wild urban Norway rats (Rattus norvegicus) trapped in Vancouver, Canada...... 115

xiii

Table 5.3 Rare non-proliferative lesions in a sample of wild urban Norway rats (Rattus norvegicus) trapped in Vancouver, Canada ...... 117 Table 5.4 Parasites and applicable lesions identified using histopathology in a sample of wild urban Norway rats (Rattus norvegicus) trapped in Vancouver, Canadaa ...... 121 Table 5.5 Bacteria isolated from macroscopic lesions in a sample of wild urban Norway rats (Rattus norvegicus) trapped in Vancouver, Canada ...... 123 Table 6.1 Descriptions of major macroscopic and microscopic lesions observed at >10% prevalence and demographic characteristics among wild Norway rats (Rattus norvegicus) from Vancouver, Canada...... 162 Table 6.2 Major macroscopic and microscopic lesions found at >10% prevalence and their univariable associations with Bartonella tribocorum infection in 390 wild Norway rats (Rattus norvegicus) from Vancouver, Canada...... 165 Table 6.3 Major macroscopic and histological lesions found at >10% prevalence and their univariable associations with Clostridium difficile carriage in 672 wild Norway rats (Rattus norvegicus) from Vancouver, Canada...... 167 Table 6.4 Major macroscopic and histological lesions found at >10% prevalence and their univariable associations with Leptospira interrogans carriage in 581 wild Norway rats (Rattus norvegicus) from Vancouver, Canada...... 169 Table 6.5 Results of multivariable mixed logistic regression model with random effect for block assessing associations among lesions and parasites with Leptospira interrogans carriage in 581 Norway rats (Rattus norvegicus) captured in Vancouver Canada.a ...... 171 Table 6.6 Patterns of infections with zoonotic pathogens among 331 Norway rats (Rattus norvegicus) from Vancouver, Canada...... 172 Table 6.7 Results of logistic regression model with random effect for block assessing associations among Bartonella tribocorum, Clostridium difficile and Leptospira interrogans carriage in Norway rats (Rattus norvegicus) captured in Vancouver Canada...... 173

xiv

LIST OF FIGURES Figure 1.1 Factors across hierarchical levels of biological organization may influence the carriage of zoonotic pathogens in rats. At a high level of the scale, populations of rats reside in the environment. Within individual rats at lower levels of the scale, organs are made up of tissues, which in turn consist of cells that contain molecules. Co-infecting pathogens inhabit cells and tissues within individuals. The environment and processes within rats (co- infections with macroparasites and microparasites, and disease at the cellular, tissue and organ levels) may have important influences on the ecology of zoonotic pathogens in urban ecosystems (inspired by Ezenwa et al. 2015; designed by J. Rothenburger; illustrated by Jessica Thiessen)...... 8 Figure 2.1 Examples of abiotic, biotic and anthropogenic factors that may influence zoonotic pathogens in urban wildlife ...... 31 Figure 3.1 Causal diagram of sexual maturity, environmental and weather factors potentially affecting Bartonella tribocorum infection status in Norway rats...... 59 Figure 3.2 Predicted probability curve demonstrating a quadratic relationship between Bartonella tribocorum infection status in Norway rats and mean precipitation in the 7 days prior to rat capture with 95% confidence intervals. This association was no longer significant after controlling for the confounding effect of season...... 60 Figure 3.3 Predicted probability curve demonstrating a quadratic relationship between Bartonella tribocorum infection status in Norway rats and relative rat abundance with 95% confidence intervals. This relationship was no longer significant after controlling for the confounding effect of the proportion of the block occupied by housing over commercial buildings, human loitering, season and mean minimum temperatures (°C) on days 84-90 prior to capture...... 61 Figure 5.1 Sample selection protocol for studying the pathology of urban Norway rats (Rattus norvegicus) from Vancouver, British Columbia, Canada. Note that only 341 rats with no gross lesions were assessed for microscopic lesions due to budgetary constraints that prohibited examination of all tissues from all rats...... 124 Figure 5.2-5.12 Microscopic lesions, Norway rats (Rattus norvegicus). Hematoxylin and Eosin (HE)...... 125

Figure 6.1 Causal diagram depicting the theoretical relationships among nematode parasite infections, lesions and rat demographic factors with zoonotic pathogen status as the outcome. Lesions and nematode parasites hypothetically influence zoonotic pathogen status via disease-associated behaviour changes and immune system modulation...... 174

xvi

LIST OF ABBREVIATIONS

AMR – Antimicrobial Resistant BLUPS – Best Linear Unbiased Predictors CAB – Commonwealth Agricultural Bureaux Direct Database CAR Bacillus – Cilia Associated Respiratory Bacillus CI – Confidence Interval CPN – Chronic Progressive Nephropathy DNA – Deoxyribonucleic Acid HE – Hematoxylin and Eosin HRFS – Hemorrhagic Fever with Renal Syndrome IBALT – Inducible Bronchus Associated Lymphoid Tissue IQR – Interquartile Range IUCN – International Union for Conservation of Nature JR – Jamie Rothenburger JSTOR – Journal Storage Database LCMV – Lymphocytic Choriomeningitis Virus MRSA – Methicillin-resistant Staphylococcus aureus NA – Not Applicable or Not Available OR – Odds Ratio PCR – Polymerase Chain Reaction RAZ – Rat Associated Zoonoses USA – United States of America VRP – Vancouver Rat Project WNV – West Nile Virus

xvii

CHAPTER 1

1. INTRODUCTION & RESEARCH OBJECTIVES

1.1 WILD ANIMALS AS A SOURCE OF HUMAN INFECTIOUS DISEASES Wildlife-associated zoonotic diseases (i.e., infections transmitted from animals to people) pose a current and growing threat to human health. Over 75% of emerging infectious diseases are zoonotic (Taylor et al. 2001). With most of the world’s human population living in cities (United Nations, Department of Economic and Social Affairs, Population Division 2015), zoonotic risks associated with urban wildlife are of increasing concern and deserve scientific scrutiny. The increased risk of prolonged and frequent contact between people and urban wildlife may exacerbate the risk of zoonotic disease transmission (Bradley and Altizer 2007). High densities of certain urban wildlife species may facilitate pathogen transmission among animal hosts and to people (Bradley and Altizer 2007). For example, a study in Asian–Pacific regions observed that emerging diseases were 15 times more likely to arise from wildlife that are ecologically associated with people, including rats and mice, compared to those that were not (McFarlane et al. 2012). Areas within cities where people provide supplemental food resources, either deliberately or not, may also be locations of particularly high risk (Bradley and Altizer 2007).

1.2 BACKGROUND ON NORWAY RATS Norway rats (Rattus norvegicus; hereafter, “rats”) are among the most common urban wildlife species, and are especially adept at living in close association with people. They are taxonomically classified as rodents, an order comprised of roughly 2,255 known species and that represents over 40% of all mammals (IUCN 2017).

Rats are omnivores, although they show preference for specific types of food including seeds, cereal grains, meat and household waste (Brooks and Jackson 1973). They are cannibalistic, particularly when food resources are scarce, which may be an important mechanism for pathogen transmission (Calhoun 1963; Carr et al. 1979; Rothenburger et al. 2014). 1

Like other rodents, rats are notorious for their tremendous reproductive capacity. Individuals reach sexual maturity at three months of age (Feng and Himsworth 2014). Gestation is three weeks and litter size can range from four to eleven (Davis 1953; Feng and Himsworth 2014). Females can produce up to five litters per year (Davis 1953). Altogether, a mature female can contribute 35-55 offspring per year to the population (Davis 1953).

Rats live in colonies and have relatively small home ranges. These tend to be less than 18 m in diameter and typically encompass a city block with very limited movement across city streets (Davis 1953; Feng and Himsworth 2014; Montes de Oca et al. 2017). Social hierarchy within colonies is important. When colonies are disrupted by events like human interventions, habitat change and/or natural phenomena, aggressive interactions among rats tend to increase (Blanchard et al. 1985; Feng and Himsworth 2014). This may also be an important factor that facilitates disease transmission through increased fighting and animal-to-animal contact.

From their suspected origins in Asian steppes, Norway rats invaded every continent on the planet except Antarctica. Global maritime shipping was the primary mechanism for this gargantuan range expansion (Brooks and Jackson 1973; Feng and Himsworth 2014). In North America, rats spread from sites of coastal introduction throughout the interior. The province of Alberta, Canada is an exception to this spread—it remains free of rats through a vigorous extermination policy (Brooks and Jackson 1973; Government of Alberta 2017).

This extensive invasion was also facilitated by rats’ ability to live in close proximity to humans (Feng and Himsworth 2014). According to the classification system of McKinney (2006), rats are an urban exploiter species: species that thrive in built environments near urban cores and subsist on food and habitat associated with human presence rather than naturally-occurring resources. Using a more recent urban wildlife classification system, urban rats are prototypical urban dwellers (Fischer et al. 2015). Rats create their own burrow habitats in human-modified environments and built structures, making them especially suited to urban environments (Brooks and Jackson 1973; Feng and Himsworth 2014).

There are many negative consequences that arise from the close association between rats and people. Throughout the world, rats damage infrastructure and provoke social stigma (Wyman 1910; Feng and Himsworth 2014). In addition, rats spoil food, ravage crops and destroy agricultural infrastructure, contributing to food insecurity and annual food losses sufficient to feed millions of people (Wyman 1910; Singleton et al. 2007; Meerburg et al. 2009b). Humans have introduced rats into many ecologically sensitive habitats, where they represent a major threat to species of conservation concern (Towns et al. 2006). Rats also harbour many zoonotic and potentially zoonotic pathogens (Himsworth et al. 2013b).

1.3 RAT-ASSOCIATED ZOONOTIC AND POTENTIALLY-ZOONOTIC PATHOGENS Rats carry a range of viral, bacterial and parasitic zoonotic pathogens that can cause severe illnesses in people (Meerburg et al. 2009a; Himsworth et al. 2013b). Perhaps most notoriously, rats were implicated in outbreaks of highly fatal bubonic plague—the so-called Black Death in 14th century Europe killed approximately one third of the continent’s human population (Gage and Kosoy 2005).

People are exposed to rat-associated zoonoses (RAZ) through a variety of ways: bites (e.g., Streptobacillus moniliformis, the causative agent of rat bite fever), inhaled aerosols (e.g., Seoul hantavirus, the causative agent of hemorrhagic fever with renal syndrome) and urine- contaminated water (e.g., Leptospira spp., the causative agents of leptospirosis and Weil’s Disease; Meerburg et al. 2009a; Himsworth et al. 2013b). In addition to Y. pestis, fleas transmit several other rat-associated zoonotic pathogens (e.g., Bartonella spp. and Rickettsia typhi; the causative agents of bartonellosis and murine typhus, respectively; Himsworth et al. 2013b). There is also indirect transmission through contaminated food. For instance, the rat lungworm, Angiostrongylus cantonensis, is transmitted to people via consumption of infected intermediate hosts (mollusks) or mucus-contaminated produce; infected people may develop potentially fatal eosinophilic meningitis (Himsworth et al. 2013b).

Since rats are present in cities worldwide, RAZ are an important global public health issue. The burden is especially severe in developing countries and their urban slums (Himsworth et al.

2013b; Costa et al. 2017). These areas tend to experience heavy rat infestations with limited means for active rat control and extermination. Inadequate, informal or non-existent housing and sanitary infrastructure (i.e., sewer and drinking water) contribute to rat infestations, as well as increase the risk of pathogen transmission (Costa et al. 2017). Public health resources may be limited and/or prioritized for other concerns. Additionally, many of these countries face concurrent tropical diseases such as dengue fever and malaria. These diseases present with the non-specific illnesses that also characterize many RAZ, leading to the possibility of under- recognition and treatment (Himsworth et al. 2013b). Nonetheless, the impact of these pathogens is profound. An estimate for the annual global burden of leptospirosis is over one million cases and approximately 60,000 deaths, although the exact number directly attributable to rat vs. other mammalian reservoirs is unknown (Costa et al. 2015).

Even in resource-rich nations, rats may pose a significant public health threat. Disadvantaged neighbourhoods suffer not only from increased infestation burden, but also increased risk of RAZ (Clinton 1969; Himsworth et al. 2013b; Feng and Himsworth 2014; Johnson et al. 2016; Leibler et al. 2016; Rothenburger et al. 2017). Infections from rat exposures are a serious health concern among marginalized urban human populations (e.g., people who are impoverished, homeless or addicted to alcohol and/or drugs; Leibler et al. 2016). Similar to developing countries, the true burden of RAZ in developed countries is likely underestimated due to misdiagnosis and under recognition (Himsworth et al. 2013b; Leibler et al. 2016). Adding to global public health concerns is the recognition that rats carry a suite of previously-unknown pathogens that may cross into people (Firth et al. 2014). There are also increased incidences of known pathogens, like A. cantonensis, that are expanding in their geographical distribution (York et al. 2015; Stockdale Walden et al. 2017).

Worldwide, the epidemiology of RAZ is changing. The prevalence and severity of these diseases are likely to increase in response to widespread urbanization and subsequent increased contact between rats and impoverished urban human populations (Bradley and Altizer 2007; Meerburg et al. 2009a; Himsworth et al. 2013b; Costa et al. 2017). Global climate change is expected to have its most severe impacts on coastal cities. Subsequent infrastructure damage, flooding and other

harmful effects may also impact urban rat populations and potentially increase the risk of pathogen transmission (Lau et al. 2010; Himsworth et al. 2013b).

1.4 BACKGROUND ON THE VANCOUVER RAT PROJECT Understanding the relationships between reservoir hosts and their zoonotic pathogens is needed to assess whether those hosts pose health risks to people and to develop strategies to monitor and reduce those risks (Mills and Childs 1998; Wolfe et al. 2007; Morse et al. 2012; Cunningham et al. 2017). Despite this requisite knowledge, relatively little is known about rats and RAZ in Canada—studies of this nature are rare, geographically dispersed and dated. In the 1930s, an investigation in Vancouver, British Columbia implicated rat mites (Liponyssus bacoti) in an outbreak of mite bites, skin irritation and itching among employees of a retail store (Spencer 1936). Research in the 1920s and 1940s identified Leptospira spp. and Trichinella spiralis, respectively, among rats sampled in Toronto, Ontario (Cameron and Irwin 1929; Kuitunen- Ekbaum and Webster 1947). Finally, a small study of 43 rats from the Richmond, British Columbia waste dump identified the presence of several parasitic pathogens, as well as seropositivity for Leptospira spp. (Harvey and MacNeill 1984). Since these studies, there have been considerable scientific advances in diagnostic techniques, pathogen identification, disease ecology and epidemiology.

The growing need to understand RAZ combined with the deficiency of Canadian data led to the initiation of the Vancouver Rat Project (VRP; www.vancouverratproject.com). This is a multidisciplinary study of the epidemiology and ecology of rat-associated zoonotic pathogens set in Vancouver’s Downtown Eastside. The initial phase of the VRP found that rats in Vancouver carried several species of zoonotic bacteria. These included pathogens traditionally associated with rats—Leptospira interrogans and Bartonella tribocorum (Himsworth et al. 2013a; 2015a). Rats in this population also carried environment-associated and potentially zoonotic bacteria for which they are not considered to be traditional reservoirs, including Escherichia coli, Salmonella spp., Clostridium difficile, methicillin-resistant Staphylococcus aureus, and methicillin-resistant Staphylococcus pseudintermedius (Himsworth et al. 2013c; 2014a; 2014b; 2015b).

The initial results of the VRP have generated new questions requiring further study. Notably, all zoonotic and rat-specific pathogens detected in Vancouver rats (e.g., L. interrogans and Capillaria hepatica) were unevenly distributed in the urban landscape (Himsworth et al. 2013a; Rothenburger et al. 2014). Using L. interrogans as an example of this heterogeneity, the overall prevalence was 11%, but varied between 0% and 67%, depending on the city block of origin (Himsworth et al. 2013a). These results suggest that clustering of infected hosts may represent varying risk of infection pressure to humans and other rats. It also represents a crucial knowledge gap that requires further exploration.

1.5 TOWARDS A BROADER UNDERSTANDING OF FACTORS INFLUENCING THE ECOLOGY OF RAT-ASSOCIATED ZOONOTIC PATHOGENS Like many prior studies of rat-associated zoonotic pathogens, the VRP identified season and several rat-level factors associated with pathogen carriage: body mass, nutritional condition, sex, sexual maturity and bite wounds (Easterbrook et al. 2007; Himsworth et al. 2013a; 2015a). The approach to focus on demographic and seasonal effects is typical of most studies of zoonotic pathogens in their host.

Recognition of the host factors associated with pathogen carriage is important for understanding of host-pathogen ecology. However, the epidemiological triad emphasizes that infectious diseases occur at the complex intersection of host-agent-environment. And many factors across hierarchical levels of biological organization can influence pathogens in rat hosts (Figure 1.1; Estrada-Peña et al. 2014; Ezenwa et al. 2015). Specifically, the environment may impact zoonotic pathogen prevalence through effects on environmental pathogen survival and host abundance, nutrition and immunity. At a lower level of this biological hierarchy, intra-host factors including co-infections and host disease may alter disease epidemiology, host immune responses and increase zoonotic pathogen transmission between reservoir hosts (Ezenwa et al. 2010; Gibson et al. 2011; Garza-Cuartero et al. 2014).

Given that RAZ are a persistent and increasing problem in cities worldwide, an enhanced understanding of the complex factors that contribute to pathogen carriage in rats is needed. This

information would lead to a better understanding of why pathogens are heterogeneously distributed and offer potential mechanisms of risk mitigation for people and targeted surveillance.

1.6 RESEARCH OBJECTIVES The purpose of my research is to investigate the role of environmental and intra-host factors (host disease and co-infections) in the epidemiology and ecology of zoonotic pathogen carriage in urban rats.

The specific objectives of my thesis are to: 1. Summarize and synthesize existing information in the published scientific literature about the influence of the environment on pathogen carriage in specific urban wildlife species including rats (Chapter 2); 2. Determine if environmental and weather factors are associated with carriage of zoonotic pathogens (Bartonella tribocorum, Clostridium difficile, antimicrobial resistant Escherichia coli and methicillin-resistant Staphylococcus aureus) in rats and to compare and contrast differences in associated factors among these pathogens (Chapters 3 & 4); 3. Describe and summarize macroscopic and microscopic lesions in rats to establish a range of rare and common host diseases in this population (Chapter 5); 4. Determine the impact of intra-host disease by examining associations among lesions, co- infections and zoonotic pathogen carriage (Bartonella tribocorum, Clostridium difficile and Leptospira interrogans; Chapter 6).

1.7 FIGURES

Figure 1.1 Factors across hierarchical levels of biological organization may influence the carriage of zoonotic pathogens in rats. At a high level of the scale, populations of rats reside in the environment. Within individual rats at lower levels of the scale, organs are made up of tissues, which in turn consist of cells that contain molecules. Co-infecting pathogens inhabit cells and tissues within individuals. The environment and processes within rats (co- infections with macroparasites and microparasites, and disease at the cellular, tissue and organ levels) may have important influences on the ecology of zoonotic pathogens in urban ecosystems (inspired by Ezenwa et al. 2015; designed by J. Rothenburger; illustrated by Jessica Thiessen).

1.8 REFERENCES

Blanchard RJ, Pank L, Fellows D, Blanchard DC. 1985. Conspecific wounding in free-ranging R. norvegicus from stable and unstable populations. The Psychological Record 35:329–335.

Bradley CA, Altizer S. 2007. Urbanization and the ecology of wildlife diseases. Trends in Ecology and Evolution 22:95–102.

Brooks JE, Jackson WB. 1973. A review of commensal rodents and their control. CRC Critical Reviews in Environmental Control 3:405–453.

Calhoun JB. 1963. The Ecology and Sociology of the Norway Rat. US Department of Health, Education, and Welfare, Public Health Service Publication No. 1008, Bethesda, Maryland, U.S.A, pp. 237–239.

Cameron GC, Irwin DA. 1929. Leptospira icterohæmorrhagiæ occurrence in wild rats at Toronto. Canadian Public Health Journal 20:386–392.

Carr WJ, Hirsch JT, Campellone BE, Marasco E. 1979. Some determinants of a natural food aversion in Norway rats. Journal of Comparative and Physiological Psychology 93:899–906.

Clinton JM. 1969. Rats in Urban America. Public Health Reports 84:1–7.

Costa F, Carvalho-Pereira T, Begon M, Riley L, Childs J. 2017. Zoonotic and vector-borne diseases in urban slums: opportunities for intervention. Trends in Parasitology 33:660–662.

Costa F, Hagan JE, Calcagno J, Kane M, Torgerson P, Martinez-Silveira MS, Stein C, Abela- Ridder B, Ko AI. 2015. Global morbidity and mortality of leptospirosis: a systematic review. PLoS Neglected Tropical Diseases 9:e0003898–20.

Cunningham AA, Daszak P, Wood JLN. 2017. One Health, emerging infectious diseases and wildlife: two decades of progress? Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 372:20160167.

Davis DE. 1953. The characteristics of rat populations. Quarterly Review of Biology 28:373–401.

Easterbrook JD, Kaplan JB, Vanasco NB, Reeves WK, Purcell RH, Kosoy MY, Glass GE, Watson J, Klein SL. 2007. A survey of zoonotic pathogens carried by Norway rats in Baltimore, Maryland, USA. Epidemiology and Infection 135:1192–1199.

Estrada-Peña A, Ostfeld RS, Peterson AT, Poulin R, la Fuente de J. 2014. Effects of environmental change on zoonotic disease risk: an ecological primer. Trends in Parasitology 30:205–214.

Ezenwa VO, Etienne R S, Luikart G, Beja-Pereira A, Jolles AE. 2010. Hidden consequences of living in a wormy world: nematode-induced immune suppression facilitates tuberculosis invasion in African buffalo. American Naturalist 176:613–624.

Ezenwa VO, Prieur-Richard A-H, Roche B, Bailly X, Becquart P, García-Peña GE, Hosseini PR, Keesing F, Rizzoli A, Suzán G, Vignuzzi M, Vittecoq M, Mills JN, Guégan J-F. 2015. Interdisciplinarity and infectious diseases: an Ebola case study. PLoS Pathogens 11:e1004992.

Feng AYT, Himsworth CG. 2014. The secret life of the city rat: a review of the ecology of urban Norway and black rats (Rattus norvegicus and Rattus rattus). Urban Ecosystems 17:149–162.

Firth C, Bhat M, Firth MA, Williams SH, Frye MJ, Simmonds P, Conte JM, Ng J, Garcia J, Bhuva NP, Lee B, Xiaoyu C, Quan RL, Lipkin WI. 2014. Detection of zoonotic pathogens and characterization of novel viruses carried by commensal Rattus norvegicus in New York City. mBio 5:e01933-14.

Fischer JD, Schneider SC, Ahlers AA, Miller JR. 2015. Categorizing wildlife responses to urbanization and conservation implications of terminology. Conservation Biology 29:1246–1248.

Gage KL, Kosoy MY. 2005. Natural history of plague: perspectives from more than a century of research. Annual Reviews of Entomology 50:505–528.

Garza-Cuartero L, Garcia-Campos A, Zintl A, Chryssafidis A, O'Sullivan J, Sekiya M, Mulcahy G. 2014. The worm turns: trematodes steering the course of co-infections. Veterinary Pathology 51:385–392.

Gibson AK, Raverty S, Lambourn DM, Huggins J, Magargal SL, Grigg ME. 2011. Polyparasitism is associated with increased disease severity in Toxoplasma gondii-infected marine sentinel species. PLoS Neglected Tropical Diseases 5:e1142.

Government of Alberta. Alberta Agriculture and Forestry 2017. History of rat control in Alberta. http://www1.agric.gov.ab.ca/$department/deptdocs.nsf/all/agdex3441. Accessed September 17, 2017.

Harvey DA, MacNeill AC. 1984. A survey of zoonotic diseases and arthropod vectors isolated from live-trapped Norway rats (Rattus norvegicus) in the municipality of Richmond, British Columbia. Canadian Journal of Public Health 75:374–378.

Himsworth CG, Bai Y, Kosoy MY, Wood H, DiBernardo A, Lindsay R, Bidulka J, Tang P, Jardine C, Patrick D. 2015a. An Investigation of Bartonella spp., Rickettsia typhi, and Seoul Hantavirus in Rats (Rattus spp.) from an inner-city neighborhood of Vancouver, Canada: is pathogen presence a reflection of global and local rat population structure? Vector Borne and Zoonotic Diseases 15:21–26.

Himsworth CG, Bidulka J, Parsons KL, Feng AYT, Tang P, Jardine CM, Kerr T, Mak S, Robinson J, Patrick DM. 2013a. Ecology of Leptospira interrogans in Norway rats (Rattus norvegicus) in an inner-city neighborhood of Vancouver, Canada. PLoS Neglected Tropical Diseases 7:e2270.

Himsworth CG, Miller RR, Montoya V, Hoang L, Romney MG, Al-Rawahi GN, Kerr T, Jardine CM, Patrick DM, Tang P, Weese JS. 2014a. Carriage of methicillin-resistant Staphylococcus aureus by wild urban Norway rats (Rattus norvegicus). PLoS ONE 9:e87983.

Himsworth CG, Parsons KL, Jardine C, Patrick DM. 2013b. Rats, cities, people, and pathogens: a systematic review and narrative synthesis of literature regarding the ecology of rat-associated zoonoses in urban centers. Vector Borne and Zoonotic Diseases 6:349–359.

Himsworth CG, Patrick DM, Mak S, Jardine CM, Tang P, Weese JS. 2014b. Carriage of Clostridium difficile by wild urban Norway rats (Rattus norvegicus) and black rats (Rattus rattus). Applied and Environmental Microbiology 80:1299–1305.

Himsworth CG, Zabek E, Desruisseau A, Parmley EJ, Reid-Smith R, Jardine CM, Tang P, Patrick DM. 2015b. Prevalence and characteristics of Escherichia coli and Salmonella spp. in the feces of wild urban Norway and black rats (Rattus norvegicus and Rattus rattus) from an inner- city neighborhood of Vancouver, Canada. Journal of Wildlife Diseases 51:589–600.

Johnson S, Bragdon C, Olson C, Merlino M, Bonaparte S. 2016. Characteristics of the built environment and the presence of the Norway rat in New York City: results from a neighborhood rat surveillance program, 2008-2010. Journal of Environmental Health 78:22–29.

Kuitunen-Ekbaum E, Webster D. 1947. Trichinosis in wild rats in Toronto. Canadian Journal of Public Health 38:76–78.

Lau CL, Smythe LD, Craig SB, Weinstein P. 2010. Climate change, flooding, urbanization and leptospirosis: fuelling the fire? Transactions of the Royal Society of Tropical Medicine and Hygiene 104:631–638.

Leibler JH, Zakhour CM, Gadhoke P, Gaeta JM. 2016. Zoonotic and vector-borne infections among urban homeless and marginalized people in the United States and Europe, 1990–2014. Vector-Borne and Zoonotic Diseases 16:435–444.

Morse SS, Mazet JAK, Woolhouse M, Parrish CR, Carroll D, Karesh WB, Zambrana-Torrelio C, Lipkin WI, Daszak P. 2012. Prediction and prevention of the next pandemic zoonosis. Lancet 380:1956–1965.

Mills JN, Childs JE. 1998. Ecologic studies of rodent reservoirs: their relevance for human health. Emerging Infectious Diseases 4:529–537.

McFarlane R, Sleigh A, McMichael T. 2012. Synanthropy of wild mammals as a determinant of emerging infectious diseases in the Asian-Australasian region. EcoHealth 9:24–35.

McKinney ML. 2006. Urbanization as a major cause of biotic homogenization. Biological Conservation 127:247–260.

Meerburg BG, Singleton GR, Kijlstra A. 2009a. Rodent-borne diseases and their risks for public health. Critical Reviews in Microbiology 35:221–270.

Meerburg BG, Singleton GR, Leirs H. 2009b. The Year of the Rat ends–time to fight hunger! Pest Management Science 65:351–352.

Montes de Oca DP, Lovera R, Cavia R. 2017. Where do Norway rats live? Movement patterns and habitat selection in livestock farms in Argentina. Wildlife Research 44:324–10.

IUCN. International Union for Conservation of Nature Red List of Threatened Species. www.iucnredlist.org. Accessed September 11, 2017.

Rothenburger JL, Himsworth CG, Chang V, Lejeune M, Leighton FA. 2014. Capillaria hepatica in wild Norway rats (Rattus norvegicus) from Vancouver, Canada. Journal of Wildlife Diseases 50:628–633.

Rothenburger JL, Himsworth CG, Nemeth NM, Pearl DL, Jardine CM. 2017. Environmental factors and zoonotic pathogen ecology in urban exploiter species. EcoHealth 14:630-641.

Singleton GR, Brown PR, Jacob J, Aplin KP. 2007. Unwanted and unintended effects of culling: a case for ecologically-based rodent management. Integrative Zoology 2:247–259.

Spencer GJ. 1936. The menace of rat parasites in Vancouver in 1936. Proceedings of the Entomological Society of British Columbia 33:44–45.

Stockdale Walden HD, Slapcinsky JD, Roff S, Mendieta Calle J, Diaz Goodwin Z, Stern J, Corlett R, Conway J, McIntosh A. 2017. Geographic distribution of Angiostrongylus cantonensis in wild rats (Rattus rattus) and terrestrial snails in Florida, USA. PLoS ONE 12:e0177910–13.

Taylor LH, Latham SM, Woolhouse MEJ. 2001. Risk factors for human disease emergence. Philosophical Transactions of the Royal Society B: Biological Sciences 356:983–989.

Towns DR, Atkinson IAE, Daugherty CH. 2006. Have the harmful effects of introduced rats on islands been exaggerated? Biological Invasions 8:863–891.

United Nations, Department of Economic and Social Affairs, Population Division. 2015. World Urbanization Prospects: The 2014 Revision. United Nations, New York City, U.S.A.

Wolfe ND, Dunavan CP, Diamond J. 2007. Origins of major human infectious diseases. Nature 447:279–283.

Wyman W. (Ed.). 1910. The Rat And Its Relation To The Public Health. Treasury Department,

Public Health and Marine-Hospital Service of the United States, Washington, D.C., U.S.A.

York EM, Creecy JP, Lord WD, Caire W. 2015. Geographic range expansion for rat lungworm in North America. Emerging Infectious Diseases 21:1234–1236.

CHAPTER 2

2. ENVIRONMENTAL FACTORS AND ZOONOTIC PATHOGEN ECOLOGY IN URBAN EXPLOITER SPECIES*

2.1 ABSTRACT Knowledge of pathogen ecology, including the impacts of environmental factors on pathogen and host dynamics, is essential for determining the risk that zoonotic pathogens pose to people. This review synthesizes the scientific literature on environmental factors that influence the ecology and epidemiology of zoonotic microparasites (bacteria, viruses and protozoa) in globally-invasive urban exploiter wildlife species (i.e., rock doves [Columba livia domestica], European starlings [Sturnus vulgaris], house sparrows [Passer domesticus], Norway rats [Rattus norvegicus], black rats [R. rattus] and house mice [Mus musculus]). Pathogen ecology, including prevalence and pathogen characteristics, are influenced by geographic location, habitat, season and weather. The prevalence of zoonotic pathogens in mice and rats varies markedly over short geographic distances, but tends to be highest in ports, disadvantaged (e.g., low-income) and residential areas. Future research should use epidemiological approaches, including random sampling and robust statistical analyses, to evaluate a range of biotic and abiotic environmental factors at spatial scales suitable for host home range sizes. Moving beyond descriptive studies to uncover the causal factors contributing to uneven pathogen distribution among wildlife hosts in urban environments may lead to targeted surveillance and intervention strategies. Application of this knowledge to urban maintenance and planning may reduce the potential impacts of urban wildlife-associated zoonotic diseases on people.

* A version of this chapter is published as: Rothenburger JL, Himsworth CG, Nemeth NM, Pearl DL, Jardine CM. Environmental factors and zoonotic pathogen ecology in urban exploiter species - a review. EcoHealth. 14(3):630- 641. DOI: 10.1007/s10393-017-1258-5. The final publication is available at: https://link.springer.com/article/10.1007%2Fs10393-017-1258-5

2.2 INTRODUCTION Human activity often leads to severe and large-scale environmental modifications, with cities representing an extreme example. Concurrent with exponential human population growth, the United Nations projects that proportion of urbanites worldwide will reach 66% by 2050 (2015). Cities are characterized by reduced biodiversity, while favoring specific free-ranging wild animal species (hereafter referred to as wildlife; Grimm et al. 2008). When classified by their responses to urbanization, “urban exploiters” are wildlife species that depend on anthropogenic resources (i.e., food and habitat) and demonstrate peak abundance in urban core areas (McKinney 2006).

A subset of urban exploiter wildlife species flourishes in cities worldwide— rock doves (Columba livia domestica), European starlings (Sturnus vulgaris), house sparrows (Passer domesticus), Norway rats (Rattus norvegicus), black rats (R. rattus) and house mice (Mus musculus; McKinney 2006). Cities share similar environmental characteristics that favor these highly-adaptive species: high-density human populations, buildings and roads, heat island effects and fragmented vegetation (McKinney 2006). Also, global transportation, such as air travel, shipping and long-distance trucking, has provided the means for their introduction.

Besides effects on biodiversity and the environment, urbanization increases contact between certain wildlife species and people, creating the potential for zoonotic pathogen transmission. Most emerging infectious diseases are zoonotic, with a large proportion arising from wildlife (Jones et al. 2008). These emerging diseases in Asia-Pacific regions were 15 times more likely to arise from wildlife that are ecologically associated with people including rats and mice (McFarlane et al. 2012), leading to concerns over the sustained and frequent contact among people in cities and urban wildlife.

Environmental factors, particularly anthropogenic modifications, are strong drivers of zoonotic disease emergence (Daszak et al. 2001; Engering et al. 2013). Yet traditional host-pathogen studies often ignore environmental influences (Gortazar et al. 2014; Barrett and Bouley 2015), likely due to systemic complexity (Estrada-Peña et al. 2014). An animal’s environment may directly impact it with indirect influences on the pathogen(s) it carries, depending on the pathogen’s characteristics and transmission route (i.e., direct, environmental and vector-borne). 15

For instance, abundant food resources contribute to good nutritional condition and enhanced immunity (Bradley and Altizer 2007). This may diminish pathogen load and persistence in the host, while enhancing reproductive success and population growth (Bradley and Altizer 2007). Other environmental factors that could influence pathogen ecology are land use, soil characteristics, and floral and faunal biodiversity, including community composition among hosts (Estrada-Peña et al. 2014; Barrett and Bouley 2015). Weather, including precipitation, humidity and temperature, as well as climate, may also influence pathogen ecology (Bradley and Altizer 2007).

Despite the dynamic nature of cities, investigations of environmental influences of zoonotic pathogens in this habitat are limited. A better understanding of how urban and other environments impact pathogen ecology will allow us to track and reduce associated public health threats. This review: 1) examines and synthesizes knowledge of the environmental factors that influence the ecology and epidemiology of zoonotic microparasites (bacteria, viruses and protozoa) among globally-invasive urban exploiter wildlife species; and 2) provides directions for future studies.

2.3 MATERIALS AND METHODS A wide variety of wild animals occupy urban habitats and the species assemblages vary depending on geographical location (e.g., coyotes [Canis latrans] in North America, rhesus macaques [Macaca mulatta] in India). For the broadest applicability and to consider those species that have the potential for prolonged and frequent contact with people, which may translate to increased risk of zoonotic pathogen transmission, we chose to limit this review to the species that are classified as urban exploiters with global distribution (i.e., rock doves, European starlings, house sparrows, Norway rats, black rats and house mice; McKinney 2006). From January-April 2016, we systematically searched Agricola, Web of Science, CAB Direct and JSTOR databases with keyword combinations of the following concepts: urban, environment, zoonotic and wildlife species (Appendix A Table 1). We selected studies from the English peer- reviewed scientific literature that considered zoonotic pathogens in their wildlife host along with environmental factors (e.g., weather, habitat). We excluded studies of pathogens that are not directly shed by animals (e.g. Cryptococcus spp. associated with pigeon feces), those in

rural/natural areas without an urban component and those with low sample sizes (<25 individuals). We added references through citation searching and evaluated papers using a structured abstracting matrix and synthesis technique (Garrard 2014; Appendix A Table 2). Of the 1400 manuscripts identified in the search, we retained 69 that included 7 viral, 15 bacterial and two protozoal pathogens. Among the studies examined, approximately one third occurred in North America (21/69; 30%) and most focused on rats (41/69; 59% [Appendix A Tables 3, 4]).

Table 2.1 summarizes the emergent themes. Most studies considered location as the primary environmental characteristic of interest. Among these, a subset compared specific habitat types, either within cities or among urban and non-urban locations. Season was an occasional environmental factor of interest. In contrast, very few studies examined comparatively novel environmental factor (e.g., host heavy metal exposure).

2.4 PATHOGEN ECOLOGY VARIES AMONG LOCATIONS

2.4.1 Varying Pathogen Prevalence By Location

Pathogen prevalence may have extreme variability, even over small geographical areas. This phenomenon is best demonstrated by mouse-associated pathogens. For example, the prevalence of Toxoplasma gondii in house mice in cities can range from 0-100% among houses and from 0- 93% among blocks (Murphy et al. 2008). The prevalence of lymphocytic choriomeningitis virus (LCMV) can range from 0-50% among houses, from 0-23% among neighborhood streets and 4- 13% among broader locations within the same city (Childs et al. 1992).

A study of rats in Vancouver, Canada found that the prevalence of zoonotic pathogens (i.e., Leptospira spp., Bartonella spp., Escherichia coli, Salmonella spp., Clostridium difficile, methicillin-resistant Staphylococcus aureus) varied significantly between city blocks (Himsworth et al. 2013, 2014a, 2014b, 2015a, 2015b). The overall prevalence of Leptospira spp. was 11%, but ranged from 0-67% depending on the city block (Himsworth et al. 2013). Residual variation in prevalence after controlling for geographical clustering and covariates with multi- level multivariable modelling suggested that block characteristics, possibly microenvironmental features (e.g., land use, human refuse management) contributed to the variation. A different

approach revealed that Leptospira spp. genomic equivalents shed by Norway rats in urine varied significantly by location, demonstrating that infected hosts and the amount of pathogen shed are geographically clustered (Costa et al. 2015).

The reasons for heterogeneous pathogen distribution were not investigated in these studies. Potential explanations include direct transmission among clustered hosts (Childs et al. 1992) and/or varied exposure to pathogens in the environment (Himsworth et al. 2015b). Environmental influences on both of these mechanisms require further investigation.

2.4.2 Varying Pathogen Genetic And Phenotypic Diversity By Location

Like prevalence, pathogen characteristics vary among locations, even on smaller scales. For example, Yokoyama et al. (2007) analyzed Salmonella enterica serovar Typhimurium isolated from rats captured in two buildings across the street from one another. Rats from each respective building had similar prevalences but were infected with genetically different clones (Yokoyama et al. 2007). A study of hepatitis E virus in Norway rats also identified genetic clustering over small geographical scales (<7 km among sampling sites; Johne et al. 2012). Antimicrobial resistant (AMR) E. coli shows similar spatial heterogeneity (Allen et al. 2011; Sacristán et al. 2014), even within a neighborhood (Himsworth et al. 2015b). This suggests that exposure to antimicrobials or AMR E. coli may differ by site. Collectively, these studies provide evidence of barriers to pathogen spread and/or maintenance within the urban environment.

The tendency for pathogen characteristics to vary among locations is not universal. Most Chlamydia psittaci isolates from European pigeons were genotypically similar, despite sampling several areas (Heddema et al. 2006; Gasparini et al. 2011; Geigenfeind et al. 2012). Birds may move great distances to share pathogens, but this result may also reflect limited genetic variation in this bacterium. Conversely, genetic diversity among birds, including sparrows infected with West Nile virus, may be spatially-dependent (Bertolotti et al. 2008). Genetic diversity was low at small scales (<1 km2) but higher when sampling locations were > 4 km apart, suggesting that distance is a limitation to pathogen transmission in this system. Although birds may move across greater geographical areas to share pathogens compared to rodents, the resulting consequences

for pathogen diversity are likely dependent on the host home range size, as well as pathogen type and transmission routes.

2.4.3 Strengths And Limitations Of Locational Studies

It is important to consider scale in urban wildlife studies (Estrada-Peña et al. 2014). Scale(s) should fit the research question, animal home range size and urban geography and hierarchical arrangement (e.g., properties within blocks within neighborhoods within districts). The studies described above suggest that coarser scales may not represent pathogen distribution in complex urban environments or reflect public health risks. But coarse-scale studies can provide essential information. For example, the discovery that Seoul hantavirus caused hemorrhagic fever with renal syndrome (HFRS) in people stimulated cross-sectional studies in multiple cities in the 1980s. This research established the worldwide distribution and probable long-term carriage of Seoul hantavirus in rats, even though human clinical disease was primarily reported in Asian countries (Childs et al. 1985; LeDuc et al. 1985; Chen et al. 1986).

Studies that compare pathogen prevalence or characteristics among locations may identify a pathogen ‘hotspot’ that triggers public health interventions (Taylor et al. 2008). But these types of studies generally do not provide details or meaningful descriptions of habitat types or environmental features. Nor do these studies analyze the specific environmental features to understand if these may contribute to differences. Therefore, studies that simply compare locations do not inform us about environmental mechanisms contributing to hotspot formation. Understanding the underlying factors, including those of the environment, which contribute to varying pathogen distribution is imperative to developing targeted surveillance and intervention strategies.

2.5 PATHOGEN ECOLOGY VARIES AMONG HABITAT TYPES

2.5.1 Comparing Urban Habitats to Other Habitat Types

High densities of urban exploiter species and people in cities provide opportunity for prolonged and frequent contact between humans and animals, which may exacerbate the risk of zoonotic transmission. Thus, it is important to understand the differences between urban and other habitats, including agricultural, rural and natural areas (Brearley et al. 2013; Mackenstedt et al. 2015). 19

Despite this, consistent trends in prevalence among different habitats are not evident. For example, the prevalence of Bartonella spp. in rats and mice was lower in urban sites compared to farms, harbour and suburban sites (Inoue et al. 2008; Hsieh et al. 2010). In contrast, Halliday (2015) found a 60% prevalence of Bartonella spp. in urban black rats vs. 13% in rats from a rural community. International trade in the urban location likely introduced rats, fleas and the Bartonella spp. they carry, while the rural site was more isolated (Halliday et al. 2015).

Avian pathogens also lack a distinct pattern. Although the prevalence of ticarcillin-resistant E. coli was higher in urban vs. rural pigeons in one study (Sacristán et al. 2014), another found no pigeons carrying Salmonella spp. in 267 urban sites compared to 4% in 139 dairy farm sites (Pedersen et al. 2006). These studies attributed differences in prevalence to varying habitat exposures. Studies of West Nile virus (WNV) further highlight the complexity of habitat comparisons. Reisen et al. (2008) found the lowest WNV seroprevalence in birds (including house sparrows and pigeons) from locations near urban centers. Yet, in a different study, WNV seroprevalence was higher in urban sparrows but lower in urban pigeons compared to elsewhere (Reisen et al. 2006). Habitat type may also influence WNV genetic diversity, which was lower in birds in urban vs. natural areas (Bertolotti et al. 2008).

2.5.2 Comparing Different Habitat Types within Cities

Although studies have sampled animals in a variety of urban habitats, such as downtown/business areas, ports, commercial/industrial areas, and near waste treatment plants (Jiang et al. 2008; Taylor et al. 2008; Widén et al. 2014), the most common approach is to compare residential areas to urban green spaces. Generally, pathogen prevalence is higher in animals in residential sites. For example, Norway rats and mice in residential areas were more likely to be seropositive for hantaviruses compared to those in urban parks (Childs, Korch, et al. 1987; Childs, Glass, et al. 1987; Korch et al. 1989) or urban centers (Jiang et al. 2008). LCMV prevalence in house mice was higher in one residential area compared to other residential sites and urban parks (Childs et al. 1992). These studies suggest that habitat features in residential areas may favor the establishment and/or maintenance of rodent-borne viruses, assuming that rodents have small home ranges. Since these viruses are transmitted directly between hosts, the "dilution effect” may play a role in these habitat differences (Mills 2006). Natural areas and

urban green spaces tend to support higher species diversity but lower densities of certain species, thus, pathogen transmission and prevalence may be reduced due to decreased contact among competent hosts.

There are exceptions to the trend towards increased prevalence in animals sampled in residential areas. For instance, rats in an informal settlement and business district in Durban, South Africa had equal prevalence of T. gondii and Leptospira spp. (Taylor et al. 2008). These are protozoal and bacterial pathogens, respectively, that are mainly transmitted indirectly, while viruses trended towards higher prevalence in residential areas. Thus, habitat influences may vary depending on the pathogen type and mode of transmission.

Shipping ports provide another exception, as they tend to support high pathogen prevalence and diversity. Norway rats near a port in Yangon, Myanmar were more often seropositive for Yersinia pestis compared to non-port sites (Brooks et al. 1977). Comparably, mice and rats near a port had higher serogroup diversity among Leptospira spp. isolates compared to elsewhere (Romero-Vivas et al. 2013). Importation of animals possibly introduced novel serogroups to this focal area. These and other studies (Anholt et al. 2014) indicate that shipping ports may sustain higher pathogen diversity and prevalence than elsewhere in cities through periodic animal and pathogen introductions.

2.5.3 Strengths and Limitations of Habitat Studies

Uncovering associations between pathogen prevalence and habitat types may be particularly useful for targeted interventions, predictive modelling and surveillance (Mills and Childs 1998). While some urban habitats (e.g., ports) may support hosts with high pathogen prevalence and diversity, there are no clear trends when comparing cities to other habitats. Pathogen ecology in urban exploiter species may differ from their non-urban counterparts, a difference that also occurs among a wider range of wildlife hosts (Brearley et al. 2013). But large-scale studies that dichotomize urban vs. other habitats may over-simplify environmental complexity along the urban-rural gradient and thus influences on pathogen ecology (Beninde et al. 2015). Studying specific habitat types within cities (i.e., urban green spaces, residential neighborhoods, industrial areas) may be more insightful than arbitrary locations since habitats may be comparable between

cities and thus findings may be more generalizable. Future research should focus on specific details of these habitats to explain differences.

2.6 DISADVANTAGED URBAN AREAS ARE ASSOCIATED WITH INCREASED PATHOGEN PREVALENCE There is an apparent association between disadvantaged urban areas including low-income areas, slums and refuse dumps, and increased pathogen prevalence in urban exploiter species. For example, rats from areas in Lyon, France with dense human populations and low average incomes were more likely to be infected with Leptospira spp. compared to less populated, higher-income areas (Ayral et al. 2015). Although this study found no correlation between capture success (a proxy for population density) and pathogen status, a more robust approach would have been to include population density as a predictor and/or potential confounding variable. Also in Lyon, all rats carrying hepatitis E virus originated in a low-income area, with none testing positive from elsewhere, including a green space, waste treatment facilities and a peri-urban area (Widén et al. 2014). This study did not control for the effects of population density. Income might be functioning as a proxy for true causal factors (Ayral et al. 2015). These may include microenvironmental characteristics found in areas with low income and high human population density, such as building disrepair and inadequate refuse management.

Urban slums may be sites of increased pathogen prevalence among rats and mice. Rats from a shantytown in Buenos Aires, Argentina were more likely to be seropositive for hantaviruses compared to rats from other locations, including urban parks and residential areas (Cueto et al. 2008). Taylor (2008) identified two “hotspots” for Leptospira spp. and T. gondii in rats and mice in Durban, South Africa, one of which was an informal settlement. But the association between high prevalence and disadvantaged areas is not always consistent. Muñoz-Zanzi et al. (2014) found the lowest Leptospira spp. prevalence among rats and mice sampled from urban slums compared to villages.

There is also evidence that garbage, a prominent feature of disadvantaged urban areas (Satterthwaite 2003), may also be associated with high pathogen prevalence. For instance, Norway rats originating from dumps had higher Leptospira spp. (Hathaway and Blackmore 1981)

and Seoul hantavirus (Jiang et al. 2008) prevalence vs. other sites, including natural areas and suburbs. High population density of rats in dumps may contribute to higher pathogen prevalence (Hathaway and Blackmore 1981). In addition to location and season, Gargiulo et al. (2014) examined the effect of a waste emergency on the prevalence of enteric zoonotic pathogens in pigeons. Prevalences were significantly higher in pigeons from municipalities in which a waste emergency had occurred during the study period. Human refuse is a major food source for urban wildlife (McKinney 2006) so collectively, these findings suggest that research into this association may uncover a causal relationship leading to novel approaches to zoonotic pathogen control.

2.6.1 Strengths and Limitations of Studies in Disadvantaged Urban Areas

There are only a few studies that include disadvantaged urban areas as an environmental characteristic. And these studies examined a variety of pathogens, so the association with increased prevalence is tenuous. If true, the phenomenon that these areas support higher proportions of zoonotic pathogen-carrying animals is a double tragedy exacerbated by an increased likelihood of infestations (Feng and Himsworth 2014; Johnson 2016). Thus, people in these areas may be at an increased risk of zoonotic diseases (Oliveira et al. 2013). This important area for future research should seek to understand the causal mechanisms underlying this perceived association. For example, does poor municipal hygiene contribute to increased pathogen prevalence independent of host population density?

2.7 SEASONALITY AND WEATHER EFFECTS ON PATHOGEN CHARACTERISTICS IN WILDLIFE HOSTS There are no overall trends between season and weather with pathogen prevalence among urban exploiter species. For instance, a study of rats from Cyprus revealed that seasonal effects are pathogen-dependent (Psaroulaki et al. 2010). The prevalence of T. gondii and Leishmania infantum was highest in summer; there were no seasonal associations with Coxiella burnetii or Bartonella henselae.

Studying Seoul hantavirus infection in Norway rats, Klein et al. (2002) found no association between prevalence and season, temperature or photoperiod in a five-year study. Childs et al.

(1987) identified new infections in every season, establishing year-long virus transmission. Using a virus-carrying index (i.e., combination of rodent density and virus-carrying rate) among rodents (primarily Norway rats), Guan et al. (2009) identified a lag in the effects of temperature, precipitation and humidity on virus-carrying index and subsequently, human incidence of HFRS over 16 years. This suggests that weather factors have a major indirect effect on human disease by influencing pathogen dynamics in reservoir hosts. Collectively, these studies hint that weather and seasonal influences on hantavirus transmission in cities are complex and require further study, yet are potentially important factors that may influence prevalence in urban rat hosts and disease risk to people.

The absence or inconsistency of seasonal/weather effects, such as with C. psittaci prevalence in urban pigeons (Heddema et al. 2006; Geigenfeind et al. 2012) and some pathogens in rats may be attributed to the complex effects of weather on hosts, vectors and environmental pathogens. Year-long contact among urban animals may facilitate direct pathogen transmission regardless of season or weather (Klein et al. 2002). Impervious surfaces (e.g., concrete, asphalt) collect moisture in otherwise dry locations, facilitating pathogen survival in the environment and creating microhabitats for pathogen transmission. Shelter provided by buildings and other built structures may reduce the effects of precipitation on transmission and pathogen survival in the environment. Cities are also warmer and experience diminished seasonality compared to adjacent areas. This urban "heat-island effect” may influence pathogen ecology (Bradley and Altizer 2007). Warm cities may prevent environmentally-transmitted pathogens and vectors from freezing. But hot cities may actually decrease pathogen survival in the environment and may impede vector transmission (e.g., flea-transmitted Y. pestis, the causative agent of plague; Cavanaugh 1971). Abundant food resources that are available year-round may support high populations of urban exploiters and the pathogens they harbor (McKinney 2006). Finally, seasonal effects likely vary by geographical location (i.e., four season vs. wet/dry season climates). These and other features of the urban environment add to the complexity of understanding pathogen ecology in cities.

2.7.1 Strengths and Limitations of Seasonal and Weather Studies

Relatively few high-quality studies have examined the effects of season and weather on pathogen carriage in urban exploiter hosts. This is an important knowledge gap to fill. There is evidence that seasonality and weather may influence the prevalence of infection in other animals (e.g., Leptospira spp. in dogs; Lelu et al. 2015) and zoonotic diseases in people (e.g., leptospirosis; Benacer et al. 2016). Future studies should follow the example of Guan et al. (2009), by including a lag period for weather factors prior to animal capture since infections likely occur at undetermined time points prior to sampling, when weather conditions differed. Multi-year studies of seasonal effects and weather are preferable to single-year studies due to the ability to replicate the exposure. Public health professionals could use this information to develop predictive models, surveillance and host control strategies, develop public awareness campaigns, and devise additional measures to reduce the risk of human infections (Mills and Childs 1998). Since climate change may impact cities worldwide (Lau et al. 2010), knowledge about the influence of meteorological factors on zoonotic pathogens in urban wildlife is increasingly important.

2.8 OTHER ENVIRONMENTAL CHARACTERISTICS A range of environmental characteristics may influence pathogen-host dynamics. For example, a unique study of Y. pestis in rodents considered building material (i.e., brick, wood or thatch) and indoor vs. outdoor trapping location, finding that most seropositive rats were indoors within wooden structures (Brooks et al. 1977). This difference likely reflects where rats live in this community but also emphasizes the risk of potential zoonotic transmission within indoor environments.

Another rare study approach involved measuring soil pH while assessing Leptospira spp. in rodents in an informal settlement of Durban, South Africa (Taylor et al. 2008). Soil pH throughout the study area was optimal for Leptospira spp. survival outside of hosts. Finally, a study of heavy metal exposure in pigeons (proposed to reflect local environmental contamination) found that birds with low zinc levels in feather samples were more likely to be infected with C. psittaci (Gasparini et al. 2014). Zinc may interact with the immune system or directly with the pathogen to cause this association (Gasparini et al. 2014). Overall, these studies demonstrate how 25

researchers can incorporate environmental features beyond location and habitat to generate novel hypotheses that stimulate further research.

2.9 DIRECTIONS FOR FUTURE RESEARCH

2.9.1 Study Quality

Study quality is a major limitation of research on environmental influences on zoonotic pathogens in urban exploiter species and may result in erroneous conclusions. Thus, the true impact of the themes identified in this review remain unclear. This research field would benefit from approaches that use epidemiological principles when feasible. Animals should be collected using randomized, systematic sampling to avoid selection bias that may result from convenience/purposive sampling in sites with high animal densities. Sites could be repeatedly sampled over time (i.e., longitudinal and repeated cross-sectional studies) to analyze the impact of environmental characteristics. Alternatively, multiple sites of a particular habitat type or that have variable environmental characteristics of interest could be systematically sampled in cross- sectional studies (e.g., systematic trapping of an entire neighborhood). Aspects of host and parasite ecology (e.g., representative demographic groups, pathogen transmission routes) should inform study design. To understand the reservoir dynamics of multi-host pathogens, studies need to consider the potential hosts in a given location, which may include both wild and domestic species (Haydon et al. 2002).

Studies should also follow reporting guidelines (Sargeant et al. 2016) and add confidence intervals when possible. For instance, studies should include sufficient methodological detail to enable replication and describe results by species/location/habitat rather than aggregating data, which may hinder interpretation. Maps greatly enhance clarity of methods and findings. Researchers should use statistical analyses such as multivariable modelling that account for confounding/interacting variables (e.g., sex, species and age) and data that may be autocorrelated in space, time or by social grouping (e.g., rat in a burrow, pigeons in a flock). As well, researchers should use sample size calculations that account for autocorrelated data to design studies with adequate statistical power (Dohoo et al. 2009). Consistent and systematic

approaches to study design, analysis and results will enhance comparability across studies and may result in more definitive conclusions.

2.9.2 Impact of the Urban Environment on Pathogen Ecology

Studies have considered a range of potential environmental factors that may influence pathogen ecology in urban exploiters—location, habitat, soil pH (Taylor et al. 2008), human socioeconomic data (Ayral et al. 2015) and heavy metal exposure (Gasparini et al. 2014). There are also missed opportunities when assessments are limited to associations between environmental factors and animal abundance, but exclude comparisons among these environmental factors and pathogen prevalence (e.g., Muñoz-Zanzi et al. 2014). It would be beneficial if future studies included a range of relevant biotic and abiotic environmental factors that may influence pathogen ecology in cities (Figure 2.1).

It is also possible that the variation and themes identified in this review are not related to the environment but rather to other factors, including host population structure, pathogen transmission dynamics and genetics. Until there are standardized, high-quality studies at smaller scales that take into account features of the microenvironment, it will be difficult to tease out these factors. Table 2.2 contains suggested areas of future research to uncover the mechanisms and factors contributing to uneven pathogen distribution in cities.

2.10 CONCLUSIONS Knowledge of zoonotic pathogen ecology in urban wildlife, particularly urban exploiter species, is essential to assessing the risks of transmission to people in this age of emerging infectious diseases. The upstream environmental effects on pathogen ecology are an important component to risk evaluation. A key finding in this review is that pathogen prevalence consistently varies by location and habitat type. Future research should seek to explain this variation by exploring environmental and other factors. The apparent increased tendency for animals carrying zoonotic pathogens to originate in residential and disadvantaged urban areas is troubling and also warrants further investigation. The relationships between environmental characteristics such as seasonality, weather and others are far more tenuous with no clear trends identified in the current literature.

Well-designed epidemiological and ecological studies would inform and strengthen these conclusions.

Urban environments could be important drivers of zoonotic pathogen ecology. Research that considers causal relationships between environmental factors and pathogen ecology is essential for designing evidence-based surveillance and intervention strategies. It would also provide fundamental information that may help mitigate public health risks through urban maintenance, planning and design. Ultimately, the results may provide a comprehensive approach to cultivating healthy urban landscapes.

2.11 TABLES Table 2.1 Summary of emergent themes

Theme

1. Pathogen prevalence and other characteristics vary by location and habitat of hosts for reasons that are poorly understood

2. Rats and mice in residential sites may have higher prevalence of viral pathogens compared to elsewhere

3. Rats and mice in disadvantaged urban areas may have higher prevalence of pathogens compared to elsewhere

4. Urban shipping ports may be sites of high pathogen prevalence and diversity

5. There are no consistent trends for seasonal and weather effects

Table 2.2 Summary of future research directions

Category Suggestion

Study Design • Account for host and pathogen ecology including transmission routes

• Use randomization, sample size calculations, multivariable modelling and other applicable epidemiological techniques to maximize the validity and utility of study results

• Draw on expertise from other disciples such as urban ecology, ecohealth, disease/ecologic modelling and urban planning to systematically study the associations between environmental characteristics and zoonotic pathogens in their hosts

• Sample indoor and outdoor environments

• Examine season, meteorological and climatological factors with various lag intervals over multiple years to identify underlying weather patterns associated with pathogens in hosts and to inform predictive models

• Conduct longitudinal studies to determine if associations with environmental features are consistent over time

• Expand research to include multi-pathogen and urban-adapted hosts and the zoonotic pathogens they carry

• Include representative cities from around the world and across varied climatic regions

Interventions • Prospective studies are needed that modify the urban environment to assess if these interventions result in meaningful change in zoonotic pathogen prevalence

• Consider meteorological factors in the context of climate change

• Target interventions to pathogen "hot spots" rather than broad-sweeping population control schemes

• Funding bodies (public health, wildlife, environmental) need to recognize and support urban wildlife studies through long-term funding

2.12 FIGURE

Figure 2.1 Examples of abiotic, biotic and anthropogenic factors that may influence zoonotic pathogens in urban wildlife

2.13 REFERENCES

Allen SE, Boerlin P, Janecko N, Lumsden JS, Barker IK, Pearl DL, Reid-Smith RJ, Jardine C. 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 77:882–888.

Anholt H, Himsworth C, Rothenburger JL, Proctor H, Patrick DM. 2014. Ear mange mites (Notoedres muris) in black and Norway rats (Rattus rattus and Rattus norvegicus) from inner- city Vancouver, Canada. Journal of Wildlife Diseases 50:104–108.

Ayral F, Artois J, Zilber AL, Widén F, Pounder KC, Aubert D, Bicout DJ, Artois M. 2015. The relationship between socioeconomic indices and potentially zoonotic pathogens carried by wild Norway rats: a survey in Rhône, France (2010-2012). Epidemiology and Infection 143:586-599.

Barrett MA, Bouley TA. 2015. Need for enhanced environmental representation in the implementation of One Health. EcoHealth 12:212–219.

Benacer D, Thong KL, Min NC, Verasahib KB, Galloway RL, Hartskeerl RA, Souris M, Zain SNM. 2016. Epidemiology of human leptospirosis in Malaysia, 2004-2012. Acta Tropica 157:162–168.

Beninde J, Veith M, Hochkirch A. 2015. Biodiversity in cities needs space: a meta-analysis of factors determining intra-urban biodiversity variation. Ecology Letters 18:581–592.

Bertolotti L, Kitron UD, Walker ED, Ruiz MO, Brawn JD. 2008. Fine-scale genetic variation and evolution of West Nile virus in a transmission ‘hot spot’ in suburban Chicago, USA. Virology 374:381-389.

Bradley CA, Altizer S. 2007. Urbanization and the ecology of wildlife diseases. Trends in Ecology and Evolution 22:95–102.

Brearley G, Rhodes J, Bradley A, Baxter G, Seabrook L, Lunney D, Liu Y, McAlpine C. 2013 Wildlife disease prevalence in human-modified landscapes. Biological Reviews 88:427–442.

Brooks JE, Naing UH, Walton DW, Myint DS, Tun UM, Thaung U, Kyi DO. 1977. Plague in small mammals and humans in Rangoon, Burma. The Southeast Asian Journal of Tropical Medicine and Public Health 8(3):335–344.

Cavanaugh DC. 1971. Specific effect of temperature upon transmission of the plague bacillus by the oriental rat flea, Xenopsylla cheopis. American Journal of Tropical Medicine and Hygiene 20:264–273.

Chen HX, Qiu FX, Dong BJ, Ji SZ, Li YT, Wang Y, Wang HM, Zuo GF, Tao XX, Gao SY. 1986. Epidemiological studies on hemorrhagic fever with renal syndrome in China. Journal of Infectious Diseases 154:394–398.

Childs JE, Glass GE, Korch GW, Ksiazek TG, LeDuc JW. 1992. Lymphocytic choriomeningitis virus infection and house mouse (Mus musculus) distribution in urban Baltimore. American Journal of Tropical Medicine and Hygiene 47:27–34.

Childs JE, Glass GE, Korch GW, LeDuc JW. 1987. Prospective seroepidemiology of hantaviruses and population dynamics of small mammal communities of Baltimore, Maryland. American Journal of Tropical Medicine and Hygiene 37:648–662.

Childs JE, Korch GW, Glass GE, LeDuc JW, Shah KV. 1987. Epizootiology of hantavirus infections in Baltimore: isolation of a virus from Norway rats, and characteristics of infected rat populations. American Journal of Epidemiology 126:55-68.

Childs JE, Korch GW, Smith GA, Terry AD, LeDuc JW. 1985. Geographical distribution and age related prevalence of antibody to Hantaan-like virus in rat populations of Baltimore, Maryland, USA. American Journal of Tropical Medicine and Hygiene 34:385–387.

Costa F, Wunder EA, De Oliveira D, Bisht V, Rodrigues G, Reis MG, Ko AI, Begon M, Childs JE. 2015. Patterns in Leptospira shedding in Norway rats (Rattus norvegicus) from Brazilian slum communities at high risk of disease transmission. PLoS Neglected Tropical Diseases 9:e0003819.

Cueto GR, Cavia R, Bellomo C, Padula PJ, Suárez OV. 2008. Prevalence of hantavirus infection in wild Rattus norvegicus and R. rattus populations of Buenos Aires City, Argentina. Tropical Medicine and International Health 13:46–51.

Daszak P, Cunningham AA, Hyatt AD. 2001. Anthropogenic environmental change and the emergence of infectious diseases in wildlife. Acta Tropica 78:103–16.

Dohoo, IR, Martin SW, Stryhn H. 2009. Veterinary Epidemiological Research 2nd Edition, Charlottetown, P.E.I.: VER, Inc.

Engering A, Hogerwerf L, Slingenbergh J. 2013. Pathogen-host-environment interplay and disease emergence. Emerging Microbes and Infections 2:e5.

Estrada-Peña A, Ostfeld RS, Peterson AT, Poulin R, la Fuente J de. 2014. Effects of environmental change on zoonotic disease risk: an ecological primer. Trends in Parasitology 30:205–214.

Feng AYT, Himsworth CG. 2014. The secret life of the city rat: a review of the ecology of urban Norway and black rats (Rattus norvegicus and Rattus rattus). Urban Ecosystems 17:149–162.

Gargiulo A, Russo TP, Schettini R, Mallardo K, Calabria M, Menna LF, Paia P, Pagnini U, Caputo V, Fioretti A, Dipineto L. 2014. Occurrence of enteropathogenic bacteria in urban pigeons (Columba livia) in Italy. Vector Borne and Zoonotic Diseases 14:251–255.

Garrard J. 2014. Health Sciences Literature Review Made Easy: The Matrix Method 4th Edition, Burlington: Jones & Bartlett Learning.

Gasparini J, Jacquin L, Laroucau K, Vorimore F, Aubry E, Castrec-Rouëlle M, Frantz A. 2014. Relationships between metals exposure and epidemiological parameters of two pathogens in urban pigeons. Bulletin of Environmental Contamination and Toxicology 92:208–212.

Geigenfeind I, Vanrompay D, Haag-Wackernagel D. 2012. Prevalence of Chlamydia psittaci in the feral pigeon population of Basel, Switzerland. Journal of Medical Microbiology 61:261–265.

Gortazar C, Reperant LA, Kuiken T, la Fuente J de, Boadella M, Martínez-Lopez B, Ruiz-Fons F, Estrada-Peña A, Drosten C, Medley G, Ostfeld R, Peterson T, VerCauteren KC, Menge C, Artois M, Schultsz C, Delahay R, Serra-Cobo J, Roulin R, Keck F, Aguirre AA, Henttonen H, Dobson AP, Kutz S, Lubroth J, Mysterud A. 2014. Crossing the interspecies barrier: opening the door to zoonotic pathogens. PLoS pathogens 10:e1004129.

Grimm NB, Faeth SH, Golubiewski NE, Redman CL, Wu J, Bai X, Briggs JM. 2008. Global change and the ecology of cities. Science 319:756–760.

Guan P, Huang D, He M, Shen T, Guo J, Zhou B. 2009. Investigating the effects of climatic variables and reservoir on the incidence of hemorrhagic fever with renal syndrome in Huludao City, China: a 17-year data analysis based on structure equation model. BMC Infectious Diseases 9:109.

Halliday JEB, Knobel DL, Agwanda B, Bai Y, Breiman RF, Cleaveland S, Njenga MK, Kosoy M. 2015. Prevalence and diversity of small mammal-associated Bartonella species in rural and urban Kenya. PLoS Neglected Tropical 9:e0003608.

Hathaway SC, Blackmore DK. 1981. Ecological aspects of the epidemiology of infection with leptospires of the Ballum serogroup in the black rat (Rattus rattus) and the brown rat (Rattus norvegicus) in New Zealand. The Journal of Hygiene 87:427–436.

Haydon DT, Cleaveland S, Taylor LH, Laurenson MK. 2002. Identifying reservoirs of infection: a conceptual and practical challenge. Emerging Infectious Diseases 8:1468–1473.

Heddema ER, Sluis S Ter, Buys JA, Vandenbroucke-Grauls CMJE, van Wijnen JH, Visser CE. 2006. Prevalence of Chlamydophila psittaci in fecal droppings from feral pigeons in Amsterdam, The Netherlands. Applied and Environmental Microbiology 72:4423–4425.

pathogen presence a reflection of global and local rat population structure? Vector Borne and Zoonotic Diseases 15:21–26.

Himsworth CG, Bidulka J, Parsons KL, Feng AYT, Tang P, Jardine CM, Kerr T, Mak S, Robinson J, Patrick DM. 2013. Ecology of Leptospira interrogans in Norway rats (Rattus norvegicus) in an inner-city neighborhood of Vancouver, Canada. PLoS Neglected Tropical Diseases 7:e2270.

Himsworth CG, Miller RR, Montoya V, Hoang L, Romney MG, Al-Rawahi GN, Kerr T, Jardine CM, Patrick DM, Tang P, Weese S. 2014a. Carriage of methicillin-resistant Staphylococcus aureus by wild urban Norway rats (Rattus norvegicus). PLoS ONE 9:e87983.

Himsworth CG, Zabek E, Desruisseau A, Parmley EJ, Reid-Smith R, Jardine CM, Tang P, Patrick DM. 2015b Prevalence and characteristics of Escherichia coli and Salmonella spp. in the feces of wild urban Norway and black rats (Rattus norvegicus and Rattus rattus) from an inner- city neighborhood of Vancouver, Canada. Journal of Wildlife Diseases 51:589–600.

Hsieh J-W, Tung KC, Chen W-C, Lin J-W, Chien L-J, Hsu Y-M, Wang HC, Chomel BB, Chang CC. 2010. Epidemiology of Bartonella infection in rodents and shrews in Taiwan. Zoonoses and Public Health 57:439–446.

Inoue K, Maruyama S, Kabeya H, Yamada N, Ohashi N, Sato Y, Yukawa M, Masuzawa T, Kawamori F, Kadosaka T, Takada N, Fujita H, Kawabata H. 2008. Prevalence and genetic diversity of Bartonella species isolated from wild rodents in Japan. Applied and Environmental Microbiology 74:5086–5092.

Jiang J-F, Zuo S-Q, Zhang W-Y, Wu X-M, Tang F, De Vlas SJ, Zhao WJ, Zhang PH, Dun Z, Wang RM, Cao WC. 2008. Prevalence and genetic diversities of hantaviruses in rodents in Beijing, China. American Journal of Tropical Medicine and Hygiene 78:98–105.

Johne R, Dremsek P, Kindler E, Schielke A, Plenge-Bönig A, Gregersen H, Wessels U, Schmidt K, Rietschel W, Groschup MH, Guenther S, Heckel G, Ulrich RG. 2012. Rat hepatitis E virus: geographical clustering within Germany and serological detection in wild Norway rats (Rattus norvegicus). Infection, Genetics and Evolution 12:947–956.

Jones KE, Patel NG, Levy MA, Storeygard A, Balk D, Gittleman JL, Daszak P. 2008. Global trends in emerging infectious diseases. Nature 451:990–993.

Klein SL, Bird BH, Nelson RJ, Glass GE. 2002. Environmental and physiological factors associated with Seoul virus infection among urban populations of Norway rats. Journal of Mammalogy 83:478-488.

Korch GW, Childs JE, Glass GE, Rossi CA, LeDuc JW. 1989. Serologic evidence of hantaviral infections within small mammal communities of Baltimore, Maryland: spatial and temporal patterns and host range. The American Journal of Tropical Medicine and Hygiene 41:230-40.

LeDuc JW, Smith GA, Pinheiro FP, Vasconcelos PF, Rosa ES, Maiztegui JI. 1985. Isolation of a Hantaan-related virus from Brazilian rats and serologic evidence of its widespread distribution in South America. American Journal of Tropical Medicine and Hygiene 34:810–815.

Lelu M, Muñoz-Zanzi C, Higgins B, Galloway R. 2015. Seroepidemiology of leptospirosis in dogs from rural and slum communities of Los Rios Region, Chile. BMC Veterinary Research 11:31.

Mackenstedt U, Jenkins D, Romig T. 2015. The role of wildlife in the transmission of parasitic zoonoses in peri-urban and urban areas. International Journal for Parasitology: Parasites and Wildlife 4:71–79.

McFarlane R, Sleigh A, McMichael T. 2012. Synanthropy of wild mammals as a determinant of emerging infectious diseases in the Asian-Australasian region. EcoHealth 9:24–35.

McKinney ML. 2006. Urbanization as a major cause of biotic homogenization. Biological Conservation 127:247–260.

Mills JN. 2006. Biodiversity loss and emerging infectious disease: an example from the rodent- borne hemorrhagic fevers. Biodiversity 7:9–17.

Mills JN, Childs JE. 1998. Ecologic studies of rodent reservoirs: their relevance for human health. Emerging Infectious Diseases 4:529–537.

Muñoz-Zanzi C, Mason M, Encina C, Gonzalez M, Berg S. 2014. Household characteristics associated with rodent presence and Leptospira infection in rural and urban communities from Southern Chile. The American Journal of Tropical Medicine and Hygiene 90:497–506.

Murphy RG, Williams RH, Hughes JM, Hide G, Ford NJ, Oldbury DJ. 2008. The urban house mouse (Mus domesticus) as a reservoir of infection for the human parasite Toxoplasma gondii: an unrecognised public health issue? International Journal of Environmental Health Research 18:177–185.

Oliveira DSC, Guimarães MJB, Portugal JL, Medeiros Z. 2013. The socio–demographic, environmental and reservoir factors associated with leptospirosis in an urban area of north– eastern Brazil. Annals of Tropical Medicine and Parasitology 103:149–157.

Pedersen K, Clark L, Andelt WF, Salman MD. 2006. Prevalence of shiga toxin-producing Escherichia coli and Salmonella enterica in rock pigeons captured in Fort Collins, Colorado. Journal of Wildlife Diseases 42:46–55.

Psaroulaki A, Antoniou M, Toumazos P, Mazeris A, Ioannou I, Chochlakis D, Christophi N, Loukaides P, Patsais A, Moschandrea I, Tselentis Y. 2010. Rats as indicators of the presence and dispersal of six zoonotic microbial agents in Cyprus, an island ecosystem: a seroepidemiological study. Transactions of the Royal Society of Tropical Medicine and Hygiene 104:733–739.

Reisen WK, Barker CM, Carney R, Lothrop HD, Wheeler SS, Wilson JL, Madon MB, Takahashi R, Carroll B, Garcia S, Fang Y, Shafii M, Kahl N, Ashtari S, Kramer V, Glaser C, Jean C. 2006. Role of corvids in epidemiology of West Nile virus in southern California. Journal of Medical Entomology 43:356–367.

Reisen WK, Lothrop HD, Wheeler SS, Kennsington M, Gutierrez A, Fang Y, Garcia S, Lothrop B. 2008. Persistent West Nile virus transmission and the apparent displacement St. Louis encephalitis virus in southeastern California, 2003-2006. Journal of Medical Entomology 45:494–508.

Romero-Vivas CME, Cuello-Pérez M, Agudelo-Flórez P, Thiry D, Levett PN, Falconar AKI. 2013. Cross-sectional study of Leptospira seroprevalence in humans, rats, mice, and dogs in a main tropical sea-port city. The American Journal of Tropical Medicine and Hygiene 88:178– 183.

Sacristán C, Esperón F, Herrera-León S, Iglesias I, Neves E, Nogal V, Muño MJ, Torre A. 2014. Virulence genes, antibiotic resistance and integrons in Escherichia coli strains isolated from synanthropic birds from Spain. Avian Pathology 43:172–175.

Sargeant JM, O’Connor AM, Dohoo IR, Erb HN. STROBE-Vet Statement. Available: https://strobevet-statement.org/ [Accessed November 9, 2016]

Satterthwaite D. 2003. The links between poverty and the environment in urban areas of Africa, Asia, and Latin America. Annals of the American Academy of Political and Social Science 590:73–92.

Taylor PJ, Arntzen L, Hayter M, Iles M, Frean J, Belmain S. 2008. Understanding and managing sanitary risks due to rodent zoonoses in an African city: beyond the Boston Model. Integrative Zoology 3:38–50.

United Nations, Department of Economic and Social Affairs, Population Division (2015) World Urbanization Prospects: The 2014 Revision, New York: United Nations. Available: https://esa.un.org/unpd/wup/ [accessed November 2, 2016]

Widén F, Ayral F, Artois M, Olofson AS, Lin J. 2014. PCR detection and analyzis of potentially zoonotic Hepatitis E virus in French rats. Virology Journal 11:90.

Yokoyama E, Maruyama S, Kabeya H, Hara S, Sata S, Kuroki T, Yamamoto T. 2007. Prevalence and genetic properties of Salmonella enterica serovar typhimurium definitive phage type 104 isolated from Rattus norvegicus and Rattus rattus house rats in Yokohama City, Japan. Applied and Environmental Microbiology 73:2624–2630.

CHAPTER 3

3. BEYOND ABUNDANCE: HOW MICROENVIRONMENTAL FEATURES AND WEATHER INFLUENCE BARTONELLA TRIBOCORUM INFECTION IN WILD NORWAY RATS (RATTUS NORVEGICUS)*

3.1 ABSTRACT Norway rats (Rattus norvegicus) inhabit cities worldwide and carry a number of zoonotic pathogens. Although many studies have investigated rat-level risk factors, there is limited research on the effects of weather and environment on zoonotic pathogen transmission ecology in rats. The objective of this study is to use a disease ecology approach to understand how abiotic (weather and urban microenvironmental features) and biotic (relative rat population abundance) factors affect Bartonella tribocorum prevalence in urban Norway rats. This potentially zoonotic pathogen is primarily transmitted by fleas and is common among rodents, including rats, around the world. During a systematic trap and removal study of rats, city blocks were evaluated for 48 environmental variables related to waste, land/alley use and property condition, as well as rat abundance. We constructed 32 weather (temperature and precipitation) variables with time lags prior to the date we captured each rat. We fitted multivariable logistic regression models with rat pathogen status as the outcome. The odds of a rat testing positive for B. tribocorum were significantly lower for rats in city blocks with one or more low-rise apartment buildings compared to blocks with none (OR = 0.20; 95% CI: 0.04-0.80; p=0.02). The reason for this association may be related to unmeasured factors that influence pathogen transmission and maintenance, as well as flea vector survival. Bartonella tribocorum infection in rats was positively associated with high minimum temperatures for several time periods prior to rat capture. This finding suggests that a baseline minimum temperature may be necessary for flea vector survival and B. tribocorum transmission among rats. There was no significant association with rat abundance, suggesting a lack of density-dependent pathogen transmission. This study is

an important first step to understanding how environment and weather impacts rat infections including zoonotic pathogen ecology in urban ecosystems.

* A version of this chapter was accepted for publication by Zoonoses and Public Health on November 19, 2017 as: Rothenburger JL, Himsworth CG, Nemeth NM, Pearl DL, Jardine CM. Beyond abundance: how microenvironmental features and weather influence Bartonella tribocorum infection in wild Norway rats (Rattus norvegicus).

3.2 INTRODUCTION Cities are drastically altered environments that support certain wildlife species. These so-called urban exploiter species capitalize on abundant food resources and human-built environmental niches (McKinney, 2006). Among the globally-invasive urban exploiter species, Norway and black rats (Rattus norvegicus and R. rattus, respectively) are the most harmful to humans. In cities worldwide, rats create several issues including social stigma, infrastructure damage and food contamination (Feng and Himsworth, 2014). Rats are also the source of zoonotic bacterial pathogens responsible for disease in people (Meerburg et al., 2009; Himsworth et al., 2013). These include plague (caused by Yersinia pestis) and leptospirosis (caused by Leptospira spp.). As a consequence of the unprecedented rate of global urbanization, the impacts of zoonotic diseases associated with rats are expected to increase (Himsworth et al., 2013).

Among the many pathogens carried by rats are several species of Bartonella (Heller et al., 1998). Rat-associated Bartonella spp. likely originated in Southeast Asia, and then dispersed globally with invasion activities (Hayman et al., 2013). These Gram-negative bacteria adhere to erythrocytes in a variety of mammalian hosts without causing clinical signs (Schulein et al., 2001; Meerburg et al., 2009). A chronically-infected primary location (likely endothelial cells) periodically releases bacteria (Schulein et al., 2001). Persistently-infected, circulating erythrocytes provide ample opportunity for fleas and other arthropod vectors to ingest infected blood meals and transmit the bacterium to more hosts (Schulein et al., 2001; Gutiérrez et al., 2015).

There is growing evidence that rat-associated Bartonella spp. are zoonotic, causing lymphadenopathy, neuroretinitis, endocarditis, myocarditis, acute febrile illness, anemia and chronic fatigue in people (Kosoy et al., 2010; Buffet et al., 2013; Kandelaki et al., 2016; Vayssier-Taussat et al., 2016). This association has prompted researchers to investigate individual-level factors associated with Bartonella spp. carriage in rodents that include age, body mass, sex, sexual maturity and flea abundance (Jardine et al., 2006; Gutiérrez et al., 2015; Himsworth et al., 2015).

The prevalence of Bartonella spp. varies among geographical locations, even at small scales (Himsworth et al., 2015; Rothenburger et al., 2017). The reason(s) for this variability is not understood, although it may be related to weather and features of the microenvironment (Ayral et al., 2015; Rothenburger et al., 2017). The microenvironment is the small-scale habitat in which an organism lives and is part of the larger environment. Further, the microenvironment includes features influenced by home range size, which in the context of rats within cities, is generally less than 30 m in diameter (Davis et al., 1948) and is typically a city block (Feng and Himsworth, 2014). Examples of microenvironmental features include land use type (e.g., building type, green space), anthropogenic activities (e.g., human use of the area, rat control activities, property maintenance) and waste disposal (e.g., recycling, garbage, hygiene). Some researchers have investigated associations between microenvironmental features and rat abundance. Muñoz-Zanzi et al. (2014) analyzed rodent counts (including rats), finding positive associations with rainfall and signs of rodent infestation, and negative associations with the number of household cats. Himsworth et al. (2014) found that rat presence and abundance were associated with specific types of land use, amount of human refuse and building conditions. Yet, few studies have analyzed specific environmental features for associations with pathogen status in rats (Rothenburger et al., 2017).

Weather (i.e., precipitation and temperature) is another key environmental feature that may influence pathogen ecology in rats (Rothenburger et al., 2017). The prevalence of Bartonella spp. in many rodent species is highest in summer and fall compared to cooler seasons (Gutiérrez et al., 2015). Despite the pervasiveness of urban rats throughout the world and the potentially dangerous pathogens they carry, the influence of the microenvironment and weather on pathogen prevalence has not been well studied (Rothenburger et al., 2017). Furthermore, the exclusion of environmental characteristics in many studies of host-pathogen systems has prompted some researchers to call for increased integration of environmental factors in the application of One Health problem solving (Barrett and Bouley, 2015). This is important because information about the kinds of environments and weather patterns that support infected rats could be used to develop targeted surveillance and interventions for both people and rats, which may also reduce the risk of rat-associated zoonoses in people.

The objective of this study is to understand how abiotic (weather and environmental features) and biotic (rat abundance) factors affect B. tribocorum prevalence in urban Norway rats based on a causal diagram framework (Figure 3.1). Specifically, we tested the following questions: 1) Do impermanent environmental variables (e.g., human loitering) impact B. tribocorum in rats when controlled for confounding variables such as weather, season and permanent environmental variables? 2) Do permanent environmental variables (e.g., land use) impact B. tribocorum infection in rats when season is controlled for as a confounding variable? 3) Do weather variables for various time lags prior to rat capture (i.e., temperature and precipitation) impact B. tribocorum infection in rats when season is controlled for as a potential confounding variable? 4) Does rat abundance impact B. tribocorum infection when controlled for confounding variables such as impermanent and permanent environmental variables, season and weather?

3.3 MATERIALS AND METHODS

3.3.1 Study Design

Data were collected for this study as part of the Vancouver Rat Project (www.vancouverratproject.com), a trap and removal cross-sectional study of zoonotic pathogens in rats from Vancouver, British Columbia, Canada. Himsworth et al. (2014) describes details of the study design and trapping protocol. Briefly, rats were trapped in the back alleys of 43 city blocks that were randomly allocated to a two-week trapping period between September 2011 and August 2012. We completed systematic autopsy and tissue collection for each of the euthanized rats. Blood clot samples from 393 rats were cultured for Bartonella spp. and the species identity was confirmed as B. tribocorum by PCR (Himsworth et al., 2015). The University of British Columbia's Animal Care Committee (A11-0087) approved this study.

3.3.2 Environmental and Weather Characteristics

During the trapping interval, the research team used a systematic environmental observation tool to collect information about the city block as previously described (Himsworth et al., 2014). The tool consisted of 58 components (hereafter referred to as “environmental variables”) in the following categories: land use, green space and alley surface characteristics, alley use by people, property condition and waste. The observers examined the block street-front, alleyway and aerial photographs to score each environmental variable. Of the 58 variables, we eliminated 14 from

analysis that were redundant (i.e., other variables included the same information), less informative compared to other variables and/or lacked variability (i.e., more than 95% of the observations had the same score). Therefore, we considered 44 environmental variables for statistical modelling (Appendix B Table 1).

We acquired historical weather data (temperature and precipitation) from Environment Canada (http://climate.weather.gc.ca) for the dates of interest from the Vancouver Harbour CS site, which is near the trapping locations. We filled in missing data from the next closest sites (i.e., North Vancouver Wharves, then Vancouver International Airport locations). We constructed 32 weather variables organized in three sets to represent several time lags prior to when rats were captured. The first set of weather variables consisted of the mean of the mean daily temperatures (°C) in the 5, 10, 30, 60 and 90 days prior to capture. The second set consisted of the total precipitation (mm) in each of the 5, 10, 30, 60 and 90 days prior to capture. The third set was calculated by taking the mean of a week’s weather values for 7, 14, 30, 60 and 90 days prior to capture. This set included minimum, maximum and mean daily temperatures (°C), as well as mean daily precipitation (mm). An example of this type of variable is the mean of the total precipitation on days 54-60 prior to capture (Appendix B Table 2). We used relative trap success during the study period as a proxy for rat abundance by city block as described by Himsworth et al. (2014). In this previous study, rat abundance was calculated using the total trap effort (i.e., number of traps set multiplied by the number of trapping nights) with an adjustment to account for sprung traps (i.e., total trap effort minus 1/2 unit for each trap sprung by any cause) to calculate relative trap success.

3.3.3 Statistical Analysis

We fitted multi-level univariable logistic regression models with rat B. tribocorum infection status (positive or negative) as the outcome and environmental, weather, season and rat abundance as predictor variables. We included random effects to account for autocorrelation (i.e., clustering) among rats collected from the same city block (multiple rats were captured in each city block).

3.3.4 Variable Assessment and Construction

For data that were collected as categorical variables, we collapsed categories with fewer than 40 observations per category and/or those that only represented data from ≤ 2 blocks to construct either categorical or dichotomous variables. We individually assessed linearity between the log odds of B. tribocorum infection status against continuous independent variables (i.e., counted vs. scored environmental and weather variables) using lowess curves (i.e., locally weighed regression). If the variable was non-linear, we assessed the significance of a quadratic term and its main effect in a logistic regression model with a random intercept for block. If the quadratic term was significant in the model and a lowess curve revealed a quadratic relationship, then we used the quadratic term and its main effect for multivariable modelling. If it was not appropriate to model a quadratic relationship, we explored log transformation of the variable (i.e., natural log), replacing zero and negative numbers with half the value of the lowest positive observation, noting that few days were < 0°C). If we could not linearize the relationship, we then categorized the variable by quartiles. We categorized season by spring (March-June), summer (June-August), fall (September-November) and winter (December-February).

3.3.5 Statistical Modeling

Following any variable transformations, restructuring or the addition of a quadratic term, we fit each variable using multi-level logistic regression models with rat B. tribocorum infection status (positive or negative) as the outcome and city block as a random intercept. We considered variables with a statistically significant association with B. tribocorum for inclusion in two variable models. Due to the large number of constructed weather variables, we estimated the correlation between all weather variables that were significant with univariable modeling using Pearson’s correlation coefficients. If the correlation between variables was high (i.e., |ρ > 0.8|), we preferentially analyzed and presented the data of continuous vs. categorical variables for two-variable modeling. If both correlated variables were continuous, we selected the variable with the strongest univariable association for further analysis.

We constructed two variable models to assess the impact of potential confounding variables. We first constructed individual causal diagrams for all significant and non-correlated variables based

on the main causal diagram (Figure 3.1; Dohoo et al., 2009). Using the diagrams, we fitted two variable models for each potential confounding relationship and considered variables as confounders if they were non-intervening variables that resulted in ≥ 30% change to the coefficients when added to the model. Since B. tribocorum was significantly associated with sexual maturity and season in a previous study of this population (Himsworth et al., 2015), we also considered these variables in two variable models as appropriate based on causal diagrams. We decided a priori to test for biologically-plausible interactions (between temperature and precipitation variables within the same lag time frame and between sexual maturity and abundance). We used R (R Development Core Team, Vienna, Austria) for all statistical analyses except for graphs, log transformations, correlation analyses and when models failed to converge in R, which were fitted in Stata (Stata 14, College Station, Texas, USA). We set the significance level at alpha=0.05.

3.4 RESULTS The prevalence of B. tribocorum based on culture results was 25.7% (101/393; 95% Confidence Interval [CI]: 21.4%, 30.3%). These rats were captured from 32 city blocks. Based on univariable analysis, significant variables included five environmental variables, 14 weather variables and rat abundance. Due to the large number of constructed weather variables that were collinear and captured similar information, we eliminated ten weather variables from further analyses, bringing the total number of variables considered to 10 (Tables 3.1 and 3.2).

3.4.1 Environmental Variables

Four of the environmental risk factors identified with univariable analysis included features of the built environment, while only one related to human behavior (Table 3.1). The proportion of the block occupied by low-rise apartments was the only environmental variable that was not affected by any confounding relationships. The odds of a rat testing positive for B. tribocorum were significantly lower for rats in city blocks with one or more low-rise apartment buildings compared to blocks with none (OR = 0.20; 95% CI: 0.04-0.80; p=0.02; Table 3.3).

3.4.2 Weather and Season Variables

In univariable analyses, the odds of a rat being B. tribocorum-positive were significantly decreased in winter, spring and summer compared to fall (Himsworth et al., 2015). Significant

weather variables related to minimum temperatures on the 0-7, 24-30 and 84-90 days prior to when the rats were captured (Table 3.2). Among these, the odds of a rat testing positive for B. tribocorum were significantly higher with increased mean minimum temperatures (°C) on days 24-30 and days 84-90 prior to capture. Neither relationship was affected by confounding variables, although the mean minimum temperatures (°C) on days 24-30 became non-significant when season was included in the model (Table 3.3). The odds of a rat testing positive for B. tribocorum were significantly decreased when the mean minimum temperatures on the 0-7 days prior to capture were between 6.1-10.1°C compared to when temperatures were <3.6°C. This association remained significant after controlling for the confounding effect of season (Table 3.3). When controlling for season, there was a significant difference between when temperatures ranged from 3.6-6.0°C compared to 6.1-10.1°C (OR=9.11; 95% CI =1.62-51.37; p-value=0.01) and 6.1-10.1°C compared to ≥ 10.2°C (OR=0.14; 95% CI =0.03-0.64; p-value=0.01); there was no significant difference between 3.6-6.0°C compared to ≥ 10.2°C (OR=1.23; 95% CI =0.28- 5.40; p-value=0.78).

Only one precipitation variable was significant (mean of total precipitation on 0-7 days prior to capture; Table 3.2). The relationship was quadratic with an increased probability of a rat testing positive for B. tribocorum until precipitation reached approximately 5 mm; the probability decreased when precipitation was >6 mm (Figure 3.2). However, this association was no longer significant after controlling for the confounding effect of season (Table 3.3). There was no significant interaction between minimum temperature and mean precipitation in the seven days prior to capture (p=0.25).

3.4.3 Rat Abundance

There was a significant quadratic relationship between rat abundance (i.e., relative trap success) and testing positive for B. tribocorum. The probability of a rat being infected with B. tribocorum decreased until abundance reached approximately 0.4, and subsequently increased when abundance was more than 0.6 (Figure 3.3). This relationship was confounded by the proportion of the block occupied by housing over commercial buildings, human loitering, season and mean of minimum temperatures (°C) on days 84-90 prior to capture with each relationship becoming non-significant in the adjusted models. There was no significant interaction or confounding

relationships between B. tribocorum infection status and sexual maturity or abundance (Table 3.3).

3.5 DISCUSSION In the present study, univariable analyses revealed that several physical features of the urban environment, specifically building type and layout (which may create rat corridors among adjacent buildings) were associated with B. tribocorum infection in rats. However, only the proportion of the block occupied by low-rise apartments remained significant after controlling for season. When low-rise apartments were present in a given block, rats were significantly less likely to be infected with B. tribocorum. The reason(s) for this association is uncertain but may relate to unmeasured factors that influence pathogen transmission and maintenance, such as vector ecology. In particular, blocks lacking low-rise apartments may provide better habitats (e.g., rat burrows) that facilitate pathogen transmission by flea vectors. Although we did not analyze flea abundance in this study, there is evidence to suggest that ectoparasite abundance, including fleas, varies significantly between specific locations within the urban environment (Frye et al., 2015), which may have implications for B. tribocorum transmission among rats. For example, > 90% of Oriental rat fleas (Xenopsylla cheopis) recovered in a study of Norway rats in Manhattan, New York originated from one residential building (Frye et al., 2015). Perhaps specific areas sustain rat populations with higher Bartonella spp. prevalence because fleas may be able to maintain environmental transmission, even in the absence of rats (Gutiérrez et al., 2015).

We did not observe significant associations between B. tribocorum infection in rats and characteristics related to green space, garbage and property upkeep. Yet these factors may seem to be conducive to rat and flea survival. Higher Bartonella spp. prevalence among rodents, including Rattus sp., is associated with habitats containing abundant organic material (e.g., agricultural areas and fragmented green space) compared to human settlements (Hsieh et al., 2010; Jiyipong et al., 2015; Morand et al., 2015). It is possible that we measured the environmental variables examined in our study at too coarse of a scale to uncover meaningful associations with specific environmental features (Estrada-Peña et al., 2014). The protective

association with low-rise apartments requires further study, but hints that physical features of the urban environment may contribute to infection status in rats.

We found an association between B. tribocorum infection and high minimum temperatures for several time periods (approximately one week, one month and three months) before we captured the rats. This finding suggests that a baseline minimum temperature may be necessary for flea vector survival and B. tribocorum transmission among rats. Other studies of Bartonella spp. in rodents corroborate this observation. Prevalence tends to be highest in the summer and fall, when minimum temperatures are expected to be higher in the Northern Hemisphere (Gutiérrez et al., 2015). Our use of time lags prior to rat capture when considering the influence of weather is an important technique (Guan et al., 2009). Assuming chronic infections are the norm (Gutiérrez et al., 2015), it is more biologically meaningful to analyze time lags since the weather on the day we capture a rat is unlikely to reflect the weather when it became infected.

There were no significant associations with any precipitation variables in the present study. This contrasts with a study of rodents in a variety of habitats in Southeast Asia, in which a higher Bartonella sp. prevalence was observed in the wet vs. dry season (Jiyipong et al., 2015). Precipitation may be less important for Bartonella sp. transmission in an urban ecosystem, possibly related to year-round food availability that maintains host populations and minimizes changes in vector ecology (Bradley and Altizer, 2007). There is limited information about the effects of weather on Bartonella spp. in rodent hosts. However, the effects of weather on flea- transmitted Yersinia pestis, the causative agent of plague, in rodent hosts are better understood. In general, the risk of plague is increased following periods of high precipitation, which stimulates rodent and flea populations to increase in number (Gage and Kosoy, 2005). In contrast to our finding that increased minimum temperatures are associated with B. tribocorum infection, hot temperatures (>27°C) decrease the risk of plague via its negative effects on flea survival and subsequent blockage of the flea gastrointestinal tract by Y. pestis (Gage and Kosoy, 2005). Hot temperatures are not expected to have as profound an effect on Bartonella spp. transmission since flea-feces inoculation is the suspected mode of transmission rather than blocking within fleas, as occurs with Y. pestis (Eisen and Gage, 2012). In the context of Vancouver, British

Columbia, temperatures rarely exceed 25°C, so hot temperatures in this city are unlikely to negatively affect flea survival.

Although season was significant in univariable analyses in terms of B. tribocorum infection in rats, the strength of this finding is limited based on a single year study that included one replication of each season. Multi-year studies are needed to conclusively demonstrate seasonal effects. We included season as a potential confounding variable for weather; however, this may not have been appropriate since season and weather variables may be capturing similar information (i.e., winter is cooler than summer). Similarly, it was unexpected that season would have confounding relationships with permanent environmental variables like building type. The effect of season on the significance of the other two building-type variables (i.e., the proportion of the block occupied by grocery stores and housing over commercial buildings) likely resulted from the study design, in which blocks with these features were sampled in particular seasons.

The collective influences of season, weather and climate are an important consideration in the ecology of zoonotic diseases. In particular, understanding weather patterns associated with infected hosts may be useful for predictive modelling (Mills and Childs, 1998; Fisman, 2007). Knowing when hosts are most likely to be infected may be especially pertinent for coastal cities, such as Vancouver, where many of the detrimental effects of global climate change, including rising temperatures and flooding, are likely to be intensified (Lau et al., 2010). With climate change, higher minimum temperatures are expected to increase the prevalence of infected rats in this system, assuming extremely high temperatures do not negatively impact the flea or the bacterium. Climate change may also increase the risk of flooding, which may force rats to move into new areas (Gubler et al., 2001). Urban planners should consider the effects of climate change on rats and the ecology of their associated zoonotic pathogens.

We identified a non-linear relationship between abundance and infection status; but the association was not significant after controlling for confounding by building type, human use and weather variables, and season. Some studies suggest that host abundance is a key factor influencing Bartonella spp. prevalence (Telfer et al., 2006). Others observed a lack of association between host abundance and Bartonella spp. prevalence, which is consistent with the

present study. For instance, an experimental field study in which fenced enclosures excluded large mammals revealed an increased Bartonella spp. abundance due to increased numbers of small hosts (e.g., rodents) and vectors rather than a change to prevalence (Young et al., 2014). It is likely that the influence of host abundance is species- and ecosystem-specific. This relationship requires further research in urban ecosystems since it is hard to compare natural sites to the highly modified urban environment.

The present study has several limitations. The sample size was small (i.e., low replication number of city blocks with similar features) since environmental variables were initially collected to assess associations with rat abundance (Himsworth et al., 2014). Small sample size together with a large number of variables and confounding relationships prevented us from fitting a single multivariable model. We overcame this limitation by fitting two variable models based on causal diagrams. Since this study contained a single year and city, generalizability may be limited to elsewhere in the world. Future studies should examine rats and associated environment and weather characteristics over many years and in multiple locations.

In the absence of similar previous research, we undertook an exploratory analysis. Considering numerous variables, we investigated the impact of abiotic and biotic factors on B. tribocorum prevalence in rats. Given the logistical challenges of wildlife research, this study is a step toward uncovering environmental characteristics associated with infections in rats. We observed increased odds of B. tribocorum infection in rats associated with higher minimum temperatures and decreased odds of infection associated with low-rise apartment buildings. This study analyzed data across several levels of biological organization in an application of One Health problem solving (Estrada-Peña et al. 2014; Barrett and Bouley, 2015). Future studies could apply similar techniques to other pathogens in rats, as well as urban and non-urban wildlife species. Longitudinal and experimental studies that modify environmental features would be useful to elucidate causal relationships and mechanisms behind varying pathogen prevalence. Understanding how the environment and weather influences zoonotic pathogens in rats is important for creating active surveillance programs in rats and people. If specific environmental features are consistently associated with high pathogen prevalence in rats, urban planning and maintenance efforts could modify the environment. It is possible that changing urban

environments to reduce the prevalence of pathogens in rats may also reduce the risk of rat- associated zoonoses in people.

3.6 TABLES Table 3.1 Characteristics and univariable associations of significant microenvironmental features among 393 Norway rats (Rattus norvegicus) tested for Bartonella tribocorum in Vancouver, Canada.

Variable Description Sub-category Number Number Number of B. Univariable Associations of Rats of tribocorum Blocks positive rats in Odds 95% CI P Overall P for each category Ratio Categorical (%) (n=101) Variablesa Proportion of block occupied by 0 217 12 86 (39.6) ref low-rise apartments > 0 176 20 15 (8.5) 0.20 0.04-0.80 0.02

Proportion of block occupied by < 0.25 150 18 37 (24.7) ref < 0.01 housing over commercial buildingsb 0.25-0.50 160 11 13 (8.1) 0.12 0.05-1.07 0.06

> 0.5 83 3 51 (61.4) 12.61 1.88-156.35 0.01

Proportion of block occupied by 0 43 7 10 (23.3) ref 0.03 grocery stores < 0.25 195 20 19 (9.7) 0.34 0.05-1.97 0.21

≥ 0.25 155 5 72 (46.5) 3.64 0.49-38.50 0.20

Number of rat corridors at the 0 249 21 58 (23.3) ref alley face ≥ 1 144 11 43 (29.9) 6.25 1.44-39.51 0.02

Amount of human loitering None to light 200 19 74 (37.0) ref

Moderate to 193 13 27 (14.0) 0.17 0.03-0.79 0.03 heavy a Calculated with likelihood-ratio test b Refers to buildings in which the ground floor is dedicated to commercial businesses and upper floors are residential.

Table 3.2 Characteristics and univariable associations of significant season, abundance and weather variables among 393 Norway rats (Rattus norvegicus) tested for Bartonella tribocorum in Vancouver, Canada.

Rat Bartonella tribocorum status Univariable Associations

Variable Variable Format Number positive Number negative Odds 95% CI P Overall P (%) or (%) or Ratio for Median (IQR) Median (IQR) Categorical Variables Mean of minimum temperatures on 7 < 3.6°C 22 (22.0%) 78 (26.7%) ref 0.01 days prior to capture (categorical by 3.6-6.0°C 42 (41.6%) 48 (16.4%) 1.11 0.44-2.67 0.82 quartiles) 6.1-10.1°C 7 (7.0%) 93 (31.8%) 0.09 0.01-0.52 0.01 ≥ 10.2°C 30 (29.7%) 73 (25.0%) 0.84 0.12-4.17 0.84 Mean of minimum temperatures on continuous 10.61°C (7.98- 7.11°C (3.23- 1.23 1.08-1.44 < 0.01 days 24-30 prior to capture 12.66) 11.36) Mean of minimum temperatures on continuous 13.97°C (12.76- 5.03°C (3.11- 1.28 1.14-1.48 < 0.01 days 84-90 prior to capture 14.20) 13.36)

Mean of total precipitation (mm) on 7 main effect 4.00°C (3.01-5.29) 3.49°C (2.03- 1.72 1.11-2.80 0.02 days prior to capture 5.27) quadratic term 0.94 0.89-0.99 0.03 Season fall 75 (74.3%) 83 (28.4%) ref 0.02 winter 9 (8.9%) 80 (27.4%) 0.26 0.08-0.91 0.03 spring 12 (11.9%) 93 (31.8%) 0.13 0.02-0.54 < 0.01 summer 5 (5.0%) 36 (12.3%) 0.10 0.01-0.63 0.03 Rat abundance main effect 0.55 (0.21-0.86) 0.39 (0.18-0.54) 1.6x10-5 3.1x10-10-0.13 0.02 quadratic term 2.0x106 25.40-2.02 0.01 x1012

Table 3.3 Results from two-variable models assessing confounding based on causal diagrams among 393 Norway rats (Rattus norvegicus) tested for Bartonella tribocorum in Vancouver, Canada. Adjusted values in bold indicated a confounding relationship (>30% change in coefficients when confounder was added to the model).

Unadjusted Adjusted Confounders (bold) and independent variable(s) OR 95% CI P OR 95% CI P Environmental Proportion of block occupied by housing over Variables commerciala Human loitering 0.17 0.03-0.79 0.030 0.53 0.10-2.44 0.407 Rat abundance main effect < 0.01 0.00-0.14 0.017 0.37 <0.01-9.59 0.858 x104 quadratic term 2.0 x106 25-2.02x1012 0.012 1.46 <0.01- 0.959 5.85x106 Number of rat corridors 6.25 1.44-39.51 0.019 16.39 3.05-24.39 <0.001 Sexual maturity 5.18 1.66-22.83 0.011 4.84 1.56-21.28 0.014 Proportion of block occupied by grocery stores Human loitering 0.17 0.03-0.79 0.030 0.21 0.04-0.78 0.003 Rat abundance main effect < 0.01 0.00-0.14 0.017 < 0.01 <0.01-0.03 0.004 quadratic term 2.0 x106 25-2.02x1012 0.012 2.0 x106 29- 0.007 1.10x1010 Number of rat corridors 6.25 1.44-39.51 0.019 3.60 0.95-17.89 0.063 Sexual maturity 5.18 1.66-22.83 0.011 5.15 1.66, 22.60 0.011 Proportion of block occupied by low-rise apartments Human loitering 0.17 0.03-0.79 0.030 0.16 0.04-0.48 0.001 Rat abundance main effect < 0.01 0.00-0.14 0.017 < 0.01 <0.01-0.02 0.002 quadratic term 2.0 x106 25-2.02x1012 0.012 2.1x105 71-7.56 0.003 x109 Number of rat corridors 6.25 1.44-39.51 0.019 3.46 0.75-23.84 0.129 Sexual maturity 5.18 1.66-22.83 0.011 5.26 1.70-23.07 0.010 Number of rat corridors

Unadjusted Adjusted Confounders (bold) and independent variable(s) OR 95% CI P OR 95% CI P Rat abundance main effect < 0.01 0.00-0.14 0.017 0.01 <0.01- < 0.001 <0.01 quadratic term 2.0 x106 25-2.02x1012 0.012 2.01x107 7478- < 0.001 6.511011 Sexual maturity 5.18 1.66-22.83 0.011 5.12 1.65-22.54 0.012 Loitering Rat abundance main effect 0.17 0.03-0.79 0.030 0.31 0.05-1.46 0.131 quadratic term 2.0 x106 25-2.02x1012 0.012 5.9x104 0.32-5.48 0.065 x1010 Rat abundance Sexual maturity 5.18 1.66-22.83 0.011 5.20 1.67-22.90 0.011 Season & Season Weather Mean of minimum temperatures (°C) on 7 days prior to 0.015b 0.042b capture (categorical) < 3.6 ref 3.6-6.0 1.11 0.44-2.67 0.818 0.59c 0.19-1.65 0.330 6.1-10.1 0.09 0.01-0.52 0.013 0.09 0.01-0.54 0.014 ≥ 10.2 0.84 0.12-4.17 0.836 0.56 0.09-3.20 0.505 Mean of minimum temperatures (°C) on days 24-30 1.23 1.08-1.44 0.003 1.22 1.00-1.54 0.062 prior to capture Mean of minimum temperatures (°C) on days 84-90 1.28 1.34-1.48 < 0.001 1.27 1.03-1.60 0.029 prior to capture Mean of total precipitation (mm) on 7 days prior to 1.72 1.11-2.80 0.021 1.38 0.82-2.55 0.258 capture quadratic term 0.94 0.89-0.99 0.029 0.96 0.90-1.02 0.202 Rat abundance main effect 0.00 0.00-0.14 0.017 < 0.01 <0.01-5.03 0.115 quadratic term 2.13x106 23.05- 0.012 4932 0.21- 0.097 1.97x1011 1.15x108 Human loitering 0.17 0.03-0.79 0.030 0.20 0.05-0.62 0.007 Sexual maturity 5.18 1.66-22.83 0.011 5.20 1.67-22.97 0.011 Proportion of block occupied by housing over commercialc 56

Unadjusted Adjusted Confounders (bold) and independent variable(s) OR 95% CI P OR 95% CI P < 0.25 ref 0.25-0.50 0.12 0.05-1.07 0.062 0.33 0.08-1.26 0.090 > 0.5 12.61 1.88-156.35 0.014 4.89 0.89-48.89 0.081 Proportion of block occupied by grocery stores none ref < 0.25 0.34 0.05-1.97 0.208 0.39 0.08-1.84 0.206 ≥ 0.25 3.64 0.49-38.55 0.200 2.44 0.42-22.74 0.337 Proportion of block occupied by low-rise apartments none ref > 0 0.20 0.04-0.80 0.017 0.31 0.07-1.13 0.069 Mean of minimum temperatures (°C) on days 84-90 prior to capture Rat abundance main effect < 0.01 0.00-0.14 0.017 0.01 <0.01- 0.164 11.55 quadratic term 2.13x106 23.05- 0.012 807.79 0.05-3.49 0.145 1.97x1011 x107 Human loitering 0.17 0.03-0.79 0.030 0.30 0.08-0.92 0.040 Mean of total precipitation (mm) on 7 days prior to 1.72 1.11-2.80 0.021 1.52 1.01-2.41 0.056 capture quadratic term 0.94 0.89-0.99 0.029 0.96 0.91-1.00 0.075 Sexual maturity 5.18 1.66-22.83 0.011 5.19 1.68-22.78 0.011 Mean of minimum temperatures (°C) on days 24-30 prior to capture Rat abundance main effect < 0.01 0.00-0.14 0.017 < 0.01 <0.01-0.05 0.006 quadratic term 2.13x106 23.05- 0.012 357990 53.93-2.90 0.004 1.97x1011 x1010 Mean of total precipitation (mm) on 7 days prior to 1.72 1.11-2.80 0.021 1.41 0.89-2.31 0.153 capture quadratic term 0.94 0.89-0.99 0.029 0.96 0.91-1.01 0.150 Human loitering 0.17 0.03-0.79 0.030 0.18 1.09-1.42 0.016 Sexual maturity 5.18 1.66-22.83 0.011 5.31 1.71-23.31 0.010

Unadjusted Adjusted Confounders (bold) and independent variable(s) OR 95% CI P OR 95% CI P Mean of minimum temperatures (°C) on 7 days prior to capture Human loiteringd 0.17 0.04-0.85 0.030 0.31 0.64-1.49 0.145 Rat abundance main effect < 0.01 0.00-0.14 0.017 < 0.01 <0.01-0.07 0.007 quadratic term 2.13x106 23.05- 0.012 3.6 x105 42.27-3.36 0.005 1.97x1011 x1011 Mean of total precipitation (mm) on 7 days prior to 1.72 1.11-2.80 0.021 1.78 1.07-3.14 0.033 capture quadratic term 0.94 0.89-0.99 0.029 0.94 0.88-0.99 0.031

Mean of total precipitation (mm) on 7 days prior to capture Rat abundance main effect < 0.01 0.00-0.14 0.017 < 0.01 <0.01-0.65 0.038 quadratic term 2.13x106 23.05- 0.012 7.2x105 3.27-2.18 0.030 1.97x1011 x1012 Human loitering 0.17 0.03-0.79 0.030 0.13 0.02-0.38 0.010 a Refers to buildings in which the ground floor is dedicated to commercial businesses and upper floors are residential. b Significant variable that was affected by a confounding relationship and overall p-value for categorical variable. c Coefficient changed from a positive to negative value d There are slightly different values for human loitering between potential confounding variables due to model non-convergence in R for mean of minimum temperatures (°C) on 7 days prior to capture. Models that did not converge in R were run in Stata.

3.7 FIGURES

Figure 3.1 Causal diagram of sexual maturity, environmental and weather factors potentially affecting Bartonella tribocorum infection status in Norway rats.

Figure 3.2 Predicted probability curve demonstrating a quadratic relationship between Bartonella tribocorum infection status in Norway rats and mean precipitation in the 7 days prior to rat capture with 95% confidence intervals. This association was no longer significant after controlling for the confounding effect of season.

Figure 3.3 Predicted probability curve demonstrating a quadratic relationship between Bartonella tribocorum infection status in Norway rats and relative rat abundance with 95% confidence intervals. This relationship was no longer significant after controlling for the confounding effect of the proportion of the block occupied by housing over commercial buildings, human loitering, season and mean minimum temperatures (°C) on days 84-90 prior to capture.

3.8 REFERENCES

Ayral F, Artois J, Zilber A-L, Widén F, Pounder KC, Aubert D, Bicout DJ, Artois M. 2015. The relationship between socioeconomic indices and potentially zoonotic pathogens carried by wild Norway rats: a survey in Rhône, France (2010-2012). Epidemiology and Infection 143:586–599. Barrett MA, Bouley TA. 2015. Need for enhanced environmental representation in the implementation of One Health. EcoHealth 12:212–219. Bradley CA, Altizer S. 2007. Urbanization and the ecology of wildlife diseases. Trends in Ecology and Evolution 22:95–102. Buffet JP, Kosoy M, Vayssier-Taussat M. 2013. Natural history of Bartonella-infecting rodents in light of new knowledge on genomics, diversity and evolution. Future Microbiology 8:1117- 1128. Davis DE, Emlen JT, Stokes AW. 1948. Studies on home range in the Brown Rat. Journal of Mammalogy 29:207-225. Dohoo IR, Martin W, Stryhn HE. 2009. Veterinary Epidemiologic Research (2nd ed.). Charlottetown, PEI: VER Inc. Eisen RJ, Gage KL. 2012. Transmission of flea-borne zoonotic agents. Annual Review Entomology 57:61–82. Estrada-Peña A, Ostfeld RS, Peterson AT, Poulin R, de la Fuente J. 2014. Effects of environmental change on zoonotic disease risk: an ecological primer. Trends in Parasitology 30:205–214. Feng AYT, Himsworth CG. 2014. The secret life of the city rat: a review of the ecology of urban Norway and black rats (Rattus norvegicus and Rattus rattus). Urban Ecosystems 17:149–162. Fisman DN. 2007. Seasonality of infectious diseases. Annual Review of Public Health. 28:127- 143. Frye MJ, Firth C, Bhat M, Firth MA, Che X, Lee D, Williams SH, Lipkin WI. 2015. Preliminary survey of ectoparasites and associated pathogens from Norway Rats in New York City. Journal of Medical Entomology 52:253–259. Gage KL, Kosoy MY. 2005. Natural history of plague: perspectives from more than a century of research. Annual Review of Entomology 50:505–528. Guan P, Huang D, He M, Shen T, Guo J, Zhou B. 2009. Investigating the effects of climatic variables and reservoir on the incidence of hemorrhagic fever with renal syndrome in Huludao City, China: a 17-year data analysis based on structure equation model. BMC Infectious Diseases 9:109. Gubler DJ, Reiter P, Ebi KL, Yap W, Nasci R, Patz, J. A. 2001. Climate variability and change in the United States: potential impacts on vector- and rodent-borne diseases. Environmental Health Perspectives 109 (Suppl 2):223–233. 62

Gutiérrez R, Krasnov B, Morick D, Gottlieb Y, Khokhlova IS, Harrus S. 2015. Bartonella infection in rodents and their flea ectoparasites: an overview. Vector Borne Zoonotic Diseases, 15:27–39. Hayman DTS, McDonald KD, Kosoy MY. 2013. Evolutionary history of rat-borne Bartonella: the importance of commensal rats in the dissemination of bacterial infections globally. Ecology and Evolution 3:3195–3203. Heller R, Riegel P, Hansmann Y, Delacour G, Bermond D, Dehio C, Lamarque F, Monteil H, Chomel B, Piémont Y. 1998. Bartonella tribocorum sp. nov., a new Bartonella species isolated from the blood of wild rats. International Journal of Systematic Bacteriology 48:1333–1339. Himsworth CG, Bai Y, Kosoy MY, Wood H, DiBernardo A, Lindsay R, Bidulka J, Tang P, Jardine C, Patrick D. 2015. An investigation of Bartonella spp., Rickettsia typhi, and Seoul hantavirus in rats (Rattus spp.) from an inner-city neighborhood of Vancouver, Canada: is pathogen presence a reflection of global and local rat population structure? Vector Borne and Zoonotic Diseases 15:21–26. Himsworth CG, Parsons KL, Feng AYT, Kerr T, Jardine CM, Patrick DM. 2014. A mixed methods approach to exploring the relationship between Norway rat (Rattus norvegicus) abundance and features of the urban environment in an inner-city neighborhood of Vancouver, Canada. PLoS ONE 9:e97776. Himsworth CG, Parsons KL, Jardine C, Patrick DM. 2013. Rats, cities, people, and pathogens: a systematic review and narrative synthesis of literature regarding the ecology of rat-associated zoonoses in urban centers. Vector Borne Zoonotic Diseases 6:349–359. Hsieh JW, Tung KC, Chen W-C, Lin J-W, Chien L-J, Hsu YM, Wang HC, Chomel BB, Chang CC. 2010. Epidemiology of Bartonella infection in rodents and shrews in Taiwan. Zoonoses and Public Health 57:439–446. Jardine C, Waldner C, Wobeser G, Leighton FA. 2006. Demographic features of Bartonella infections in Richardson’s ground squirrels (Spermophilus richardsonii). Journal of Wildlife Diseases 42:739-749. Jiyipong T, Morand S, Jittapalapong S, Rolain JM. 2015. Bartonella spp. infections in rodents of Cambodia, Lao PDR, and Thailand: identifying risky habitats. Vector Borne Zoonotic Diseases 15:48–55. Kandelaki G, Malania L, Bai Y, Chakvetadze N, Katsitadze G, Imnadze P, Nelson CA, Harrus S, Kosoy MY. 2016. Human lymphadenopathy caused by ratborne Bartonella, Tbilisi, Georgia. Emerging Infectious Diseases 22:544-546. Kosoy M, Bai Y, Sheff K, Morway C, Baggett H, Maloney SA, Boonmar S, Bhengsri S, Dowell SF, Sitdhirasdr A, Lerdthusnee K. 2010. Identification of Bartonella infections in febrile human patients from Thailand and their potential animal reservoirs. American Journal of Tropical Medicine Hygiene 82:1140-1145.

Lau CL, Smythe LD, Craig SB, Weinstein P. 2010. Climate change, flooding, urbanisation and leptospirosis: fuelling the fire? Transactions of the Royal Society of Tropical Medicine and Hygiene 104:631-638. McKinney ML. 2006. Urbanization as a major cause of biotic homogenization. Biological Conservation 127:247–260. Meerburg BG, Singleton GR, Kijlstra A. 2009. Rodent-borne diseases and their risks for public health. Critical Reviews in Microbiology 35:221–270. Mills JN, Childs JE. 1998. Ecologic studies of rodent reservoirs: their relevance for human health. Emerging Infectious Diseases 4:529–537. Morand S, Bordes F, Blasdel K, Pilosof S, Cornu JF, Chairsiri K, Chaval Y, Cosson JF, Claude J, Feyfant T, Tran A. 2015. Assessing the distribution of disease-bearing rodents in human- modified tropical landscapes. Journal of Applied Ecology 52:784–794. Muñoz-Zanzi C, Mason M, Encina C, Gonzalez M, Berg S. 2014. Household characteristics associated with rodent presence and Leptospira infection in rural and urban communities from Southern Chile. American Journal of Tropical Medicine and Hygiene 90:497–506. Rothenburger JL, Himsworth CH, Nemeth NM, Pearl DL, Jardine CM. 2017. Environmental factors and zoonotic pathogen ecology in urban exploiter species. EcoHealth 14:630-641. Schulein R, Seubert A, Gille C, Lanz C, Hansmann Y, Piémont Y, Dehio C. 2001. Invasion and persistent intracellular colonization of erythrocytes. A unique parasitic strategy of the emerging pathogen Bartonella. Journal of Experimental Medicine 193:1077–1086. Telfer S, Begon M, Bennett M, Bown KJ, Burthe S, Lambin X, Telford G, Birtles R. 2006. Contrasting dynamics of Bartonella spp. in cyclic field vole populations: the impact of vector and host dynamics. Parasitology 134:413–13. Vayssier-Taussat M, Moutailler S, Féménia F, Raymond P, Croce O, La Scola B, Fournier PE, & Raoult D. 2016. Identification of novel zoonotic activity of Bartonella spp., France. Emerging Infectious Diseases 22:457–462.

Young HS, Dirzo R, Helgen KM, McCauley DJ, Billeter SA, Kosoy MY, Osikowicz LM, Salkeld DJ, Young TP, Dittmar K. 2014. Declines in large wildlife increase landscape-level prevalence of rodent-borne disease in Africa. Proceedings of the National Academy of Sciences of the United States of America 111:7036–7041.

CHAPTER 4

4. ENVIRONMENTAL FACTORS ASSOCIATED WITH THE CARRIAGE OF BACTERIAL PATHOGENS IN NORWAY RATS*

4.1 ABSTRACT Worldwide, Norway rats (Rattus norvegicus) carry a number of zoonotic pathogens. Many studies have identified rat-level risk factors for pathogen carriage. The objective of this study was to examine associations between abundance, microenvironmental and weather features and Clostridium difficile, antimicrobial resistant (AMR) Escherichia coli and methicillin-resistant Staphylococcus aureus (MRSA) carriage in urban rats. We assessed city blocks for rat abundance and 48 microenvironmental variables during a trap and removal study of rats, then constructed 32 time-lagged temperature and precipitation variables and fitted multivariable logistic regression models. The odds of C. difficile positivity were significantly lower when mean maximum temperatures were high (≥ 12.89°C) approximately 3 months before rat capture. Alley pavement condition was significantly associated with AMR E. coli. Rats captured when precipitation was low (< 49.40mm) in the 15 days before capture and those from blocks that contained food gardens and institutions had increased odds of testing positive for MRSA. Different factors were associated with each pathogen, which may reflect varying pathogen ecology including exposure and environmental survival. This study adds to the understanding how the microenvironment and weather impacts the epidemiology and ecology of zoonotic pathogens in urban ecosystems, which may be useful for surveillance and control activities.

* A version of this chapter was submitted to EcoHealth on September 6, 2017 as: Rothenburger JL, Himsworth CG, Nemeth NM, Pearl DL, Jardine CM. Environmental factors associated with the carriage of bacterial pathogens in Norway rats.

4.2 INTRODUCTION Cities are locations with radically altered environments that are non-permissive to most wild animals. Yet urban areas support large populations of so-called urban exploiter species through provision of food resources and human-built environmental niches (McKinney, 2006). Perhaps the most obtrusive urban exploiter species is the Norway rat (Rattus norvegicus). These globally- invasive rodents contaminate food, damage infrastructure and carry several zoonotic pathogens, including Yersinia pestis and Leptospira interrogans, the causative agents of plague and leptospirosis, respectively (Himsworth et al., 2013a; Feng and Himsworth, 2014). Rats also carry potentially zoonotic pathogens for which they are not considered the traditional host (e.g., Clostridium difficile, antimicrobial resistant [AMR] Escherichia coli and methicillin-resistant Staphylococcus aureus [MRSA]; Himsworth et al., 2014a; Himsworth et al., 2014c; Himsworth et al., 2015). In people and animals, MRSA causes serious opportunistic infections (Weese, 2010; Fitzgerald, 2012), while Clostridium difficile and E. coli are fecal bacteria that can cause enteric illness (Moono et al., 2016; Warriner et al., 2016). There is growing concern about antimicrobial resistance in E. coli, including as a potential source of antimicrobial-resistant genes to other pathogenic bacteria (Ewers et al., 2012). A wide range of healthy animal species, including rats, may carry all three of these bacteria (Weese, 2010; Fitzgerald, 2012; Hensgens et al., 2012; Szmolka and Nagy, 2013; Warriner et al., 2016). However, the role of rats in the epidemiology of these bacteria is incompletely understood.

In rat populations, rat-level risk factors affect the ecology of zoonotic pathogens. These include sex, nutritional condition, body mass, sexual maturity and bite wounds (Easterbrook et al., 2007; Himsworth et al., 2013a; Himsworth et al., 2015). Also, zoonotic pathogen characteristics (e.g., antimicrobial resistance patterns, genotype) and prevalence vary considerably among geographical locations, even at small scales (Yokoyama et al., 2007; Himsworth et al., 2015; Rothenburger et al., 2017a). The reason for this variation remains unknown; however, environmental characteristics including the rat’s microenvironment, rat abundance and weather may contribute (Himsworth et al., 2013a; Ayral et al., 2015; Rothenburger et al., 2017a). The microenvironment represents the small-scale habitat in which an organism lives that is part of the larger environment. Examples include anthropogenic activities (e.g., rat control, human use, maintenance), land use (e.g., green space, buildings), waste disposal (e.g., garbage, recycling)

and other environmental features within an area <30 m in diameter or the limits of a city block (their typical home range; Davis et al., 1948; Feng and Himsworth, 2014). There are associations between microenvironmental features and rat abundance, including signs of rodent infestation, number of household cats, land use, building condition and human refuse (Himsworth et al., 2014b; Muñoz-Zanzi et al., 2014). Yet the effect of rat abundance, weather and microenvironmental characteristics on pathogen prevalence are poorly understood (Rothenburger et al., 2017a), even though inclusion of these types of environmental characteristics are important to One Health problem solving approaches (Barrett and Bouley, 2015). In particular, information about weather patterns and microenvironments associated with infected rats could be used to create targeted surveillance and intervention protocols for both people and rats, which may also reduce the risk of rat-associated zoonoses in people.

The purpose of this study is to understand how abiotic (weather and urban microenvironmental features) and biotic (rat population density) factors affect C. difficile, AMR E. coli, and MRSA carriage in urban Norway rats.

4.3 METHODS

4.3.1 Study Design

This study occurred in the Downtown Eastside of Vancouver, British Columbia, Canada and is part of the Vancouver Rat Project (www.vancouverratproject.com). Himsworth et al. (2014b) describe study design and trapping protocol details. Rats were collected from September 2011- August 2012 from 43 back alleys that were randomly allocated to a three-week trapping period. For each euthanized rat, we performed systematic autopsy and tissue collection. We cultured oropharyngeal swabs for MRSA and colon contents for C. difficile and E. coli (Himsworth et al., 2013b; Himsworth et al., 2014c; Himsworth et al., 2015). Escherichia coli isolates were analyzed for resistance to 15 antimicrobial drugs using minimum inhibitory concentrations (Himsworth et al., 2015). We classified resistant E. coli as those designated as intermediate or resistant to >= 1 antimicrobial drug. We followed all applicable institutional and/or national guidelines for the care and use of animals and the University of British Columbia Animal Care Committee (A11- 0087) approved this study.

4.3.2 Microenvironmental and Weather Characteristics

Two observers used a systematic environmental observation tool to compile microenvironmental data for each city block during the rat collection period (Himsworth et al., 2014b). The tool’s 58 items (hereafter referred to as “microenvironmental variables”) belonged to waste, land use, alley use and property condition categories. The observers scored each item by examining the street front, alleyway and aerial photographs of each block. We eliminated variables from analysis because of redundancy (i.e., other variables include the same information) or lack of variation (i.e., < 5% of the observations had different scores), depending on the pathogen of interest. Thus, we considered up to 48 microenvironmental variables for statistical modelling of each pathogen (Appendix C Tables 1-6).

For the required dates related to the rat trapping period, we acquired historical temperature and precipitation data for the Vancouver Harbour CS site from Environment Canada (http://climate.weather.gc.ca) and added missing values from the next closest weather stations (North Vancouver Wharves then Vancouver International Airport). We combined data from time periods before trapping to create 32 weather variables, grouped into three sets (Appendix C Table 2, 4 & 6): 1) mean of the mean daily temperatures in the 5, 10, 30, 60 and 90 days before rat capture; 2) total precipitation in the 5, 10, 30, 60 and 90 days before rat capture; 3) the mean of a week’s daily weather values (mean precipitation and minimum, maximum and mean daily temperatures) for 7, 14, 30, 60 and 90 days before rat capture.

We used relative trap success in each block to approximate rat abundance, as calculated by Himsworth et al. (2014b). They adjusted the total trap effort (number of traps set multiplied by the number of trapping nights) for sprung traps (total trap effort minus 1/2 unit for each trap sprung by any cause).

4.3.3 Statistical Analyses

We fitted multi-level univariable logistic regression models with pathogen carriage (positive or negative) as the outcomes and microenvironmental, weather and rat abundance as predictor variables. We included a random intercept for city block to account for autocorrelation in pathogen status among rats in the same block.

For categorical microenvironmental data, we combined categories that represented data from ≤ 2 blocks and/or with < 40 observations per category to create either fewer categorical or dichotomous variables. Using lowess curves (i.e., locally weighted regression), we assessed linearity between the log odds of pathogen status against continuous independent variables (i.e., abundance, weather and counted vs. scored microenvironmental variables). For non-linear variables, we determined the significance of a quadratic term and its main effect in a multi-level logistic regression model (with a random effect for city block) and used these for multivariable modelling if it was significant and the lowess curve revealed a quadratic relationship. Otherwise, we assessed a log transformation of the variable (natural log with zero/negative values replaced by half the value of the lowest positive observation; few days were < 0°C). When we could not linearize the relationship, we categorized the variable. We categorized season as winter (December-February), spring (March-June), summer (June-August) and fall (September- November).

After we restructured, transformed or added a quadratic term, we fit each variable using multi- level logistic regression models with rat pathogen status as the outcome and random effects for city block. We considered variables with a statistically significant association (α=0.05) with the pathogen for inclusion in multivariable models. Because there were many constructed weather variables, we used Pearson’s and Spearman correlation coefficients to estimate the correlation between all weather variables that were significant with univariable modelling. When correlation was high (i.e., |ρ > 0.8|), we preferentially retained continuous vs. categorical variables for multivariable modeling except when categorical variables that captured narrower vs. broad timeframes. Between two correlated continuous variables, we chose the variable with the lowest p-value with univariable analysis.

For multivariable multi-level models, we included rat-level characteristics that were significantly associated with the pathogen in previous studies: C. difficile with body mass (Himsworth et al., 2014c), MRSA with body fat score (Himsworth et al., 2014a) and none with AMR E. coli (Himsworth et al., 2015). We created a full multivariable model with all variables that were significant with univariable analysis, as well as significant rat-level variables, and removed non-

significant variables using manual backwards selection. If the full model did not converge, we used step-wise manual forward selection by sequentially adding the most to least significant variables into the base model. We then removed variables that were not significant in the multivariable model, while retaining those that were. For both techniques, we used a likelihood ratio test for the overall significance of categorical variables before removal.

Using the base model, we then assessed confounding variables and interactions. We individually added each significant variable in the univariable analyses to the preliminary model to check for confounding. Confounding variables were non-intervening variables that resulted in ≥ 30% change to the coefficients of statistically significant variables when removed from the model. We decided to test for biologically-plausible interactions a priori (e.g., between precipitation and temperature within the same lag time frame). The final model for each pathogen contained variables that were statistically significant (α=0.05), part of a significant interaction term or acted as a confounding variable. We used variance estimates from each model to estimate the percentage variance at the two levels in the models (individual rat and city block) using the latent variable technique (Dohoo et al., 2009). To assess multi-level model fit, we examined the assumptions of normality using normal quantile plots and homoscedasticity of the best linear unbiased predictors (BLUPS) by plotting the BLUPS against the predicted log odds of the outcome. We examined Pearson residuals to identify outlying observations. If the random intercept was not significant and the variance component estimate was < 1 X 10-3, we fit an ordinary logistic regression model, used Pearson residuals and deviance residuals to identify potential outliers and assessed model fit using Hosmer-Lemeshow Goodness of Fit tests.

For most statistical analyses, we used R (R Development Core Team, Vienna, Austria). We used Stata (Stata 14, College Station, Texas, USA) for graphs, log transformations, correlation analyses, residuals, variance estimates and when models failed to converge in R.

4.4 RESULTS Tables 4.1-4.3 provide descriptive statistics and univariable associations for each pathogen and Table 4.4 provides multivariable model results.

4.4.1 Clostridium difficile

The prevalence of C. difficile was 12.8% (86/673; 95% Confidence Interval [CI]: 10.3%-15.5%); rats originated from 32 city blocks. With univariable analyses, three microenvironmental and 17 weather variables were significant (Appendix C Tables 1 and 2). Thirteen weather variables were collinear/captured similar information and were thus eliminated from further analyses, leaving seven variables for the multivariable model (Table 4.1). The final model, resulting from manual backwards elimination, included body mass and mean maximum daily temperatures (°C) on days 84-90 before capture (Table 4.4). The odds of a rat testing positive for C. difficile were significantly decreased with increasing body mass. The odds of a rat testing positive for C. difficile were significantly lower when the mean maximum temperatures were ≥ 12.89°C on days 84-90 before capture compared to <12.89°C.

4.4.2 Antimicrobial Resistant Escherichia coli

The prevalence of AMR E. coli was 12.3% (46/374; 95% CI: 9.1%-16.1%); rats originated from 32 city blocks. With univariable analyses, there were six significant microenvironmental variables (Table 4.2 and Appendix C Table 3) but no significant weather variables (Appendix C Tables 4). The final model that resulted from manual backwards elimination included the proportion of alley face in fair condition and the proportion of alley bordered by non-paved surfaces (Table 4.4). The odds of a rat testing positive for AMR E. coli were significantly lower when ≥ 25% of the alley face was in fair condition (moderate amount of superficially cracked pavement that is unlikely to permit rat burrowing) compared to when none of the alley face was in fair condition. There was no significant difference between when > 0 to 25% vs. ≥ 25% of the alley surface was in fair condition (OR=1.55; 95% CI=0.60-3.96; p=0.36). The odds of a rat testing positive for AMR E. coli were significantly greater when any portion of the alley was bordered by non-paved surfaces compared to when all of the alley surfaces were paved.

4.4.3 Methicillin-resistant Staphylococcus aureus

The prevalence of MRSA was 3.7% (22/590; 95% CI: 2.4%-5.6%); rats originated from 29 city blocks. With univariable analyses, six microenvironmental and six weather variables were significant (Table 4.3, Appendix C Table 5 & 6). Three weather variables were collinear/captured similar information and were thus eliminated from further analysis, leaving four variables for the multivariable model (Table 4.3). The final model that resulted from manual forward selection included the proportion of the block occupied by institutional parcels and food gardens, total precipitation in the 15 days before capture and rat internal fat score (Table 4.4). Rats with an internal fat score of 2 (ample internal fat) had significantly increased odds of testing positive for MRSA compared to rats with an internal fat score of 0. There was no difference between rats with a body condition score of 1 vs. 2 (OR=0.29; 95% CI=0.08-1.01; p=0.052). The odds of a rat testing positive for MRSA were significantly increased when the proportion of the block occupied by institutional parcels exceeded 25% compared to < 25%, when blocks contained any food gardens compared to none and when the total precipitation in the 15 days before capture was < 49.40 mm compared to 49.40-59.29 mm (OR=13.19; 95% CI=1.60-108.48; p=0.02), 59.30-79.59 mm (OR=4.28; 95% CI=1.18-15.55; p=0.03) and ≥ 79.60 mm.

4.4.4 Model Diagnostics

For C. difficile and MRSA, BLUPS were normally distributed with constant variance. Observations with large Pearson residuals were not recording errors and their removal did not impact the model. Most of the variation in pathogen occurrence was explained at the individual rat level; only 11% (95% CI: 3.8%-27.2%) of variance in the C. difficile model and 5% (95% CI: 0.02%-92.3%) of variance in the MRSA model was explained at the block level. For AMR E. coli, the variance component for block was small and the Akaike’s Information Criteria was equal in models with and without a random effect for block. Therefore, we examined model fit using a regular logistic regression model. Observations with large Pearson and deviance residuals were not recording errors. Based on the Hosmer-Lemshow Goodness of Fit test, we did not reject the null hypothesis that the model fit the data (degrees of freedom=2; χ2=0.14; p=0.93).

4.5 DISCUSSION The impact of abiotic (microenvironment and weather) and biotic (abundance) factors on the occurrence of C. difficile, AMR E. coli and MRSA in urban Norway rats varied by pathogen.

4.5.1 Clostridium difficile

Rats had significantly reduced odds of testing positive for C. difficile when maximum daily temperatures exceeded 12.89°C in days 84-90 before capture. A large number of weather variables were significant with univariable analyses but these were not retained during multivariable analyses. These results suggest that weather may be an important factor in the epidemiology of this pathogen in rats.

The association between lower odds of positivity and high temperatures was significant for the longest time lag we evaluated (84-90 days before capture). Time lags are a biologically meaningful approach since the weather on the day a rat is captured is unlikely to reflect the weather on the day it acquired the pathogen at an unknown prior time point. However, evidence from longitudinal studies of C. difficile in other wild species suggests that animals are only transiently colonized, with shedding lasting <5 weeks (Bondo et al., 2015). If this is also true for rats, then an 84-90 day time lag may be too large to reflect a reasonable period for C. difficile carriage.

Previous research of this sample identified a significant association between low body mass (proxy for young age) and positivity for C. difficile (Himsworth et al., 2014c), so we included this variable in our multivariable model. Observing this bacterium in younger rats is not surprising given the epidemiology of this pathogen in other animals (Moono et al., 2016). Since rats are typically weaned at 1 month of age (Feng and Himsworth, 2014), all rats had weaned at the time of trapping and young rats are predisposed to carrying C. difficile, 84-90 days may surpasses the lifespan of these rats at sampling. This variable may be acting as a proxy for the time of year when a cohort of rats were born that were more likely to acquire C. difficile. It is also possible that lower environmental survival of C. difficile may contribute to the time lag identified.

Clostridium difficile survives up to 5 months on dry surfaces (Kramer et al., 2006). Spores may remain viable despite prolonged storage at 4°C and through freeze-thaw cycles (Deng et al., 2015). In general, cooler temperatures facilitate bacterial pathogen survival in the environment (Kramer et al., 2006), but high temperatures may be detrimental to C. difficile spores (Usui et al., 2017). The capacity for environmental survival may explain the time lag between high maximum temperatures and fewer positive rats. It may take time following high maximum temperatures for C. difficile spores to reach sufficient numbers in the environment to support transmission.

4.5.2 Antimicrobial resistant Escherichia coli

The occurrence of antimicrobial resistant E. coli was significantly associated with microenvironmental variables related to dilapidated alley surface condition. Exposed soil due to absent or cracked pavement may promote AMR E. coli accumulation and survival in the environment. Non-paved surfaces permit digging and burrowing, which may facilitate exposure (Feng and Himsworth, 2014). Soil contains bacteria with antimicrobial resistance genotypes that may exchange resistance genes with E. coli (Forsberg et al., 2012). As well, soil contains antimicrobial substances including those produced by soil bacteria, which may exert selection pressure on AMR E. coli and promote the survival of resistant genotypes (Allen et al., 2010). There was no association with green spaces, so perhaps exposure to soil is not the only factor at play in this system. The absence of associations among weather variables is consistent with a study of AMR E. coli carriage by wild raccoons (Bondo et al., 2016).

These observations suggest that improved alley maintenance may be an important and attainable intervention. Cities that are concerned about AMR E. coli maintenance and transmission by rats could intervene by paving alley surfaces. Future studies should clarify the role of non-paved surfaces in AMR E. coli transmission among hosts in urban ecosystems by including environmental samples.

4.5.3 Methicillin-resistant Staphylococcus aureus

Increased odds of MRSA carriage in rats was significantly associated with city blocks with food gardens and higher proportions of institutional parcels. Institutions, including soup kitchens and homeless shelters, are areas where disadvantaged people congregate; these people may be at increased risk of carrying MRSA (Loewen, 2017). In a previous study, the most common 74

genotype of rat-associated MRSA was indistinguishable from MRSA isolated from people who lived in this area (Al-Rawahi, 2008; Himsworth et al., 2014a). Although the directionality of transmission is uncertain, it is likely that rats acquire human-associated MRSA from the environment. Both food gardens and institutions are places where people aggregate and may shed MRSA into the environment, supporting the idea that people may be the source (Dancer, 2008). MRSA can survive on surfaces for weeks and is resistant to desiccation (Beard-Pegler et al., 1988; Dancer, 2008). MRSA-colonized gardeners could shed the bacterium to plants, soil and other materials within the garden environment and MRSA is a known contaminant of fresh produce (Huijbers et al., 2015). There was no significant association between MRSA in rats and human loitering, suggesting that human behavior, including interactions with the environment, may be necessary for MRSA transmission among the environment, rats and people. Body fat score was also significantly associated with MRSA positivity in rats (Himsworth et al., 2014a). Perhaps the same mechanism that leads to MRSA exposure in rats also supports higher levels of nutrition. Fatter rats, through social dominance, may have increased exposure to MRSA by controlling access to food resources contaminated by people and may engage in more exploratory behaviors compared to subordinate rats (Feng and Himsworth, 2014; Himsworth et al., 2014a).

Rats had increased odds of testing positive for MRSA when there were lower amounts of precipitation in the 15 days before capture. Heavy rainfall may dilute or eliminate MRSA from the environment and may induce weather-associated changes in behavior (Feng and Himsworth, 2014). Rats that seek shelter may be less likely to acquire MRSA from their environment. Although human MRSA colonization and infections follow variable seasonal effects, with more skin and soft tissue infections in warmer seasons, limited data exist on associations between precipitation and human infections (Leekha et al., 2014).

4.5.4 Conclusions

Some factors anticipated to be associated with pathogens had no such association in the present study. For instance, none of the pathogens were significantly associated with rat abundance, suggesting that density-dependent transmission is not a factor in the ecology of these organisms in rats. However, the lack of association between pathogen carriage and abundance is not

surprising since all three pathogens are likely acquired from the environment (Himsworth et al., 2014a; Himsworth et al., 2014c; Himsworth et al., 2015) and transmission of environmental pathogens is usually not density-dependent (McCallum et al., 2001). It is also possible that rat densities were not sufficiently variable in the city blocks we studied to uncover significant associations. While our multi-level models accounted for autocorrelation among rats within the same block, our estimates of the percent of variation explained at the block-level were relatively imprecise after accounting for rat and block-level characteristics for MRSA and C. difficile. For AMR E. coli, two block-level factors explained essentially all the block-level variance in our model. Our results suggest that block-level vs. individual rat-level interventions may be promising at reducing pathogen prevalence, but our confidence in their impact remains uncertain.

Our approach allowed us to examine and compare multiple pathogens in the same system. Clostridium difficile, MRSA and Bartonella tribocorum (a vector-borne bacteria evaluated in this population; Rothenburger et. al., 2017b in review) all had significant associations with weather variables. There was no association between weather and AMR E. coli, even though it is spread via fecal-oral transmission like C. difficile. There are likely key differences in the ecology of AMR E. coli vs. these other pathogens that account for the differing associations with weather. Weather is key feature that may influence pathogen ecology in rats (Rothenburger et al., 2017a). For example, one study observed a time-lag effect of weather characteristics on a Seoul hantavirus-carrying index (i.e., combination of virus-carrying rate and rodent density) in rodents (primarily Norway rats) and subsequent rates of hemorrhagic fever with renal syndrome in people (Guan et al., 2009).

There were significant associations between AMR E. coli, MRSA and B. tribocorum and microenvironmental characteristics, although these differed by pathogen. The lack of association between microenvironmental features and C. difficile is unique among the pathogens examined in this system and may related to the propensity of this bacterium to form highly-resistant spores (Deng et al., 2015). It is also possible that the environmental features we assessed did not adequately capture sites of C. difficile exposure since underground (i.e., sewers) and indoor locations were excluded (Himsworth 2014b). Study results would have been strengthened by the

inclusion of a larger number of city blocks with similar features to increase the sample size, as well as multi-year and multi-city data collection.

This study and previous work on B. tribocorum (Rothenburger et. al., 2017b in review) suggest there is no single microenvironmental or weather factor associated with zoonotic pathogen carriage in rats. Differing pathogen ecology, including environmental survival and transmission routes, likely contribute to the lack of consistency among pathogens. This means that a single focus of control and/or surveillance is not feasible. However, this research is an important first step to understanding how environment and weather relate to zoonotic pathogen ecology in urban rats. By examining microenvironmental features and weather, we undertook an ‘upstream' and integrative approach to understand zoonotic pathogen ecology in rats. This method may be applicable to zoonotic pathogens in other hosts and systems.

4.6 TABLES Table 4.1 Characteristics and results of univariable associations of microenvironmental, weather and abundance variables considered for a multivariable model among 673 Norway rats (Rattus norvegicus) tested for Clostridium difficile in Vancouver, Canada.

Proportion of block Dichotomous 0 12 32 (8.9) 327 (91.1) refb occupied by low rise ≥ 0.25 20 54 (17.2) 260 (82.8) 2.23 1.04, 4.80 0.030 apartment buildings

Proportion of block Dichotomous 0 24 66 (11.2) 525 (88.8) ref occupied by ≥ 0.25 8 20 (24.4) 62 (75.6) 2.32 0.96, 5.28 0.043 industrial food establishments

Mean temperature Continuous NA NA 8.92 (6.50- 10.82 (6.88- 0.91 0.83, 0.99 0.022 (°C) in 15 days 11.12) 12.70) before capture

Mean of minimum Main effect NA NA 6.09 (3.71-8.47) 8.39 (4.41- 1.15 0.91, 1.45 0.224 temperatures (°C) on 11.54) days 24-30 before capture Quadratic NA NA NA NA 0.98 0.96, 1.00 0.020 term Total precipitation Categorical by < 192.8 7 6 (3.7) 156 (96.3) ref 0.033 (mm) in 60 days quartiles c > 192.8- 7 21 (13.1) 139 (86.9) 3.90 1.33, 12.41 0.013 before capture 216.79 78

Variable Variable Sub- Number Rat Clostridium difficile Status Univariable Associations Description Format category of Blocks Number Number Odds 95% CI P Overall P positive negative Ratio for (%)/Median (%)/Median Categorical (IQRa) (n=86) (IQR) (n=587) Variables 216.80- 10 29 (16.5) 147 (83.5) 4.88 1.69, 15.20 0.003 298.09 ≥ 298.10 8 30 (17.1) 145 (82.9) 4.87 1.62, 15.33 0.004

Mean of maximum Dichotomous < 12.89 21 59 (17.6) 277 (82.4) ref temperatures (°C) on ≥ 12.89 14 27 (8.0) 310 (92.0) 0.31 0.13, 0.66 0.003 days 84-90 before capture

a IQR = Interquartile Range; for continuous variables only b ref = referent category c This was the only precipitation variable with univariable analysis, so it was checked in multivariable modeling after mintemp30 (a collinear variable) was dropped during model building

Table 4.2 Characteristics and univariable associations of microenvironmental, weather and abundance variables considered for a multivariable model among 374 Norway rats (Rattus norvegicus) tested for antimicrobial-resistant Escherichia coli in Vancouver, Canada.

Proportion of block Dichotomous None 24 30 (10.1) 268 (89.9) refb occupied by green > 0 8 16 (21.1) 60 (78.9) 2.78 1.16, 7.45 0.022 space parcels

Proportion of block Dichotomous None 6 32 (10.5) 274 (89.5) ref occupied by grounds in > 0 26 14 (20.6) 54 (79.4) 2.68 1.10, 8.31 0.039 excellent condition

Proportion of alley face Categorical None 11 19 (22.6) 65 (77.4) ref 0.009 in fair condition < 0.25 5 7 (15.6) 38 (84.4) 0.61 0.19, 1.70 0.352

≥ 0.25 16 20 (8.2) 225 (91.8) 0.29 0.12, 0.62 0.002 Proportion of alley face Categorical None 2 1 (1.9) 53 (98.1) ref 0.003 in good condition < 0.25 17 25 (10.7) 208 (89.3) 6.37 1.29, 115.01 0.073

0.25-0.75 8 10 (21.7) 36 (78.3) 14.72 2.65, 276.08 0.012 ≥ 0.75 5 10 (24.4) 31 (75.6) 17.10 3.06, 321.29 0.008 Proportion of alley Dichotomous None 6 4 (4.1) 94 (95.9) ref bordered by non-paved > 0 26 42 (15.2) 234 (84.8) 4.22 1.59, 14.31 0.007 surface

Number of rat corridors Categorical None 21 34 (14.5) 201 (85.5) ref 0.040

Variable Description Variable Sub- Number Rat MRSA Status Univariable Associations Format category of Blocks Number Number Odds 95% CI P Overall P positive negative Ratio for (%)/Median (%)/Median categorical (IQRa) (IQR) variables (n=22) (n=568) at the alley face 1 5 8 (17.0) 39 (11.9) 1.21 0.49, 2.71 0.654 ≥ 2 6 4 (4.3) 88 (95.7) 0.27 0.08, 0.71 0.016

a IQR = Interquartile Range; for continuous variables only b ref = referent category

Table 4.3 Characteristics and results of univariable associations of microenvironmental, weather and abundance variables considered for a multivariable model among 590 Norway rats (Rattus norvegicus) tested for methicillin-resistant Staphylococcus aureus in Vancouver, Canada.

Proportion of block Dichotomous < 0.25 23 13 (2.6) 486 (97.4) refb occupied institutional parcels ≥ 0.25 6 9 (9.9) 82 (90.1) 5.17 1.23, 35.49 0.024

Proportion of block Dichotomous 0 21 12 (2.5) 464 (97.5) ref occupied by green space parcels > 0 8 10 (8.8) 104 (91.2) 4.75 1.12, 34.09 0.051

Proportion of block Dichotomous 0 22 13 (2.6) 483 (9.7) ref occupied by food gardens > 0 7 9 (9.6) 85 (90.4) 4.87 0.93, 34.51 0.054

Amount of overflowing Dichotomous None to 22 18 (6.1) 275 (93.9) ref garbage receptacles a little Some to 7 4 (1.3) 293 (98.7) 0.22 0.04, 1.08 0.034 a lot

Number of commercial Categorical ≤ 1 12 5 (4.5) 106 (95.5) ref 0.0338 recycling receptacles 2-3 8 13 (9.0) 131 (91.0) 2.10 0.60, 6.73 0.170 4 6 2 (1.0) 193 (99.0) 0.22 0.03, 1.04 0.073 ≥ 6 3 2 (1.4) 138 (98.6) 0.31 0.04, 1.45 0.163

Amount of human Categorical None to 16 3 (1.2) 243 (98.8) ref 0.019 loitering a little Some 6 9 (11.0) 73 (89.0) 10.96 2.24, 18.00 0.004 A lot 7 10 (3.8) 252 (96.2) 3.00 0.54, 18.00 0.176

Total precipitation Categorical by < 49.40 8 14 (10.4) 121 (89.6) ref <0.000 (mm) in 15 days before quartiles 49.40- 7 1 (0.6) 158 (99.4) 0.03 0.01, 0.22 0.003 capture 59.29 59.30- 12 4 (2.8) 137 (97.2) 0.16 0.04, 0.54 0.006 79.59 ≥ 79.60 2 3 (1.9) 152 (98.1) 0.11 0.02, 0.41 0.003

Mean of total Main effect NA NA 2.17 (1.31- 4.09 (2.46-6.23) 0.36 0.19, 0.67 0.001 precipitation (mm) on 3.66) 8-14 days before Quadratic NA NA NA NA 1.08 1.02, 1.15 0.007 capture term

Mean of total Categorical by < 1.01 5 1 (1.3) 147 (98.7) ref 0.008 precipitation (mm) on quartiles 1.01- 12 1 (0.7) 144 (99.3) 0.56 0.02, 10.49 0.683 84-90 days before 4.53 capture 4.54-7.19 6 4 (3.2) 121 (96.8) 2.63 0.30, 42.52 0.386 ≥ 7.20 6 13 (8.8) 156 (91.2) 8.12 1.18, 136.56 0.043

Mean of maximum Continuous NA NA 7.47 (7.19- 9.24 (7.39- 0.88 0.73, 0.99 0.050 temperatures (°C) on 8.38) 21.03) days 84-90 before capture a IQR = Interquartile Range; for continuous variables only b ref = referent category

Table 4.4 Results of multivariable mixed logistic regression models with random effect for block assessing associations among environmental and weather variables with Clostridium difficile, methicillin-resistant Staphylococcus aureus and antimicrobial resistant Escherichia coli carriage in Norway rats (Rattus norvegicus) captured in Vancouver Canada.

Unadjusted Adjusted Odds Independent Variables Sub-category Odds Ratio 95% CI P Ratio 95% CI P

Clostridium difficile Body mass (g)* 0.02 0.00, 0.23 0.002 0.03 0.00, 0.39 0.008 Mean of maximum temperatures on days 84- < 12.89 ref 90 before capture (°C) ≥ 12.89 0.31 0.13, 0.66 0.003 0.36 0.15, 0.77 0.011 AMR Escherichia coli Proportion of alley face in fair condition None ref > 0 - 0.25 0.61 0.19, 1.70 0.352 0.51 0.18, 1.30 0.175 ≥ 0.25 0.29 0.12, 0.62 0.002 0.33 0.16, 0.67 0.002 Proportion of alley bordered by non-paved None ref surface > 0 4.22 1.59, 14.31 0.007 3.79 1.44, 13.03 0.015 MRSA Internal fat scorea 0 (no fat) ref 1 (some fat) 1.47 0.33, 6.57 0.601 1.79 0.40, 8.15 0.436 2 (ample fat) 4.96 1.56, 19.46 0.011 5.96 1.78, 24.37 0.006 Proportion of block occupied institutional < 0.25 ref parcels ≥ 0.25 5.17 1.23, 35.49 0.024 6.30 1.84, 27.98 0.003 Proportion of block occupied by food gardens None ref > 0 4.87 0.93, 34.51 0.054 6.32 1.68, 27.20 0.002

Total precipitation (mm) in 15 days before ≥ 79.60 ref capture 59.30-79.59 1.43 0.28, 8.22 0.67 1.23 0.25, 6.82 0.795

Unadjusted Adjusted Odds Independent Variables Sub-category Odds Ratio 95% CI P Ratio 95% CI P 49.40-59.29 0.30 0.01, 2.86 0.33 0.44 0.02, 3.88 0.498 < 49.40 9.21 2.44, 49.46 0.003 5.39 1.53, 29.50 0.016 a Rat-level factors identified in previous studies that were included the models.

4.7 REFERENCES

Allen HK, Donato J, Wang HH, Cloud-Hansen KA, Davies J, Handelsman J. 2010. Call of the wild: antibiotic resistance genes in natural environments. Nature Reviews Microbiology 8:251– 259.

Al-Rawahi GN, Schreader AG, Porter SD, Roscoe DL, Gustafson R, Bryce EA. 2008. Methicillin-resistant Staphylococcus aureus nasal carriage among injection drug users: six years later. Journal of Clinical Microbiology 46:477–479.

Barrett MA, Bouley TA. 2015. Need for enhanced environmental representation in the implementation of One Health. EcoHealth 12:212–219.

Beard-Pegler MA, Stubbs E, Vickery AM. 1988. Observations on the resistance to drying of staphylococcal strains. Journal of Medical Microbiology 26:251–255.

Bondo KJ, Pearl DL, Janecko N, Boerlin P, Reid-Smith RJ, Parmley J, Jardine CM. 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:e0165303–16.

Bondo KJ, Weese JS, Rouseau J, Jardine CM. 2015. Longitudinal study of Clostridium difficile shedding in raccoons on swine farms and conservation areas in Ontario, Canada. BMC Veterinary Research 11:1039–7.

Dancer SJ. 2008. Importance of the environment in meticillin-resistant Staphylococcus aureus acquisition: the case for hospital cleaning. The Lancet Infectious Diseases 8:101–113.

Davis DE, Emlen JT, Stokes AW. 1948. Studies on home range in the brown rat. Journal of Mammalogy 29:207–225.

Deng K, Plaza-Garrido A, Torres JA, Paredes-Sabja D. 2015. Survival of Clostridium difficile spores at low temperatures. Food Microbiology 46:218–221.

Dohoo IR, Martin W, Stryhn, HE. 2009. Veterinary Epidemiologic Research (2nd ed.). Charlottetown, PEI: VER Inc.

Ewers C, Bethe A, Semmler T, Guenther S, Wieler LH. 2012. Extended-spectrum -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 18:646–655.

Feng AYT, Himsworth CG. 2014. The secret life of the city rat: a review of the ecology of urban Norway and black rats (Rattus norvegicus and Rattus rattus). Urban Ecosystems 17:149–162.

Fitzgerald JR. 2012. Livestock-associated Staphylococcus aureus: origin, evolution and public health threat. Trends in Microbiology 20:192–198.

Forsberg KJ, Reyes A, Wang B, Selleck EM, Sommer MOA, Dantas G. 2012. The shared antibiotic resistome of soil bacteria and human pathogens. Science 337:1107–1111.

Hensgens MPM, Keessen EC, Squire MM, Riley TV, Koene MGJ, de Boer E, Lipman LJA, Kuijper EJ. 2012. Clostridium difficile infection in the community: a zoonotic disease? Clinical Microbiology and Infection 18:635–645.

Himsworth CG, Parsons KL, Feng AYT, Kerr T, Jardine CM, Patrick DM. 2014b. A mixed methods approach to exploring the relationship between Norway rat (Rattus norvegicus) abundance and features of the urban environment in an inner-city neighborhood of Vancouver, Canada. PLoS ONE 9:e97776.

Himsworth CG, Parsons KL, Jardine C, Patrick DM. 2013a. Rats, cities, people, and pathogens: a systematic review and narrative synthesis of literature regarding the ecology of rat-associated zoonoses in urban centers. Vector Borne and Zoonotic Diseases 6:349–359.

Himsworth CG, Patrick DM, Mak S, Jardine CM, Tang P, Weese JS. 2014c. Carriage of Clostridium difficile by wild urban Norway rats (Rattus norvegicus) and black rats (Rattus rattus). Applied and Environmental Microbiology 80:1299–1305.

Himsworth CG, Patrick DM, Parsons K, Feng A, Weese JS. 2013b. Methicillin-resistant Staphylococcus pseudintermedius in rats. Emerging Infectious Diseases 19:169–170.

Himsworth CG, Zabek E, Desruisseau A, Parmley EJ, Reid-Smith R, Jardine CM, Tang P, Patrick DM. 2015. Prevalence and characteristics of Escherichia coli and Salmonella spp. in the

feces of wild urban Norway and black rats (Rattus norvegicus and Rattus rattus) from an inner- city neighborhood of Vancouver, Canada. Journal of Wildlife Diseases 51:589–600.

Huijbers PMC, Blaak H, de Jong MCM, Graat EAM, Vandenbroucke-Grauls CMJE, de Roda Husman AM. 2015. Role of the environment in the transmission of antimicrobial resistance to humans: a review. Environmental Science and Technology 49:11993–12004.

Kramer A, Schwebke I, Kampf G. 2006. How long do nosocomial pathogens persist on inanimate surfaces? A systematic review. BMC Infectious Diseases 6:318–8.

Leekha S, Diekema DJ, Perencevich EN. 2014. Seasonality of staphylococcal infections. Clinical Microbiology and Infection 18:927–933.

Loewen K, Schreiber Y, Kirlew M, Bocking N, Kelly L. 2017. Community-associated methicillin-resistant Staphylococcus aureus infection: Literature review and clinical update. Canadian Family Physician 63:512–520.

McCallum H, Barlow N, Hone J. 2001. How should pathogen transmission be modelled? Trends in Ecology and Evolution 16:295-300.

McKinney ML. 2006. Urbanization as a major cause of biotic homogenization. Biological Conservation 127:247–260.

Moono P, Foster NF, Hampson DJ, Knight DR, Bloomfield LE, Riley TV. 2016. Clostridium difficile infection in production animals and avian species: a review. Foodborne Pathogens and Disease 13:647–655.

Muñoz-Zanzi C, Mason M, Encina C, Gonzalez M, Berg S. 2014. Household characteristics associated with rodent presence and Leptospira infection in rural and urban communities from Southern Chile. American Journal of Tropical Medicine and Hygiene 90:497–506.

Psaroulaki A, Antoniou M, Toumazos P, Mazeris A, Ioannou I, Chochlakis D, Christophi N, Loukaides P, Patsias A, Moschandrea I, Tselentis Y. 2010. Rats as indicators of the presence and dispersal of six zoonotic microbial agents in Cyprus, an island ecosystem: a seroepidemiological study. Transactions of the Royal Society of Tropical Medicine and Hygiene 104:733–739.

Rothenburger JL, Himsworth CH, Nemeth NM, Pearl DL, Jardine CM. 2017a. Environmental factors and zoonotic pathogen ecology in urban exploiter species. EcoHealth 14:630-641.

Rothenburger JL, Himsworth CG, Nemeth NM, Pearl DL, Jardine CM. 2017b. Beyond abundance: how microenvironmental features and weather influence Bartonella tribocorum prevalence in urban Norway rats (Rattus norvegicus). Zoonoses and Public Health. Submitted June 29, 2017.

Szmolka A, Nagy B. 2013. Multidrug resistant commensal Escherichia coli in animals and its impact for public health. Frontiers in Microbiology 4. DOI: 10.3389/fmicb.2013.00258

Usui M, Kawakura M, Yoshizawa N, San LL, Nakajima C, Suzuki Y, Tamura Y. 2017. Survival and prevalence of Clostridium difficile in manure compost derived from pigs. Anaerobe 43:15– 20.

Warriner K, Xu C, Habash M, Sultan S, Weese SJ. 2016. Dissemination of Clostridium difficile in food and the environment: significant sources of C. difficile community-acquired infection? Journal of Applied Microbiology 1–12.

Weese JS. 2010. Methicillin-resistant Staphylococcus aureus in animals. ILAR Journal 51:233– 244.

Yokoyama E, Maruyama S, Kabeya H, Hara S, Sata S, Kuroki T, Yamamoto T. 2007. Prevalence and genetic properties of Salmonella enterica serovar Typhimurium definitive phage type 104 isolated from Rattus norvegicus and Rattus rattus house rats in Yokohama City, Japan. Applied and Environmental Microbiology 73:2624–2630.

CHAPTER 5

5. PATHOLOGY IN WILD NORWAY RATS (RATTUS NORVEGICUS)

5.1 ABSTRACT To achieve a modern understanding of the common and rare lesions that affect wild, urban Norway rats (Rattus norvegicus), we conducted a detailed pathology analysis of 672 rats from Vancouver, Canada. Grossly-evident lesions were relatively rare yet severe. The most common and medically-significant diseases were infectious and inflammatory (most frequently infections of the respiratory tract and those associated with bite wounds). We identified many microscopic lesions in a variety of organ systems. Among these, the most frequent lesions that could impact individual rat health included cardiomyopathy, pulmonary inducible bronchus-associated lymphoid tissue aggregates indicative of respiratory tract infections and thyroid follicular hyperplasia. Of the 21 bacterial species isolated from purulent lesions, the most frequent were Escherichia coli, Enterococcus sp. and Staphylococcus aureus. Rats were often infected with several invasive nematodes: Capillaria hepatica of the liver, Eucoleus sp. of the upper gastrointestinal tract and Trichosomoides crassicauda of the urinary bladder. Neoplastic, congenital and degenerative lesions were rare, which likely reflects their adverse effect on survival. Our results establish a baseline of expected disease in urban rats, which may have implications for urban rat and zoonotic pathogen ecology, as well as rat control in cities worldwide.

5.2 INTRODUCTION "Very little work has been conducted on the subject of normative biology or natural disease of wild rats. It is presumed that much of the data derived from laboratory rats can be extrapolated to wild rats." –Hulin and Quinn 2006

Laboratory rats are among the most ubiquitous animals used in biomedical research and their wild conspecifics are equally pervasive. Norway rats (Rattus norvegicus, hereafter, “rats”)

inhabit every continent on Earth with the exception of Antarctica, leading to their designation as one of the most successful free-ranging animal species. As one of six designated globally- invasive “urban exploiter species,” rats thrive in human-modified environments including cities, where they capitalize on the built environmental habitats with abundant year-round food sources (McKinney 2006). The presence of rats are associated with substantial infrastructure damage and social stigma (Wyman 1910; Feng and Himsworth 2014). They also destroy crops, food and agricultural infrastructure, leading to food insecurity and annual losses sufficient to feed millions of people (Wyman 1910; Singleton et al. 2007; Meerburg et al. 2009). Rats also threaten many populations of endangered wildlife species and carry numerous zoonotic pathogens, some of which cause significant illness in people (Towns et al. 2006; Himsworth et al. 2013b). These include Yersinia pestis, which causes plague, Seoul hantavirus, which causes hemorrhagic fever with renal syndrome and Leptospira spp., which causes Weil’s disease.

Despite their impact and interwoven association with human societies, the biology, ecology and diseases of wild rats are understudied. As the above quote from Hulin and Quinn (2006) emphasizes, equivalence between rats in the wild and in laboratories is assumed, but there are few data to support this assumption. Previous studies of disease in wild rats have been limited by small sample sizes and/or a narrow focus on specific pathogens or disease processes. A modern, holistic disease assessment of a large sample is lacking. Due to the wide distribution and thriving numbers of wild rats worldwide and their close association with humans, there is an urgent need for more information on wild rat diseases.

A nuanced understanding of any wild animal species’ life history should include its naturally- occurring diseases (Karesh and Cook 1995; Wobeser 2006). The ecology and biology of urban rats must be understood to enable integrated population control and management strategies (Singleton et al. 1999; Feng and Himsworth 2014). There are also applications to biomedical research and human health. Understanding spontaneous disease in urban rats may be crucial in the event of a bioterrorism/agroterrorism event, since rats have the capacity to carry weaponized pathogens (Lõhmus et al. 2013). Observing the presence or absence of certain diseases in wild rats may provide insight into diseases of laboratory rats used in biomedical research (e.g., cardiomyopathy; Chanut et al. 2013; Rothenburger et al. 2015b). Finally, naturally-occurring

pathogens and diseases in rats may influence zoonotic pathogen ecology and public health (Himsworth et al. 2013b; Vaumourin et al. 2015).

The aim of this study was to establish the spectrum and characteristics of natural pathology in urban rats. Specifically, our objective was to describe and summarize the macroscopic and microscopic lesions in an urban population of wild rats.

5.3 MATERIALS AND METHODS

5.3.1 Rat Collection

As part of a trap and removal study of urban rats and their associated zoonoses (www.vancouverratproject.com), we collected 685 Norway rats from an inner city neighborhood in Vancouver, British Columbia, Canada from September 2011-August 2012. Himsworth et al. (2014b) describes the trapping technique in detail. Briefly, we randomly assigned city blocks to a single two-week trapping period and actively trapped rats in back alleyways. A professional pest control company also collected rats within an international shipping port adjacent to the study area using snap-type lethal traps. Following general anesthesia with isoflurane and blood collection via cardiac puncture, we euthanized rats with intracardiac pentobarbital.

5.3.2 Autopsy and Sample Collection

We stored rats at -30°C, then thawed carcasses for a systematic autopsy and tissue collection protocol conducted at the Animal Health Centre, British Columbia Ministry of Agriculture, Abbotsford, British Columbia. We collected data on demographic characteristics (sex, sexual maturity, body mass and body condition score) either at the time of capture or during autopsy (Himsworth et al. 2014b).

5.3.3 Pathology Analyses

For each rat, we recorded macroscopic lesions. We trimmed tissues for histopathology from most rats with grossly-evident lesions except for six rats that were erroneously missed and those rats with only liver lesions that were consistent with Capillaria hepatica because infection with this parasite was known to be common in this population (Rothenburger et al. 2014a). For grossly- evident purulent lesions, we collected sterile swabs or tissue samples for routine aerobic and

anaerobic bacterial isolation and identification at the Animal Health Centre. We also randomly selected 341 rats from the 601 rats with no gross lesions for histopathology (Fig. 5.1). We typically examined the following tissues: adrenal gland, esophagus, intestines, liver, lungs, lymph node, kidney, pharynx, skeletal muscle, spleen, stomach, tongue, thyroid gland, trachea and urinary bladder, as well as any additional tissues in which gross lesions were observed. Not all tissues were available from each rat due to collection errors or autolysis (see Table 5.1 for number of rats examined for tissues with main microscopic lesions). For 25 randomly-selected rats, we examined a transverse section through the decalcified head including both eyes. Using light microscopy, we examined 4 !m-thick sections of formalin-fixed, paraffin-embedded tissues stained with hematoxylin and eosin. As required, additional sections were stained with Gram’s, Steiner’s silver and Masson’s trichrome stains to enhance diagnostic interpretation. We performed immunohistochemistry on thyroid tissues from six rats using monoclonal antibodies against thyroglobulin (Abcam, EPR9730/ab-156008, monoclonal rabbit anti-mouse, -rat, -human, 1:15,000) and calcitonin (Dako, A0576, polyclonal rabbit anti-human, 1:30,000).

5.3.4 Lesion Categorization

One observer (JR) initially screened all tissues. From these data, we developed a binary classification scheme to conservatively categorize tissues for the presence or absence of the most frequently observed microscopic lesions (Table 5.1). This scheme was based on observed lesions and numerous laboratory rat pathology references (Boorman 1990; Dungworth et al. 1992; Suckow et al. 2006; Percy and Barthold 2007; Renne et al. 2009; McInnes 2012). JR then re- examined each specimen to confirm the presence or absence of the lesions in the defined categories without reference to the autopsy results, and noted infrequent lesions that were not captured by the categories. Microscopic lesions that corresponded to the macroscopic lesions noted at autopsy were separately evaluated.

5.4 RESULTS Of the 685 Norway rats trapped, we excluded 13 from analysis due to advanced autolysis and/or incomplete data that resulted from data collection or entry errors (Fig. 5.1). Among the remaining 672 rats, 374 (55.7%) were male, 290 (43.2%) were female and sex could not be determined for eight rats. Most rats were sexually mature (n=390; 57.9%; open vaginal orifices in females and scrotal testes in males); maturity could not be determined for 64 rats. The median 93

weight was 145.4 g (range = 20.0–466.2 g). There were macroscopic lesions in 71 rats (10.6%), of which six rats were not examined histologically due to human errors.

Table 5.1 summarizes and describes the most frequent histopathological findings in the 406 rats examined for microscopic lesions. Summaries and descriptions of neoplastic and proliferative lesions, rare lesions, as well as parasitic and bacterial infections are provided in Tables 5.2 to 5.5, respectively.

5.4.1 Cardiovascular Lesions

The most common cardiovascular lesion was cardiomyopathy (31.5%) consisting of mononuclear cell myocarditis, fibrosis and/or myocardial degeneration (Fig. 5.2). Myocardial mineralization, which was not typically associated with cardiomyopathy, was evident in 8.6% of rats and 4.5% of rats had right ventricular hypertrophy. Vascular lesions were most prominent in the lungs and included pulmonary arteriolar medial hypertrophy in 22.3% of rats (Fig. 5.3) and intimal mineralization in pulmonary blood vessels in 15.3% of rats. Rothenburger et al. (2015b) described statistical associations among lesions defined as cardiomyopathy using a subset of rats in the current study.

5.4.2 Digestive Tract Lesions

A capillarid nematode, Eucoleus sp., was in the upper gastrointestinal tract of many rats (41.1%) and was associated with hyperkeratosis, mucosal hyperplasia and submucosal inflammation in the forestomach, as previously described in a subset of rats that are included in the current study (Fig. 5.4; Rothenburger et al. 2014b). Rarely, rats were co-infected with Eucoleus sp. and Gongylonema neoplasticum (Brown and Hardisty 1990). Nematode and cestodes (species unidentified) were within the small intestine lumens in 41.9% and 3.0% of rats, respectively, and a few rats were infected with coccidian protozoa (6.6%). Intestinal contents of at least two rats contained bright turquoise-green material consistent with “bait” containing anticoagulant rodenticide (Supplemental Fig. S1; see section on hemorrhage in “Other Lesions and General Conditions” below).

Dental disease was infrequent and most often included mild lower incisor overgrowth or other malocclusion (Supplemental Fig. S2). In the most severe case of malocclusion, multiple missing 94

teeth created diastemata and alveolar bones were affected by suppurative osteomyelitis and/or fracture. Several rats were affected by variably severe stomatitis, glossitis, pharyngitis, and/or esophagitis. Rare lesions included a focal papilloma on the tongue (Table 5.2), a 5mm-long Meckel's diverticulum and a double esophagus, each in one rat (Table 5.3; 5.5).

The most common hepatic lesions associated with Capillaria hepatica (syn. Calodium hepatium) infection in 36.0% of rats, as described in a previous study (Table 5.4; Fig. 5.6; Supplemental Fig. S3; Rothenburger et al. 2014a). Rarely, peripancreatic fat was saponified and/or infiltrated by neutrophils and macrophages.

5.4.3 Endocrine Lesions

Thyroid glands of 50.9% of rats had diffuse follicular hyperplasia (i.e., hyperplastic goiter) consistent with dietary iodine deficiency (Fig. 5.7). Follicles were lined by tall cuboidal to columnar epithelial cells. Follicular lumens were not evident or barely appreciable and lacked colloid. This lesion was differentiated from C-cell hyperplasia using immunohistochemistry for calcitonin and thyroglobulin in six exemplar rats. Immunoreactivity to thyroglobulin was weak, diffuse and intracytoplasmic in the majority of hyperplastic cells (consistent with follicular cells). Calcitonin immunoreactivity was variably strong and punctate to weak and diffuse within the cytoplasm of interstitial cells located centrally within the thyroid gland (consistent with C-cells). There was mineralization of residual colloid in 47.1% of rats. Adrenal glands of several rats were subjectively enlarged (suspected stress-related adrenomegaly).

5.4.4 Hemolymphatic Lesions

Generalized or locally-enlarged lymph nodes were common (exact number of affected rats not recorded) and were typically associated with skin wounds or abscesses. Granulocytic lymphadenitis was observed in 3.2% of rats.

5.4.5 Integumentary Lesions

Skin wounds were the most frequent lesions (33.1%) and likely resulted from conspecific bites. The mean number of wounds was 0.6 per rat (range=0-15). Wounds were most often over the caudal back and base of the tail (Supplemental Fig. S4); facial and tail wounds were less frequent. Wounds ranged from focal puncture wounds (1 mm) to large (≤2 cm) and sharply demarcated 95

lacerations. Chronicity ranged from acute to completely healed with scar tissue. In some cases, bite wounds were associated with severe myositis, cellulitis, and/or osteomyelitis, particularly those occurring on the dorsal lumbar area and hind limbs. Some rats with bite wounds (and some without) had subcutaneous abscesses along the ventral head and neck, and lateral thorax that were often associated with local or generalized lymph node enlargement. A previous study in these rats describes the bacterial culture results from infected bite wounds (Himsworth et al. 2014c).

Many rats were affected by bilaterally-symmetrical alopecia over the dorsal lumbar region (exact number not recorded; Supplemental Figs. S4 and S5). In severe cases, alopecia was accompanied by yellow-tan epidermal crusting and extended cranially to the neck and down the thighs. Histologically, these areas featured decreased numbers of follicles, mild mononuclear cell inflammation, sebaceous gland atrophy, epidermal hyperplasia and orthokeratotic hyperkeratosis. In severe cases, inflammation extended to the deep dermis and into the panniculus muscle. Many rats were infested by ectoparasites—fleas (Nosopsyllus fasciatus), mites (Ornithonyssus bacoti) and lice (Polyplax spinulosa). Two rats were affected by proliferative dermatitis associated with the ear mange mite, Notoedres muris (Table 5.4; Supplemental Fig. S6; Anholt 2014). Two rats had a proliferative skin lesion consistent with poxvirus infection (Table 5.2).

5.4.6 Musculoskeletal, Adipose and Connective Tissue Lesions

Occasionally, rats were affected by neutrophilic to lymphoplasmacytic myositis in one or more of the following skeletal muscle groups: tongue, larynx, diaphragm, triceps or quadriceps. Some rats had simultaneous infections or traumatic injuries. For example, one rat with pyelonephritis and another with chronic traumatic lesions on the distal tail had multifocal lymphoplasmacytic myositis of the tongue, diaphragm, triceps and quadriceps. Glossal myositis was often perineural. Severe and deep skin wounds were often associated with suppurative myositis and/or abscesses in the subjacent musculature.

5.4.7 Respiratory Lesions

The most common histological lesion in the upper respiratory tract was lymphoplasmacytic or neutrophilic tracheitis (51.6%; Fig. 5.8), followed by lymphoplasmacytic rhinitis (40%) and 96

lymphoplasmacytic or neutrophilic laryngitis/epiglossitis (39.1%). Cilia-associated respiratory bacillus (CAR bacillus), which is characterized histologically as mats of filamentous bacteria among respiratory epithelium lining the nasal cavity or trachea, was present in 24.4% of rats. The most frequent histological lesions in the lower respiratory tract were cuffs of lymphocytes and plasma cells that surrounded bronchioles and/or pulmonary blood vessels (consistent with inducible bronchus-associated lymphoid tissue [iBALT]) in 67.0% of rats (Figs. 5.9 and 5.10). Less frequently, a mixed inflammatory cell population that included granulocytes, lymphocytes and plasma cells surrounded pulmonary blood vessels (19.8%; Fig. 5.11).

Rothenburger et al. (2015a) described grossly-evident respiratory pathology for all rats in the current study, including abscesses/bronchiectasis and pulmonary neoplasms (Table 5.2), and rare microscopic lesions in a subsample of rats in the current study.

4.8 Reproductive Lesions

Lesions of the male reproductive tract were rare and are listed in Table 5.3. Lesions of the female reproductive tract were also infrequent (exact number not recorded unless explicitly stated). Mucohemorrhagic vaginal discharge was present in a few sexually-mature rats that were pregnant or lactating, or had lesions consistent with anticoagulant rodenticide toxicity or mild neutrophilic endometritis. A few had clear to hemorrhagic fluid in the uterine lumen. Endometritis of varying severity was present in at least seven rats and was typically granulocytic. Placental scars from previous pregnancies were frequently observed (Supplemental Fig. S7). Two female rats with no reproductive tract lesions were pregnant in only one uterine horn.

5.4.9 Urinary Lesions

The most common urinary tract lesions were intratubular renal crystals present in 33.8% of rats. Lymphoplasmacytic interstitial nephritis occurred in 29.9% of rats and was typically mild and perivascular in distribution (Fig. 5.11). Interstitial lymphoplasmacytic pyelitis immediately adjacent to the renal pelvis occurred in 33.3%.

The nematode parasite Trichosomoides crassicauda was in the lumen or superficial mucosa of the urinary bladder, or less frequently, of the renal pelvis, of 30.4% rats (Fig. 5.12). In a few infected rats, the submucosa and/or muscularis also contained mild lymphoplasmacytic 97

inflammation. Bladders of nine male rats (7.6%; 9/119) contained agonal proteinaceous plugs (Supplemental Fig S8).

5.4.10 Other Lesions and General Conditions

Hemorrhages consistent with anticoagulant rodenticide toxicity were in 8.8% (59/672) of rats (Supplemental Figs. S1, S9-S12). Most often, hemorrhages were within the subcutaneous tissues. Periarticular, parenchymal (lung, intestine, testis/scrotum) and intracranial hemorrhages, as well as hemothorax and hemoabdomen, were less frequent. Hemorrhagic discharge from penis or vagina, epistaxis and retroperitoneal hemorrhages were rare. Lungs of two rats with pulmonary hemorrhages contained hemosiderin-laden macrophages, suggesting the hemorrhagic events were chronic but ongoing. In at least two rats, stomach and intestines contained bright turquoise green material that resembled anticoagulant rodenticide “bait” (Supplemental Fig. S1).

Eight rats were affected by severe, chronic traumatic injuries. For example, a large mediastinal abscess in the thorax of one rat contained a metal airgun pellet. There were associated scapulae abscesses (presumably from the pellet entrance wound; Rothenburger et al. 2015a). The skin of one rat contained a sharply-demarcated, deep, linear wound with associated crusting and alopecia that surrounded the neck and continued caudally to the right front leg (Supplemental Fig. S13). One rat with debris accumulation and alopecia on the caudal portion of the body had proliferation of the dorsal spinous processes of the cranial thorax and thickening of the spine at the base of the tail, collectively suggestive of paralysis. Five rats had femoral fractures or amputations affecting the right hind limb. The above lesions were frequently associated with wounds, suppurative cellulitis and osteomyelitis. Tail amputations were frequent (exact number not recorded; Supplemental Fig. S14).

Acute traumatic injuries in at least 36 rats were likely related to the live trapping activities we undertook to collect rats: hyperemia of the distal extremities (Supplemental Fig. S15) and less often, nail devolving, tail tip amputations and nostril lacerations. The whiskers, facial fur and ears of two rats were singed (suspected as malicious burning while rats were in the trap).

Additional rare conditions are described in Table 5.3 and depicted in Supplemental Figs. S16- S19.

5.5 DISCUSSION The objective of this study was to describe and summarize gross and microscopic lesions in a population of wild, urban Norway rats. A range of infectious and non-infectious lesions occurred in a variety of organ systems and these data help establish a baseline of both common and rare lesions. Baseline data are especially important since studies of wild rats often focus on specific systems and pathogens of interest or include relatively small sample sizes (Gregson et al. 1979; Sterling and Thiermann 1981; MacKenzie et al. 1981; Lewis 1982; Brogden et al. 1993; Webster and Macdonald 1995; Giusti et al. 1998; Kakrada et al. 2002; Easterbrook et al. 2008). Many historical studies of wild rat disease describe obvious and common gross lesions (e.g., chronic lung disease and Capillaria hepatica infection of the liver) rather than undertaking a systematic histopathological evaluation of major organ systems (Davis 1951; Schiller 1956). Investigators during historical plague eradication efforts examined thousands of rats; but non-plague- associated pathology was of low priority and thus, was not documented in detail (Wyman 1910). Further, lesion descriptions and ancillary testing lacked the rigor of modern standards, and pathogen nomenclature and knowledge of disease pathogenesis have since advanced considerably.

Our sampling technique may have impacted the estimated prevalence of histological lesions since we sampled most rats with grossly-evident lesions and roughly half of rats with no grossly- evident lesions. If the presence of macroscopic lesions was related to the major histological lesions, then the prevalence estimates are likely artificially high. Similarly, we did not account for clustering at the block level, which could have impacted our estimated prevalences and confidence intervals. Rats originating from the same city block were probably more similar to each other than rats in surrounding blocks, creating the potential for lesions to cluster by geographical location. Accounting for geographical clustering by block may have also reduced the estimated prevalences we report here. Prevalence and confidence interval estimates should be only considered in the context of the rats examined; they provide a relative ranking various lesion frequencies from the sampled population.

5.5.1 Lesions Associated with Bacterial Infections

Our results indicate that a wide range of bacterial pathogens and associated diseases frequently affected wild rats. The most common bacterium isolated from purulent lesions was Escherichia coli, followed by Enterococcus spp. and Staphylococcus spp. The latter was frequently isolated from abscesses in our study, a finding that is consistent with rats examined during the San Francisco, California, USA plague control effort and a survey of infections in New Zealand rats (Wyman 1910; Carter and Cordes 1980). Other studies have also identified a diverse array of bacteria in wild rats associated with infections (mainly abscesses) including Bordetella bronchiseptica and Corynebacterium sp. (Carter and Cordes 1980). Conspecific bite wounds were among the most frequent lesions associated with bacterial infections in this study (see “Traumatic and Anthropogenic Lesions” below). In total, 4% of rats (28/672) had abscesses in a variety of systems, which may have resulted from bacteria inoculated into bite wounds.

The other most common bacterial disease in this study was respiratory tract lesions associated with M. pulmonis and CAR bacillus. These pathogens are associated with perivascular and/or peribronchiolar lymphoplasmacytic cuffs and tracheitis in rats (Rothenburger et al. 2015a). These infections were not microscopically distinguishable and many rats were co-infected in the previous study. Easterbrook et al. (2008) identified similar co-infections in wild rats, although they did not describe associated lesions. In laboratory and pet rats, respiratory infections with these pathogens can progress to severe bronchiectasis, bronchopneumonia and pulmonary abscesses (Giddens et al. 1971; Davis and Cassell 1982; Ganaway et al. 1985; Weisbroth 1996; Kling 2011). We observed these severe lesions in a small number of rats, which is compatible with this disease progression. Our results are also in agreement with other studies of wild rats, which document severe respiratory disease in the form of bronchiectasis/pulmonary abscesses (Habermann et al. 1954; Laurain 1955; Schiller 1956; Kilham et al. 1962; Owen 1976; MacKenzie et al. 1981; Brogden et al. 1993). Furthermore, Calhoun (1963) observed clinical signs attributable to respiratory disease in wild rats: wheezing, sneezing, coughing and increased respiratory effort. Of all the lesions we observed, those associated with the respiratory system likely had the most impact on rat health.

100

Bacteria also infected the reproductive and urinary tracts. Purulent lesions in the seminal vesicles and uterus were rare occurrences in other studies of wild rats (bacterial culture results not given; Wyman 1910; Balfour 1922). Among our results, there was an unusual Salmonella enteritidis- associated metritis, perhaps from an ascending infection, although Salmonella spp. was only isolated from the colon contents of 3/633 (0.5%) rats in this population (this rat was not tested for fecal Salmonella spp. carriage; Himsworth et al. 2015). Within the urinary system were rare cases of pyelonephritis and perirenal abscesses. These lesions were also rare in other wild rat studies (Gray et al. 1974; Tucunduva de Faria et al. 2007).

No rats had lesions consistent with plague (e.g., subcutaneous hyperemia, multifocal hepatic necrosis, splenomegaly and bilateral pleural effusions; Wyman 1910). We did observe abscessed lymph nodes, but these were purulent rather than caseous, as is typical of plague in rats. Albeit rarely, rats may be infected Yersinia pestis without lesions; however, bacteriological results yielded no evidence of Y. pestis infection, indicating that rats in Vancouver are free of plague during the time frame when rats were collected. However, as a major port city, there is the ongoing potential for plague introductions (Gage and Kosoy 2005).

5.5.2 Parasitic Lesions

Parasites and associated lesions were common. These included nematode infections in the liver (C. hepatica), upper gastrointestinal system (Eucoleus sp.), urinary bladder and kidney (T. crassicauda) and intestines; ectoparasites on the skin (N. muris, N. fasciatus, O. bacoti and P. spinulosa); and unidentified cestodes and suspected coccidian protozoa in the intestines. Of the most common parasites, Eucoleus sp., and C. hepatica were consistently associated with host responses to infection. Rats infected with Eucoleus sp. had hyperkeratosis and mucosal hyperplasia in the non-glandular stomach, while those infected with C. hepatica had fibrosis and hepatic parenchyma destruction (Rothenburger et al. 2014a; 2014b). In general, C. hepatica infection is not associated with clinical signs of hepatic dysfunction or mortality, even when severe (Farhang-Azad 1977). And although the biology of Eucoleus sp. is less understood than C. hepatica, it is similarly unlikely to directly cause clinical disease or mortality. But in addition to these lesions, nematodes may also negatively affect rat health through their influence on the

101

immune system and response to microparasite infections (Garza-Cuartero et al. 2014). Other studies have identified a similarly diverse array of parasites in wild rats (Milazzo et al. 2010; McGarry et al. 2015). However, histopathology is required to detect the full array and distribution of parasites along with associated lesions (e.g., Eucoleus sp. in the oral cavity and esophagus; T. crassicauda in the kidney).

We identified T. crassicauda in 30% of rats, which is in agreement with previous studies in North America in which 25%-56% of rats were infected (Harkema 1936; Firlotte 1948). We most often observed adults and/or eggs in the urinary bladder—they were only rarely in the renal pelvis. The rarity of renal infection differs from Gray et al. (1974), who observed either adults or eggs in approximately 25% of rat kidneys examined. This discrepancy may reflects our tissue sampling methods. The prevalence of urinary and enteric parasites may be underestimated in our study due to freeze-thaw of carcasses and the use of histopathology rather than direct examination. Both factors may decrease parasite detection (Schiller 1956).

Four zoonotic parasites were notably absent in this study: Angiostrongylus cantonensis, Echinococcus multilocularis, Trichinella spiralis and Toxoplasma gondii. The range of the globally-emerging parasite, A. cantonensis, is also expanding in the southern United States; therefore, documenting its apparent current absence in Vancouver may be useful if it continues its northern spread (York et al. 2015; Stockdale Walden et al. 2017; Cowie 2017). The absence of E. multilocularis is not surprising since rats are only rarely intermediate hosts and likely have no to minimal roles in the epidemiology of this pathogen (Okamoto et al. 1992; Antoniou et al. 2010). The absence of T. spiralis is also not surprising since T. spiralis is rare in Canadian pigs. Cases of trichinellosis in Canada are usually attributed to consumption of infected wildlife, particularly bears, walrus and seal, but there are no free-ranging large wild mammal populations or pig farms near the study site (Appleyard and Gajadhar 2000). Despite a historical study documenting T. spiralis in 25% of wild rats (n=260) from British Columbia, Canada (most of which originated on swine farms), studies of wild rats elsewhere also document the absence of this parasite (Moynihan and Musfeldt 1949). Collectively, these results support the theory that the domestic transmission cycle in rats requires infected pigs (Appleyard and Gajadhar 2000). We did not observe tissues cysts consistent with T. gondii in any of the rats examined, even

102

though it is a common infection in rats elsewhere (Webster 1994). The apparent absence may be a reflection of our diagnostic technique since histopathology is less sensitive compared to other diagnostic methods (Dubey and Frenkel 1998). All four pathogens are important causes of zoonotic diseases, so it is helpful to document their current apparent absence in Vancouver.

Larva of Taenia taeniaformis were also absent, despite being present in relatively high numbers in rats elsewhere (e.g., up to 45% prevalence in a historical study of rats from Montreal; Firlotte 1948). This result may be related to modern deworming of domestic cats and limiting their outdoor access.

5.5.3 Idiopathic Inflammatory Lesions

Idiopathic inflammatory lesions were present in nearly every body system of rats in this study. With the exception of cardiomyopathy and interstitial nephritis, most of these lesions were likely incidental. These included sialoadenitis, peripancreatic fat saponification and inflammation, adrenalitis, orchitis, cerebral gliosis and pustular stomatitis/esophagitis that was distinct from lesions typically associated with Eucoleus sp. infection (i.e., hyperkeratosis, mucosal hyperplasia and submucosal inflammation; Rothenburger et al. 2014b)

Cardiomyopathy, consisting of myocarditis, fibrosis and/or myocardial degeneration, was present in approximately a third of rats in our study. In a subsample of rats, lesions of cardiomyopathy were associated with heavier (presumably older) male rats, which mirrors the epidemiology of this condition in laboratory populations (Rothenburger et al. 2015b). Cardiomyopathy is a frequent spontaneous lesion in laboratory rats, where it most often affects aged males fed standard commercial diets ad libitum (Keenan et al. 1995b; Kemi et al. 2000; Chanut et al. 2013). The disease can be fatal in laboratory rats, which suggests that this condition may be similarly important to wild rat health (Keenan et al. 1995b).

We identified interstitial nephritis in approximately a third of rats examined, which is consistent with previous studies in wild rats (Wyman 1910; Woolley and Wherry 1911; Laurain 1955; Kilham et al. 1962; Gray et al. 1974; Ceruti et al. 2002). This lesion may be incidental, an early manifestation of chronic progressive nephropathy or the result of chronic Leptospira spp.

103

infection (Sterling and Thiermann 1981; Hard et al. 1999). Although roughly 11% of the rats in this sample tested positive for Leptospira spp., the association between interstitial nephritis and Leptospira sp. remains unclear (Himsworth et al. 2013a). Results of other studies are often contradictory (Cameron and Irwin 1929; Sterling and Thiermann 1981; Athanazio et al. 2008; Monahan et al. 2009). For instance, one study observed no difference in the prevalence of interstitial nephritis based on Leptospira spp. culture status in wild rats (Tucunduva de Faria et al. 2007). In experimental studies in laboratory rats, acute infections are not associated with lesions, but chronic infections lead to interstitial nephritis (Monahan et al. 2009). The inconsistent association may be due to differing pathogen detection methods (i.e., culture, immunohistochemistry, polymerase chain reaction) and statistical analyses. Pyelitis is a specific type of interstitial nephritis that is characterized by lymphoplasmacytic inflammation immediately adjacent to the renal pelvis and has been identified previously in wild rats (Gray et al. 1974). Since interstitial nephritis and pyelitis tended to surround renal blood vessels, it is also possible this inflammatory infiltrate could be the result of migrating T. crassicauda larva (Hard et al. 1999).

5.5.4 Traumatic and Anthropogenic Lesions

Conspecific bite wounds were common and seemingly important lesions. These wounds ranged in severity, were frequently associated with lymphadenopathy and were often infected. Infected wounds harbored 22 species of aerobic and anaerobic bacteria, including polymicrobial infections (Himsworth et al. 2014c). Based on the bacterial culture similarities, bite wounds were the likely portals of entry for many internal infections. Wounds and associated infections likely had an important impact on individual rat health. Indeed, fighting-induced wounds and those that become infected can be fatal (Calhoun 1963). Most bite wounds occurred in the lumbar sacral region near the base of the tail, which is consistent with previous observations in wild rats (Calhoun 1963; Blanchard et al. 1985). Dominant male rats tend to preferentially bite the back of subordinate conspecifics, while females bite the head (Takahashi and Blanchard 1982). Disruption of a colony’s established dominance hierarchy may increase fighting (Blanchard et al. 1985; Feng and Himsworth 2014). Fighting and the resulting bite wounds may also be an important mechanism for transmitting zoonotic and other pathogens among rats (Glass et al. 1988; Himsworth et al. 2013a).

104

Although conspecific bite wounds were the most prevalent traumatic lesions, human activities likely resulted in other traumatic injuries. One rat was shot with a metal pellet in the thorax. We suspected that 7/672 of rats (1.0%) with chronic wounds, limb fractures and/or amputations survived snap-type traps based on wound appearance and location of chronic wounds. Although other researchers have attributed chronic amputations in wild rats to frost bite, temperatures in Vancouver are rarely below freezing (Schiller 1956; Government of Canada 2017). Most of these chronic limb injuries occurred to the right hind limb, which may be a coincidence. Alternatively, injuries to the right hind limb may reflect how traps are set or perhaps how rats approach and attempt to flee from traps. It is remarkable that these individuals survived, despite the apparent severity of these lesions since animals that survive limb fractures are rarely reported in wildlife (Woodman 2013). Collectively, these results suggest that rats experience non-lethal trauma in urban environments and raise animal welfare concerns about trapping methodologies for rat control. Although snap-type traps are generally considered to be humane, there are serious welfare implications if traps are improperly set and/or fail to kill (Mason and Littin 2003).

There were also incidences of acute trauma that were attributable to our live-trapping techniques. The cage-type traps likely caused injury to the nose and toes of a small number of rats. There were also two instances in which we suspected that members of the public intentionally burnt rats in traps. Due to the overall objectives of the project, live trapping was necessary to obtain the required biological samples (serum, fresh tissues); therefore, alternative methods were not considered. Future studies could potentially avoid these types of injuries by lining cage traps with a solid bottom to prevent pinched toes and avoid startling the rats to prevent nose injuries. Although reducing time in traps is important for rat welfare during trapping, the standard approach for nocturnal mammals such as rats is to leave traps set overnight, as was done in this study (Mason and Littin 2003; Gannon and Sikes 2007).

Anticoagulant rodenticide exposure was suspected in many of the rats based on the presence of typical “bait” in the stomach and/or intestinal contents, as well as subcutaneous, periarticular and internal hemorrhages (Mason and Littin 2003). Toxicology testing was not included in the present study, and information about rodenticide use in the study location is not available.

105

Poisoned rats can live for at least several days, depending on the dosage and type of anticoagulant rodenticide (Mason and Littin 2003). Our results show that poisoned rats are active within the urban landscape and continue to forage. Depending on the anticoagulant rodenticide (e.g., warfarin vs. second generation anticoagulants), there may be a substantial risk of secondary poisonings to cats, dogs, predatory and scavenging birds and mammals (Stone and Okoniewski 1999; Mason and Littin 2003). This result contradicts the assumption that poisoned rats die in burrows and are not readily recovered (Conlogue et al. 1979).

5.5.5 Degenerative Conditions

Degenerative conditions were infrequent in wild rats in the present study as compared to laboratory rats and included chronic progressive nephropathy (CPN), alopecia, mineralization, dental disease, right ventricular cardiac hypertrophy and coxofemoral arthritis. Degenerative conditions are also infrequently observed in other free-ranging wild animals (Munson et al. 2005; Fenton et al. 2017).

The rarity of CPN (identified in one rat) is surprising. It is one of the most commonly encountered background lesions of laboratory rats and other studies of wild rats have also identified lesions consistent with this disease (Wyman 1910; Hard et al. 1999; Tucunduva de Faria et al. 2007). It may be that the early lesions of CPN, which are subtle (i.e., tubular basophilia and basement membrane thickening, hyaline casts and glomerulosclerosis) were obscured by tissue artifacts such as autolysis (Hard et al. 1999). The low prevalence of CPN may also be a function of environmental and dietary differences between wild and laboratory populations. For example, rat burrows may be more humid than typical laboratory animal facilities and thus may support maintenance of hydration and kidney function. Additionally, diets of free-ranging rats may consist of less protein and fewer calories compared to commercial rations. Low protein and calorie-restricted diets reduce the prevalence of CPN in laboratory rats (Keenan et al. 1995a; Hard et al. 1999). Alternatively, it is possible that CPN in wild rats does not progress to the severity seen in laboratory rats if affected individuals are selectively removed from the population or die of other reasons prior to its development.

106

Many rats had bilaterally symmetrical alopecia. Previous trauma is a frequent cause of alopecia in laboratory rats (Mecklenburg et al. 2013), and although alopecia was often concurrent with bite wounds in the present study, many of the alopecic rats had no evidence of previous skin injury. In the absence of traumatic injury, other potential causes include age-related senescence, endocrine diseases and ectoparasites. Calhoun’s observational study of wild rats maintained in an outdoor enclosure associated alopecia with normal senescence, as well as social rank among male rats (1963). He also noted that hair thinned over the lumbosacral area before progressing to generalized alopecia in most rats. Since many rats in the present study were affected by hyperplastic goiter, endocrine-associated alopecia is a possibility. Finally, ectoparasitism was common in this population and may have contributed to hair loss.

A range of tissues had mild metastatic mineralization, which was most likely incidental; these included pulmonary blood vessels, heart, adrenal glands, kidney and thyroid glands. Although vitamin D toxicity may cause tissue mineralization, vitamin D analogue rodenticides are not licensed for pest control use in Canada, making exposure to these substances unlikely (Government 2017). Similarly, mineralization is often associated with CPN in laboratory rats, but CPN was rare in the present study (Hard et al. 1999). Other wild rat studies describe mineralization in the kidneys and mammary glands (Tucunduva de Faria et al. 2007; De Oliveira et al. 2016).

There were several rare degenerative lesions that likely had an important impact on individual rat health. Although rare in our study, dental disease is frequently observed in laboratory rat populations, affecting approximately one third of rats (n=200) in a chronic disease study (Losco 1995). Right ventricular hypertrophy was rarely identified but was not associated with heart failure (Rothenburger et al. 2015b). We observed arthritis in the coxofemoral joint of one rat that may have resulted from a traumatic injury. In general, laboratory rats do not develop osteoarthritis and are not used as spontaneous models of this disease (Gerwin et al. 2010).

5.5.6 Neoplastic, Proliferative and Congenital Lesions

The most frequent proliferative lesion was diffuse thyroid follicular hyperplasia in more than half of the rats. Although the cause is unknown, iodine deficiency is likely and may have

107

contributed to alopecia. Consumption of certain anthropogenic food sources may lead to an iodine-deficient diet. It is also possible rats were exposed to a goitrogenic substance in the urban environment or that there is seasonal variability in thyroid gland morphology. There were several rare tumors and proliferative lesions in this sample; however, only the round cell lung tumor was deemed malignant and was associated with severe cachexia. The rarity of neoplasia is consistent with previous studies and the absence of other malignant neoplasia may be a function of reduced survival of affected individuals in nature. For instance, a study of 23,000 rats examined during the 1907-1908 plague eradication effort in San Francisco, California, describes 14 epithelial and 8 mesenchymal tumors; half of these were malignant (Woolley and Wherry 1911). The estimated prevalence of tumors among these Californian rats was 0.1%, which is much lower than the prevalence of proliferative/neoplastic lesions in our study at 1.5% (10/672; Wyman 1910). This is likely because half of the lesions in our study were only identified via histopathology. Considering those that were grossly identifiable, the prevalence is closer to this historical estimate at 0.7%. Interestingly, among this historical sample, neoplastic lesions were most often identified in the mammary tissue (Wyman 1910), which were not observed in the present study. Another observational study of 444 wild Norway rats did not describe any neoplastic changes, suggesting that large sample sizes are needed to identify these rare lesions (Balfour 1922). Other neoplasia described in wild rats include uterine fibroma, leiomyosarcoma, lipoma, renal carcinoma and testicular angiosarcoma (Wyman 1910; Kilham et al. 1962; Jeon et al. 2013). Benign neoplasia and proliferative lesions (e.g., pulmonary mucous metaplasia, hibernoma, adrenal cortical hyperplastic nodules) were likely incidental observations in the present study. Similarly, two rats had cutaneous pox lesions of likely minimal health significance.

We observed two congenital anomalies: a double esophagus and a Meckel’s diverticulum. Both are similarly rare in laboratory rats and likely of no clinical significance (Gupta 1973; Canpolat et al. 1998). The low prevalence of congenital anomalies may be due to high genetic diversity and/or poor survival of rats with severe anomalies.

5.6 CONCLUSIONS Overall, the most severe diseases of wild rats in Vancouver appear to be infectious and inflammatory conditions in a variety of systems. Lesions in the cardiovascular, respiratory,

108

endocrine, skin and digestive systems from a variety of causes are also common and apparently important based on their severity. Grossly-evident lesions were relatively rare yet severe. And although individual microscopic lesions may have limited significance, the cumulative effects of multi-systemic lesion patterns (i.e., total disease burden) may have negative impacts on individual rat health and longevity (Snyder et al. 2016). Given the breadth and severity of lesions we observed, it is possible that many individuals in this sample were in poor overall health. However, we were unable to directly link lesions with evidence of ill health such as emaciation and decreased fecundity.

There are similarities between the lesions in this sample of wild rats and those in laboratory settings. Historically, infectious diseases were a major cause of morbidity and mortality in laboratory rats prior to their elimination through cesarean derivation and enhanced biosecurity measures, so it is not surprising to find similar infections in wild rats (Weisbroth 1996; Baker 1998). Yet several lesions were notably rare or absent in our study as compared to their captive conspecifics; these include chronic progressive nephropathy, gastric ulcerations and malignant neoplasia. Similarly, we did not identify viral diseases, with the exception of skin pox, even though these are important in laboratory rat populations. This may be due to our lack of viral testing. It is probable that severely-diseased individuals are rapidly removed from the population through conspecific aggression, starvation and predation. If this is true, then the prevalence of all lesions in this study is likely an underestimate. Thus, it cannot be assumed that results in laboratory rats are directly applicable to wild rats (Hulin and Quinn 2006).

Considering that we assessed live-trapped rats rather than collected carcasses, the frequency of lesions and disease afflicting our study population was striking. It is unclear whether live trapping biased the sample towards either diseased or healthy individuals (Mitchell 1976). Diseased rats may have been more prone to enter traps and consume bait. Conversely, healthy rats may have foraged more widely and thus encountered traps. Regardless, our live trapping efforts seem to have sampled diverse individuals along the spectrum of health and disease, as well as body mass, sex and sexual maturity. This range is consistent with the mass rat trapping efforts in San Francisco, California, in which apparently sick rats were frequently captured (Wyman 1910). It would be advantageous to study disease in rats that die naturally; however,

109

carcass recovery is impeded by their small body size, rapid decomposition, and subterranean habitat preference (i.e., burrows and infrastructure like sewers; Wyman 1910; Calhoun 1963; Feng and Himsworth 2014).

Despite the potential impacts of disease on individual rats, there was evidence of robust reproduction, suggesting that population-level effects were unlikely (Balfour 1922; Davis 1953; Karesh and Cook 1995). Specifically, among sexually mature female rats in this study, at least 21% were pregnant with a median of 9 embryos per pregnancy; 50% were lactating and over 60% were parous (Himsworth et al. 2014a). Leptospira sp., Salmonella sp. and Capillaria hepatica have no apparent effect on populations (Davis 1951). Even plague outbreaks—the most virulent infectious disease to afflict rats—fail to suppress populations (Wyman 1910).

Our large, live-trapped sample and in-depth analysis of lesions and causative agents differs from typical wildlife pathology studies, which tend to be retrospective cause-of-death analyses of sick or dead specimens collected through passive surveillance (Mörner et al. 2002; Fenton et al. 2017). The scale and systematic approach of our study is analogous to disease surveys undertaken in fish, which are often actively caught rather than collected through passive surveillance (Lang et al. 2017). The lack of analogous wild mammal studies is likely due to logistical and ethical limitations since histopathology requires invasive and/or lethal sampling that is expensive, requires skilled personnel, may be only ethically feasible for certain species such as pests (e.g., rats and mice) and those sustainably harvested by subsistence hunting (Sullivan et al. 2003; Kutz et al. 2013; Carnegie et al. 2016). One of the most comprehensive health studies in any terrestrial mammal included active sample collection from 835 reindeer and caribou (Rangifer sspp.) and the primary focus was on body condition indices, pathogen detection and heavy metals/contaminants rather than a systematic pathology assessment. Yet, the use of general diagnostic processes including histopathology for wild animals is an invaluable tool to detect novel and emerging diseases in this era of molecular diagnostics and provides important baseline disease data (Carnegie et al. 2016; Karesh and Cook 1995).

Our cross-sectional study captured the occurrence of lesions during a single year in one rat population. It is unknown if this amount and type of disease is reflective of rats in other cities or

110

environments (e.g., farms and natural areas). Future studies should compare our results to other rat populations. And given that rats carry many zoonotic pathogens, understanding the influence of co-morbidities and co-infections on zoonotic pathogen status in individual rats is an intriguing area of future research. Studies that document and assess the impact of disease on rat survival would also increase our general knowledge of these large and persistent populations. Ultimately, accounting for disease will allow for a more nuanced understanding of urban rat ecology.

111

5.7 TABLES Table 5.1 Description and prevalence of major histological lesions in a sample of wild urban Norway rats (Rattus norvegicus) trapped in Vancouver, Canada.

Category Name Histological Description n/Na Prevalence 95% CI b

Cardiovascular Cardiomyopathy Lymphoplasmacytic myocarditis, fibrosis and/or myocardial 128/406 31.5 27.0-36.0 degeneration

Myocardial mineralization Finely granular basophilic material within interstitium and myocytes 35/406 8.6 6.1-11.7

Mineralization of pulmonary blood Finely granular basophilic material within the tunica intima 62/404 15.3 12.0-19.2 vessels Medial hypertrophy of pulmonary Thickening of the smooth muscle within the tunica media such that 90/404 22.3 18.3-26.7 arterioles percent medial thickness was ³ 50% of the vessel diameterc

Right ventricular hypertrophy Right and left ventricle free walls are of equal width 18/403 4.5 2.7-7.0

Digestive Non-glandular stomach lesionsd Hyperkeratosis,e mucosal hyperplasia,f keratin pustules and/or 231/388 59.5 54.4-64.5 submucosal inflammationg

Endocrine Adrenal gland mineralization Finely granular basophilic material within adrenal cortex or medulla 5/365 1.4 0.4-3.2

Thyroid gland mineralization Finely granular basophilic material within interstitium and follicles 132/280 47.1 41.1-53.2

Thyroid gland follicular hyperplasia Small to absent follicular lumens that are devoid of colloid with 142/279 50.9 44.9-56.9 hypertrophic follicular epithelial cells (tall cuboidal to columnar) present in > 50% of the thyroid gland and with ≤ 10 normal follicles

112

Category Name Histological Description n/Na Prevalence 95% CI b

Hemolymphatic Lymphadenitis Severe accumulations of granulocytes and/or abscess(s) 11/342 3.2 1.6-5.7

Integumentary Dermatitis Inflammation in the dermis, panniculus and/or hair follicles not 26/355 7.3 4.8-10.5 associated with a skin wound Respiratory Cilia Associated Respiratory Bacillus Mats of filamentous bacteria on ciliated epithelium of the trachea or 66/270 24.4 19.4-30.0 nasal cavity Epiglossitis/Laryngitis Lymphoplasmacytic and/or granulocytic submucosal inflammation 79/202 39.1 32.3-46.2

Inducible BALTh Peribronchiolar and/or perivascular lymphoplasmacytic cuffsi 270/403 67.0 62.2-71.6

Perivascular mixed inflammation Perivascular cuffs of lymphocytes, plasma cells and granulocytes 80/404 19.8 16.0-24.0 surrounding ³3 blood vessels

Rhinitis Lymphoplasmacytic and/or neutrophilic inflammation in the nasal 10/25 40.0 21.1-61.3 submucosa with suppurative debris in nasal cavity lumen

Tracheitis Lymphoplasmacytic and/or granulocytic submucosal/periglandular 192/372 51.6 46.6-56.8 inflammation

Tracheal gland ectasia Dilation of ³3 tracheal submucosal glands 68/364 18.7 14.8-23.1

Tracheal gland adenitis Granulocytes and/or necrotic debris within ³3 submucosal glands 25/364 6.9 4.5-10.0

Urinary

Crystals Birefringent crystals within ³3 renal tubules 137/405 33.8 29.2-38.7

Interstitial nephritis Lymphoplasmacytic inflammation in ³2 locations within the renal 121/405 29.9 25.5-34.6 interstitium, excluding areas immediately adjacent to the renal pelvis

Mineralization Finely granular basophilic material within renal tubules/interstitium 85/405 21.0 17.1-25.3

Pyelitis Lymphoplasmacytic inflammation in interstitium immediately adjacent 85/255 33.3 27.6-39.5 to the renal pelvis

113

a The number of rats examined for each lesion varies because not all tissues were available for each individual rat due to sampling error, tissue artifacts and/or autolysis. b 95% confidence interval c % medial thickness = 2 (Medial thickness) x 100/external vessel diameter d Lesions were associated with Eucoleus sp. infection (Rothenburger et al. 2014b) e Keratin layer >45 µm thick in areas without artificial separation and not adjacent to the limiting ridge (junction with glandular stomach) f Mucosa is >4-6 cells thick, affecting mainly the basal layers and >25% of the section g Granulocytes infiltrating >2 foci not adjacent to the limiting ridge h Inducible Bronchus Associated Lymphoid Tissue i Surrounds ³3 secondary or tertiary bronchioles or 1 primary bronchiole and/or ³3 blood vessels

114

Table 5.2 Neoplastic and proliferative lesions in a sample of wild urban Norway rats (Rattus norvegicus) trapped in Vancouver, Canada.

System Diagnosis Gross Description Histological Description Number Reference Affecteda Digestive Squamous Proliferative mass on the A focal, discrete area of hyperkeratosis, moderate mucosal 1 n/a papilloma dorsal tongue hyperplasia with rete peg formation

Squamous Pedunculated mass arising Severe hyperkeratosis, mucosal hyperplasia and 1 Rothenburger papilloma from the non-glandular submucosal edema with intramucosal Eucoleus sp. et al. 2014b stomach nematodes and eggs Endocrine Cortical n/a Focal, discrete, expansile, encapsulated nodules comprised 4b n/a hyperplastic of cords and bundles of polygonal epithelial cells nodules consistent with adrenal cortical cells within the adrenal gland capsule Integumentary Cutaneous Proliferative mass spanning A focal, discrete area of epithelial proliferation 2 n/a pox lesions the nail bed to the distal characterized by severe acanthosis, dyskeratotic interphalangeal joint on the hyperkeratosis and scalloping of the epidermal-dermal second finger (in one rat); junction. Keratinocytes in the stratum spinosum contain Proliferative plague on the large, amorphous amphophilic intracytoplasmic inclusions left front first metacarpal (in that contained multiple pox-like virons (enveloped ovoid a different rat) to brick-shaped viral particles, approximately 200 X 300 nm)b Connective Tissue Hibernoma n/a A focal, expansile, pseudoencapsulated mass in the 1b n/a perirenal fat. Cells were round to polygonal and contained single or multiple variably-sized lipid vacuoles.

Respiratory Mucous cell Focal, circular nodule with Focal dilated, respiratory epithelium- lined alveoli 1 Rothenburger metaplasia depressed center on the containing mucin, inflammatory cell debris, mononuclear et al. 2015a surface of the right cranial cells, giant cells. (Supplemental lung lobe materials)

Round cell Multiple neoplastic nodules Locally-extensive effacement of lung parenchyma by 1 Rothenburger tumor arise from the cranial and sheets of neoplastic round cells (possibly histiocytes) with et al. 2015a middle lung lobes moderate anisokaryosis and copious, foamy cytoplasm and (Supplemental lymphocytes. Similar neoplastic cells were within multiple materials) lymph nodes.

115

a 672 individuals for grossly-evident lesions; lesions detected by histological examination were assessed in up to 406 individuals and are indicated with b. b Up to 406 individuals were examined for microscopic lesions. The exact number of tissues examined for these lesions was not systematically recorded due to the rarity of these lesions. c Visualized with transmission electron microscopy

116

Table 5.3 Rare non-proliferative lesions in a sample of wild urban Norway rats (Rattus norvegicus) trapped in Vancouver, Canada

System Lesion Description Number Affecteda Digestive Double esophagus Two distinct esophageal lumens in the proximal esophagus at the level of the thyroid glands 1 (Fig. 5.5) Fungal gastritis Yeast and fungal hyphae in the superficial keratin layer of the non-glandular stomach with 3 associated neutrophilic inflammation in the keratin layer, mucosa and submucosa. There was a focal ulceration in one of these Gastric ulceration Focal ulceration of the glandular stomach 1

Meckel’s A 5mm saccular appendage arising from the antimesenteric side of the jejunum 1* diverticulum

Sialadenitis Salivary glands were swollen and red with associated microscopic focal necrosis, 5* neutrophilic and/or mononuclear inflammation and/or fibrosis and atrophy. In one affected rat, there were microscopic abscesses within the adjacent lymph node.

Endocrine Adrenalitis Lymphoplasmacytic inflammation of the adrenal cortex and medulla; severe architectural 3 effacement with necrosis and mineralization in one

Adrenal gland Finely granular basophilic material the adrenal cortex and medulla 4 mineralization

Thyroiditis Bilateral, multifocal lymphoplasmacytic cellular infiltrate affecting approximately 50% of 1 the thyroid gland

Hemolymphatic Lymph node Purulent material effacing lymph node architecture and surrounded by a fibrous capsule 8* abscess (mesenteric and retropharyngeal) Splenic fibrosis A band of fibrous connective tissue focally constricted the spleen 2*

Hepatic Multifocal hepatic Multifocal, random areas of hepatic necrosis throughout the parenchyma 1 necrosis

117

System Lesion Description Number Affecteda Lympho- Infiltrate of lymphocytes and plasma cells within portal regions (etiology unknown) 1 plasmacyitic periportal hepatitis Hepatic lipidosis Diffusely, hepatocytes of a gravid female in good nutritional condition contained single 1 clear vacuoles (consistent with fat)

Integument Pinnal fungal Multifocal white crusts on the margin of the right pinna with histologic mild parakeratotic 1* dermatitis hyperkeratosis, focal intrakeratin pustules, moderate epidermal hyperplasia, subcutaneous lymphoplasmacytic inflammation and septate fungal hyphae within hair follicles

Pinnal Grossly-evident severe bilateral pinnal hyperkeratosis (Supplemental Fig. S16) 1* hyperkeratosis

Chemical Pinnae were shrunken, deformed, hyperemic and ulcerated along the edges of turquoise 1* dermatosis and green material (suspected paint) and had linear excoriations (suspected from scratching thermal injury trauma). Whiskers of this rat were singed (suspected as malicious burning while rat was in the trap). Histologically, pinnae were affected by orthokeratotic hyperkeratosis, epidermal hyperplasia, congestion, neutrophils, lymphocytes and plasma cells and serum lakes in the dermis (suspect chronic response to paint). In multifocal areas of the pinna, the epidermis was ulcerated and dermal collagen in was pale and lacked distinct fibers. Hair follicles, adnexal structures and blood vessels were necrotic with multifocal colonies of coccoid bacteria throughout the dermis (suspect acute thermal injury) Ocular Bilateral Crusting mucopurulent discharge rimmed eyes and corneas were opaque (Supplemental Fig. 1* panophthalmitis S17). Histologically, corneal melting and perforations were covered in necrotic cell debris, bacterial colonies, fibrin and neutrophils that also extended into the anterior and posterior chambers. Free hair shafts were present deep within both globes (suggestive of trauma). The iris was prolapsed and remnant lens capsules were coiled with no other visible lens structures. Optic nerves contained neutrophils, lymphocytes and plasma cells. Bacterial culture results: Pasteurella dagmatis and Streptococcus gallinaceus. This rat also had multiple wounds to the left front limb, possibly related to its lack of vision Musculoskeletal, Limb ascending A chronic footpad wound of the right hind limb was associated with an ascending infection 1* Adipose and infection characterized by suppurative cellulitis, subcutaneous abscesses and edema. Bacterial culture Connective Tissue results: Staphylococcus aureus, Staphylococcus sp. and Enterobacter sp.

Generalized muscle Grossly-evident skeletal muscles were shrunken and pale with prominent bony 1* atrophy protuberances likely rated to concurrent malocclusion, edentulism and pulmonary neoplasia

118

System Lesion Description Number Affecteda Coxofemoral Grossly-evident left femoral head cartilage erosion and acetabulum deformation with 1* arthritis associated thickened joint capsule, osteophytes and synovial hyperplasia (Supplemental Fig. S18) Nervous Cerebral gliosis Multifocal gliosis in the cerebral grey matter with no associated meningitis 1

Respiratorya Laryngeal Severe hyperkeratosis with multiple bacterial colonies and mucosal hyperplasia of the 1 hyperkeratosis squamous epithelium in the larynx

Female Reproductive Neutrophilic Grossly-evident purulent exudate from the clitoral glands was associated histologically with 2* clitoral adenitis multifocal neutrophilic inflammation and necrosis of the clitoral gland. Bacterial culture not performed Endometritis Variable amounts of erythrocytes, sloughed endometrial cells and/or granulocytes within the 4 uterine lumen, granulocytic with or without lymphoplasmacytic inflammation in the endometrium, mesometrium and/or broad ligament and lymphoplasmacytic cuffing of blood vessels in the broad ligament. Endometritis with Multifocal abscesses in the uterine horns (Supplemental Fig. S7). Lumen of affected areas 1* abscesses contained sloughed endometrial cells, viable and degenerative granulocytes, keratinized squamous cells, necrotic cell debris and colonies of coccobacilli. The endometrium was multifocally sloughed. Urine glands were ectatic and contained granulocytes. Granulocytes infiltrated the edematous endometrium and myometrium. Hemosiderphages were within the deep myometrium. Bacterial culture results: Enterococcus faecalis, Escherichia coli, Proteus sp. and Bacteroides sp. Necrotizing Right uterine horn was focally congested and dilated by a firm coagulum that consisted of 1* endometritis necrotic and mineralized tissues with no identifiable fetal structures microscopically (possibly necrotic placenta or mineralized necrotic cellular debris). The adjacent lumen contained granulocytes and proteinaceous fluid. The myometrium was diffusely edematous and contained a mixed population of inflammatory cells, mainly granulocytes and macrophages. Concurrently, one kidney had pyelonephritis. Bacterial culture results: Salmonella enteriditis serogroup D and Proteus sp. Hematometra and Externally, there was a mucohemorrhagic vaginal discharge. Uterine horns were bilaterally 1* endometritis dilated and contained hemorrhagic fluid. Histologically, the endometrium was infiltrated by moderate numbers of granulocytes

119

System Lesion Description Number Affecteda Male Reproductive Orchitis Lymphoplasmacytic cell infiltrate and fibrosis within the testes. In one rat, the right testicle 3* was grossly atrophied; microscopically, a large amount of fat surrounded the few remaining seminiferous tubules, which were frequently mineralized and had no visible spermatogenesis. In another rat, granulocytic orchitis was associated with testicular necrosis and accessory sex glands were inflamed, including one preputial gland, which was grey- green and contained purulent material. Bacterial culture results: Enterococcus faecalis and Escherichia coli Urinary Pyelonephritis Kidney(s) were severely shrunken with an irregular surface (Supplemental Fig. S19). 2* Histologically, tubules contained intralumenal neutrophils, protein and necrotic debris that radiated from the renal pelvis to the cortex. In some areas, lymphoplasmacytic inflammation and severe fibrosis surrounded atrophic tubules, glomeruli or obliterated normal renal architecture. Pyelonephritis was associated with severe metritis in one rat. Renal cyst A single cystic structure lined by squamous epithelium within the renal cortex. 1 Peri-renal abscess A focal abscess in the retroperitoneal space adjacent to kidney. Bacterial culture results: 2* Staphylococcus aureus and Escherichia coli

Chronic progressive Bilateral kidneys had multifocal to coalescing, tan areas with microscopic evidence of 1* nephropathy depressions in the renal capsule and cortical lymphoplasmacytic interstitial nephritis and fibrosis. Tubules were occasionally dilated and contained proteinaceous fluid.

a Up to 406 individuals were examined for microscopic lesions. The exact number examined for these lesions was not systematically recorded due to the rarity of these lesions. Grossly-evident lesions (indicated by an asterisk [*] following the number affected) were assessed in 672 individuals. b Rare microscopic lesions of the respiratory tract are described in Rothenburger et al. 2015a

120

Table 5.4 Parasites and applicable lesions identified using histopathology in a sample of wild urban Norway rats (Rattus norvegicus) trapped in Vancouver, Canadaa

System Location Organism Associated Lesion n/Nb Prevalence 95% CIc Reference

Digestive Tongue, Eucoleus spp. Hyperkeratosis, mucosal hyperplasia, 164/399 41.1 36.2-46.1 Rothenburger oropharynx, keratin pustules and submucosal et al. 2014b esophagus, non- inflammation in the non-glandular glandular stomach stomach Non-glandular Gongylonema None Rare n/ad n/a n/ae stomach neoplasticum Small intestine Nematode None 83/198 41.9 35.0-49.1 n/ae

Small intestine Cestode None 6/198 3.0 1.1-6.5 n/ae

Small intestine Coccidia None 13/198 6.6 3.5-11.0

Hepatic Liver Capillaria Multifocal, white, tortuous parasite tracts 242/672 36.0 32.4-39.8 Rothenburger hepatica on the liver surface and subcapsular et al. 2014a parenchyma characterized microscopically by viable and necrotic capillarid nematodes and/or eggs surrounded by granulocytes or mononuclear cell inflammatory infiltrates and fibrosis, sometimes with mineralization Integumentary Ears, nose, distal Notoedres Proliferative and crusting dermatitis of 2/672 0.3 0.0-1.1 Anholt et al. limbs, tail muris the ears (primarily) and poorly-haired 2014 skin of the nose, distal limbs and tail, characterized by hyperkeratosis with intracorneal pustules and adult mites and/or eggs, epidermal hyperplasia and lymphoplasmacytic dermatitis Skin Nosopsyllus Possible alopecia and exudative n/af n/a n/a n/ae fasciatus dermatitis Skin Ornithonyssus Possible alopecia and exudative n/af n/a n/a n/ae bacoti dermatitis

121

System Location Organism Associated Lesion n/Nb Prevalence 95% CIc Reference

Skin Polyplax Possible alopecia and exudative n/af n/a n/a n/ae spinulosa dermatitis

Urinary Urinary bladder Trichosomoide At least one nematode or egg cross- 59/194 30.4 24.0-37.4 n/ae (rarely renal s crassicauda section in urinary bladder or renal pelvis. pelvis) Variably accompanied by mild submucosal lymphoplasmacytic inflammation

a We found no evidence of Angiostrongylus cantonensis, Echinococcus multilocularis, Sarcocystis spp., Toxoplasma gondii or Trichinella spiralis infection in any of the rats b Rats were positive if there was at least one characteristic egg and/or adult cross-section in the sections examined. The number of rats examined for each lesion varies because not all tissues were available for each individual rat due to sampling error, tissue artifacts and/or autolysis c 95% confidence interval d Exact numbers were not assessed due to rarity of infection. e There are no previous studies of these parasite in this population of rats. f Prevalence data are not available for these ectoparasites because they are being assessed by another collaborator on this research project who will present these data in a separate future study.

122

Table 5.5 Bacteria isolated from macroscopic lesions in a sample of wild urban Norway rats (Rattus norvegicus) trapped in Vancouver, Canada

Bacterial Speciesa Associated Lesionsb Number of Isolates

Bacteroides sp. Uterine abscess 1 Bordetella bronchiseptica Pneumonia, pulmonary round cell neoplasia 2 Clostridium sordellii Subcutaneous abscess 1 Corynebacterium kutscheri Subcutaneous abscess 1 Enterobacter sp. Ascending hind limb infection 1 Enterococcus cloacae Cervical lymph node abscess 1 Enterococcus faecalis Cervical lymph node abscess, mesenteric lymph node abscess, preputial gland abscess, purulent penile 12 discharge, subcutaneous abscess, urinary bladder concretion, uterine abscess, vaginal discharge Escherichia coli Ascending hind limb infection, cervical lymph node abscess, perirenal abscess, lung abscess, mesenteric 15 lymph node abscess, preputial discharge, pneumonia, subcutaneous abscess, pulmonary round cell neoplasia, seminal vesicle, uterine abscess, vaginal discharge Gemella morbillorum Pulmonary round cell neoplasia 1 Klebsiella pneumoniae Cervical lymph node abscess 1 Lactobacillus animalis Vaginal discharge 1 Pasteurellaceaec Subcutaneous abscess 1 Pasteurella dagmatis Panophthalmitis 1 Pasteurella pneumotropica Submandibular abscess 1 Proteus sp. Preputial discharge, seminal vesicle, suppurative metritis, uterine abscess, vaginal discharge 5 Salmonella enteritidis Suppurative metritis 1 Staphylococcus aureus Ascending hind limb infection, perirenal abscess, mediastinal abscess with metal projectile pellet, purulent 13 penile discharge, pulmonary round cell neoplasia, retropharyngeal abscess, skeletal muscle abscess, subcutaneous abscess, subscapular abscess, vaginal discharge Staphylococcus sp. Ascending hind limb infection, preputial gland abscess, mesenteric lymph node abscess, skeletal muscle, 6 subcutaneous abscess Streptococcus sp. Vaginal discharge 1 Streptobacillus moniliformis Cervical lymph node abscess, subcutaneous abscess 3 Streptococcus gallinaceus Panophthalmitis 1 a Himsworth (2014) describes bacteria isolated from bite wounds and associated soft tissue infections; Rothenburger (2015) describes ancillary testing that identified Mycoplasma pulmonis and cilia-associated respiratory bacillus in the lungs of a sub-sample of rats in the current study; this table does not include these results b Many infections were polymicrobial, therefore lesions may be listed for multiple bacterial species c Isolates from a potentially novel genus that grouped most closely with Haemophilus sp., Actinobacillus sp., Pasteurella sp. and Aggregatibacter sp. using 16s DNA sequencing

123

5.8 FIGURES

Figure 5.1 Sample selection protocol for studying the pathology of urban Norway rats (Rattus norvegicus) from Vancouver, British Columbia, Canada. Note that only 341 rats with no gross lesions were assessed for microscopic lesions due to budgetary constraints that prohibited examination of all tissues from all rats.

124

Figure 5.2-5.12 Microscopic lesions, Norway rats (Rattus norvegicus). Hematoxylin and Eosin (HE).

Figure 5.2. Cardiomyopathy, heart. Lymphocytic myocarditis, myocyte necrosis and fibrosis. Figure 5.3. Medial hypertrophy of pulmonary blood vessels, lungs. The tunica media is thickened by disorganized smooth muscle cells. Figure 5.4. Non-glandular stomach. Eucoleus sp. adults and eggs in the mucosa in association with hyperkeratosis and mucosal hyperplasia. Figure 5.5. Double esophagus, esophagus and trachea. There are two distinct esophageal lumens adjacent to the trachea. Figure 5.6. Capillaria hepatica, liver. An adult in oblique section and eggs efface hepatocytes and are surrounded by granulomatous inflammation and fibrosis. 125

Figure 5.7. Diffuse thyroid follicular hyperplasia, thyroid gland. Thyroid follicular cells are cuboidal to tall columnar; follicles lack colloid. Inset: Immunohistochemistry for thyroglobulin. Immunoreactivity is weak, diffuse and intracytoplasmic, consistent with follicular cells. Figure 5.8. Lymphoplasmacytic tracheitis, trachea. There is a severe infiltration by lymphocytes and plasma cells within the submucosa. Few granulocytes are adherent to the mucosal surface. Figure 5.9. Inducible Bronchus Associated Lymphoid Tissue (IBALT), lungs. Bronchiole is cuffed by asymmetrical clusters of lymphocytes. Figure 5.10. Perivascular mixed inflammation, lungs. Blood vessels are cuffed by a mixed population of granulocytes and lymphocytes. Figure 5.11. Interstitial nephritis, kidney. The interstitium adjacent to glomeruli and tubules contains a moderate infiltrate of lymphocytes and plasma cells. Figure 5.12. Trichosomoides crassicauda, urinary bladder. Multiple adults in cross- and longitudinal sections are within the urinary bladder lumen. 126

5.9 SUPPLEMENTAL FIGURES

Figure S1. Anticoagulant rodenticide toxicity, gastrointestinal tract. Turquoise material within the stomach and segments of the small intestine consistent with anticoagulant rodenticide “bait”. Figure S2. Malocclusion, oral cavity. The lower incisors are severely overgrown and the occlusive surface of upper incisors is slanted. Figure S3. Capillaria hepatica, liver. Multiple tortuous, tan parasite tracts along the liver surface. Figure S4. Alopecia and bite wounds, dorsal skin. Hair is thin and there are and multiple, small bite wounds. Figure S5. Alopecia, skin. Multifocal areas of alopecia over the trunk. Figure S6. Notoedres muris, ear. Nodular areas of proliferative dermatitis of the pinna.

127

Figure S7. Abscess, uterus. Focal abscess in the middle of the left horn. There are multiple, black placental scars (normal in parous rats). Figure S8. Prostatic concretions, urinary bladder. Firm, yellow concretions of prostatic discharge are an agonal change in male rats Figure S9. Anticoagulant rodenticide toxicity, whole body. Periarticular and subcutaneous hemorrhages affect multiple distal limbs. Figure S10. Anticoagulant rodenticide toxicity, whole body. There are severe, locally-extensive subcutaneous hemorrhages. Figure S11. Anticoagulant rodenticide toxicity, hind limbs. Severe subcutaneous hemorrhage affecting the right lower limb; the left is unaffected. Figure S12. Anticoagulant rodenticide toxicity, left hind limb. Severe subcutaneous and periarticular hemorrhages.

128

Figure S13. Chronic wound, neck and cranial thorax. Linear wound with associated crusting and alopecia. Suspected traumatic injury from a snap-type trap. Figure S14. Chronic tail amputation. Figure S15. Cutaneous hyperemia, toes. Multifocal areas of hyperemia affecting the toes. Suspected acute trauma from cage-type traps. Figure S16. Severe hyperkeratosis, left pinna. Figure S17. Panophthalmitis, left eye. Cornea is opaque, globe is collapsed and there is mucopurulent exudate on the eyelids. Figure S18. Coxofemoral arthritis, left femur.

129

Figure S19. Pyelonephritis, kidney. Affected kidney is shrunken with an irregular capsular surface.

130

5.9 REFERENCES Anholt H, Himsworth C, Rothenburger JL, Proctor H, Patrick DM. 2014. Ear mange mites (Notoedres muris) in black and Norway rats (Rattus rattus and Rattus norvegicus) from inner- city Vancouver, Canada. Journal of Wildlife Diseases 50:104–108.

Antoniou M, Psaroulaki A, Toumazos P, Mazeris A, Ioannou I, Papaprodromou M, Georgiou K, Hristofi N, Patsias A, Loucaides F, Moschandreas, J, Tsatsaris A, Tselentis Y. 2010. Rats as indicators of the presence and dispersal of pathogens in Cyprus: ectoparasites, parasitic helminths, enteric bacteria, and encephalomyocarditis virus. Vector-Borne and Zoonotic Diseases 10:867–873.

Appleyard GD, Gajadhar AA. 2000. A review of trichinellosis in people and wildlife in Canada. Canadian Journal of Public Health 91:293–297.

Athanazio DA, Silva EF, Santos CS, Rocha GM, Vannier-Santos MA, McBride AJA, Ko AI, Reis MG. 2008. Rattus norvegicus as a model for persistent renal colonization by pathogenic Leptospira interrogans. Acta Tropica 105:176–180.

Baker DG. 1998. Natural pathogens of laboratory mice, rats, and rabbits and their effects on research. Clinical Microbiology Reviews 11:231–266.

Balfour A. 1922. Observations on wild rats in England, with an account of their ecto- and endoparasites. Parasitology 14:282.

Blanchard RJ, Pank L, Fellows D, Blanchard DC. 1985. Conspecific wounding in free-ranging R. norvegicus from stable and unstable populations. The Psychological Record 35:329–335.

Boorman GA. 1990. Lung. In: Pathology of the Fischer Rat, GA Boorman, editor. Elsevier Academic Press, San Diego. pp. 339–367.

Brogden KA, Cutlip RC, Lehmkuhl HD. 1993. Cilia-associated respiratory bacillus in wild rats in central Iowa. Journal of Wildlife Diseases 29:123–126.

Brown HR, Hardisty JF. 1990. Oral cavity, esophagus, and stomach. In: Pathology of the Fischer Rat, GA. Boorman, editor. Elsevier Academic Press, San Diego. pp. 15–30.

Calhoun JB. 1963. The Ecology and Sociology of the Norway Rat. US Department of Health, Education, and Welfare, Public Health Service Publication No. 1008, Bethesda, Maryland, U.S.A. pp. 237–239.

Cameron GC, Irwin DA. 1929. Leptospira Icterohæmorrhagiæ occurrence in wild rats at Toronto. Canadian Public Health Journal 20:386–392.

Canpolat L, Sağiroğlu O, Kükner A. 1998. Case report: a rare congenital esophageal malformation on double esophagus in the rat. Kaibogaku Zasshi. Journal of Anatomy 73:13-17. 131

Carnegie RB, Arzul I, Bushek D. 2016. Managing marine mollusc diseases in the context of regional and international commerce: policy issues and emerging concerns. Philosophical transactions of the Royal Society of London. Series B, Biological sciences 371:20150215–11.

Carter ME, Cordes DO. 1980. Leptospirosis and other infections of Rattus rattus and Rattus norvegicus. New Zealand Veterinary Journal 28:45–50.

Ceruti R, Ghisleni G, Ferretti E, Cammarata S, Sonzogni O, Scanziani E. 2002. Wild rats as monitors of environmental lead contamination in the urban area of Milan, Italy. Environment and Pollution 117:255–259.

Chanut F, Kimbrough C, Hailey R, Berridge B, Hughes-Earle A, Davies R, Roland K, Stokes A, Casartelli A, York M, Jordan H, Crivellente F, Cristofori P, Thomas H, Klapwijk J, Adler R. 2013. Spontaneous cardiomyopathy in young Sprague-Dawley rats: evaluation of biological and environmental variability. Toxicologic Pathology 41:1126-1136.

Conlogue G, Foreyt W, Adess M, Levine H. 1979. Capillaria hepatica (Bancroft) in select rat populations of Hartford, Connecticut, with possible public health implications. The Journal of Parasitology 65:105–108.

Davis DE. 1951. The relation between the level of population and the prevalence of Leptospira, Salmonella, and Capillaria in Norway rats. Ecology 32:465–468.

Cowie RH. 2017. Angiostrongylus cantonensis: agent of a sometimes fatal globally emerging infectious disease (rat lungworm disease). ACS Chem Neurosci. Published online September 13, 2017. DOI: 10.1021/acschemneuro.7b00335.

Davis DE. 1953. The characteristics of rat populations. Quarterly Review of Biology 28:373–401.

Davis JK, Cassell GH. 1982. Murine respiratory mycoplasmosis in LEW and F344 rats: strain differences in lesion severity. Veterinary Pathology 19:280–293.

De Oliveira D, Figueira CP, Zhan L, Pertile AC, Pedra GG, Gusmão IM, Wunder EA, Rodrigues G, Ramos EAG, Ko AI, Childs JE, Reis MG, Costa F. 2016. Leptospira in breast tissue and milk of urban Norway rats (Rattus norvegicus). Epidemiology and Infection 144:2420–2429.

Dubey JP, Frenkel JK. 1998. Toxoplasmosis of rats: a review, with considerations of their value as an animal model and their possible role in epidemiology. Veterinary Parasitology 77:1–32.

Dungworth DL, Ernst H, Nolte T, Mohr U. 1992. Nonneoplastic lesions in the lungs. In: Pathobiology of the Aging Rat, U. Mohr, DL Dungworth, and CC Capen, editors. ILSI Press, Washington, D.C., pp. 143–160.

132

Easterbrook JD, Kaplan JB, Glass GE, Watson J, Klein SL. 2008. A survey of rodent-borne pathogens carried by wild-caught Norway rats: a potential threat to laboratory rodent colonies. Lab Animal 42:92–98.

Farhang-Azad A. 1977. Ecology of Capillaria hepatica (Bancroft 1893) (Nematoda). 1; Dynamics of infection among Norway rat populations of the Baltimore Zoo, Baltimore, Maryland. Journal of Parasitology 63:117–122.

Feng AYT, Himsworth CG. 2014. The secret life of the city rat: a review of the ecology of urban Norway and black rats (Rattus norvegicus and Rattus rattus). Urban Ecosystems 17:149–162.

Fenton H, Daoust PY, Forzán MJ, Vanderstichel RV, Ford J, Spaven L, Lair S, Raverty S. 2017. Causes of mortality of harbor porpoises Phocoena phocoena along the Atlantic and Pacific coasts of Canada. Diseases of Aquatic Organisms 122:171–183.

Firlotte WR. 1948. Parasites of the brown Norway Rat. Canadian Journal of Comparative Medicine and Veterinary Science 12:187.

Fuehrer HP, Igel P, Auer H. 2011. Capillaria hepatica in man—an overview of hepatic capillariosis and spurious infections. Parasitologic Research 109:969–979.

Gage KL., Kosoy MY. 2005. Natural history of plague: perspectives from more than a century of research. Annual Reviews Entomology 50:505–528.

Ganaway JR, Spencer TH, Moore TD, Allen AM. 1985. Isolation, propagation, and characterization of a newly recognized pathogen, cilia-associated respiratory bacillus of rats, an etiological agent of chronic respiratory disease. Infection and Immunity 47:472–479.

Gannon WL, Sikes RS. 2007. Guidelines of the American Society of Mammalogists for the use of wild mammals in research. Journal of Mammalogy 88:809–823.

Gerwin N, Bendele AM, Glasson S, Carlson CS. 2010. The OARSI histopathology initiative - recommendations for histological assessments of osteoarthritis in the rat. Osteoarthritis and Cartilage 18:S24–S34.

Giddens WE, Whitehair CK, Carter GR. 1971. Morphologic and microbiologic features of trachea and lungs in germfree, defined-flora, conventional, and chronic respiratory disease- affected rats. American Journal of Veterinary Research 32:115–129.

Giusti AM, Crippa L, Bellini O, Luini M, Scanziani E. 1998. Gastric spiral bacteria in wild rats from Italy. Journal of Wildlife Diseases 34:168–172.

133

Glass GE, Childs JE, Korch GW, LeDuc JW. 1988. Association of intraspecific wounding with hantaviral infection in wild rats (Rattus norvegicus). Epidemiology and Infection 101:459–472.

Government of Canada. Health Canada: Pesticides and Pest Management. http://www.hc- sc.gc.ca/cps-spc/pest/index-eng.php. Accessed September 20, 2017.

Government of Canada. Historical Climate Data. http://climate.weather.gc.ca. Accessed September 20, 2017.

Gray JE, Weaver RN, Connor ND. 1974. Observations on the kidneys and urine of the wild Norway rat, Rattus norvegicus. Veterinary Pathology 11:144–152.

Gregson RL, Davey MJ, Prentice DE. 1979. Bronchus-associated lymphoid tissue (BALT) in the laboratory-bred and wild rat, Rattus norvegicus. Lab Animal 13:239–243.

Gupta BN. 1973. Meckel's diverticulum in a rat. Lab Animal Science 23:426–427.

Habermann RT, Williams FP, Thorp WT. 1954. Common infections and disease conditions observed in wild Norway rats kept under simulated natural conditions. American Journal of Veterinary Reseasearch 15:152–156.

Hard GC, Alden CL, Bruner RH, Frith CH. 1999. Non-proliferative lesions of the kidney and lower urinary tract in rats. Guides for Toxicologic Pathology, STP/ARP/AFIP, Washington, DC. 1-32.

Harkema R. 1936. The parasites of some North Carolina rodents. Ecological Monographs. 6:151- 232.

Himsworth CG, Jardine CM, Parsons KL, Feng AYT, Patrick DM. 2014a. The characteristics of wild rat (Rattus spp.) populations from an inner-city neighborhood with a focus on factors critical to the understanding of rat-associated zoonoses. PLoS ONE 9:e91654.

134

Himsworth CG, Zabek E, Desruisseau A, Parmley EJ, Reid-Smith R, Jardine CM, Tang P, Patrick DM. 2015. Prevalence and characteristics of Escherichia coli and Salmonella spp. in the feces of wild urban Norway and black rats (Rattus norvegicus and Rattus rattus) from an inner- city neighborhood of Vancouver, Canada. Journal of Wildlife Diseases 51:589–600.

Himsworth CG, Zabek E, Tang P, Parsons KL, Koehn M, Jardine CM, Patrick DM. 2014c. Bacteria isolated from conspecific bite wounds in Norway and Black Rats: implications for rat bite-associated infections in people. Vector-Borne and Zoonotic Diseases 14:94–100.

Hulin MS, Quinn R. 2006. Wild and Black Rats. In: The Laboratory Rat, MA Suckow, S H Weisbroth, and CL Franklin, editors. Elsevier, Burlington, MA. pp. 865–882.

Jeon B-S, Kim H-G, Lee B-W, Han J-H, Yoon B-I. 2013. Uterine leiomyosarcoma in a wild rat (Rattus norvegicus): usefulness of Ki-67 labeling index for diagnosis. Lab Animal Research 29:127–4.

Kakrada MK, Lumsden JS, Lee EA, Collett MG. 2002. Cilia-associated respiratory bacillus infection in rats in New Zealand. New Zealand Veterinary Journal 50:81–82.

Karesh WB, Cook RA. 1995. Applications of veterinary medicine to in situ conservation efforts. Oryx 29:244–252.

Keenan KP, Soper KA, Hertzog PR, Gumprecht LA, Smith PF, Mattson BA, Ballam GC, Clark RL. 1995a. Diet, overfeeding, and moderate dietary restriction in control Sprague-Dawley rats: II. Effects on age-related proliferative and degenerative lesions. Toxicologic Pathology 23:287–302.

Keenan KP, Soper KA, Smith PF, Ballam GC, Clark RL. 1995b. Diet, overfeeding, and moderate dietary restriction in control Sprague-Dawley rats: I. Effects on spontaneous neoplasms. Toxicologic Pathology 23:269–286.

Kemi M, Keenan KP, Soper KA, Hertzog PR, Gumprecht LA, Smith PF, Mattson BA, Ballam GC, Clark RL. 2000. The relative protective effects of moderate dietary restriction versus dietary modification on spontaneous cardiomyopathy in male Sprague-Dawley rats. Toxicologic Pathology 28:285–296.

Kilham L, Low RJ, Conti SF, Dallenbach FD. 1962. Intranuclear inclusions and neoplasms in the kidneys of wild rats. Journal of the National Cancer Institute 29:863–885.

Kling MA. 2011. A review of respiratory system anatomy, physiology, and disease in the mouse, rat, hamster, and gerbil. Veterinary Clinics of North America: Exotic Animal Practice 14:287– 337.

Kutz S, Ducrocq J, Cuyler C, Elkin B, Gunn A, Kolpashikov L, Russell D, White RG 2013. Standardized monitoring of Rangifer health during International Polar Year. Rangifer 33:91–114.

135

Lang T, Feist SW, Stentiford GD, Bignell JP, Vethaak AD, Wosniok W. 2017. Diseases of dab (Limanda limanda): Analysis and assessment of data on externally visible diseases, macroscopic liver neoplasms and liver histopathology in the North Sea, Baltic Sea and off Iceland. Marine Environmental Research 124:61–69.

Laurain AR. 1955. Lesions of skeletal muscle in leptospirosis: review of reports and an experimental study. American Journal of Pathology 31:501–519.

Lewis DJ. 1982. A comparison of the pathology of the larynx from SPF, germ-free, conventional, feral and mycoplasma-infected rats. Journal of Comparative Pathology 92:149–160.

Losco PE. 1995. Dental dysplasia in rats and mice. Toxicologic Pathology 23:677–688.

Lõhmus M, Janse I, van de Goot F, van Rotterdam BJ. 2013. Rodents as potential couriers for bioterrorism agents. Biosecurity and Bioterrorism. Supplemental 1:S247–57.

MacKenzie WF, Magill LS, Hulse M. 1981. A filamentous bacterium associated with respiratory disease in wild rats. Veterinary Pathology 18:836–839.

Mason GJ, Littin KE. 2003. The humaneness of rodent pest control. Animal Welfare 12:1–37.

McGarry JW, Higgins A, White NG, Pounder KC, Hetzel U. 2015. Zoonotic helminths of urban brown rats (Rattus norvegicus) in the UK: neglected public health considerations? Zoonoses and Public Health 62:44–52.

McInnes EF. 2012. Wistar and Sprague-Dawley rats. In: Background Lesions in Laboratory Animals: A Color Atlas. Saunders Elsevier, China. pp. 17–36.

McKinney ML. 2006. Urbanization as a major cause of biotic homogenization. Biological Conservation 127:247–260.

Mecklenburg L, Kusewitt D, Kolly C, Treumann S, Adams ET, Diegel K, Yamate J, Kaufmann W, Müller S, Danilenko D, Bradley A. 2013. Proliferative and non-proliferative lesions of the rat and mouse integument. Journal of Toxicological Pathology 26:27S–57S.

Meerburg BG, Singleton GR, Leirs H. 2009. The year of the rat ends–time to fight hunger! Pest Management Science 65:351–352.

Milazzo C, Ribas A, Casanova JC, Cagnin M, Geraci F, Bella C. 2010. Helminths of the brown rat (Rattus norvegicus) (Berkenhout, 1769) in the city of Palermo, Italy. Helminthologia 47:238– 240.

Mitchell D. 1976. Experiments on neophobia in wild and laboratory rats: a reevaluation. Journal of Comparative and Physiological Psychology 90:190–197.

136

Monahan AM, Callanan JJ, Nally JE. 2009. Host-pathogen interactions in the kidney during chronic leptospirosis. Veterinary Pathology 46:792–799.

Moynihan IW, Musfeldt IW. 1949. A study of the incidence of trichinosis in rats in British Columbia. Canadian Journal of Medicine and Veterinary Science 13:152–155.

Mörner T, Obendorf DL, Artois M. 2002. Surveillance and monitoring of wildlife diseases. Revue Scientifique et Technique (International Office of Epizootics) 21:67-76.

Munson L, Terio KA, Worley M, Jago M, Bagot-Smith A, Marker L. 2005. Extrinsic factors significantly affect patterns of disease in free-ranging and captive cheetah (Acinonyx jubatus) populations. Journal of Wildlife Diseases 41:542–548.

Okamoto M, Fujita O, Arikawa J, Kurosawa T, Oku Y, Kamiya M. 1992. Natural Echinococcus multilocularis infection in a Norway rat, Rattus norvegicus, in southern Hokkaido, Japan. International Journal for Parasitology 22:681–684.

Owen D. 1976. Some parasites and other organisms of wild rodents in the vicinity of an SPF unit. Lab Animals 10:271–278.

Percy DH, Barthold SW. 2007. Rat, 3rd edition. In: Pathology of Laboratory Rodents and Rabbits. Blackwell Publishing, Ames, Iowa. pp. 125–177.

Renne R, Brix A, Harkema J, Herbert R, Kittel B, Lewis D, March T, Nagano K, Pino M, Rittinghausen S, Rosenbruch M, Tellier P, Wohrmann T. 2009. Proliferative and non- proliferative lesions of the rat and mouse respiratory tract. Toxicologic Pathology 37:5S–73S.

Rothenburger JL, Himsworth CG, Chang V, Lejeune M, Leighton FA. 2014a. Capillaria hepatica in wild Norway rats (Rattus norvegicus) from Vancouver, Canada. Journal of Wildlife Diseases 50:628–633.

Rothenburger JL, Himsworth CG, Clifford CB, Ellis J, Treuting PM, Leighton FA. 2015a. Respiratory pathology and pathogens in wild urban rats (Rattus norvegicus and Rattus rattus). Veterinary Pathology 52:1210–1219.

Rothenburger JL, Himsworth CG, Lejeune M, Treuting PM, Leighton FA. 2014b. Lesions associated with Eucoleus sp. in the non-glandular stomach of wild urban rats (Rattus norvegicus). International Journal for Parasitology: Parasites and Wildlife 3:95–101.

Rothenburger JL, Himsworth CG, Treuting PM, Leighton FA. 2015b. Survey of cardiovascular pathology in wild urban Rattus norvegicus and Rattus rattus. Veterinary Pathology 52:201–208.

Schiller EL. 1956. Ecology and health of Rattus at Nome, Alaska. Journal of Mammalogy 37:181–188.

137

Singleton GR, Brown PR, Jacob J, Aplin KP. 2007. Unwanted and unintended effects of culling: a case for ecologically-based rodent management. Integrative Zoology 2:247–259.

Singleton GR, Hinds LA, Leirs H, Zhang ZB. 1999. Ecologically-based management of rodent pests. Australian Centre for International Agriculture Research, Canberra. pp. 17–29.

Snyder JM, Ward JM, Treuting PM. 2016. Cause-of-death analysis in rodent aging studies. Veterinary Pathology 53:233–243.

Sterling C, Thiermann A. 1981. Urban rats as chronic carriers of leptospirosis: an ultrastructural investigation. Veterinary Pathology 18:628.

Stone WB, Okoniewski JC. 1999. Poisoning of wildlife with anticoagulant rodenticides in New York. Journal of Wildlife Diseases 35:187-193.

Suckow MA, Weisbroth SH, Franklin CL. (Eds.). 2006. The Laboratory Rat, 2nd edition. Elsevier Academic Press, Burlington, MA.

Sullivan TP, Sullivan DS, Ransome DB. 2003. Impact of removal-trapping on abundance and diversity attributes in small-mammal communities. Wildlife Society Bulletin 31:464-474.

Takahashi LK, Blanchard RJ. 1982. Attack and defense in laboratory and wild Norway and black rats. Behavioural Processes 7:49–62.

Towns DR, Atkinson IAE, Daugherty CH. 2006. Have the harmful effects of introduced rats on islands been exaggerated? Biological Invasions 8:863–891.

Tucunduva de Faria M, Athanazio DA, Gonçalves Ramos EA, Silva EF, Reis MG, Ko AI. 2007. Morphological alterations in the kidney of rats with natural and experimental Leptospira infection. Journal of Comparative Pathology 137:231–238.

Vaumourin E, Vourc’h G, Gasqui P, Vayssier-Taussat M. 2015. The importance of multiparasitism: examining the consequences of co-infections for human and animal health. Parasites and Vectors 8:4–13.

Webster, JP. 1994. Prevalence and transmission of Toxoplasma gondii in wild brown rats, Rattus norvegicus. Parasitology 108:407-411.

Webster JP, Macdonald DW. 1995. Parasites of wild brown rats (Rattus norvegicus) on UK farms. Parasitology 111:247–255.

138

Weisbroth SH. 1996. Post-indigenous disease: changing concepts of disease in laboratory rodents. Lab Animal 25:25–33.

Wobeser GA. 2006. Essentials of Disease in Wild Animals. Blackwell Publishing, Ames.

Woodman N. 2013. Survival of the less-fit: a least shrew (Mammalia, Soricidae, Cryptotis parvus) survives a separated leg fracture in the wild. Journal of Wildlife Diseases 49:735–737.

Woolley PG, Wherry WB. 1911. Notes on twenty-two Spontaneous tumors in wild rats (M. norvegicus). J Med Res 25:205–216.

Wyman W. (Ed.). 1910. The Rat and Its Relation to the Public Health. Treasury Department, Public Health and Marine-Hospital Service of the United States, Washington.

York EM, Creecy JP, Lord WD, Caire W. 2015. Geographic range expansion for rat lungworm in North America. Emerging Infectious Diseases 21:1234–1236.

139

CHAPTER 6

6. THE DEVIL IS IN THE DETAILS—HOST DISEASE AND CO-INFECTIONS ASSOCIATED WITH ZOONOTIC PATHOGEN CARRIAGE IN NORWAY RATS (RATTUS NORVEGICUS)

6.1 ABSTRACT Traditionally, zoonotic pathogen ecology studies in wildlife have focused on the interplay among hosts, their demographic characteristics and their pathogens. But pathogen ecology is also influenced by factors that traverse the hierarchical scale of biological organization, ranging from within host factors at the molecular, cellular, tissue and organ levels, all the way to the host population within a larger environment. The influence of host disease and co-infections on zoonotic pathogen carriage in hosts is important because these factors may be key to a more holistic understanding of pathogen ecology in wildlife hosts, which are a major source of emerging infectious diseases in humans. Using wild, urban Norway rats (Rattus norvegicus) as a model species, the purpose of this study was to investigate how host disease and co-infections impact the carriage of zoonotic pathogens. Following a systematic trap and removal study, we tested the rats for the presence of three potentially zoonotic bacterial pathogens (Bartonella tribocorum, Clostridium difficile and Leptospira interrogans) and assessed them for host disease not attributable to these bacteria (i.e., nematode parasite infections, and macroscopic and microscopic lesions). We fitted multi-level multivariable logistic regression models with pathogen status as the outcome, lesions and parasites as predictor variables, and city block as a random effect. Rats had significantly increased odds of being infected with B. tribocorum if they had a concurrent nematode infection in one or more organ systems. The odds of a rat testing positive for C. difficile were significantly decreased when rats had thyroid goiter. Finally, rats with bite wounds, any macroscopic lesion, cardiomyopathy or tracheitis had significantly increased odds of being infected with L. interrogans. These results suggest that host disease may have an important role in the ecology and epidemiology of rat-associated zoonotic pathogens. We also assessed rats for patterns of co-infections with the three zoonotic pathogens, observing 140

that co-infections were infrequent. However, individuals carrying multiple zoonotic pathogens may pose a disproportionately high risk to people. Our multi-scale approach to assessing complex intra-host factors in relation to zoonotic pathogens carriage may be applicable to future studies in rats and other wildlife hosts.

141

6.2 INTRODUCTION Based on the high proportion of emerging infectious diseases that originate in wild animals, knowledge of zoonotic pathogen ecology in their wildlife hosts is a critical component to understanding pathogen spread among hosts and to humans (Daszak et al. 2001; Mills and Childs 1998; Taylor et al. 2001; Wolfe et al. 2007; Morse et al. 2012; Cunningham et al. 2017). Many studies of zoonotic and potentially zoonotic pathogens in wildlife hosts test animals with no or limited analyses of risk factors associated with host infection. When such studies consider risk factors, they tend to limit their analyses to host demographic characteristics (e.g., age and sex).

These studies often do not incorporate the influence of other factors that may affect pathogen ecology such as the host’s environment and within-host factors (Barrett and Bouley 2015; Ezenwa et al. 2015; Rothenburger et al. 2017). Host diseases—both infectious and non- infectious—may influence behaviour and immune system function with downstream influences on pathogen carriage (Bordes and Morand 2011). Sickness behaviour (i.e., behaviours that occur in response to injury or infection), typically include fever, inappetence, weight loss and lethargy through the systemic actions of pro-inflammatory cytokines (e.g., Interleukin-1 and Tumor Necrosis Factor alpha; Tizard 2008). These disease-associated behavioural changes may also affect foraging activities, contact rates, social interactions (e.g., mating, fighting and grooming) and dispersal patterns, which may in turn, alter pathogen exposure (Hart 1988; Bouwman and Hawley 2010; Ghai et al. 2015). Even individual microscopic lesions that suggest minor disease may substantially affect behavior and immune responses through their cumulative effects (i.e., total disease burden; Snyder et al. 2016). Collectively, behaviour and immune responses to disease are a mechanism for how intra-host factors may influence pathogen transmission and persistence.

A subtype of host disease, namely co-infection (also called polyparasitism and multiparasitism), may have a profound effect on infectious disease epidemiology (Vaumourin et al. 2015). Co- infecting pathogens may include both microparasites (i.e., viruses, bacteria and protozoa) and macroparasites (i.e., nematodes, cestodes, trematodes, fleas, mites and lice). Co-infections may lead to more severe disease manifestations compared to single-agent infections. For example,

142

marine mammals co-infected with Toxoplasma gondii and Sarcocystis neurona had more severe encephalitis and increased mortality compared to those individuals that were infected with either protozoan parasite alone (Gibson et al. 2011). Even when not associated with overt disease, co- infections may exert systemic influences. For instance, nematode infections down regulate immune responses to microparasites and modulate auto-immune diseases (Bordes and Morand 2011; Garza-Cuartero et al. 2014). This modulating effect was demonstrated by a study in African water buffalo in which anthelmintic treatment for intestinal worms increased the survival of Mycobacterium tuberculosis (TB)-infected individuals, which may in turn increase TB shedding and transmission (Ezenwa 2016). In addition to effects on the immune system, macro- and microparasites may directly modify host behaviour to increase their chances of survival and reproduction (Webster 2001; Ezenwa et al. 2016). This phenomenon is evident in the lifecycles of many zoonotic parasites in which the parasite causes behaviour changes in the infected prey animal that increases susceptibility to predation (e.g., Echinococcus multilocularis and Toxoplasma gondii; Webster 2001; Raoul et al. 2015).

The influence of host disease and co-infections on zoonotic pathogen carriage in hosts is important because these factors may be key to a holistic understanding of pathogen ecology in wildlife hosts. Given that the factors contributing to disease emergence on a global scale are persistent and increasing, there is an urgent need to understand how pathogens spread from wildlife to human populations (Wolfe et al. 2007; Morse et al. 2012). Understanding anthropogenic factors in this spread is essential, but host factors such as disease and co- infections may be equally important to zoonotic pathogen ecology (Morse et al. 2012). A nuanced appreciation of how host factors influence pathogen ecology in wildlife host systems may be particularly useful for surveillance and disease prevention activities.

Using wild, urban Norway rats (Rattus norvegicus) as a model species, the purpose of this study was to investigate how host disease and co-infections relate to the carriage of potentially zoonotic pathogens (Bartonella tribocorum, Clostridium difficile and Leptospira interrogans) based on a causal diagram framework (Figure 6.1). Urban rats are an ideal model species for this type of study since they carry many important zoonotic pathogens (e.g., L. interrogans, the causative agent of Weil’s disease and B. elizabethae, B. tribocorum, and other Bartonella spp.,

143

which are causative agents of bartonellosis). Also, large urban rat populations live in close proximity to people in cities, which may enhance the risk of zoonotic pathogen transmission. Because these zoonotic bacterial pathogens are not known to directly cause disease in rats, our objectives were to investigate associations between 1) lesions (both macroscopic and microscopic) and 2) nematode parasites with the carriage of zoonotic pathogens in an urban population of wild rats. Additionally, we aimed to determine how frequently rats were co- infected with two or more zoonotic pathogens and describe the patterns of those co-infections.

6.2 METHODS

6.2.1 Study Design

This study builds on the initial phase of the Vancouver Rat Project (www.vancouverratproject.com), a cross-sectional trap and removal study of rat-associated zoonotic pathogens in Vancouver, British Columbia, Canada. Himsworth et al. (2014a) described the study design and trapping protocol in detail. In short, 43 city blocks were randomly allocated to a two-week trapping period between September 2011 and August 2012. Rats were trapped in the back alleys that bisected the block. Additionally, a professional pest control company collected rats within an international shipping port adjacent to the study area using snap-type lethal traps. The University of British Columbia's Animal Care Committee approved this study (A11-0087).

6.2.2 Autopsy, Pathology and Zoonotic Pathogen Analyses

We collected tissues during a systematic autopsy for each of the 672 euthanized rats at the Animal Health Centre, British Columbia Ministry of Agriculture, Abbotsford, British Columbia. We collected demographic data (body mass, sexual maturity [mature rats had open vaginal orifices for females; scrotal testes for males], sex, internal body fat score) either at the time of capture or during autopsy. We recorded any visible (macroscopic) lesions during the autopsy examination. Then, we assessed tissues from a subsample of rats for microscopic lesions and parasites. We typically examined the following tissues: adrenal gland, esophagus, intestines, liver, lungs, lymph node, kidney, pharynx, skeletal muscle, spleen, stomach, tongue, thyroid gland, trachea and urinary bladder. Not all tissues were available for each rat due to collection errors or autolysis. Using a light microscope, one observer (JR) examined 4 !m-thick sections of

144

formalin-fixed, paraffin-embedded tissues stained with hematoxylin and eosin. We created a binary classification system to systematically and conservatively evaluate tissues for the presence/absence of the most frequent lesions (Chapter 5).

To identify Bartonella tribocorum carriage, blood clot samples from 393 rats were cultured and then the species identity was confirmed by polymerase chain reactions (PCR; Himsworth et al. 2015). Colon contents from 672 rats were cultured for Clostridium difficile (Himsworth et al. 2014b). Kidney tissues from 581 rats were tested by PCR to identify Leptospira interrogans carriage (Himsworth et al. 2013a).

6.2.3 Statistical Analyses

We fitted multi-level univariable logistic regression models with zoonotic pathogen status (positive or negative) as the outcome. We included predictor variables for macroscopic and microscopic lesions that had more than 10% prevalence in the sample and were assessed in >100 rats. Although small foci of tissues were frequently mineralized in this sample (Chapter 5), we did not analyze these observations since they were mild and likely an incidental finding with limited potential impact on individual rat health. We excluded lesion variables that were potentially caused by the pathogen of interest (e.g., L. interrogans with interstitial nephritis and pyelitis). In total, we considered up to 19 lesion variables for statistical modeling. We also included random effects to account for autocorrelation (i.e., clustering) among rats collected from the same city block since multiple rats were captured in each city block. For all statistical testing, we used a significance level of 5% (α= 0.05).

We considered variables for the multivariable model based on their significance in univariable analysis. Due to the potential for correlation between lesions, we estimated the correlation between all independent variables that were significant with univariable modeling using Phi coefficients and considered variables as highly correlated if |ρ > 0.8|.

We constructed a causal diagram to account for the relationships among lesions, parasites, zoonotic pathogens and rat demographic characteristics (Figure 6.1). Lesions and nematode parasites hypothetically influence zoonotic pathogen status via disease-associated behaviour

145

changes and immune system modulation. It is probable that disease in rats influenced both internal body fat and body mass; therefore, these were deemed intervening variables in the relationship between lesions and zoonotic pathogen status and were excluded from multivariable models. We did not include related variables (e.g., lesions in the same organ system attributable to the same cause) or those that were constructs of each other (e.g., single nematode infection in one organ vs. nematode in any organ) within the same model. Instead, we selected the variable with the lowest p-value on univariable analysis for multivariable modeling. We then constructed a full multivariable model with all selected variables that were significant in the univariable analyses and potential confounding variables (i.e., sex and sexual maturity), then sequentially removed non-significant variables using manual backwards selection. Prior to removing sex and sexual maturity, we assessed the potential confounding effects of these variables on the significant associations. We defined confounding variables as non-intervening variables that resulted in ≥ 30% change to other model coefficients when removed from the model. Once we had achieved a base multivariable model, we assessed for interactions between sex and sexual maturity with the remaining significant covariates. The final model consisted of variables that were statistically significant, part of a significant interaction term or acted as a confounding variable.

Using the latent variable technique, we estimated the variance partition coefficient for each hierarchical level (i.e., rat- and city-block levels; Dohoo et al., 2009). To assess model fit, we examined the assumptions of normality and homoscedasticity of the best linear unbiased predictors (BLUPS) by using normal quantile plots and plotting the BLUPS against the predicted log odds of the outcome, respectively. We examined Pearson residuals to identify outlying observations.

To assess co-infections with all three zoonotic pathogens, we extracted patterns for carriage of the three zoonotic pathogens among rats that were tested for all three (n=331). Then, using each pathogen as the independent variable, we fitted univariable multi-level logistic regression models with the other two pathogens as predictors. For most statistical analyses, we used R (R Development Core Team, Vienna, Austria). We used Stata (Stata 14, College Station, Texas, USA) for variance estimates, tests of model fit and to extract pathogen carriage patterns.

146

6.3 RESULTS Table 6.1 provides descriptions of the major macroscopic and microscopic lesions found at >10% prevalence in this sample, as well as demographic characteristics included in the analyses.

6.3.1 Bartonella tribocorum

The prevalence of Bartonella tribocorum was 25.6% (100/390; 95% Confidence Interval [CI]: 21.4%-30.3%). Based on univariable analysis, the following variables were significantly associated with the isolation of B. tribocorum in rats: inducible bronchus-associated lymphoid tissue (IBALT) in the lungs, enteric nematodes, Eucoleus sp. and “nematodes in any organ” (Table 6.2). These variables were not highly correlated (|ρ < 0.8|). Data for enteric nematodes and Eucoleus sp. were captured by the nematode in any organ variable, so of these three variables, we chose only nematode in any organ for multivariable modeling because it had the strongest univariable association with the outcome. Thus, we included IBALT and nematode in any organ for multivariable modeling. When modeled with nematode in any organ in a multivariable model, IBALT was no longer significant. Sex did not confound the relationship for either variable nor was it part of a significant interaction term. Of the 100 rats that tested positive for B. tribocorum, only three were sexually immature; therefore, we could not assess the confounding or interaction effects of sexual maturity on nematode in any organ in the model. In the final model, the odds of a rat testing positive for B. tribocorum were significantly increased when rats had any nematode infections (Odds Ratio [OR]=3.32; 95% CI=1.57-7.40; p=0.002).

6.3.2 Clostridium difficile

The prevalence of C. difficile was 12.9% (87/672; 95% CI: 10.5%-15.7%). Based on univariable analysis, thyroid goiter and interstitial nephritis were significantly associated with the isolation of C. difficile in rats (Table 6.3). When modeled with thyroid goiter in a multivariable model, interstitial nephritis was no longer significant. Neither sex nor sexual maturity confounded the relationship between thyroid goiter or interstitial nephritis and C. difficile positivity. Neither sex nor sexual maturity were part of a significant interaction term. In the final model, the odds of a rat testing positive for C. difficile were significantly decreased when rats had thyroid goiter (OR=0.33; 95% CI=0.13-0.79; p=0.017).

147

6.3.3 Leptospira interrogans

The prevalence of L. interrogans was 11.4% (66/581; 95% CI: 8.8%-14.2%). Based on univariable analysis, the following variables were significantly associated with L. interrogans positivity: Capillaria hepatica in the liver, bite wounds, other macroscopic lesion(s), cardiomyopathy, non-glandular stomach lesions, thyroid goiter, IBALT, tracheitis, urinary crystals and nematode in any organ (Table 6.4). Tracheitis and IBALT in the lungs are related variables since both lesions are attributable to bacterial respiratory infections (cilia-associated respiratory bacillus [CAR bacillus] and/or Mycoplasma pulmonis; Rothenburger et al. 2015a). Similarly, data for C. hepatic in the liver are captured by nematode in any organ variable; therefore, we did not model these variables together in our multivariable model. Instead, we chose to model tracheitis and nematode in any organ, both of which had the strongest univariable associations with the outcome. Sex did not confound any relationships and was not part of a significant interaction term. Of the 66 rats that tested positive for L. interrogans, only one was sexually immature; therefore, we could not assess the confounding or interaction effects of sexual maturity on the significant covariates in the multivariable model. In the final model, the odds of a rat testing positive for L. interrogans were significantly increased when rats had bite wounds, any other macroscopic lesion(s), cardiomyopathy or tracheitis (Table 6.5).

6.3.4 Co-infections Among Zoonotic Pathogens

Table 6.6 includes the patterns of zoonotic pathogen carriage among 331 rats tested for all three pathogens. The most frequent pattern was when rats tested negative for all three pathogens, which occurred in 59.5% of rats. The most common co-infection was with L. interrogans and B. tribocorum, occurring in 10 rats (3.0%). The second most common co-infection pattern was with B. tribocorum and C. difficile, in seven rats (2.1%). A mature male rat was infected with L. interrogans, B. tribocorum and C. difficile. The observed frequency of infection patterns did not differ from the expected values. The only statistically significant associations were between B. tribocorum and L. interrogans carriage; the presence of one pathogen significantly decreased the odds of rat testing positive for the other (Table 6.7).

148

6.3.5 Model Diagnostics

For all models, BLUPS were normally distributed with constant variance. Observations with large Pearson residuals were not recording errors and their removal did not impact the models. For the B. tribocorum multivariable model, the four rats with the largest Pearson residuals were males that tested positive for B. tribocorum and negative for nematode in any organ. For the C. difficile multivariable model, the four rats with the largest Pearson residuals were sexually mature, tested positive for C. difficile and had thyroid goiter. For both B. tribocorum and C. difficile multivariable models, most of the variation in pathogen occurrence was explained at the individual rat level; 40.3% (95% CI: 18.5%-66.7%) of the variance in the B. tribocorum multivariable model and 12.8% (95% CI: 1.7%-55.4%) of variance in the C. difficile multivariable model was explained at the block level. Approximately half the variation in L. interrogans occurrence was explained at the individual rat level, while 50% of variance in the multivariable model was explained at the block level (95% CI: 21.3%-78.6%).

6.4 DISCUSSION The objectives of this study were to assess the relationships between host disease and co- infections with zoonotic pathogens using wild, urban Norway rats as a model species. We observed associations with lesions for three bacterial pathogens, but the associated lesions differed by pathogen. We also identified patterns of co-infections among zoonotic pathogens.

6.4.1 Bartonella tribocorum

Bartonella spp. are potentially zoonotic pathogenic bacteria that establish chronic, sub-clinical infections in rodents, including rats, worldwide (Schulein et al. 2001; Meerburg et al. 2009). In infected hosts, Bartonella spp. adhere to erythrocytes, which facilitates transmission by fleas and other hemophagocytic arthropod vectors (Schulein et al. 2001; Gutiérrez et al. 2015). There is accumulating evidence that rat-associated Bartonella spp. are zoonotic, causing lymphadenopathy, neuroretinitis, endocarditis, myocarditis, acute febrile illness, anemia and chronic fatigue in people (Kosoy et al. 2010; Buffet et al. 2013; Kandelaki et al. 2016; Vayssier- Taussat et al. 2016).

149

In our study, rats had increased odds of being infected with B. tribocorum if they had one or more nematode infections (i.e., Eucoleus sp. in the upper gastrointestinal system, C. hepatica in the liver, nematodes in the small intestine and/or Trichosomoides crassicauda in the urinary bladder). As with any cross-sectional study, the time sequence of infection is unknown, and thus, observed nematode infections may have occurred before or after B. tribocorum infection. In the former case, immune modulation by nematodes could impact the acquisition of Bartonella spp. and the establishment of chronic persistent infections in hosts like rats. If immune modulation as described in other species works similarly in rats, then down regulation of T-Helper type 1 (TH1) immunity in response to nematode infections may alter immune responses to infections with microparasites such as Bartonella spp. (Garza-Cuartero et al. 2014).

Our results contradict previous research that observed a lack of association between macroparasites and microparasites that occupy tissues in different organ systems. In these previous studies, the strongest interactions occurred when parasites co-existed in the same tissues (Knowles et al. 2013; Henrichs et al. 2016). For instance, an experimental study of micro- and macroparasites in wood mice (Apodemus sylvaticus) revealed that anthelmintic treatment had no effect on parasites of the cardiovascular system (i.e., Bartonella sp. and Trypanosoma grosi) but resulted in 15-fold increase of the number of intestinal Eimeria spp. oocysts shed compared to untreated controls (Knowles et al. 2013). The authors speculate that the observed effect is related to the fact that both Eimeria spp. and helminth intestinal parasites occupy similar intestinal niches and thus the removal of the latter enhanced the replication and shedding of the former. The results of our study suggest that there may be effects across organ systems in rats that are co- infected with B. tribocorum and nematode parasites. Alternatively, similar behaviours may facilitate rat exposure to both nematodes and Bartonella-carrying fleas. In addition, environmental factors such as moderate weather may concurrently enhance flea and nematode transmission (Gutiérrez et al. 2015).

6.4.2 Clostridium difficile

Clostridium difficile is an enteric bacterium that is associated with diarrheal disease in a variety of animals and people (Warriner et al. 2016). A wide range of apparently healthy animals, including rats, may carry C. difficile and it is a potenial food-borne pathogen (Hensgens et al.

150

2012; Warriner et al. 2016). Yet the role of rats in the epidemiology of this bacterium is uncertain (Himsworth et al. 2014b).

Rats with thyroid goiter had significantly decreased odds of carrying C. difficile. The reason for this association is not unclear but may include indirect protective effects of goiter against C. difficile carriage. The cause(s) of the thyroid lesions in these rats is also unknown and may include iodine deficiency or exposure to goitrogenic substances. Perhaps areas of the city that have iodine deficient rat diets/goitrogens correspond to those with increased environmental C. difficile spores. It is also possible that thyroid goiter is a proxy for an unmeasured factor that contributes to reduced carriage of this pathogen. Finally, thyroid function has a major impact on the immune system, so the goiter in these rats may represent a non-infectious host factor that could influence pathogen ecology indirectly via immune system effects (De Vito et al. 2011).

6.4.3 Leptospira interrogans

In terms of global disease burden, Leptospira spp. are among the most common and important zoonotic pathogens, as they cause Weil’s Disease, pulmonary hemorrhage syndrome and non- specific febrile illnesses in infected people (Costa et al. 2015). In reservoir hosts like rats, the bacteria colonize renal tubules without causing clinical signs and is shed in the urine (Bonilla- Santiago and Nally 2011). Rats are an emerging and important source of Leptospira sp. infection in urban environments where people become infected primarily through contact with urine- contaminated water (Evangelista and Coburn 2010).

Rats with bite wounds, macroscopic lesions, tracheitis or cardiomyopathy had increased odds of testing positive for L. interrogans in the present study. On an individual animal level, these lesions have the potential to substantially impact rat health. Collectively, the impacts are expected to compound, leading to a high overall total disease burden that may have an effect on rat health and subsequent impacts on disease-associated behaviour (Snyder et al. 2016). Macroscopic lesions – fractures, internal abscesses and tumors, for example – had to be severe and grossly evident to be considered present for this variable. Cardiomyopathy is a known cause of mortality in laboratory rats and is suspected to similarly impact wild rat health (Keenan et al. 1995; Rothenburger et al. 2015b). Tracheitis, associated with respiratory pathogens (CAR

151

bacillus and M. pulmonis), is indicative of severe respiratory disease (Rothenburger et al. 2015a). Bite wounds ranged in severity but were often infected (Himsworth et al. 2014c). Himsworth et al. (2013a) identified an association between L. interrogans and the number of bite wounds, as well as body mass and increased internal fat, in a previous study of this sample of rats.

Of the three pathogens we examined, the relationship between host disease and L. interrogans is apparently the most biologically complex. Furthermore, the lesions that were associated with L. interrogans-positivity were more severe than those associated with C. difficile and B. tribocorum (thyroid goiter and nematode infections, respectively). Although we cannot deduce the time sequence of these invents in a cross-sectional study, if the lesions occurred prior to L. interrogans infection, they may have led to sickness behaviours that altered the likelihood of exposure to this pathogen (Hart 1988; Bouwman and Hawley 2010; Ghai et al. 2015). It is possible that disease-related changes to foraging behaviour and dominance may have influence exposure to environmental sources of L. interrogans. Similarly, changes to thirst and water consumption are a feature of sickness behaviour, so perhaps sick rats are more likely to drink from L. interrogans-contaminated water sources compared to those that lack these specific lesions (Tizard 2008).

6.4.4 Co-infections

Association analysis is a valid technique to assess relationships among a limited number of co- infecting pathogens, as was the case in our study (Vaumourin et al. 2015). While only one rat, a mature male, concurrently carried L. interrogans, B. tribocorum and C. difficile, rats were occasionally co-infected with two of these pathogens, most often L. interrogans and B. tribocorum. Since the observed frequency of infection patterns did not differ from the expected values, our results suggest that co-infections with these pathogens may occur less frequently than expected by chance. This conclusion is supported by a study of Brazilian rats that observed co- infections with Leptospira sp. and Bartonella sp. in approximately 7% of rats (Costa et al. 2014). They observed that the prevalence of co-infections was not statistically different than expected, suggesting that the observed infections occurred independently rather than as a result of a biological interaction. Using a logistic regression model, we observed that rats infected with B. tribocorum were significantly less likely to be infected with L. interrogans and vice versa. This

152

result may be due to differing transmission routes. While L. interrogans is transmitted via contact with infected urine in the environment, B. tribocorum is primarily transmitted by fleas (Evangelista and Coburn 2010; Gutiérrez et al. 2015).

Other studies of rats have also identified co-infections with a variety of zoonotic pathogens, although it is often not clear if these are the result of pathogen interactions (Kim et al. 2007; Easterbrook et al. 2007; Zhao et al. 2013; Laudisoit et al. 2014). For instance, a study of pathogens carried by rats in New York City, USA, observed that approximately 9% of rats (10/133) were co-infected with two or more bacterial pathogens, which is similar to our results (Firth et al. 2014). In the same study, rats were occasionally co-infected with flaviviruses and 5% (6/114) of rats were infected with ≥ 3 flaviviruses. The rates of co-infection are likely dependent on the pathogens themselves as well as other factors. For instance, a study set in Rhône, France, observed that rats that were infected with L. interrogans had significantly increased odds of being co-infected with either Seoul hantavirus or hepatitis E virus (Ayral et al. 2015). Whether co-infections arise by coincidence or through a biological interaction, these so-called “super- carrier” rats may pose a disproportionate risk to public health via their ability to transmit multiple pathogens through a variety of transmission routes.

6.4.5 Implications

Our multi-scale approach analyzed host disease associated with zoonotic pathogen carriage. This type of study is challenging since histopathology requires specialized training and invasive/lethal sampling and thus, may only be ethical under specific circumstances (e.g., invasive/pest species and animals harvested for subsistence hunting; Sullivan et al. 2003; Kutz et al. 2013; Carnegie et al. 2016). Some studies have examined associations between zoonotic pathogen carriage and non-disease physiological states. For instance, bats that carry paramyxoviruses and Leptospira sp. are more likely to test positive for both pathogen types during pregnancy and immediately post- partum, possibly as a result of pregnancy-associated immunosuppression (Dietrich et al. 2015). In another study of bats, shedding of Hendra virus, a zoonotic paramyxovirus, is associated with pregnancy, lactation and nutritional stress (determined by the combination of low body mass, poor body condition, food shortages, abnormal feeding behaviour and lack of reproduction; Plowright et al. 2008). The pathogens that we included in the present study and those

153

investigated in previous studies of co-infections in other wild rodents do not cause clinical disease (Himsworth et al. 2013b; 2014b; Knowles et al. 2013). Our results suggest that host disease and co-infections is an important intra-host factor worthy of consideration in these systems. But despite the fundamental role of co-infections in pathogen dynamics, there appear to be few studies that examine co-infections in relation to zoonotic pathogen carriage in rats and other species (Vaumourin et al. 2015).

We did not assess all possible microparasite infections nor did we evaluate the entire microbiome of commensal organisms, which is a common criticism of many co-infection studies (Bordes and Morand 2011). However, it is currently difficult to account for and analyze all infections in a meaningful way. Histopathology provides a way to crudely assess host responses to a range of infections. For example, several pathogens including CAR bacillus, M. pulmonis, other bacteria and several viruses all cause respiratory disease in laboratory rats, and there is evidence that many of these are also important pathogens of wild rats (Baker 1998; Rothenburger et al. 2015a). We included the variable for IBALT changes in the lung to account for the collective impact of any respiratory tract infections, which may be more meaningful for this type of analysis since it is an objective host response compared to using molecular or serological tests for all possible respiratory pathogens without accounting for the host response to infection. Standard pathology techniques like autopsy examinations and histopathology have utility in this age of molecular diagnostics (Carnegie et al. 2016). Furthermore, identifying a pathogen within a lesion (e.g., Eucoleus sp. associated with hyperkeratosis, mucosal hyperplasia and submucosal inflammation; Rothenburger et al. 2014) adds to the causal assumptions of the role of that pathogen in a disease process (Morse et al. 2012).

Despite the utility of histopathology, it is probable that its use to identify the presence of parasite infections likely underestimated their true prevalence. Direct examination of tissues for parasites would have been preferable. And in the present study, histopathology alone did not allow us to definitively identify enteric parasite species.

Future studies could experimentally manipulate rats in an urban setting to better understand the associations we found. For instance, it would be informative to explore how specific

154

anthelmintic treatment that does not affect fleas may impact the acquisition of Bartonella sp. in urban rats. Prospective studies that seek to document the acquisition of infections may be facilitated by new trapping methodologies that selectively target specific individual rats (Vaumourin et al. 2015; Parsons et al. 2015). Since rats are a ubiquitous laboratory animal, field studies that uncover relationships between co-infections and zoonotic parasites could be verified in laboratory settings (Ezenwa 2016).

6.5 CONCLUSIONS This study is an example of a multi-faceted approach to understanding how aspects of intra-host variation, namely host disease and co-infections, may impact pathogen ecology. While other studies have examined associations with co-infections, our results provide evidence that zoonotic pathogen carriage in rats varies with co-infection status and a variety of lesions indicative of disease. Some rats were co-infected with two or more zoonotic pathogens; these individuals may be particularly important for pathogen transmission to other rats and to people should a suitable contact with a susceptible host or environment occur. These concepts and techniques could be applied to other hosts and zoonotic pathogens to gain a better understanding of the ecology of zoonotic pathogens in their host—information that may be crucial for the study and control of emerging infectious diseases.

155

6.6 REFERENCES Ayral F, Artois J, Zilber AL, Widén F, Pounder KC, Aubert D, Bicout DJ, Artois M. 2015. The relationship between socioeconomic indices and potentially zoonotic pathogens carried by wild Norway rats: a survey in Rhône, France (2010-2012). Epidemiology and Infection 143:586-599.

Baker DG. 1998. Natural pathogens of laboratory mice, rats, and rabbits and their effects on research. Clinical Microbiology Reviews 11:231–266.

Barrett MA, Bouley TA. 2015. Need for enhanced environmental representation in the implementation of One Health. EcoHealth 12:212–219.

Bonilla-Santiago R, Nally JE. 2011. Rat model of chronic leptospirosis. Current Protocols in Microbiology 12:12E.1-12E.3.8.

Bordes F, Morand S. 2011. The impact of multiple infections on wild animal hosts: a review. Infection Ecology and Epidemiology 1:7346.

Bouwman KM, Hawley DM. 2010. Sickness behaviour acting as an evolutionary trap? Male house finches preferentially feed near diseased conspecifics. Biology Letters 6:462–465.

Buffet JP, Kosoy M, Vayssier-Taussat M. 2013. Natural history of Bartonella-infecting rodents in light of new knowledge on genomics, diversity and evolution. Future Microbiology 8:1117- 1128.

Costa F, Porter FH, Rodrigues G, Farias H, de Faria MT, Wunder EA, Osikowicz LM, Kosoy MY, Reis MG, Ko AI, Childs JE. 2014. Infections by Leptospira interrogans, Seoul virus, and Bartonella spp. among Norway rats (Rattus norvegicus) from the urban slum environment in Brazil. Vector Borne and Zoonotic Diseases 14:33–40.

Daszak P, Cunningham AA, Hyatt AD. 2001. Anthropogenic environmental change and the emergence of infectious diseases in wildlife. Acta Tropica 78:103–16.

156

De Vito P, Incerpi S, Pedersen JZ, Luly P, Davis FB, Davis PJ. 2011. Thyroid hormones as modulators of immune activities at the cellular level. Thyroid 21:879–890.

Dietrich M, Wilkinson DA, Benlali A, Lagadec E, Ramasindrazana B, Dellagi K, Tortosa P. 2015. Leptospira and paramyxovirus infection dynamics in a bat maternity enlightens pathogen maintenance in wildlife. Environmental Microbiology 17:4280–4289.

Evangelista KV, Coburn J. 2010. Leptospira as an emerging pathogen: a review of its biology, pathogenesis and host immune responses. Future Microbiology 5:1413–1425.

Ezenwa VO. 2016. Helminth-microparasite co-infection in wildlife: lessons from ruminants, rodents and rabbits. Parasite Immunology 38:527–534.

Ezenwa VO, Archie EA, Craft ME, Hawley DM, Martin LB, Moore J, White L. 2016. Host behaviour-parasite feedback: an essential link between animal behaviour and disease ecology. Proceedings Biological Sciences 283:20153078–9.

Ezenwa VO, Prieur-Richard A-H., Roche B, Bailly X, Becquart P, García-Peña GE, Hosseini PR, Keesing F, Rizzoli A, Suzán G, Vignuzzi M, Vittecoq M, Mills JN, Guégan JF. 2015. Interdisciplinarity and infectious diseases: an Ebola case study. PLoS Pathogens 11:e1004992.

Ghai RR, Fugère V, Chapman CA, Goldberg TL, Davies TJ. 2015. Sickness behaviour associated with non-lethal infections in wild primates. Proceedings. Biological Sciences 282:20151436–8.

Gutiérrez R, Krasnov B, Morick D, Gottlieb Y, Khokhlova IS, Harrus S. 2015. Bartonella infection in rodents and their flea ectoparasites: an overview. Vector Borne Zoonotic Diseases, 15:27–39.

157

Hart BL. 1988. Biological basis of the behavior of sick animals. Neuroscience and Biobehavioral Reviews 12:123–137.

Henrichs B, Oosthuizen MC, Troskie M, Gorsich E, Gondhalekar C, Beechler BR, Ezenwa VO, Jolles AE. 2016. Within guild co-infections influence parasite community membership: a longitudinal study in African Buffalo. Journal of Animal Ecology 85:1025–1034.

Himsworth CG, Bai Y, Kosoy MY, Wood H, DiBernardo A, Lindsay R, Bidulka J, Tang P, Jardine C, Patrick D. 2015. An Investigation of Bartonella spp., Rickettsia typhi, and Seoul Hantavirus in Rats (Rattus spp.) from an inner-city neighborhood of Vancouver, Canada: is pathogen presence a reflection of global and local rat population structure? Vector Borne and Zoonotic Diseases 15:21–26

Himsworth CG, Parsons KL, Feng AYT, Kerr T, Jardine CM, Patrick DM. 2014a. A mixed methods approach to exploring the relationship between Norway rat (Rattus norvegicus) abundance and features of the urban environment in an inner-city neighborhood of Vancouver, Canada. PLoS ONE 9:e97776.

Kandelaki G, Malania L, Bai Y, Chakvetadze N, Katsitadze G, Imnadze P, Nelson CA, Harrus S, Kosoy MY. 2016. Human lymphadenopathy caused by ratborne Bartonella, Tbilisi, Georgia. Emerging Infectious Diseases 22:544-546.

158

Kim HC, Klein TA, Chong ST, Collier BW, Yi SC, Song KJ, Baek LJ, Song JW. 2007. Seroepidemiological survey of rodents collected at a US military installation, Yongsan garrison, Seoul, Republic of Korea. Military Medicine 172:759–764.

Knowles SCL, Fenton A, Petchey OL, Jones TR, Barber R, Pedersen AB. 2013. Stability of within-host-parasite communities in a wild mammal system. Proceedings Biological Sciences 280:20130598–20130598.

Kosoy M, Bai Y, Sheff K, Morway C, Baggett H, Maloney SA, Boonmar S, Bhengsri S, Dowell SF, Sitdhirasdr A, Lerdthusnee K. 2010. Identification of Bartonella infections in febrile human patients from Thailand and their potential animal reservoirs. American Journal of Tropical Medicine and Hygiene 82:1140-1145.

Kutz S, Ducrocq J, Cuyler C, Elkin B, Gunn A, Kolpashikov L, Russell D, White RG 2013. Standardized monitoring of Rangifer health during International Polar Year. Rangifer 33:91–114.

Laudisoit A, Falay D, Amundala N, Akaibe D, de Bellocq JG, Van Houtte N, Breno M, Verheyen E, Wilschut L, Parola P, Raoult D. 2014. High prevalence of Rickettsia typhi and Bartonella species in rats and fleas, Kisangani, Democratic Republic of the Congo. American Journal of Tropical Medicine and Hygiene 90:463-8.

Meerburg BG, Singleton GR, Kijlstra A. 2009a. Rodent-borne diseases and their risks for public health. Critical Reviews in Microbiology 35:221–270.

Mills JN, Childs JE. 1998. Ecologic studies of rodent reservoirs: their relevance for human health. Emerging Infectious Diseases 4:529–537.

Morse SS, Mazet JAK, Woolhouse M, Parrish CR, Carroll D, Karesh WB, Zambrana-Torrelio C, Lipkin WI, Daszak P. 2012. Prediction and prevention of the next pandemic zoonosis. Lancet 380:1956–1965.

Parsons MH, Sarno RJ, Deutsch MA. 2015. Jump-starting urban rat research: conspecific pheromones recruit wild rats into a behavioral and pathogen-monitoring assay. Frontiers in Ecolology and Evolution 3:487.

Plowright RK, Field HE, Smith C, Divljan A, Palmer C, Tabor G, Daszak P, Foley JE. 2008. Reproduction and nutritional stress are risk factors for Hendra virus infection in little red flying foxes (Pteropus scapulatus). Proceedings of the Royal Society B: Biological Sciences 275:861– 869.

Raoul F, Hegglin D, Giraudoux P. 2015. Trophic ecology, behaviour and host population dynamics in Echinococcus multilocularis transmission. Veterinary Parasitology 213:162–171.

159

Rothenburger JL, Himsworth CG, Clifford CB, Ellis J, Treuting PM, Leighton FA. 2015. Respiratory pathology and pathogens in wild urban rats (Rattus norvegicus and Rattus rattus). Veterinary Pathology 52:1210–1219.

Rothenburger JL, Himsworth CG, Treuting PM, Leighton FA. 2015b. Survey of cardiovascular pathology in wild urban Rattus norvegicus and Rattus rattus. Veterinary Pathology 52:201–208.

Rothenburger JL, Himsworth CG, Nemeth NM, Pearl DL, Jardine CM. 2017. Environmental factors and zoonotic pathogen ecology in urban exploiter species. EcoHealth 14:630-641.

Schulein R, Seubert A, Gille C, Lanz C, Hansmann Y, Piémont Y, Dehio C. 2001. Invasion and persistent intracellular colonization of erythrocytes. A unique parasitic strategy of the emerging pathogen Bartonella. Journal of Experimental Medicine 193:1077–1086.

Snyder JM, Ward JM, Treuting PM. 2016. Cause-of-death analysis in rodent aging studies. Veterinary Pathology 53:233–243.

Sullivan TP, Sullivan DS, Ransome DB. 2003. Impact of removal-trapping on abundance and diversity attributes in small-mammal communities. Wildlife Society Bulletin 31:464-474.

Taylor LH, Latham SM, Woolhouse MEJ. 2001. Risk factors for human disease emergence. Philosophical Transactions of the Royal Society B: Biological Sciences 356:983–989.

Vayssier-Taussat M, Moutailler S, Féménia F, Raymond P, Croce O, La Scola B, Fournier PE, Raoult D. 2016. Identification of novel zoonotic activity of Bartonella spp., France. Emerging Infectious Diseases 22:457–462.

Tizard I. 2008. Sickness behavior, its mechanisms and significance. Animal Health Research Reviews 9:87–99.

Webster JP. 2001. Rats, cats, people and parasites: the impact of latent toxoplasmosis on behaviour. Microbes and Infection 3:1037–1045.

160

Wolfe ND, Dunavan CP, Diamond J. 2007. Origins of major human infectious diseases. Nature 447:279–283.

Zhao X-G, Li H, Sun Y, Zhang Y-Y, Jiang J-F, Liu W, Cao W-C. 2013. Dual infection with Anaplasma phagocytophilum and Babesia microti in a Rattus norvegicus, China. Ticks and Tick- borne Diseases 4:399–402.

161

6.7 TABLES Table 6.1 Descriptions of major macroscopic and microscopic lesions observed at >10% prevalence and demographic characteristics among wild Norway rats (Rattus norvegicus) from Vancouver, Canada.

Variable Description n/Na Prevalence 95% CI b Major Macroscopic Lesions Capillaria hepatica in White, tortuous tracts on surface of liver from C. hepatica infection 242/672 36.0 32.3-40.0 liver Bite wounds Wounds on the skin 167/671 24.9 21.7-28.3

Other macroscopic lesion Major macroscopic lesion in any organs system excluding above lesions 71/672 10.6 8.3-13.1 (e.g., fractures, internal abscesses, tumors) Cardiovascular

Cardiomyopathy Heart muscle inflammation and degeneration (lymphoplasmacytic 128/406 31.5 27.0-36.0 myocarditis, fibrosis and/or myocardial degeneration)

Medial hypertrophy of Thickening of the smooth muscle within the blood vessel wall 90/404 22.3 18.3-26.7 pulmonary arterioles

Digestive

Non-glandular stomach Hyperkeratosis, mucosal hyperplasia, keratin pustules and/or submucosal 231/388 59.5 54.4-64.5 lesions inflammation

Eucoleus sp. At least one egg or nematode cross-section in the mucosal of the tongue, 164/399 41.1 36.2-46.1 oropharynx, esophagus or stomach

Enteric nematodes At least one nematode cross-section in the intestines 83/198 41.9 35.0-49.1

162

Variable Description n/Na Prevalence 95% CI b Endocrine

Thyroid goiter Small to absent thyroid follicular lumens that are devoid of colloid and 142/279 50.9 44.9-56.9 hypertrophic follicular epithelial cells

Respiratory

Cilia Associated Mats of filamentous bacteria on ciliated epithelium of the trachea or nasal 66/270 24.4 19.4-30.0 Respiratory Bacillus cavity

Epiglossitis/Laryngitis Inflammation of the epiglottis and/or larynx (lymphoplasmacytic and/or 79/202 39.1 32.3-46.2 granulocytic submucosal inflammation)

Inducible bronchus- Cuffs of immune cells (lymphocytes and plasma cells) surrounding 270/403 67.0 62.2-71.6 associated lymphoid tissue bronchioles and/or pulmonary blood vessels

Perivascular mixed Cuffs of mixed inflammatory cells (lymphocytes, plasma cells and 80/404 19.8 16.0-24.0 inflammation granulocytes) surrounding pulmonary blood vessels

Tracheitis Inflammation of the trachea (lymphoplasmacytic and/or granulocytic 192/372 51.6 46.6-56.8 submucosal/periglandular inflammation)

Tracheal gland ectasia Dilation of tracheal submucosal glands 68/364 18.7 14.8-23.1

Urinary

Crystals Crystals within renal tubules 137/405 33.8 29.2-38.7

Interstitial nephritis Inflammation in the kidney interstitium (lymphoplasmacytic), excluding 121/405 29.9 25.5-34.6 areas immediately adjacent to the renal pelvis Pyelitis Inflammation in kidney interstitium (lymphoplasmacytic) immediately 85/255 33.3 27.6-39.5 adjacent to the renal pelvis

Trichosomoides At least one nematode or egg cross-section in urinary bladder or renal 59/194 30.4 24.0-37.4 crassicauda pelvis

163

Variable Description n/Na Prevalence 95% CI b Demographic Characteristics

Sex Male 374/664 56.3 52.5-60.1

Female 290/664 43.7 39.9-47.5

Sexual maturity Mature (females: open vaginal orifice; males: scrotal testes) 389/608 64.0 60.0-67.8

Immature (females: closed vaginal orifice; males: inguinal/abdominal 219/608 36.0 32.2-40.0 testes)

a The number of rats examined for each lesion/demographic characteristic varies because not all tissues were available for each individual rat due to collection error, tissue artifacts and/or autolysis. b 95% confidence interval

164

Table 6.2 Major macroscopic and microscopic lesions found at >10% prevalence and their univariable associations with Bartonella tribocorum infection in 390 wild Norway rats (Rattus norvegicus) from Vancouver, Canada.

Variable Sub-category Number of Bartonella Odds Ratio 95% CIb P tribocorum positive rats in each category (%) (n = 100)a

Major Gross Lesions Capillaria hepatica in liver No 32/212 (15.1) ref Yes 68/178 (38.2) 1.95 1.00-3.83 0.051 Bite wounds No 74/283 (26.1) ref Yes 26/106 (24.5) 0.97 0.50-1.87 0.935 Other macroscopic lesion No 84/341 (24.6) ref Yes 16/49 (32.7) 1.27 0.57-2.81 0.553 Cardiovascular

Cardiomyopathy No 50/228 (21.9) ref Yes 37/114 (32.5) 1.50 0.79-2.83 0.212 Medial hypertrophy of pulmonary arterioles No 56/261 (21.5) ref

Yes 30/79 (38.0) 1.82 0.96-3.49 0.068

Digestive Non-glandular stomach lesions No 25/132 (18.9) ref Yes 59/197 (29.9) 0.97 0.48-1.92 0.922

Eucoleus sp. No 36/193 (18.7) ref Yes 51/144 (35.4) 2.04 1.12-3.74 0.020 Enteric nematodes No 17/74 (23.0) ref Yes 25/63 (39.7) 2.46 1.02-6.16 0.047 Endocrine Thyroid goiter No 31/123 (25.2) ref Yes 27/115 (23.5) 0.82 0.39-1.66 0.577 Respiratory Cilia Associated Respiratory Bacillus No 39/182 (21.4) ref 165

Variable Sub-category Number of Bartonella Odds Ratio 95% CIb P tribocorum positive rats in each category (%) (n = 100)a Yes 19/51 (37.3) 1.44 0.63-3.25 0.381 Epiglossitis/Laryngitis No 26/111(23.4) ref Yes 21/68 (30.9) 0.92 0.39-2.11 0.853 Inducible bronchus-associated lymphoid No 15/109 (13.8) ref tissue Yes 71/231 (30.7) 2.35 1.12-5.19 0.028 Perivascular mixed inflammation No 66/277 (23.8) ref Yes 20/63 (31.7) 1.20 0.56-2.56 0.636 Tracheitis No 25/157 (15.9) ref Yes 56/161 (34.8) 1.40 0.67-2.92 0.366 Tracheal gland ectasia No 58/257 (22.6) ref Yes 20/54 (37.0) 2.15 0.99-4.70 0.053 Urinary Crystals No 58/224 (25.9) ref Yes 29/117 (24.8) 1.14 0.62-2.11 0.675 Interstitial nephritis No 63/241 (26.1) ref Yes 24/100 (24.0) 1.13 0.57-2.21 0.724 Pyelitis No 33/142 (23.2) ref Yes 19/71 (26.8) 0.90 0.40-1.99 0.797 Trichosomoides crassicauda No 25/120 (20.8) ref Yes 18/54 (33.3) 1.57 0.63-3.92 0.325 Composite Variable Nematode in any organc No 12/141 (8.5) ref Yes 88/249 (35.3) 3.32 1.57- 7.40 0.002

a Not all tissues were available to assess each lesion, therefore the numbers may not add up to 100%. b 95% confidence interval c We classified rats as positive for this variable if they were infected with one or more of Capillaria hepatica, Eucoleus sp., enteric nematodes and Trichosomoides crassicauda.

166

Table 6.3 Major macroscopic and histological lesions found at >10% prevalence and their univariable associations with Clostridium difficile carriage in 672 wild Norway rats (Rattus norvegicus) from Vancouver, Canada.

Variable Sub-category Number of Clostridium difficile positive rats Odds 95% CIb P in each category Ratio (%) (n = 87) a

Major Gross Lesions Capillaria hepatica in liver No 60/430 (14.0) ref Yes 27/242 (11.2) 0.86 0.48-1.53 0.605 Bite wounds No 68/504 (13.5) ref Yes 19/167 (11.4) 0.69 0.38-1.21 0.205 Other macroscopic lesion No 79/601 (13.1) ref Yes 8/71 (11.3) 0.87 0.36-1.88 0.743 Cardiovascular

Cardiomyopathy No 28/278 (10.1) ref Yes 12/128 (9.4) 0.87 0.42-1.83 0.715

Medial hypertrophy of pulmonary arterioles No 29/314 (9.2) ref Yes 9/90 (10.0) 1.23 0.51-2.74 0.625

Digestive Non-glandular stomach lesions No 17/157 (10.8) ref Yes 19/231 (8.2) 0.75 0.36-1.57 0.439 Eucoleus sp. No 21/235 (8.9) ref Yes 17/164 (10.4) 1.16 0.57-2.32 0.685 Enteric nematodes No 11/115 (9.6) ref Yes 7/83 (8.4) 0.94 0.31-2.73 0.913 Endocrine Thyroid goiter No 19/137 (13.9) ref Yes 8/142 (5.6) 0.33 0.13-0.79 0.017 Respiratory Cilia Associated Respiratory Bacillus No 23/204 (11.3) ref

167

Variable Sub-category Number of Clostridium difficile positive rats Odds 95% CIb P in each category Ratio (%) (n = 87) a Yes 5/66 (7.6) 0.65 0.21-1.65 0.395 Epiglossitis/Laryngitis No 15/123 (12.2) ref Yes 7/79 (8.9) 0.67 0.22-1.84 0.442 Inducible bronchus-associated lymphoid No 16/133 (12.0) ref tissue Yes 22/270 (8.1) 0.62 0.30-1.33 0.213 Perivascular mixed inflammation No 31/324 (9.6) ref Yes 7/80 (8.7) 1.04 0.39-2.54 0.928 Tracheitis No 19/180 (10.6) ref Yes 15/192 (7.8) 0.72 0.35-1.46 0.361 Tracheal gland ectasia No 29/296 (9.8) ref Yes 5/68 (7.4) 0.73 0.24-1.81 0.534 Urinary Crystals No 28/268 (10.4) ref Yes 11/137 (8.0) 0.77 0.35-1.58 0.489 Interstitial nephritis No 33/284 (11.6) ref Yes 6/121 (5.0) 0.33 0.11-0.80 0.022 Pyelitis No 19/170 (11.2) ref Yes 6/85 (7.1) 0.49 0.16-1.32 0.178 Trichosomoides crassicauda No 14/135 (10.4) ref Yes 5/59 (8.5) 0.80 0.25-2.23 0.683 Composite Variable Nematode in any organc No 52/352 (14.8) ref Yes 35/320 (10.9) 0.71 0.41-1.22 0.214

168

Table 6.4 Major macroscopic and histological lesions found at >10% prevalence and their univariable associations with Leptospira interrogans carriage in 581 wild Norway rats (Rattus norvegicus) from Vancouver, Canada.

Variable Sub- Number of Leptospira interrogans positive Odds Ratio 95% CIb P category rats in each category (%) (n = 66) a Major Gross Lesions Capillaria hepatica in liver No 22/386 (5.7) ref Yes 44/195 (22.6) 2.07 1.01-4.30 0.047 Bite wounds No 28/437 (6.4) ref Yes 38/144 (26.4) 5.08 2.60-10.25 <0.001 Other macroscopic lesion No 48/526 (9.1) ref Yes 18/55 (32.7) 5.39 2.26-13.40 <0.001 Cardiovascular Cardiomyopathy No 20/258 (7.8) ref Yes 34/119 (31.6) 3.95 1.91-8.42 <0.001 Medial hypertrophy of pulmonary No 40/294 (13.6) ref arterioles Yes 14/81 (17.3) 1.61 0.68-3.78 0.275 Digestive Eucoleus sp. No 24/219 (11.0) ref Yes 29/151 (19.2) 1.47 0.73-3.00 0.280 Enteric nematodes No 22/96 (22.9) ref Yes 16/73 (21.9) 0.89 0.35-2.18 0.793 Non-glandular stomach lesions No 9/150 (6.0) ref Yes 43/209 (20.6) 2.85 1.23-7.15 0.018 Endocrine Thyroid goiter No 12/131 (9.2) ref Yes 27/132 (20.5) 2.84 1.16-7.50 0.027 Respiratory Cilia Associated Respiratory Bacillus No 25/194 (12.9) ref Yes 15/56 (26.8) 1.26 0.49-3.28 0.629 Epiglossitis/Laryngitis No 14/118 (11.9) ref Yes 18/70 (25.7) 2.24 0.88-5.81 0.090

169

Variable Sub- Number of Leptospira interrogans positive Odds Ratio 95% CIb P category rats in each category (%) (n = 66) a Inducible bronchus-associated No 1/130 (0.8) ref lymphoid tissue Yes 52/244 (21.3) 21.18 4.11-389.88 0.004 Perivascular mixed inflammation No 49/304 (16.1) ref Yes 5/71 (7.0) 0.36 0.12-1.02 0.069 Tracheitis No 6/175 (3.4) ref Yes 46/169 (27.2) 7.90 2.93-24.96 <0.001 Tracheal gland ectasia No 37/282 (13.1) ref Yes 14/55 (25.5) 2.27 0.94-5.51 0.066 Urinary Crystals No 28/251 (11.2) ref Yes 26/125 (20.8) 2.36 1.14-4.92 0.020

Trichosomoides crassicauda No 15/132 (11.4) ref Yes 11/52 (21.2) 1.23 0.41-3.42 0.702 Composite Variable Nematode in any organc No 10/310 (3.2) ref Yes 56/271 (20.7) 3.47 1.56-8.21 0.003

170

Table 6.5 Results of multivariable mixed logistic regression model with random effect for block assessing associations among lesions and parasites with Leptospira interrogans carriage in 581 Norway rats (Rattus norvegicus) captured in Vancouver Canada.a

Independent Variables Sub-category Odds Ratio 95% CIb P

Bite wounds No ref Yes 2.54 1.01-6.41 0.041 Other macroscopic lesion No ref Yes 5.92 2.09-18.49 0.001 Cardiomyopathy No ref Yes 4.08 1.76- 9.99 0.001 Tracheitis No ref Yes 5.03 1.65-17.83 0.007

a Intraclass correlation at the block level=50% (95% CI: 21.3%-78.6%). Sex did not confound any relationships and sexual maturity could not be assessed for confounding because only one immature rat was infected with L. interrogans. b 95% confidence interval

171

Table 6.6 Patterns of infections with zoonotic pathogens among 331 Norway rats (Rattus norvegicus) from Vancouver, Canada.

Observed a Patterns of pathogen detection number of Observed rats with prevalence Leptospira each of each Expected prevalence interrogansb Bartonella tribocorumc Clostridium difficiled pattern pattern (%) Observed 95% CIe for each pattern – – – 197 59.5 54.0-64.8 198 – + – 61 18.4 14.4-23.0 62 + – – 32 9.7 6.7-13.4 30 – – + 22 6.6 4.2-9.9 20 + + – 10 3.0 1.5-5.5 10 – + + 7 2.1 0.9-4.3 6 + – + 1 0.3 0.0-1.7 3 + + + 1 0.3 0.0-1.7 1 a “–” indicates the absence and “+” represents the presence of the pathogen b Prevalence in this sample = 13.3%; 95% Confidence Interval = 9.8%-17.4% c Prevalence in this sample = 23.9%; 95% Confidence Interval = 19.4%-28.8% d Prevalence in this sample = 9.4%; 95% Confidence Interval = 6.5%-13.0% e 95% Confidence Interval

172

Table 6.7 Results of logistic regression model with random effect for block assessing associations among Bartonella tribocorum, Clostridium difficile and Leptospira interrogans carriage in Norway rats (Rattus norvegicus) captured in Vancouver Canada.

Dependent Variable Independent Variables Sub-category Odds Ratio 95% CIa P Bartonella tribocorum Clostridium difficile No ref Yes 1.28 0.50-3.25 0.599 Leptospira interrogansb No ref Yes 0.21 0.07-0.59 0.004 Clostridium difficile Bartonella tribocorum No ref Yes 1.27 0.54-2.88 0.572 Leptospira interrogans No ref Yes 0.38 0.12-0.97 0.060 Leptospira interrogans Bartonella tribocorumc No ref Yes 0.21 0.07-0.62 0.005 Clostridium difficile No ref Yes 0.37 0.12-0.98 0.061

a 95% confidence interval b Intraclass correlation at the block level=55% (95% CI: 29.9%-77.6%) c Intraclass correlation at the block level=60% (95% CI: 30.3%-84.3%)

173

6.8 FIGURE

174

CHAPTER 7

7. GENERAL DISCUSSION

7.1 SUMMARY OF MAIN RESULTS Norway rats are globally-invasive mammals that carry a number of zoonotic pathogens that pose a potential health risk to people, especially in urban locations. The overall objective of this work was to understand the role of environmental and intra-host factors in the epidemiology and ecology of zoonotic pathogen carriage in urban Norway rats. Collectively, these studies traverse hierarchical levels of biological organization to describe how distal factors may influence the ecology of zoonotic pathogens in rats (Ezenwa et al. 2015).

In the first part of my research, I examined factors of the host’s environment that may impact pathogen ecology, thus working at a higher level of hierarchical biological organization than the host. As a part of a literature review and narrative synthesis, I determined that very few studies have examined fine-scale environmental differences in relation to zoonotic pathogen carriage in specific urban wildlife hosts, including rats (Chapter 2). Another important result from this review is that pathogen ecology, including prevalence and pathogen characteristics, are influenced by habitat, geographic location, weather and season. Ports, disadvantaged (e.g., low- income) and residential areas tend to be sites of high pathogen prevalence in rats and house mice (Mus musculus).

Associations between the environment and zoonotic pathogen carriage by wild animal hosts are indeed complex and challenging to study. I confirmed this intricacy when I analyzed associations among time-lagged weather data, microenvironmental features, rat abundance and several bacterial zoonotic pathogens (Chapters 3 and 4). Time-lagged weather factors (temperature or precipitation) were important for Bartonella tribocorum, Clostridium difficile and methicillin- resistant Staphylococcus aureus (MRSA) carriage in rats. Microenvironmental features were associated with B. tribocorum, MRSA and antimicrobial resistant Escherichia coli in rats. Depending on the pathogen, the factors associated with pathogens ranged from land use type (i.e., low-rise apartment buildings, food gardens and institutions) to property maintenance (alley 175

pavement condition). Rat abundance was not significantly associated with any of the pathogens I investigated.

In the second part of my research, I examined intra-host factors associated with zoonotic pathogen carriage, thus working at a lower level of hierarchical biological organization than the host. The first step necessitated that I conduct a systematic and detailed assessment of common and rare infections and lesions in this sample of rats to understand the spectrum of intra-host disease (Chapter 5). Rats had a tremendous variety of macroscopic and microscopic lesions. Among them were many examples of inflammatory and infectious diseases, most often of the respiratory tract, and associated with bite wounds or nematode infections (e.g., Eucoleus sp. and Capillaria hepatica). Of the microscopic lesions that I identified, the most frequent and likely to impact individual rat health included cardiomyopathy, pulmonary inducible bronchus-associated lymphoid tissue aggregates indicative of respiratory tract infections and thyroid follicular hyperplasia (i.e., thyroid goiter).

These results provided the necessary baseline data for my final study. I described associations among host disease, co-infections and zoonotic pathogen carriage in rats (Chapter 6). I identified co-infections among zoonotic pathogens and with nematode infections. The presence of macroscopic and/or microscopic lesions was significantly associated with C. difficile and Leptospira interrogans, while B. tribocorum was significantly associated with nematode infections. Only rarely were rats infected with multiple zoonotic pathogens. Among the pathogens that I considered in this study, the relationship between L. interrogans and lesions was the most complex.

7.2 LIMITATIONS OF STUDY DESIGN This research has several limitations that I would like to acknowledge. As part of a larger research initiative, a major overarching limitation is that I relied on data and samples that were primarily collected for other research purposes. The initial study was not specifically designed to answer the objectives of my research. This impacted data type and quality available for analyses. For instance, artifacts compromised my ability to assess certain tissues and lesions during the pathology analyses, which may have been prevented if pathology assessments were a primary

176

outcome of the initial study. Autolysis resulted from delayed carcass freezing during field collection and freeze-thaw artefacts arose because of the need to freeze carcasses due to the geographical distance between the study site and laboratory. For some rats, tissues were either not collected or missed for histopathology examination. This arose from an incomplete sampling checklist, lack of knowledge of expected pathology in wild rats, multiple people with differing levels of training collecting samples and general human error. It would have been ideal to collect data on parasite infections via direct examination of tissues. The use of histopathology for the detection of enteric parasites likely resulted in an underestimate of the true prevalence and also prevented identification of the parasite species. Ectoparasite data were not included; however, the inclusion of these data would have been an interesting addition to the host disease study, particularly for flea-transmitted B. tribocorum.

Environmental data were collected to assess how the microenvironment impacts rat abundance rather than pathogens. It is possible that I may have selected different environmental factors to assess for associations with pathogens had this been a primary objective of the initial Vancouver Rat Project Study. For instance, information on the presence and type of other animals in the study site may have been informative for multi-host pathogens (e.g., antimicrobial resistant E. coli). In addition, the outdoor trapping location in public areas may have selectively omitted segments of the rat population. Similarly, the environmental assessment of city blocks did not include all environments inhabited by rats (i.e., burrows and nests).

For all studies in this research, small sample size was an issue. In the environmental studies, there were a low number of blocks with similar characteristics. Additionally, few rats tested positive for certain pathogens, which precluded my ability to include these pathogens in certain analyses. Rats originated from a single neighbourhood in Vancouver, Canada. Thus, it is uncertain how generalizable these results are to rats in other cities or those found in agricultural or natural areas.

It may have been advantageous to take other approaches to studying co-infections, including community ecology and network analyses, as these methods may provide further insight into these complex relationships (Vaumourin et al. 2015). Additionally, as a cross-sectional study, it

177

was not possible to ascertain causality and time sequence. Finally, it is not clear how the infections in rats relate to infections in people; therefore, the true risk of rat-associated zoonotic pathogens in Vancouver remains unknown.

Despite these limitations, I was able to make use of an extensive data set collected during an intensive live-trapping study in a challenging setting (i.e., an impoverished inner-city neighbourhood whose residents suffer from high rates of homelessness, mental illness and intravenous drug use). By using previously-collected data and samples, I was able to focus my efforts on analyses and interpretation while maximizing the output of this publically-funded research project.

7.3 RESEARCH IMPLICATIONS AND SIGNIFICANCE Collectively, these studies provide an example of how infectious diseases studies can incorporate data from higher and lower levels of hierarchical biological organization (Figure 1.1; Ezenwa et al. 2015). Analyses of this type move beyond the classic host-pathogen paradigm to broaden our understanding of pathogen ecology in host systems. A unique feature of this research is the granular detail of both the environmental and host disease data that we were able to analyze for a wild animal species. Whether analyzing up a scale to the environment or down to the intra-host level, the relationships among these factors and zoonotic pathogens in rats varied by pathogen. My results provide evidence that additional factors beyond host demographics may affect zoonotic pathogen carriage in rats. These results further emphasize the complexity of infectious disease systems. The lack of consistency among pathogens is likely in part related to differing pathogen transmission routes and environmental survival. This means that surveillance and control may need to be uniquely catered to the ecology of each pathogen in rats.

Rats are an excellent model for this type of research. Since they carry numerous zoonotic pathogens, it is possible to make comparisons among pathogens in the same system to gain a deeper understanding of pathogen ecology in urban environments. Furthermore, rats have limited home ranges that constrict their environmental exposures to small geographical areas. At less than 0.5 kg, their small body size makes them relatively easy to trap, transport, handle and autopsy. Since rats are considered abundant urban pests, there is limited to no societal resistance

178

to their removal. Similarly, there are fewer ethical concerns for trap and removal studies of rats, beyond humane handling and welfare while in captivity, than would arise for threatened, charismatic or otherwise revered wild animals (Sullivan et al. 2003; Kutz et al. 2013; Turner and Paterson 2013).

7.4 FUTURE RESEARCH DIRECTIONS Although cities and their rat populations worldwide share similar characteristics, the generalizability of these studies is unknown. It would be useful to conduct similar studies in other cities and other countries. Analyzing environmental factors associated with infected hosts may be especially impactful for rat-associated zoonotic pathogens in developing countries where the burden of these diseases is high. For instance, analysis of environmental factors associated with Leptospira spp. infection in rats from urban slums may uncover interventional strategies to reduce transmission to people and target surveillance for this important zoonotic pathogen (Costa et al. 2017). It may also be useful to link time-lagged weather effects in rats to outbreaks of rat- associated zoonotic pathogens in people to create predictive models that could inform public health interventions (Mills and Childs 1998).

Prospective studies could lead to a mechanistic understanding of the associations that I identified during this research (Vaumourin et al. 2015). Future studies could apply experimental interventions to understand how the associated factors that I uncovered influence zoonotic pathogens in rats. For instance, do improvements to alley surface condition decrease the prevalence of rats carrying antimicrobial-resistant E. coli? Similarly, how does anthelmintic treatment impact the ecology of B. tribocorum since there is an apparent association with nematode infections? Is hyperplastic goiter in this population the result of iodine deficiency? And if yes, does iodine supplementation impact zoonotic pathogen dynamics?

Future studies could also focus on a narrow piece of the urban ecosystem to enhance our understanding of disease transmission routes and maintenance. The apparent associations among MRSA-carrying rats, food gardens and institutions would be amenable to this type of detailed study. Researchers could further analyze this particular microsystem within the urban habitat to

179

understand the directionality of transmission among rats, people and the environment, as well as plan for risk mitigation.

Other researchers have emphasized the utility of studying common laboratory animal species, like rats, in their natural habitats since they are genetically and immunologically diverse and are free to interact with their complex environment (Turner and Paterson 2013). So this study of disease in wild rats and the associations among disease, co-infections and zoonotic pathogens may be an important first step to studying the genetic basis of infectious diseases in wild rat populations.

Infectious disease researchers studying zoonotic pathogens in their wildlife hosts could traverse scales of biological organization through the inclusion of both environmental and intra-host variables to better understand the ecology of zoonotic pathogens in these systems. For instance, it would be interesting to evaluate how the microenvironment at the scale appropriate for house mice relates to their carriage of the zoonotic arenavirus, lymphocytic choriomeningitis virus. Analyzing intra-host factors like disease and co-infections may be similarly informative for a range of other viral, bacterial or parasitic zoonotic pathogen systems such as Hendra and/or Nipah viruses in bats, Borrelia burgdorferi in white-footed mice (Peromyscus leucopus) and Baylisascaris procyonis in raccoons (Procyon lotor).

7.5 FINAL THOUGHTS Using urban Norway rats a model, I took a multi-scale approach in this research effort to examine factors beyond simple host-pathogen relationships. By applying the disciplines of epidemiology, disease ecology and pathology, I attempted to understand a piece of the biological mosaic found in zoonotic-pathogen carrying rats from Vancouver, Canada. No doubt, these studies barely scratch the surface of the underlying convoluted influences of cross-scale factors. Nevertheless, this research offers an example of how infectious disease studies could explore factors beyond the host and pathogen to include the environment and intra-host variation (disease and co-infections). Other zoonotic disease researchers could apply these concepts and techniques to a variety of other hosts and ecosystems for a better understanding of the ecology of zoonotic

180

pathogens in their host. This information may ultimately be important for the study and control of emerging infectious diseases that originate in wild animals.

181

7.6 REFERENCES

Costa F, Carvalho-Pereira T, Begon M, Riley L, Childs J. 2017. Zoonotic and vector-borne diseases in urban slums: opportunities for intervention. Trends in Parasitology 33:660–662.

Kutz S, Ducrocq J, Cuyler C, Elkin B, Gunn A, Kolpashikov L, Russell D, White RG 2013. Standardized monitoring of Rangifer health during International Polar Year. Rangifer 33:91–114.

Mills JN, Childs JE. 1998. Ecologic studies of rodent reservoirs: their relevance for human health. Emerging Infectious Diseases 4:529–537.

Sullivan TP, Sullivan DS, Ransome DB. 2003. Impact of removal-trapping on abundance and diversity attributes in small-mammal communities. Wildlife Society Bulletin 31:464-474.

Turner AK, Paterson S. 2013. Wild rodents as a model to discover genes and pathways underlying natural variation in infectious disease susceptibility. Parasite Immunology 35: 386- 395.

182

APPENDIX A Appendix A Table 1. Details of the keywords used in the search process

Concept* Search Terms

Urban urban* OR city OR cities OR municipal* OR suburban OR exurban OR residential OR metropol* OR “human- modified landscapes” Environment ecosystem* OR landscape* OR ecolog* OR habitat* OR management OR harbourage OR environment* OR abiotic OR biotic OR climate* OR precipitation OR weather Urban wildlife species** Columba liva domestica “rock dove” OR “rock pigeon” OR “Columba livia” OR “feral pigeon*” OR pigeon* OR “columba livia domestica”

Passer domesticus “house sparrow*” OR “Passer domesticus”

Sturnus vulgaris “European starling*” OR “Sturnus vulgaris”

Mus musculus mice OR mouse OR “Mus musculus” OR “Mus domesticus” Rattus sp. “Rattus norvegicus” OR “Rattus rattus” OR “black rat” OR “Norway rat” OR “brown rat” OR “roof rat” OR rat OR rats Zoonotic pathogens Columba liva domestica Salmonell* OR “Escherichia coli” OR “E. coli” OR “Chlamydophila psittaci” OR “Histoplasma capsulatum” OR Aspergill* OR “Candida parapsilosis” OR “Cryptococcus neoformans” OR chlamyd* OR histoplasmosis OR cryoptococcosis OR zoono* OR “zoonotic disease”

Passer domesticus “west nile virus” OR Salmonell* OR “E. coli” OR “Escherichia coli” OR “Buggy creek virus” OR arbovirus* OR zoono* OR “zoonotic disease” Sturnus vulgaris Salmonell* OR “Chlamydophila psittaci” OR chlamyd* OR “E. coli” OR “Escherichia coli” OR “Histoplasma capsulatum” OR histoplasmosis OR “west nile virus” OR zoono* OR “zoonotic disease”

183

Concept* Search Terms

Mus musculus lymphocytic choriomeningitis OR Arenavir* OR salmonell* OR “E. coli” OR “Escherichia coli” OR zoono* OR “zoonotic disease” OR rickettsialpox OR “Rickettsia akari” OR “scrub typhus” OR “tsutsugamushi disease” OR “Orientia tsutsugamushi” OR “rat-bite fever” OR “Streptobacillus moniliformis”

Rattus sp. Bartonell* OR leptospir* OR “Weil’s disease*” OR Salmonell* OR “Escherichia coli” OR “E. coli” OR Yersin* OR plague OR “Streptobacillus monilliformis” OR “rat bite fever” OR “Haverhill fever” OR Rickettsia OR typhus OR “murine typhus” OR Campylobacter* OR “hepatitis E virus” OR hantavirus* OR “hemorrhagic fever with renal syndrome” OR “Seoul hantavirus” OR “Seoul virus” OR zoono* OR “zoonotic disease*”

* Concepts were combined with the Boolean operator AND ** Keywords for each species and its respective pathogens were searched together

Appendix A Table 2. Details of the topics used to create a structured abstracting matrix

Category Topics

Study design Species, non-urban exploiter species included, number in study, zoonotic pathogen(s), dates, study design, study objective, study location Methods Diagnostic test(s), number of sampling sites, sampling technique (e.g., selection criteria for sampling sites, sample size calculation), statistical analysis Environmental Was environmental component primary or secondary, scale, description of environmental factors, consideration for disease in factors people & domestic animals, weather factors Results Overall prevalence, range of prevalence, inclusion of a map & description, environmental factors associated with pathogen in host, habitat type with highest prevalence, weather factors associated with pathogen in host, reason for distribution, varying pathogen characteristics by location Study quality Subjective assessment of quality & relevance

184

Appendix A Table 3. Details of the systematic search results

Species* Number of papers in initial search** Number retained for final review***

Mus musculus 285 12 Columba liva domestica 257 13 Rattus sp. 478 41 Passer domesticus 205 2 Sturnus vulgaris 175 1

Total: 1400 Total: 69 * papers that evaluated multiple species were included in the count for the main species in the study. ** limited to peer-reviewed scientific literature written in English and excluding relevant reviews *** Studies retained for final review included consideration for zoonotic pathogens in their host and had an environmental component to the study (e.g., weather, geographical location). We excluded studies that focused exclusively on identifying zoonotic pathogen in a host without regard for environmental influences. We also excluded studies of pathogens that are not directly shed by animals (e.g. Cryptococcus spp. associated with pigeon feces), those in rural/natural areas without an urban component and those with low sample sizes (<25 individuals).

Appendix A Table 4. Summary of continental location of studies considered in the review

Continent Number (%) n = 69

Africa 3 (4.3) Asia 16 (23.2) Europe 19 (27.5) North America 21 (30.4) Oceania 1 (1.4) South America 9 (13.0)

185

APPENDIX B Appendix B Table 1. Characteristics of environmental risk factor variables and results of univariable mixed logistic regression models with random effect for block assessing associations with Bartonella tribocorum in Norway rats (Rattus norvegicus) captured in Vancouver Canada.

Category Sub- Odds 95% CI P Overall P for category Ratio Categorical Variables

Proportion of block occupied by commercial parcels < 0.25 ref 0.08 0.25-0.50 0.38 0.05, 2.49 0.30 0.5-0.75 1.51 0.15, 14.02 0.70 > 0.75 6.45 0.96, 66.02 0.06 Proportion of block occupied by industrial parcels none ref > 0 1.37 0.23, 8.88 0.71 Proportion of block occupied by green space parcels none ref > 0 0.32 0.03, 2.48 0.28 Proportion of block occupied by vacant parcels none ref > 0 1.57 0.13, 23.11 0.70 Proportion of block occupied by parcels under construction none ref > 0 2.41 0.03, 3.48 0.41 Proportion of block occupied by abandoned parcels none ref > 0 2.20 0.08, 2.30 0.31 Proportion of block occupied by open parcels none ref > 0 1.09 0.20, 6.34 0.91 Proportion of block occupied by single family houses none ref > 0 0.87 0.13, 5.25 0.87 Proportion of block occupied by low-rise apartments none ref > 0 0.20 0.04, 0.80 0.02 Proportion of block occupied by mid-rise apartments none ref 186

Category Sub- Odds 95% CI P Overall P for category Ratio Categorical Variables

> 0 0.69 0.12, 3.94 0.65 Proportion of block occupied by housing over commercial buildings* < 0.25 ref < 0.00 0.25-0.50 0.12 0.05, 1.07 0.06 > 0.5 12.61 1.88, 156.35 0.01 Proportion of block occupied by buildings not associated with food < 0.25 ref ≥ 0.25 0.31 0.04, 1.79 0.19 Proportion of block occupied by restaurants none ref > 0 1.84 0.32, 13.66 0.49 Proportion of block occupied by grocery stores none ref 0.03 < 0.25 0.34 0.05, 1.97 0.21 ≥ 0.25 3.64 0.49, 38.55 0.20 Proportion of block occupied by industrial food establishments none ref > 0 0.73 0.09, 5.22 0.75 Proportion of block occupied by other food establishments none ref > 0 1.53 0.28, 8.62 0.60 Proportion of block occupied by buildings in extremely poor condition none ref > 0 0.51 0.09, 2.73 0.39 Proportion of block occupied by buildings in fair condition < 0.25 ref 0.74 0.25-0.50 1.11 0.16, 7.85 0.91 0.5-0.75 2.37 0.20, 31.35 0.46 Proportion of block occupied by buildings in good condition ≥ 0.25 ref > 0.25 0.64 0.10, 3.73 0.61 Proportion of block occupied by buildings in excellent condition none ref 0.70 < 0.25 1.84 0.28, 14.43 0.51 ≥ 0.25 0.85 0.09, 7.54 0.87 Proportion of block occupied by grounds in extremely poor condition none ref > 0 1.36 0.22, 10.84 0.74 Proportion of block occupied by grounds in poor condition < 0.25 ref 0.25-0.50 2.90 0.41, 26.37 0.26

187

Category Sub- Odds 95% CI P Overall P for category Ratio Categorical Variables

Proportion of block occupied by grounds in fair condition < 0.25 ref 0.39 0.25-0.50 0.30 0.04, 2.25 0.21 0.50-0.85 0.19 0.01, 3.12 0.22 Proportion of block occupied by grounds in good condition < 0.25 ref ≥ 0.25 0.72 0.12, 3.76 0.69 Proportion of block occupied by grounds in excellent condition none ref > 0 0.45 0.06, 3.14 0.39 Proportion of block occupied by green space none ref > 0 0.27 0.03, 1.76 0.16 Proportion of block occupied by unkempt green space none ref > 0 0.26 0.04, 1.27 0.08 Proportion of block occupied by well-kempt green space none ref > 0 0.69 0.10, 3.69 0.65 Proportion of block occupied by food gardens none ref > 0 0.16 0.01, 1.28 0.10 Proportion of alley face in poor condition < 0.25 ref ≥ 0.25 0.44 0.03, 5.99 Proportion of alley face in fair condition < 0.25 ref 0.77 0.25-0.5 0.41 0.03, 5.35 0.49 0.5-0.75 0.64 0.09, 3.96 0.62 Proportion of alley face in good condition < 0.25 ref ≥ 0.25 1.09 0.19, 6.58 0.92 Proportion of alley bordered by non-paved surface none ref > 0 2.21 0.27, 21.29 0.44 Number of rat holes at the alley face < 2 ref 0.44 2-3 0.29 0.04, 1.91 0.18 4-9 0.68 0.08, 6.80 0.71 10-26 2.14 0.17, 27.7 0.52 Number of rat corridors at the alley face 0 ref

188

Category Sub- Odds 95% CI P Overall P for category Ratio Categorical Variables

≥ 1 6.25 1.44, 39.51 0.02 Amount of garbage/trash/junk/litter A little ref 0.20 Some 0.35 0.07-1.92 0.17 A lot 1.91 0.25-19.63 0.52 Amount of overflowing garbage receptacles A little ref Some to a 2.89 0.50, 24.05 0.52 lot Number of commercial garbage receptacles 0-9 ref 0.08 10-11 0.37 0.04, 3.03 0.32 12-18 1.36 0.18, 13.26 0.76 19-20 10.46 1.26, 209.15 0.05 ** Number of private garbage receptacles (median [IQR]) 0 (0-1) 0.86 0.61, 1.15 0.34 Number of commercial recycling receptacles < 2 ref 0.22 2-3 0.26 0.02, 2.46 0.22 4-7 1.58 0.28, 13.07 0.61 Number of private recycling receptacles 0-2 ref 0.40 3-4 3.21 0.52, 26.80 0.20 5-9 1.16 0.11, 0.54 0.89 Presence of strong odors yes ref no 0.65 0.10, 4.00 0.61 Amount of loitering None to ref Moderatelight 0.17 0.03, 0.79 0.03 to heavy Amount of transport Light ref 0.80 Moderate 0.72 0.09, 5.73 0.73 Heavy 1.36 0.14, 17.24 0.78

* Refers to buildings in which the ground floor is dedicated to commercial businesses and upper floors are residential. ** Analyzed as a continuous variable

189

Appendix B Table 2. Characteristics of weather and abundance risk factor variables and results of univariable mixed logistic regression models with random effect for block assessing associations with Bartonella tribocorum in Norway rats (Rattus norvegicus) captured in Vancouver Canada.

Overall P for Sub-category Odds Categorical Category Variable Format Ratio 95% CI P Variables Mean Temperature in 5 days prior to capture (°C) Categorized by quartiles < 6.14 ref 0.59 ≥ 6.14-8.47 1.41 0.63, 3.16 0.40 ≥ 8.48-12.45 0.66 0.20, 2.16 0.48 ≥ 12.46 0.64 0.15, 2.50 0.52

Mean Temperature in 10 days prior to capture (°C) Categorized by quartiles < 6.58 ref 0.79 ≥ 6.58-9.15 1.29 0.41, 3.87 0.65 ≥ 9.16-12.74 0.92 0.27, 3.27 0.90 ≥ 12.75 0.66 0.15, 2.67 0.56

Mean Temperature in 15 days prior to capture (°C) Categorized by quartiles < 6.55 ref 0.77 ≥ 6.55-9.55 1.00 0.32, 2.97 0.99 ≥ 9.56 – 12.74 1.28 0.37, 4.74 0.69 ≥ 12.75 0.81 0.19, 3.41 0.77

Mean Temperature in 30 days prior to capture (°C) Categorized by quartiles < 5.41 ref 0.35 ≥ 5.41-10.62 1.00 0.29, 3.01 1.00 ≥ 10.64-14.24 0.86 0.21, 3.38 0.83 ≥ 14.25 3.01 0.56, 17.77 0.19

Mean Temperature in 60 days prior to capture (°C) Continuous NA 1.16 1.02, 1.34 0.02 Mean Temperature in 90 days prior to capture (°C) Continuous NA 1.21 1.07, 1.39 < 0.01

190

Overall P for Sub-category Odds Categorical Category Variable Format Ratio 95% CI P Variables Total precipitation in 5 days prior to capture (mm) Categorized by quartiles < 7.60 ref 0.02 ≥ 7.60-17.99 1.41 0.47, 4.36 0.54 ≥ 18.00-30.59 1.80 0.61, 5.56 0.29 ≥ 30.60 4.44 1.48, 14.57 0.01

Total precipitation in 10 days prior to capture (mm) Categorized by quartiles < 26.50 ref 0.19 ≥ 25.50-35.19 0.42 0.16, 1.06 0.07 ≥ 35.20-50.99 0.84 0.33, 2.08 0.70 ≥ 51.00 0.63 0.25, 1.57 0.32

Total precipitation in 15 days prior to capture (mm) Categorized by quartiles < 48.70 ref 0.19 ≥ 48.70-58.59 1.62 0.41, 6.11 0.48 ≥ 58.60-72.59 2.20 0.58, 7.71 0.22 ≥ 72.60 0.91 0.22, 3.48 0.88

Total precipitation in 30 days prior to capture (mm) Categorized by quartiles < 107.3 ref 0.71 ≥ 107.30-119.39 1.33 0.52, 3.40 0.55 ≥ 119.40-150.19 1.57 0.55, 4.67 0.40 ≥ 150.20 0.82 0.23, 2.96 0.76

Total precipitation in 60 days prior to capture (mm) Categorized by quartiles < 191.00 ref 0.83 ≥ 191.00-216.79 0.56 0.09, 3.99 0.52 ≥ 216.80-310.79 0.48 0.08, 3.42 0.42 ≥ 310.80 0.40 0.05, 3.97 0.39

Total precipitation in 90 days prior to capture (mm) Categorized by quartiles < 227.50 ref 0.02 ≥ 227.50-358.99 0.76 0.10, 4.86 0.76 ≥ 359.00-484.39 0.23 0.03, 1.43 0.09 191

Overall P for Sub-category Odds Categorical Category Variable Format Ratio 95% CI P Variables ≥ 484.40 0.09 0.01, 0.63 0.02

Mean of minimum temperatures on 7 days prior to Categorized by quartiles < 3.61 ref 0.01 capture (°C) ≥ 3.61-6.08 1.11 0.44, 2.67 0.82 ≥ 6.09-10.19 0.09 0.01, ≥0.52 0.01 ≥ 10.20 0.84 0.12, 4.17 0.84

Mean of minimum temperatures on days 8-14 prior to Categorized by quartiles < 4.06 ref 0.18 capture (°C) ≥ 4.06-7.75 0.35 0.07, 1.39 0.16 ≥ 7.76-10.28 0.49 0.09, 2.05 0.35 ≥ 10.29 1.53 0.23, 9.40 0.64

Mean of minimum temperatures on days 24-30 prior to Continuous NA 1.23 1.08, 1.44 < 0.01 capture (°C)

Mean of minimum temperatures on days 54-60 prior to Categorized by quartiles < 3.54 ref 0.05 capture (°C) ≥ 3.54-8.19 1.09 0.31, 3.75 0.90 ≥ 8.20-13.30 6.60 1.58, 28.86 0.01 ≥ 13.31 5.18 1.21, 22.54 0.02

Mean of minimum temperatures on days 84-90 prior to Continuous NA 1.28 1.14, 1.48 < 0.01 capture (°C) Mean of maximum temperatures on 7 days prior to Categorized by quartiles < 8.96 ref 0.94 capture (°C) ≥ 8.96-10.83 0.97 0.31, 2.82 0.95 ≥ 10.84-15.15 0.71 0.20, 2.56 0.60 ≥ 15.16 0.74 0.13, 3.81 0.71

Mean of maximum temperatures on days 8-14 prior to Categorized by quartiles < 8.51 ref 0.33

192

Overall P for Sub-category Odds Categorical Category Variable Format Ratio 95% CI P Variables capture (°C) ≥ 8.51-13.28 1.31 0.34, 4.59 0.68 ≥ 13.29-16.01 0.92 0.23, 3.72 0.91 ≥ 16.01 2.04 0.46, 9.08 0.34

Mean of maximum temperatures on days 24-30 prior to Continuous with log NA 1.31 1.16, 21.5 0.03 capture (°C) transformation

Mean of maximum temperatures on days 54-60 prior to Continuous NA 1.14 1.05, 1.23 < 0.01 capture (°C)

Mean of maximum temperatures on days 84-90 prior to Categorized by quartiles < 7.47 ref 0.01 capture (°C) ≥ 7.47-13.06 0.94 0.26, 3.37 0.93 ≥ 13.07-21.02 2.61 0.56, 12.12 0.20 ≥ 21.03 11.22 2.48, 51.23 < 0.01

Mean of mean temperatures on 7 days prior to capture Categorized by quartiles < 6.44 ref 0.72 (°C) ≥ 6.44-8.33 1.38 0.47, 3.92 0.54 ≥ 8.34-12.40 0.99 0.29, 3.46 0.98 ≥ 12.41 0.68 0.16, 2.74 0.58

Mean of mean temperatures on days 8-14 prior to capture Categorized by quartiles < 6.20 ref 0.38 (°C) ≥ 6.20-10.68 1.27 0.21, 6.40 0.78 ≥ 10.69-12.73 1.08 0.17, 6.08 0.93 ≥ 12.74 2.96 0.45, 18.51 0.23

Mean of mean temperatures on days 24-30 prior to Categorized by quartiles < 5.69 ref 0.21 capture (°C) ≥ 5.69-10.85 0.97 0.20, 4.83 0.97 ≥ 10.86-14.16 2.17 0.41, 11.32 0.34 ≥ 14.17 1.75 0.21, 11.46 0.57

193

Overall P for Sub-category Odds Categorical Category Variable Format Ratio 95% CI P Variables

Mean of mean temperatures on days 54-60 prior to Continuous NA 1.18 1.08, 1.31 < 0.01 capture (°C)

Mean of mean temperatures on days 84-90 prior to Categorized by quartiles < 5.54 ref < 0.01 capture (°C) ≥ 5.54-10.16 0.67 0.18, 2.53 0.55 ≥ 10.17-17.30 4.17 1.06, 16.80 0.03 ≥ 17.31 9.50 2.14, 42.89 0.00

Mean of total precipitation on 7 days prior to capture Main effect NA 1.72 1.11, 2.80 0.02 (°C) Quadratic term NA 0.94 0.89, 0.99 0.03

Mean of total precipitation on 8-14 days prior to capture Categorized by quartiles < 2.46 ref 0.13 (°C) ≥ 2.46-4.06 0.54 0.21, 1.34 0.19 ≥ 4.07-5.85 0.77 0.32, 1.80 0.56 ≥ 5.86 0.27 0.08, 0.86 0.03

Mean of total precipitation on 24-30 days prior to Categorized by quartiles < 2.64 ref 0.65 capture (°C) ≥ 2.64-3.79 1.69 0.72, 4.09 0.24 ≥ 3.80-5.42 1.27 0.47, 3.38 0.64 ≥ 5.43 1.22 0.42, 3.69 0.72

Mean of total precipitation on 54-60 days prior to Main effect NA 0.76 0.56, 1.12 0.06 capture (°C) Quadratic term NA 1.03 1.01, 1.06 0.02

Mean of total precipitation on 84-90 days prior to Categorized by quartiles < 1.01 ref 0.51 capture (°C) ≥ 1.01-4.36 1.29 0.24, 13.83 0.80 ≥ 4.37-6.99 0.67 0.11, 7.74 0.70 ≥ 7.00 1.22 0.15, 26.46 0.88

194

Overall P for Sub-category Odds Categorical Category Variable Format Ratio 95% CI P Variables

Rat abundance Main effect NA 1.6 X10-5 3.1 X10-10, 0.02 0.13 Quadratic term NA 1.9X106 25.40, 0.01 2.0X1012

195

APPENDIX C Appendix C Table 1. Characteristics of environmental risk factor variables and results of univariable mixed logistic regression models with random effect for block assessing associations with Clostridium difficile carriage in Norway rats (Rattus norvegicus) captured in Vancouver, Canada. Significant P-values (α=0.05) are indicated in bold.

Category Sub- Odds 95% CI P Overall P for Categorical category Ratio Variables

Proportion of block occupied by commercial parcels < 0.25 ref 0.425 0.25-0.50 0.69 0.26, 1.70 0.411 0.5-0.75 0.53 0.17, 1.63 0.245 ≥ 0.75 0.42 0.14, 1.32 0.116 Proportion of block occupied by industrial parcels None ref > 0 1.87 0.79, 4.20 0.124 Proportion of block occupied by institutional parcels < 0.25 ref ≥ 0.25 2.20 0.87, 5.30 0.070 Proportion of block occupied by green space parcels None ref > 0 1.75 0.67, 4.30 0.214 Proportion of block occupied by vacant parcels None ref > 0 0.80 0.23, 2.50 0.694 Proportion of block occupied by parcels under construction None ref > 0 0.42 0.14, 1.17 0.098 Proportion of block occupied by abandoned parcels None ref > 0 2.00 0.91, 4.57 0.077 Proportion of block occupied by open parcels None ref > 0 1.65 0.75, 3.70 0.198 Proportion of block occupied by single family houses None ref > 0 1.51 0.62, 3.56 0.333 196

Category Sub- Odds 95% CI P Overall P for Categorical category Ratio Variables

Proportion of block occupied by low-rise apartments None ref > 0 2.23 1.04, 4.80 0.030 Proportion of block occupied by mid-rise apartments None ref > 0 0.65 0.28, 1.52 0.303 Proportion of block occupied by housing over commercial buildings* < 0.25 ref 0.899

0.25-0.50 0.82 0.35, 1.94 0.640 > 0.5 0.90 0.21, 3.61 0.876 Proportion of block occupied by buildings not associated with food < 0.25 ref ≥ 0.25 0.57 0.25, 1.34 0.168 Proportion of block occupied by restaurants None ref > 0 0.89 0.37, 2.18 0.793 Proportion of block occupied by grocery stores None ref 0.176

< 0.25 1.81 0.61, 6.15 0.297 ≥ 0.25 0.79 0.21, 3.20 0.724 Proportion of block occupied by industrial food establishments None ref > 0 2.32 0.96, 5.28 0.043 Proportion of block occupied by other food establishments None ref > 0 0.86 0.36, 1.92 0.700 Proportion of block occupied by buildings in extremely poor condition None ref > 0 1.03 0.46, 2.39 0.950 Proportion of block occupied by buildings in fair condition < 0.25 ref 0.152

0.25-0.50 1.27 0.56, 3.06 0.560 0.5-0.75 0.40 0.11, 1.46 0.149 Proportion of block occupied by buildings in good condition ≥ 0.25 ref > 0.25 2.20 0.90, 5.13 0.063 Proportion of block occupied by buildings in excellent condition None ref 0.496

< 0.25 1.71 0.69, 4.45 0.237 ≥ 0.25 1.40 0.49, 3.82 0.504 Proportion of block occupied by grounds in extremely poor condition None ref

197

Category Sub- Odds 95% CI P Overall P for Categorical category Ratio Variables

> 0 1.50 0.55, 4.76 0.442 Proportion of block occupied by grounds in poor condition < 0.25 ref 0.25-0.50 0.72 0.27, 1.92 0.494 Proportion of block occupied by grounds in fair condition < 0.25 ref 0.732

0.25-0.50 1.39 0.47, 4.45 0.546 0.50-0.85 0.98 0.24, 4.49 0.978 Proportion of block occupied by grounds in good condition < 0.25 ref 0.866

0.25-0.50 1.12 0.44, 2.66 0.789 ≥ 0.50 1.38 0.37, 4.08 0.585 Proportion of block occupied by grounds in excellent condition None ref > 0 1.44 0.54, 3.41 0.408 Proportion of block occupied by green space None ref > 0 1.82 0.69, 4.50 0.183 Proportion of block occupied by unkempt green space None ref > 0 1.20 0.49, 2.73 0.662 Proportion of block occupied by well-kempt green space None ref > 0 0.89 0.37, 1.97 0.769 Proportion of block occupied by food gardens None ref > 0 0.91 0.27, 2.66 0.864 Proportion of alley face in poor condition None ref 0.969

< 0.25 1.13 0.31, 4.39 0.850 0.25-0.5 1.21 0.26, 6.23 0.802 Proportion of alley face in fair condition < 0.25 ref 0.094

0.25-0.5 0.29 0.08, 0.96 0.039 0.5-0.75 0.50 0.22, 1.13 0.079 Proportion of alley face in good condition < 0.25 ref 0.332

0.25-0.75 0.85 0.28, 2.45 0.762 ≥ 0.75 2.12 0.71, 6.47 0.159 Proportion of alley bordered by non-paved surface None ref 0.877

198

Category Sub- Odds 95% CI P Overall P for Categorical category Ratio Variables

< 25 0.83 0.31, 2.28 0.693 ≥ 25 1.05 0.27, 4.03 0.937 Number of rat holes at the alley face < 2 ref 0.689

2-3 1.51 0.53, 4.67 0.437 4-9 0.74 0.25, 2.34 0.576 10-26 1.14 0.41, 3.50 0.798 Number of rat corridors at the alley face 0 ref 0.109

1 0.65 0.22, 1.75 0.388 2-3 1.41 0.38, 3.88 0.544 4 0.24 0.07, 0.75 0.010 Amount of garbage/trash/junk/litter A little ref 0.193

Some 1.07 0.47, 2.51 0.873 A lot 0.43 0.14, 1.30 0.113 Amount of overflowing garbage receptacles None ref 0.564

A little 0.84 0.30, 2.56 0.731 Some to a 0.56 0.17, 1.93 0.318 ** lot Number of commercial garbage receptacles (median [IQR]) 12 (8-19) 0.41 0.20, 0.88 0.014 Number of private garbage receptacles None ref 0.862

1 0.83 0.21, 2.82 0.770 2-4 0.71 0.21, 2.10 0.540 5-14 0.61 0.12, 2.51 0.504 Number of commercial recycling receptacles ≤ 1 ref 0.265

2-3 0.76 0.28, 1.95 0.556 4 0.40 0.16, 1.04 0.043 ≥ 6 0.87 0.28, 3.00 0.807 Number of private recycling receptacles 0 ref 0.085

1-3 2.56 1.02, 6.16 0.032 4-5 0.90 0.29, 2.44 0.840 ≥ 6 1.10 0.28, 3.62 0.884

199

Category Sub- Odds 95% CI P Overall P for Categorical category Ratio Variables

Presence of strong odors Yes ref No 0.74 0.31, 1.89 0.501 Amount of loitering None to light ref 0.092

Moderate 2.70 1.06, 6.92 0.029 Heavy 0.95 0.42, 2.29 0.905 Amount of transport Light ref 0.646

Moderate 0.78 0.24, 1.70 0.346 Heavy 0.64 0.25, 2.55 0.666

* Refers to buildings in which the ground floor is dedicated to commercial businesses and upper floors are residential. ** Analyzed as a log-transformed continuous variable in logistic regression models. IQR = interquartile range.

200

Appendix C Table 2. Characteristics of weather and abundance risk factor variables and results of univariable mixed logistic regression models with random effect for block assessing associations with Clostridium difficile carriage in Norway rats (Rattus norvegicus) captured in Vancouver, Canada. Significant P-values (α=0.05) are indicated in bold.

Category Variable Format Sub-category Odds 95% CI P Overall P Ratio for Categorical Variables Mean temperature in 5 days before capture (°C) Continuous NA 0.92 0.85, 1.00 0.035

Mean temperature in 10 days before capture (°C) Continuous NA 0.91 0.84, 0.99 0.028

Mean temperature in 15 days before capture (°C) Continuous NA 0.91 0.83, 0.99 0.022

Mean temperature in 30 days before capture (°C) Main effect NA 1.53 0.95, 0.99 0.052 0.004 Quadratic term NA 0.97 0.01, 0.31 0.015 Mean temperature in 60 days before capture (°C) Categorized by quartiles < 6.72 ref 0.039 ≥ 6.72-11.01 0.95 0.44, 1.95 0.886 ≥ 11.02-14.42 0.74 0.32, 1.76 0.475 ≥ 14.43 0.22 0.07, 0.62 0.003

Mean temperature in 90 days before capture (°C) Continuous NA 0.90 0.82, 0.98 0.010

Total precipitation in 5 days before capture (mm) Categorized by quartiles < 5.80 ref 0.529

≥ 5.80-16.99 1.61 0.74, 3.52 0.225 ≥ 17.00-30.59 1.20 0.54, 2.74 0.656 ≥ 30.60 1.03 0.43, 2.46 0.954

Total precipitation in 10 days before capture Categorized by quartiles < 24.90 ref 0.686 (mm) ≥ 24.90-35.19 0.92 0.44, 1.95 0.826

201

Category Variable Format Sub-category Odds 95% CI P Overall P Ratio for Categorical Variables ≥ 35.20-48.69 0.63 0.27, 1.44 0.275 ≥ 48.70 0.84 0.40, 1.74 0.630

Total precipitation in 15 days before capture Categorized by quartiles < 42.60 ref 0.457 (mm) ≥ 42.60-58.59 0.87 0.50, 2.61 0.742 ≥ 58.60-72.59 0.85 0.35, 2.06 0.718 ≥ 72.60 0.64 0.26, 1.52 0.315

Total precipitation in 30 days before capture Categorized by quartiles < 104.2 ref 0.825 (mm) ≥ 104.2-117.99 1.49 0.60, 3.82 0.384 ≥ 118.00-143.69 1.07 0.45, 2.53 0.875 ≥ 143.70 1.10 0.43, 2.73 0.841

Total precipitation in 60 days before capture Categorized by quartiles < 192.80 ref 0.033 (mm) ≥ 192.80-216.79 3.90 1.33, 12.41 0.013 ≥ 216.80-298.09 4.88 1.69, 15.20 0.003 ≥ 298.10 4.87 1.62, 15.33 0.004

Total precipitation in 90 days before capture Categorized by quartiles < 233.10 ref 0.057 (mm) ≥ 233.10-366.99 2.78 1.09, 7.82 0.040 ≥ 367.00-460.99 3.73 1.37, 10.77 0.010 ≥ 461.00 3.60 1.22, 10.76 0.017

Mean of minimum temperatures on 7 days before Categorized by quartiles < 4.11 ref 0.012 capture (°C) ≥ 4.11-7.55 1.01 0.54, 1.85 0.982

202

Category Variable Format Sub-category Odds 95% CI P Overall P Ratio for Categorical Variables ≥ 7.56-10.19 0.79 0.37, 1.65 0.518 ≥ 10.20 0.19 0.06-0.52 0.002

Mean of minimum temperatures on days 8-14 Categorized by quartiles < 4.59 ref 0.022 before capture (°C) ≥ 4.59-7.78 0.96 0.47, 1.89 0.899 ≥ 7.79-10.28 0.75 0.35, 1.57 0.437 ≥ 10.29 0.21 0.07, 0.57 0.002 Mean of minimum temperatures on days 24-30 Main effect NA 1.15 0.91, 1.45 0.224 0.004 before capture (°C) Quadratic term NA 0.98 0.96, 1.00 0.020

Mean of minimum temperatures on days 54-60 Categorized by quartiles < 3.27 ref 0.049 before capture (°C) ≥ 3.27-5.25 0.83 0.44, 1.54 0.546 ≥ 5.26-13.12 0.26 0.10, 0.67 0.005 ≥ 13.13 0.41 0.16, 1.03 0.054

Mean of minimum temperatures on days 84-90 Categorized by quartiles < 3.40 ref 0.032 before capture (°C) ≥ 3.40-6.62 1.29 0.63, 2.59 0.479 ≥ 6.63-13.96 0.37 0.14, 0.94 0.038 ≥ 13.97 0.41 0.14, 1.08 0.078

Mean of maximum temperatures on 7 days before Categorized by quartiles < 9.23 ref 0.443 capture (°C) ≥ 9.23-13.50 1.09 0.54, 2.25 0.808 ≥ 13.51-15.66 0.55 0.21, 1.38 0.201 ≥ 15.67 0.64 0.24, 1.66 0.347

Mean of maximum temperatures on days 8-14 Categorized by quartiles < 8.95 ref 0.059

203

Category Variable Format Sub-category Odds 95% CI P Overall P Ratio for Categorical Variables before capture (°C) ≥ 8.96-13.50 1.10 0.49, 2.46 0.819 ≥ 13.51-16.09 1.01 0.45. 2.37 0.972 ≥ 16.10 0.31 0.11, 0.88 0.023

Mean of maximum temperatures on days 24-30 Main effect NA 1.35 0.97, 1.87 0.067 0.005 before capture (°C) Quadratic term NA 0.98 0.97, 1.00 0.017

Mean of maximum temperatures on days 54-60 Categorized by quartiles < 7.57 ref 0.060 before capture (°C) ≥ 7.57-11.49 1.52 0.77, 3.04 0.225 ≥ 11.50-21.65 0.54 0.22, 1.32 0.164 ≥ 21.66 0.43 0.16, 1.12 0.080

Mean of maximum temperatures on days 84-90 Categorized by quartiles < 7.41 ref 0.001 before capture (°C) ≥ 7.41-12.88 1.49 0.77, 2.87 0.230 ≥ 12.89-20.29 0.18 0.05, 0.51 0.002 ≥ 20.30 0.76 0.29, 1.77 0.524

Mean of mean temperatures on 7 days before Categorized by quartiles < 6.70 ref 0.504 capture (°C) ≥ 6.70-10.56 0.86 0.45, 1.65 0.648 ≥ 10.57-12.52 0.61 0.25, 1.43 0.252 ≥ 12.53 0.51 0.21, 1.26 0.137

Mean of mean temperatures on days 8-14 before Categorized by quartiles < 7.11 ref 0.004 capture (°C) ≥ 7.11-10.68 1.58 0.76, 3.42 0.216 ≥ 10.69-12.79 0.99 0.48, 2.19 0.973

204

Category Variable Format Sub-category Odds 95% CI P Overall P Ratio for Categorical Variables ≥ 12.80 0.23 0.08, 0.65 0.006

Mean of mean temperatures on days 24-30 before Main effect NA 1.27 0.97, 1.78 0.085 0.003 capture (°C) Quadratic term NA 0.98 0.97, 1.00 0.013

Mean of mean temperatures on days 54-60 before Categorized by quartiles < 5.54 ref 0.067 capture (°C) ≥ 5.54-7.98 1.29 0.69, 2.42 0.422 ≥ 7.99-17.59 0.54 0.22, 1.33 0.172 ≥ 17.60 0.36 0.14, 0.92 0.032

Mean of mean temperatures on days 84-90 before Categorized by quartiles < 5.54 ref 0.001 capture (°C) ≥ 5.54-9.83 1.47 0.74, 2.94 0.258 ≥ 9.84-17.15 0.21 0.07, 0.58 0.003 ≥ 17.16 0.75 0.28, 1.90 0.544

Mean of total precipitation on 7 days before Categorized by quartiles < 2.03 ref 0.960 capture (°C) ≥ 2.03-3.65 0.91 0.42, 1.98 0.811 ≥ 3.66-5.26 0.86 0.38, 1.94 0.703 ≥ 5.27 0.81 0.35, 1.83 0.607

Mean of total precipitation on 8-14 days before Categorized by quartiles < 2.00 ref 0.935 capture (°C) ≥ 2.00-3.99 1.14 0.55, 2.40 0.715 ≥ 4.00-5.33 0.88 0.36, 2.15 0.778 ≥ 5.34 0.81 0.54, 2.30 0.790

Mean of total precipitation on 24-30 days before Categorized by quartiles < 2.64 ref 0.803

205

Category Variable Format Sub-category Odds 95% CI P Overall P Ratio for Categorical Variables capture (°C) ≥ 2.64-3.99 0.85 0.44, 1.79 0.655 ≥ 4.00-5.42 0.78 0.35, 1.73 0.528 ≥ 5.43 1.10 0.54, 2.26 0.789

Mean of total precipitation on 54-60 days before Continuous NA 1.04 0.95, 1.14 0.357 capture (°C)

Mean of total precipitation on 84-90 days before Categorized by quartiles < 1.36 ref 0.168 capture (°C) ≥ 1.36-4.36 0.77 0.30, 2.00 0.584 ≥ 4.37-7.19 0.50 0.18, 1.37 0.171 ≥ 7.20 1.19 0.44, 3.10 0.711

Rat abundance Categorized by quartiles < 0.18 ref 0.604 ≥ 0.18-0.37 0.80 0.34, 1.99 0.607 ≥ 0.38-0.54 0.46 0.14, 1.48 0.164 ≥ 0.55 1.07 0.20, 5.12 0.934

206

Appendix C Table 3. Characteristics of environmental risk factor variables and results of univariable mixed logistic regression models with random effect for block assessing associations with antimicrobial resistant Escherichia coli carriage in Norway rats (Rattus norvegicus) captured in Vancouver, Canada. Significant P-values (α=0.05) are indicated in bold.

Category Sub-category Odds 95% CI P Overall P for Ratio Categorical Variables

Proportion of block occupied by commercial parcels < 0.25 ref 0.624

0.25-0.50 0.84 0.27, 2.80 0.760 0.5-0.75 1.85 0.58, 7.47 0.304 ≥ 0.75 1.28 0.40, 5.10 0.686 Proportion of block occupied by industrial parcels None ref > 0 1.67 0.67, 4.02 0.228 Proportion of block occupied by institutional parcels < 0.25 ref ≥ 0.25 1.28 0.41, 4.10 0.654 Proportion of block occupied by green space parcels None ref > 0 2.78 1.16-7.45 0.022 Proportion of block occupied by vacant parcels None ref > 0 1.29 0.34, 4.22 0.677 Proportion of block occupied by parcels under construction None ref > 0 0.94 0.27, 2.66 0.908 Proportion of block occupied by abandoned parcels None ref > 0 0.82 0.33, 1.97 0.650 Proportion of block occupied by open parcels None ref > 0 0.98 0.41, 2.41 0.954 Proportion of block occupied by single family houses None ref > 0 1.87 0.78, 4.56 0.160 Proportion of block occupied by low-rise apartments None ref > 0 1.07 0.46, 2.47 0.875

207

Category Sub-category Odds 95% CI P Overall P for Ratio Categorical Variables

Proportion of block occupied by mid-rise apartments None ref > 0 0.84 0.31, 2.03 0.696 Proportion of block occupied by housing over commercial buildings* < 0.25 ref 0.565 0.25-0.50 0.77 0.28, 1.90 0.566 > 0.5 1.58 0.39, 7.15 0.504 Proportion of block occupied by buildings not associated with food < 0.25 ref ≥ 0.25 0.69 0.28, 1.70 0.425 Proportion of block occupied by restaurants None ref > 0 0.81 0.32, 1.98 0.640 Proportion of block occupied by grocery stores < 0.25 ref ≥ 0.25 0.67 0.26, 1.76 0.412 Proportion of block occupied by industrial food establishments None ref > 0 2.29 0.95, 5.76 0.051 Proportion of block occupied by other food establishments None ref > 0 0.47 0.19, 1.07 0.062 Proportion of block occupied by buildings in extremely poor condition None ref > 0 0.95 0.41, 2.20 0.911 Proportion of block occupied by buildings in fair condition < 0.25 ref 0.787 0.25-0.50 0.78 0.31, 1.95 0.590 0.5-0.75 0.65 0.17, 2.46 0.520 Proportion of block occupied by buildings in good condition ≥ 0.25 ref > 0.25 1.25 0.50, 3.13 0.635 Proportion of block occupied by buildings in excellent condition None ref 0.333 < 0.25 1.91 0.77, 4.71 0.162 ≥ 0.25 1.83 0.65, 5.17 0.256 Proportion of block occupied by grounds in extremely poor condition None ref > 0 0.62 0.23, 1.64 0.332 Proportion of block occupied by grounds in poor condition < 0.25 ref

208

Category Sub-category Odds 95% CI P Overall P for Ratio Categorical Variables

0.25-0.50 0.45 0.34, 1.17 0.109 Proportion of block occupied by grounds in fair condition < 0.25 ref 0.881 0.25-0.50 0.76 0.22, 2.39 0.621 0.50-0.75 0.77 0.18, 4.02 0.724 Proportion of block occupied by grounds in good condition < 0.25 ref 0.498 0.25-0.50 1.23 0.45, 3.03 0.653 ≥ 0.50 2.02 0.59, 6.09 0.210 Proportion of block occupied by grounds in excellent condition None ref > 0 2.68 1.10, 8.31 0.039 Proportion of block occupied by green space None ref > 0 2.48 0.95, 7.89 0.064 Proportion of block occupied by unkempt green space None ref > 0 2.05 0.88, 5.12 0.079 Proportion of block occupied by well-kempt green space None ref > 0 1.05 0.45, 2.43 0.090 Proportion of block occupied by food gardens None ref > 0 1.73 0.60, 5.01 0.312 Proportion of alley face in poor condition < 0.25 ref 0.25-0.5 0.53 0.17, 1.62 0.214 Proportion of alley face in fair condition None ref 0.009 < 0.25 0.61 0.19, 1.70 0.352 ≥ 0.25 0.29 0.12, 0.62 0.002 Proportion of alley face in good condition None ref 0.003 < 0.25 6.37 1.29, 115.01 0.073 0.25-0.75 14.72 2.65, 276.08 0.012 ≥ 0.75 17.10 3.06, 321.29 0.008 Proportion of alley bordered by non-paved surface None ref > 0 4.22 1.59, 14.31 0.007

209

Category Sub-category Odds 95% CI P Overall P for Ratio Categorical Variables

Number of rat holes at the alley face < 2 ref 0.534 2-3 1.32 0.46, 4.26 0.589 4-9 1.08 0.19, 2.24 0.472 9-26 0.54 0.19, 1.94 0.386 Number of rat corridors at the alley face None ref 0.040 1 1.21 0.49, 2.71 0.654 ≥ 2 0.27 0.08, 0.71 0.016 Amount of garbage/trash/junk/litter A little ref 0.115 Some 1.02 0.43, 2.75 0.969 A lot 0.35 0.09, 1.13 0.077 Amount of overflowing garbage receptacles None ref 0.154 A little 1.22 0.43, 4.42 0.722 Some to a lot 0.50 0.14, 1.96 0.268 ** Number of commercial garbage receptacles (median [IQR]) 11 (7-17) 0.95 0.87, 1.03 0.166 Number of private garbage receptacles None ref 0.735 1 1.56 0.49, 4.97 0.447 ≥ 2 0.99 0.36, 2.73 0.987 Number of commercial recycling receptacles ≤ 1 ref 0.846 2-3 1.05 0.36, 3.12 0.923 4 0.75 0.26, 2.16 0.589 ≥ 6 1.25 0.34, 4.57 0.732 Number of private recycling receptacles None ref 0.138 1-2 1.71 0.56, 5.18 0.308 3-4 0.45 0.13, 1.35 0.170 ≥ 5 0.84 0.25, 2.44 0.759 Presence of strong odors Yes ref No 0.70 0.28, 1.74 0.438 Amount of loitering None to light ref 0.442

210

Category Sub-category Odds 95% CI P Overall P for Ratio Categorical Variables

Moderate 1.41 0.49, 4.05 0.525 Heavy 0.67 0.26, 1.68 0.390 Amount of transport Light ref 0.979 Moderate 0.90 0.31, 2.76 0.838 Heavy 0.92 0.26, 3.52 0.894

* Refers to buildings in which the ground floor is dedicated to commercial businesses and upper floors are residential. ** Analyzed as a continuous variable in logistic regression models

211

Appendix C Table 4. Characteristics of weather and abundance risk factor variables and results of univariable mixed logistic regression models with random effect for block assessing associations with antimicrobial resistant Escherichia coli carriage in Norway rats (Rattus norvegicus) captured in Vancouver, Canada. Significant P-values (α=0.05) are indicated in bold.

Category Variable Format Sub-category Odds 95% CI P Overall P Ratio for Categorical Variables Mean temperature in 5 days before capture (°C) Categorized by quartiles < 6.22 ref 0.266 ≥ 6.22-10.05 1.53 0.63, 3.97 0.355 ≥ 10.06-12.37 0.56 0.17, 1.89 0.333 ≥ 12.38 1.36 0.51, 3.90 0.538

Mean temperature in 10 days before capture (°C) Categorized by quartiles < 6.87 ref 0.888 ≥ 6.87-10. 57 0.67 0.17, 1.97 0.496 ≥ 10.58-12.60 0.90 0.29, 3.00 0.847 ≥ 12.61 0.74 0.23, 2.26 0.579 Mean temperature in 15 days before capture (°C) Categorized by quartiles < 6.58 ref 0.914 ≥ 6.58-9.78 0.89 0.27, 2.46 0.829 ≥ 9.79-12.70 0.74 0.22, 2.19 0.583 ≥ 12.71 0.69 0.22, 2.16 0.507 Mean temperature in 30 days before capture (°C) Categorized by quartiles < 5.41 ref 0.604 ≥ 5.41-10.30 0.89 0.31, 2.56 0.815 ≥ 10.31-12.42 1.44 0.49, 4.12 0.462 ≥ 12.43 0.69 0.22, 2.43 0.522 Mean temperature in 60 days before capture (°C) Categorized by quartiles < 6.00 ref 0.423 ≥ 6.00-9.99 1.10 0.37, 3.25 0.869 ≥ 10.00-13.99 1.91 0.64, 5.68 0.246

212

Category Variable Format Sub-category Odds 95% CI P Overall P Ratio for Categorical Variables ≥ 14.00 0.73 0.22, 2.46 0.614 Mean temperature in 90 days before capture (°C) Categorized by quartiles < 5.67 ref 0.981 ≥ 5.67-8.66 0.88 0.26, 2.76 0.823 ≥ 8.67-14.76 1.10 0.34, 3.48 0.861 ≥ 14.77 1.02 0.31, 3.82 0.976 Total precipitation in 5 days before capture (mm) Continuous NA 0.98 0.95, 1.00 0.097

Total precipitation in 10 days before capture Categorized by quartiles < 27.40 ref 0.915 (mm) ≥ 27.40-37.79 0.77 0.26, 2.21 0.628 ≥ 37.80-53.79 1.10 0.41, 2.94 0.854 ≥ 53.80 0.99 0.37, 2.67 0.976

Total precipitation in 15 days before capture Categorized by quartiles < 46.30 ref 0.895 (mm) ≥ 46.30-58.59 1.43 0.47, 4.84 0.539 ≥ 58.60-72.59 1.41 0.50, 4.55 0.534 ≥ 72.6 1.44 0.52, 4.49 0.493

Total precipitation in 30 days before capture Categorized by quartiles < 106.80 ref 0.663 (mm) ≥ 106.80-123.64 1.44 0.53, 3.82 0.450 ≥ 123.65-150.19 0.78 0.26, 2.35 0.645 ≥ 150.20 0.87 0.28, 2.47 0.794

Total precipitation in 60 days before capture Categorized by quartiles < 202.30 ref 0.238 (mm) ≥ 202.30-224.69 2.68 0.85, 9.04 0.083 ≥ 224.70-309.99 2.00 0.64, 6.39 0.215

213

Category Variable Format Sub-category Odds 95% CI P Overall P Ratio for Categorical Variables ≥ 310.0 1.11 0.29, 3.83 0.868

Total precipitation in 90 days before capture Categorized by quartiles < 234.15 ref 0.834 (mm) ≥ 234.15-400.59 1.19 0.40, 3.37 0.744 ≥ 400.60-485.79 0.72 0.17, 2.35 0.611 ≥ 485.80 0.92 0.24, 2.96 0.887

Mean of minimum temperatures on 7 days before Categorized by quartiles < 3.84 ref 0.657 capture (°C) ≥ 3.84-7.10 1.10 0.43, 2.81 0.834 ≥ 7.11-9.82 0.94 0.32, 2.65 0.899 ≥ 9.83 0.54 0.16, 1.74 0.285

Mean of minimum temperatures on days 8-14 Categorized by quartiles < 4.05 ref 0.848 before capture (°C) ≥ 4.05-7.55 1.11 0.41, 3.01 0.837 ≥ 7.56-9.21 0.97 0.35, 2.73 0.959 ≥ 9.21 0.67 0.21, 2.12 0.496 Mean of minimum temperatures on days 24-30 Categorized by quartiles < 3.67 ref 0.852 before capture (°C) ≥ 3.67-7.15 0.79 0.22, 2.39 0.696 ≥ 7.16-10.73 1.26 0.35, 4.07 0.695 ≥ 10.74 1.13 0.34, 4.06 0.842 Mean of minimum temperatures on days 54-60 Continuous NA 1.02 0.93, 1.13 0.727 before capture (°C)

Mean of minimum temperatures on days 84-90 Categorized by quartiles < 3.12 ref 0.941 before capture (°C) ≥ 3.12-5.65 0.81 0.26, 2.28 0.706 ≥ 5.66-13.83 1.12 0.35, 3.48 0.838

214

Category Variable Format Sub-category Odds 95% CI P Overall P Ratio for Categorical Variables ≥ 13.84 1.02 0.28, 3.48 0.977

Mean of maximum temperatures on 7 days before Categorized by quartiles < 8.99 ref 0.124 capture (°C) ≥ 8.99-13.28 0.86 0.28, 2.14 0.764 ≥ 13.29-15.26 0.32 0.09, 1.02 0.050 ≥ 15.27 1.18 0.46, 3.06 0.718

Mean of maximum temperatures on days 8-14 Categorized by quartiles < 8.00 ref 0.608 before capture (°C) ≥ 8.00-12.99 0.93 0.34, 2.53 0.881 ≥ 13.00-14.99 0.88 0.31, 2.51 0.811 ≥ 15.00 0.51 0.17, 1.55 0.233

Mean of maximum temperatures on days 24-30 Categorized by quartiles < 8.20 ref 0.205 before capture (°C) ≥ 8.20-13.22 0.82 0.28, 2.44 0.714 ≥ 13.23-15.02 2.00 0.73, 5.97 0.172 ≥ 15.03 0.82 0.27, 2.81 0.731 Mean of maximum temperatures on days 54-60 Categorized by quartiles < 7.47 ref 0.778 before capture (°C) ≥ 7.47-9.82 0.97 0.33, 2.69 0.961 ≥ 9.83-22.05 1.10 0.37, 3.47 0.868 ≥ 22.06 1.72 0.55, 6.72 0.377

Mean of maximum temperatures on days 84-90 Categorized by quartiles < 7.39 ref 0.408 before capture (°C) ≥ 7.38-9.77 1.20 0.38, 3.79 0.760 ≥ 9.78-20.00 0.72 0.18, 2.89 0.639 ≥ 20.01 1.91 0.52, 7.11 0.332

215

Category Variable Format Sub-category Odds 95% CI P Overall P Ratio for Categorical Variables Mean of mean temperatures on 7 days before Categorized by quartiles < 6.49 ref 0.506 capture (°C) ≥ 6.49-10.22 1.15 0.46, 2.84 0.755 ≥ 10.23-12.35 0.53 0.17, 1.65 0.251 ≥ 12.36 1.11 0.42, 3.00 0.833

Mean of mean temperatures on days 8-14 before Categorized by quartiles < 5.89 ref 0.560 capture (°C) ≥ 5.89-10.50 1.00 0.37, 2.74 0.994 ≥ 10.51-12.66 1.13 0.41, 3.10 0.819 ≥ 12.67 0.52 0.16, 1.68 0.274

Mean of mean temperatures on days 24-30 before Categorized by quartiles < 6.20 ref 0.673 capture (°C) ≥ 6.20-10.30 1.12 0.39, 3.16 0.821 ≥ 10.31-12.85 1.64 0.54, 4.71 0.339 ≥ 12.86 0.84 0.25, 2.97 0.772

Mean of mean temperatures on days 54-60 before Categorized by quartiles < 5.37 ref 0.695 capture (°C) ≥ 5.37-7.36 1.62 0.64, 4.30 0.318 ≥ 7.37-17.59 1.23 0.39, 3.98 0.714 ≥ 17.60 0.58 0.55, 6.58 0.370

Mean of mean temperatures on days 84-90 before Categorized by quartiles < 5.46 ref 0.220 capture (°C) ≥ 5.46-7.66 1.26 0.43, 3.26 0.647 ≥ 7.67-17.59 0.65 0.20, 1.88 0.414 ≥ 17.60 2.19 0.70, 8.00 0.166

216

Category Variable Format Sub-category Odds 95% CI P Overall P Ratio for Categorical Variables Mean of total precipitation on 7 days before Main effect NA 0.82 0.55, 1.22 0.324 capture (°C) Quadratic term NA 1.01 0.98, 1.06 0.464

Mean of total precipitation on 8-14 days before Categorized by quartiles < 2.20 ref 0.428 capture (°C) ≥ 2.20-3.99 2.10 0.69, 6.45 0.193 ≥ 4.00-5.33 2.69 0.78,9.27 0.117 ≥ 5.34 2.28 0.74, 7.04 0.151

Mean of total precipitation on 24-30 days before Categorized by quartiles < 2.54 ref 0.420 capture (°C) ≥ 2.54-3.65 2.11 0.69, 6.50 0.192 ≥ 3.66-5.36 2.69 0.77, 9.38 0.121 ≥ 5.37 2.37 0.76, 7.33 0.136

Mean of total precipitation on 54-60 days before Categorized by tertiles* < 0.54 ref 0.507 capture (°C) ≥ 0.54-4.62 1.19 0.48, 2.99 0.700 ≥ 4.63 0.73 0.27, 1.95 0.522

Mean of total precipitation on 84-90 days before Continuous with quadratic term NA 0.78 0.54, 1.13 0.187 capture (°C) Quadratic term NA 1.02 0.98, 1.05 0.325

Rat abundance Categorized by quartiles < 0.16 ref 0.539 ≥ 0.16-0.24 0.85 0.33, 2.33 0.722 ≥ 0.25-0.54 0.50 0.17, 1.60 0.192 ≥ 0.55 0.51 0.11, 1.91 0.294 * Unable to calculate quartiles due to zero counts.

217

Appendix C Table 5. Characteristics of environmental risk factor variables and results of univariable mixed logistic regression models with random effect for block assessing associations with methicillin-resistant Staphylococcus aureus in Norway rats (Rattus norvegicus) captured in Vancouver, Canada. Significant P-values (α=0.05) are indicated in bold.

Category Sub-category Odds 95% CI P Overall P for Ratio categorical variables

Proportion of block occupied by commercial parcels < 0.25 ref 0.484 0.25-0.50 1.13 0.21, 7.04 0.875 ≥ 0.5 0.36 0.05, 3.03 0.272 Proportion of block occupied by industrial parcels None ref > 0 0.20 0.01, 1.52 0.174 Proportion of block occupied by institutional parcels < 0.25 ref ≥ 0.25 5.17 1.23, 35.49 0.024 Proportion of block occupied by green space parcels None ref > 0 4.75 1.12, 34.09 0.051 Proportion of block occupied by parcels under construction None ref > 0 0.60 0.02, 7.44 0.699 Proportion of block occupied by abandoned parcels None ref > 0 0.53 0.08, 2.64 0.434 Proportion of block occupied by open parcels None ref > 0 1.25 0.22, 7.06 0.774 Proportion of block occupied by single family houses None ref > 0 2.02 0.33, 10.01 0.357 Proportion of block occupied by low-rise apartments None ref > 0 2.26 0.31, 9.36 0.290 Proportion of block occupied by mid-rise apartments None ref > 0 1.20 0.18, 6.69 0.819

218

Category Sub-category Odds 95% CI P Overall P for Ratio categorical variables

Proportion of block occupied by housing over commercial buildings* < 0.25 ref 0.383 0.25-0.50 2.64 0.48, 11.13 0.172 > 0.5 0.71 0.04, 8.31 0.757 Proportion of block occupied by buildings not associated with food < 0.25 ref ≥ 0.25 0.32 0.04, 1.93 0.199 Proportion of block occupied by restaurants None ref > 0 0.40 0.06, 2.03 0.247 Proportion of block occupied by grocery stores < 0.25 ref ≥ 0.25 0.88 0.13, 13.76 0.900 Proportion of block occupied by industrial food establishments None ref > 0 0.30 0.01, 2.52 0.318 Proportion of block occupied by other food establishments None ref > 0 2.16 0.44, 19.36 0.368 Proportion of block occupied by buildings in extremely poor condition None ref > 0 1.22 0.21, 7.23 0.802 Proportion of block occupied by buildings in fair condition < 0.25 ref 0.596 0.25-0.50 0.57 0.07, 3.65 0.539 0.5-0.75 1.47 0.12, 16.57 0.721 Proportion of block occupied by buildings in good condition ≥ 0.25 ref > 0.25 1.48 0.19, 12.61 0.688 Proportion of block occupied by buildings in excellent condition None ref 0.755 < 0.25 1.96 0.32, 20.16 0.469 ≥ 0.25 1.44 0.15, 14.74 0.729 Proportion of block occupied by grounds in extremely poor condition None ref > 0 0.28 0.05, 1.40 0.083 Proportion of block occupied by grounds in poor condition < 0.25 ref 0.25-0.50 0.32 0.05, 2.21 0.172

219

Category Sub-category Odds 95% CI P Overall P for Ratio categorical variables

Proportion of block occupied by grounds in fair condition < 0.25 ref 0.073 0.25-0.50 0.18 0.02, 1.06 0.060 0.50-0.75 0.98 0.12, 9.54 0.984 Proportion of block occupied by grounds in good condition < 0.25 ref 0.628 0.25-0.50 1.75 0.21, 9.90 0.510 ≥ 0.50 2.53 0.27, 22.06 0.366 Proportion of block occupied by grounds in excellent condition None ref > 0 2.75 0.38, 12.85 0.182 Proportion of block occupied by green space None ref > 0 2.14 0.23, 11.21 0.366 Proportion of block occupied by unkempt green space None ref > 0 1.27 0.15, 6.33 0.777 Proportion of block occupied by well-kempt green space None ref > 0 1.82 0.33, 9.33 0.418 Proportion of block occupied by food gardens None ref > 0 4.87 0.93, 34.51 0.054 Proportion of alley face in poor condition None ref 0.145 < 0.25 0.17 0.03, 1.07 0.014 0.25-0.5 0.41 0.06, 3.46 3.460 Proportion of alley face in fair condition < 0.5 ref ≥ 0.5 1.31 0.25, 13.63 0.760 Proportion of alley face in good condition < 0.25 ref 0.25-0.75 1.94 0.21, 14.19 0.494 0.651 ≥ 0.75 2.26 0.17, 17.05 0.401 Proportion of alley bordered by non-paved surface None ref > 0 0.96 0.11, 6.41 0.964 Number of rat holes at the alley face None ref

220

Category Sub-category Odds 95% CI P Overall P for Ratio categorical variables

1-4 1.09 0.13, 7.30 0.927 0.860 9 1.54 0.13, 44.50 0.726 ≥ 12 0.42 0.02, 8.83 0.535 Number of rat corridors at the alley face None ref 1-3 0.50 0.04, 3.39 0.503 0.486 ≥ 4 0.28 0.12, 4.03 0.270 Amount of garbage/trash/junk/litter A little ref 0.156 Some 3.53 0.61, 23.99 0.129 A lot 0.75 0.10, 5.74 0.750 Amount of overflowing garbage receptacles None to a little ref Some to a lot 0.22 0.04, 1.08 0.034 Number of commercial garbage receptacles < 11 ref ≥ 11 0.26 0.05, 1.49 0.065 Number of private garbage receptacles None ref 0.058 1-2 4.05 0.69, 12.09 0.010 3-4 2.08 0.30, 10.95 0.370 ≥ 5 9.43 2.47, 32.18 0.000 Number of commercial recycling receptacles ≤ 1 ref 0.034

2-3 2.10 0.60, 6.73 0.170 4 0.22 0.03, 1.04 0.073 ≥ 6 0.31 0.04, 1.45 0.163 Number of private recycling receptacles None ref 0.291 1 1.52 0.08, 15.43 0.715 2-4 0.34 0.02, 3.16 0.356 ≥ 5 2.37 0.34, 22.67 0.368 Presence of strong odors Yes ref No 1.58 0.24, 13.07 0.628

221

Category Sub-category Odds 95% CI P Overall P for Ratio categorical variables

Amount of loitering None to light ref 0.019 Moderate 10.96 2.24, 18.00 0.004 Heavy 3.00 0.54, 18.00 0.176 Amount of transport Light ref 0.600 Moderate 2.09 0.28, 18.99 0.445 Heavy 0.92 0.07, 11.87 0.941

* Refers to buildings in which the ground floor is dedicated to commercial businesses and upper floors are residential.

222

Appendix C Table 6. Characteristics of weather and abundance risk factor variables and results of univariable mixed logistic regression models with random effect for block assessing associations with methicillin-resistant Staphylococcus aureus carriage in Norway rats (Rattus norvegicus) captured in Vancouver, Canada. Significant P-values (α=0.05) are indicated in bold.

Overall P for Sub-category Odds Categorical Category Variable Format Ratio 95% CI P Variables Mean temperature in 5 days before capture (°C) Categorized by quartiles < 6.14 ref 0.321 ≥ 6.14-9.43 3.58 0.82, 25.08 0.126 ≥ 9.44-12.31 4.17 0.62, 39.91 0.153 ≥ 12.32 3.39 0.47, 37.01 0.247

Mean temperature in 10 days before capture (°C) Categorized by quartiles < 6.87 ref 0.556 ≥ 6.87-10.36 1.98 0.47, 9.41 0.359 ≥ 10.37-12.31 1.73 0.17, 16.80 0.596 ≥ 12.32 3.44 0.49, 53.97 0.247 Mean temperature in 15 days before capture (°C) Categorized by quartiles < 6.58 ref 0.167 ≥ 6.58-9.56 0.21 0.03, 1.04 0.081 ≥ 9.57-12.40 1.03 0.18, 4.71 0.968 ≥ 12.41 0.34 0.04, 2.82 0.279 Mean temperature in 30 days before capture (°C) Continuous NA 0.91 0.73, 1.14 0.384

Mean temperature in 60 days before capture (°C) Categorized by quartiles < 5.72 ref 0.094 ≥ 5.72-10.41 1.58 0.40, 9.26 0.512 ≥ 10.42-13.23 0.18 0.01, 2.14 0.169 ≥ 13.24 0.26 0.02, 2.13 0.187 Mean temperature in 90 days before capture (°C) Categorized by quartiles < 6.00 ref 0.259

223

Overall P for Sub-category Odds Categorical Category Variable Format Ratio 95% CI P Variables ≥ 6.00-8.72 1.68 0.37, 12.57 0.500 ≥ 8.73-14.92 0.39 0.04, 4.56 0.381 ≥ 14.93 0.26 0.02, 2.57 0.226 Total precipitation in 5 days before capture (mm) Categorized by quartiles < 9.20 ref 0.510 ≥ 9.20-19.79 0.80 0.24, 2.60 0.709 ≥ 19.80-30.59 0.88 0.24, 3.21 0.849 ≥ 30.60 0.29 0.04, 1.57 0.163

Total precipitation in 10 days before capture Categorized by quartiles < 29.20 ref 0.085 (mm) ≥ 29.20-37.69 0.54 0.13, 2.04 0.372 ≥ 37.70-51.59 0.84 0.28, 2.50 0.750 ≥ 51.60 0.29 0.01, 0.67 0.045

Total precipitation in 15 days before capture Categorized by quartiles < 49.40 ref < 0.001 (mm) ≥ 49.40-59.29 0.03 0.01, 0.22 0.003 ≥ 59.30-79.50 0.16 0.04, 0.54 0.006 ≥ 79.60 0.11 0.02, 0.41 0.003

Total precipitation in 30 days before capture Categorized by quartiles < 108.50 ref 0.253 (mm) ≥ 108.50-124.79 0.54 0.11, 2.33 0.404 ≥ 124.80-150.19 0.21 0.03, 1.08 0.073 ≥ 150.20 0.68 0.11, 3.75 0.646

Total precipitation in 60 days before capture Categorized by quartiles < 202.30 ref 0.474 (mm) ≥ 202.30-224.19 1.05 0.12, 11.75 0.966

224

Overall P for Sub-category Odds Categorical Category Variable Format Ratio 95% CI P Variables ≥ 224.20-299.79 0.54 0.05, 5.63 0.595 ≥ 299.80 2.16 0.22, 27.82 0.500

Total precipitation in 90 days before capture Categorized by quartiles < 227.50 ref 0.050 (mm) ≥ 227.50-400.59 0.29 0.01, 5.28 0.385 ≥ 400.60-484.39 2.94 0.32, 31.83 0.287 ≥ 484.40 4.38 0.60, 60.13 0.146

Mean of minimum temperatures on 7 days before Categorized by quartiles < 3.77 ref 0.378 capture (°C) ≥ 3.77-6.62 1.89 0.49, 8.11 0.364 ≥ 6.63-9.10 3.15 0.61, 21.88 0.170 ≥ 9.11 0.91 0.09, 10.59 0.931

Mean of minimum temperatures on days 8-14 Categorized by quartiles < 4.10 ref 0.626 before capture (°C) ≥ 4.10-7.46 1.08 0.18, 9.14 0.929 ≥ 7.47-8.70 1.50 0.46, 15.60 0.662 ≥ 8.71 0.34 0.02, 4.96 0.383 Mean of minimum temperatures on days 24-30 Categorized by quartiles < 3.77 ref 0.054 before capture (°C) ≥ 3.77-7.28 0.36 0.04, 1.47 0.222 ≥ 7.29-10.60 0.06 0.00, 0.47 0.022 ≥ 10.61 0.25 0.03, 1.55 0.140 Mean of minimum temperatures on days 54-60 Categorized by quartiles < 3.13 ref 0.253 before capture (°C) ≥ 3.13-4.40 1.07 0.39, 2.92 0.893 ≥ 4.41-13.30 0.27 0.03, 1.90 0.166 ≥ 13.31 0.19 0.02, 1.19 0.092

225

Overall P for Sub-category Odds Categorical Category Variable Format Ratio 95% CI P Variables

Mean of minimum temperatures on days 84-90 Categorized by quartiles < 3.11 ref 0.022 before capture (°C) ≥ 3.11-5.36 0.29 0.05, 1.06 0.089 ≥ 5.37-14.03 0.05 0.00, 0.39 0.017 ≥ 14.04 0.19 0.01, 0.89 0.059

Mean of maximum temperatures on 7 days before Categorized by quartiles < 8.99 ref 0.695 capture (°C) ≥ 8.99-12.80 0.29 0.19, 3.79 0.878 ≥ 12.81-14.75 0.05 0.44, 19.32 0.321 ≥ 14.76 0.19 0.20, 12.08 0.769

Mean of maximum temperatures on days 8-14 Categorized by quartiles < 8.56 ref 0.972 before capture (°C) ≥ 8.56-13.18 1.18 0.18, 10.82 0.864 ≥ 13.19-15.22 1.18 0.2, 9.51 0.846 ≥ 15.23 0.79 0.10, 10.46 0.824

Mean of maximum temperatures on days 24-30 Continuous NA 0.87 0.70, 1.01 0.084 before capture (°C)

Mean of maximum temperatures on days 54-60 Categorized by quartiles < 7.47 ref 0.118 before capture (°C) ≥ 7.47-9.98 0.56 0.14, 1.73 0.345 ≥ 9.99-22.06 0.22 0.03, 1.16 0.074 ≥ 22.07 0.08 0.00, 0.67 0.041

Mean of maximum temperatures on days 84-90 Continuous NA 0.88 0.73, 0.99 0.050 before capture (°C)

226

Overall P for Sub-category Odds Categorical Category Variable Format Ratio 95% CI P Variables Mean of mean temperatures on 7 days before Categorized by quartiles < 6.49 ref 0.295 capture (°C) ≥ 6.49-9.83 2.09 0.54, 9.08 0.292 ≥ 9.84-12.26 1.64 0.17, 17.58 0.637 ≥ 12.27 4.81 0.68, 84.49 0.150

Mean of mean temperatures on days 8-14 before Categorized by quartiles < 6.44 ref 0.774 capture (°C) ≥ 6.44-10.45 1.28 0.23, 9.68 0.777 ≥ 10.46-12.40 0.97 0.16, 7.71 0.976 ≥ 12.41 0.42 0.04, 5.43 0.451

Mean of mean temperatures on days 24-30 before Categorized by quartiles < 6.20 ref 0.544 capture (°C) ≥ 6.20-10.36 0.49 0.09, 2.17 0.348 ≥ 10.37-12.79 0.30 0.04, 1.74 0.187 ≥ 12.80 0.42 0.04, 2.47 0.253

Mean of mean temperatures on days 54-60 before Categorized by quartiles < 5.40 ref 0.138 capture (°C) ≥ 5.40-7.49 1.54 0.55, 4.46 0.409 ≥ 7.50-17.82 0.15 0.01, 1.10 0.093 ≥ 17.83 0.42 0.06, 2.12 0.269

Mean of mean temperatures on days 84-90 before Categorized by quartiles < 5.45 ref 0.029 capture (°C) ≥ 5.45-7.15 0.28 0.05, 1.08 0.097 ≥ 7.16-17.33 0.08 0.00, 0.53 0.023 ≥ 17.34 0.42 0.00, 0.72 0.046

227

Overall P for Sub-category Odds Categorical Category Variable Format Ratio 95% CI P Variables Mean of total precipitation on 7 days before Categorized by quartiles < 2.63 ref 0.091 capture (°C) ≥ 2.63-3.00 1.25 0.44, 3.71 0.676 ≥ 4.00-5.33 0.50 0.10, 2.11 0.355 ≥ 5.34 0.14 0.01, 0.92 0.079

Mean of total precipitation on 8-14 days before Main effect NA 0.36 0.19, 0.67 0.001 capture (°C) Quadratic term NA 1.08 1.02, 1.15 0.007

Mean of total precipitation on 24-30 days before Categorized by quartiles < 2.40 ref 0.701 capture (°C) ≥ 2.40-3.65 1.61 0.44, 6.40 0.478 ≥ 3.66-5.42 0.85 0.16, 3.85 0.836 ≥ 5.43 1.69 0.45, 6.62 0.436

Mean of total precipitation on 54-60 days before Categorized by tertiles* < 0.31 ref 0.232 capture (°C) ≥ 0.31-4.06 1.62 0.28, 9.69 0.579 ≥ 4.07 3.46 0.62, 22.60 0.156

Mean of total precipitation on 84-90 days before Categorized by quartiles < 1.01 ref 0.008 capture (°C) ≥ 1.01-4.53 0.56 0.02, 10.49 0.683 ≥ 4.54-7.19 2.63 0.30, 42.52 0.386 ≥ 7.20 8.12 1.18, 136.56 0.043

Rat abundance Categorized by quartiles < 0.16 ref 0.934 ≥ 0.16-0.38 1.51 0.23, 11.90 0.628 ≥ 0.39-0.54 1.71 0.11, 28.87 0.631

228

Overall P for Sub-category Odds Categorical Category Variable Format Ratio 95% CI P Variables ≥ 0.55 8.12 0.03, 47.00 0.869

*Unable to calculate quartiles due to zero counts

229

230