An Abstract of the Dissertation Of

AN ABSTRACT OF THE DISSERTATION OF

Brian S. Dugovich for the degree of Doctor of Philosophy in Zoology presented on May 28, 2019.

Title: Investigating Variation in Immunity and Infection Risk in Wild Ungulates.

Abstract approved: ______Anna E. Jolles Brian P. Dolan

In the midst of the sixth mass extinction, understanding wildlife disease spillover is critical to maintaining protected wildlife areas. Studying ecoimmunology and wildlife disease ecology helps to understand immune and disease traits in an ecological context, which is invaluable in preventing pathogen spillover between livestock and wildlife. To investigate this interface, tests of innate immunity were validated in various ungulates and desert bighorn sheep, and then infectious spillover diseases in desert bighorn sheep and African buffalo were examined. In ungulates at Wildlife Safari in Oregon, innate immunity varied between species and strong positive correlations were found between innate immunity measures. In general, ecoimmunology study results are robust to the choice of innate immune defense assay. In desert bighorn sheep in California, an ELISA was developed that could measure natural antibody levels from blood samples, and differences in natural antibody levels and a bacterial killing assay were found between

Peninsular and Mojave desert bighorn sheep. Spillover of Mycoplasma ovipneumoniae from domestic animals to desert bighorn sheep can lead to region-wide extinction.

Disease prevalence, connectivity, genetic data, and immunophenotype were utilized to

show that better connected populations resisted pathogen invasion. Finally, in Kruger

National Park, South Africa, African buffalo calves infected with foot-and-mouth disease virus reinfected previously exposed animals to become carriers. Other risk factors for carrier status were identified: low body condition score, younger age, and being male.

Identifying these attributes during risk assessment of stray buffalo could be used to help avoid foot-and-mouth disease spillover to cattle on the park border.

INVESTIGATING VARIATION IN IMMUNITY AND INFECTION RISK IN WILD UNGULATES

by Brian S. Dugovich

A DISSERTATION

submitted to

Oregon State University

in partial fulfillment of the requirements for the degree of

Doctor of Philosophy

Presented May 28, 2019 Commencement June 2020

Doctor of Philosophy dissertation of Brian S. Dugovich presented on May 28, 2019.

APPROVED:

Major Professor, representing Zoology

Chair of the Department of Integrative Biology

Dean of the Graduate School

I understand that my dissertation will become part of the permanent collection of Oregon State University libraries. My signature below authorizes release of my dissertation to any reader upon request.

Brian S. Dugovich, Author

ACKNOWLEDGEMENTS

This work would not have been possible without the help of a diversity of organizations and incredible people. The Dolan Lab funded my first year of PhD studies and the Wildlife Safari Project through the Department of Biomedical Sciences at the

Carlson College of Veterinary Medicine. Special thanks to Benji Alcantar, Kirsten

Thomas, Sarah Cannizzo, and the rest of the Wildlife Safari staff for facilitating animal immobilization and sampling. I am grateful to Benji providing me a place to stay in

Roseburg. Becca Sullivan, Lucie Crane, and Claire Behnke were invaluable for helping with fieldwork and immunoassays and camping by the South Umpqua. I would also like to thank Heather Broughton for helping on the futile task to validate the LPA. The desert bighorn project was made possible by funds from the Epps Lab through the Department of Fisheries and Wildlife at Oregon State University and the Jolles Lab through the

Department of Biomedical Sciences. I am grateful to Heather Broughton, Rhea

Hanselmann, Abby Sage, Rachel Crowhurst, Brian Henrichs, Lindsay Hunter, and

Daniella Dekelaita for assisting on captures and running immunoassays and Ben

Gonzalez and Lora Konde of the California Department Fish and Wildlife for mentorship and integrating our research team into the desert bighorn sheep captures. Additional thanks go to the Sweeney Granite Mountains Desert Research Center for providing housing and laboratory space and the Department of Fisheries and Wildlife for letting the

Golden Falcon fly. I also greatly appreciate Tyler Creech and Brandon Nickerson providing their DBH genetic data for analysis. The Morris Animal Foundation made my fieldwork in South Africa possible by graciously funding me for two years. Much

gratitude is given to SANParks for approving and overseeing the study. Sincere appreciation is given to Peter Buss, Michele Miller, Jenny Hewlett and the rest of the

KNP Veterinary Wildlife Services team for tirelessly capturing buffalo every two months and making scores of sedated buffalo a tranquil affair and Louis Van Schalkwyk and Lin-

Mari deKlerk-Lorist for many more mornings of buffalo capture. I would also like to thank Katherine Scott and Eva Perez-Martin for helping me with the FMDV data. My

South Africa time will always be precious to me with lifetime memories made with the

Jolles Lab buffalo team including Courtney Coon, Julie Rushmore, Erin Gorsich, Kath

Potgieter Forssman, Hannah Tavalire, Claire Couch, Henri Combrink, Leigh Combrink,

Caroline Glidden, Robert Spaan, Johannie Spaan, and Danielle Sisson. Many thanks go to the Department of Integrative Biology at Oregon State University for funding my post- fieldwork PhD studies through teaching assistantships and to Lindsay Biga and Devon

Quick for teaching mentoring. Finally, I owe these amazing wildlife research opportunities, compelling challenges, fieldwork and lab work assistance, amazing discoveries, and exceptional mentorship to Anna Jolles, Brian Dolan, Clinton Epps, and

Brianna Beechler.

CONTRIBUTION OF AUTHORS

Authors: * designate co-authors and ** designate co-principal investigators

Chapters 1 – 4: Dr. Anna E. Jolles (A.E.J.) was my co-adviser, co-principal investigator on the Wildlife Safari project, co-principal investigator for the desert bighorn sheep project, and the principal investigator on the USDA-AFRI FMDV African buffalo project.

Chapters 1 – 3: Dr. Brian P. Dolan (B.P.D) was my co-adviser, co-principal investigator on the Wildlife Safari project, and co-principal investigator for the desert bighorn sheep project.

Chapters 2 – 3: Dr. Clinton W. Epps (C.W.E.) was co-principal investigator on the desert bighorn sheep project.

Chapters 1 – 4: Dr. Brianna R. Beechler (B.R.B.) was a post-doctoral researcher and assistant professor of research who was involved in all projects.

Chapter 1: B.P.D., A.E.J., B.R.B. conceived the project idea and B.P.D. and B.R.B designed the study’s methodology. Dr. Benji Alcantar’s role as a zoo veterinarian was responsible for ungulate capture at Wildlife Safari. Dr. Lucie L. Crane helped collect data and ran the TLR experiment. B.P.D. helped write the manuscript.

Chapter 2: Dr. Melanie J. Peel (M.J.P.)* was co-author with me on this manuscript. B.P.D., B.R.B., A.E.S., C.W.E., and M.J.P. helped conceptualize the project. Investigation was led by Amy. L. Palmer (A.L.P.), Dr. Aleksandra E. Sikora (A.E.S.), and Dr. Ryszard A. Zielke. M.J.P., A.E.S., and B.P.D. designed the study’s methodology. The project was administrated by C.W.E. and B.P.D. and supervised by A.E.S., A.E.J., and B.P.D. B.R.B. and B.P.D. helped write the original draft with review and editing by A.E.S., A.E.J., and C.W.E.

Chapter 3: A.E.J.**, C.W.E.**, B.R.B., and B.P.D. conceived the project’s idea and helped with writing. B.R.B., B.P.D., C.W.E., and Rachel S. Crowhurst designed methodology. Dr. Benjamin Gonzalez was responsible for desert bighorn sheep capture.

Chapter 4: A.E.J., B.R.B., Bryan Charleston (B.C.), and Francois F. Maree (F.F.M.) conceived the project’s idea and A.E.J., B.R.B., B.C., F.M., Eva Perez-Martin, and Katherine Scott designed the methodology. B.R.B. was the project’s administrator.

TABLE OF CONTENTS

Page INTRODUCTION ...... 1

CHAPTER 1: MULTIPLE INNATE ANTIBACTERIAL IMMUNE DEFENSE ELEMENTS ARE CORRELATED IN DIVERSE UNGULATE SPECIES .... 5

Abstract ...... 6

Introduction ...... 6

Materials and Methods ...... 9

Results ...... 16

Discussion ...... 17

Figures and Tables ...... 19

CHAPTER 2: DETECTION OF BACTERIAL-REACTIVE NATURAL IGM ANTIBODIES IN DESERT BIGHORN SHEEP (OVIS CANADENSIS NELSONI) POPULATIONS ...... 23

Abstract ...... 24

Introduction ...... 24

Materials and Methods ...... 27

Results ...... 33

Discussion ...... 37

Figures and Tables ...... 42

CHAPTER 3: POPULATION CONNECTIVITY PROTECTS DESERT BIGHORN SHEEP (OVIS CANADENSIS NELSONI) FROM INFECTIOUS PNEUMONIA...... 47

Abstract ...... 48

Introduction ...... 48

TABLE OF CONTENTS (Continued)

Page Materials and Methods ...... 50

Results and Discussion ...... 56

Figures and Tables ...... 64

CHAPTER 4: FOOT-AND-MOUTH DISEASE EPIDEMIOLOGY IN A HERD OF ITS NATURAL RESERVOIR HOST, AFRICAN BUFFALO (SYNCERUS CAFFER) ...... 72

Abstract ...... 73

Introduction ...... 73

Materials and Methods ...... 76

Results ...... 80

Discussion ...... 82

Figures and Tables ...... 88

CONCLUSION ...... 99

BIBLIOGRAPHY ...... 102

APPENDICES ...... 121

LIST OF FIGURES

Figure Page

1.1 Variation in innate immune parameters among ungulate species ...... 20

1.2 Measures of innate immunity were correlated in individual animals after accounting for species differences ...... 21

2.1 Desert bighorn sheep IgM binds to cell-surface proteins from the marine pathogen V. coralliilyticus ...... 43

2.2 ELISA reliably detects desert bighorn sheep IgM binding to V. coralliilyticus cell envelope proteins ...... 44

2.3 Natural IgM levels differ between geographically distinct desert bighorn sheep populations and correlate with bacteria killing capability of plasma .... 45

3.1 Map of desert bighorn sheep populations in the study area in the Mojave Desert ...... 65

3.2 Mycoplasma ovipneumoniae exposure was lower in well-connected populations of desert bighorn sheep in the Mojave Desert ...... 66

3.3 Well-connected populations of desert bighorn sheep in the Mojave Desert had higher neutral and adaptive-linked genetic diversity, and adaptive- linked genetic diversity was associated with lower Mycoplasma ovipneumoniae prevalence in populations...... 67

3.4 Principal component analysis of individual immunophenotype represented immune response strength and the innate and adaptive immunity axis in desert bighorn sheep in the Mojave Desert ...... 69

4.1 Foot-and-mouth disease virus infection class prevalence in a herd of African buffalo over the course of the study by capture number ...... 89

4.2 A low body condition score in adult African buffalo is associated with carrier status that includes seasonal and sex patterns and increased numbers of foot-and-mouth disease virions recovered ...... 90

4.3 Age and sex influence foot-and-mouth disease virus infection class in a herd of African buffalo in Kruger National Park ...... 92

4.4 Stray African buffalo events in the vicinity of Kruger National Park (1998 – 2013) ...... 93

LIST OF TABLES

Table Page

1.1 Adult ungulates included in the study ...... 22

3.1 Population-level relationships of connectivity, Mycoplasma ovipneumoniae exposure, and genetic diversity in desert bighorn sheep in the Mojave Desert by linear regression...... 70

3.2 Infected individual desert bighorn sheep in the Mojave Desert had upregulated innate immune responses, but stronger adaptive responses were associated with lower active Mycoplasma ovipneumoniae prevalence; individual variation in immune function was associated with immunogenetic variation at the BL4 locus but not adaptive-linked heterozygosity ...... 71

4.1 Summary of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park ...... 94

4.2 Whole herd models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park ...... 95

4.3 Calf models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park ...... 96

4.4 Adult models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park ...... 97

4.5 Adult reinfection models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park ...... 98

LIST OF APPENDIX FIGURES

Figure Page

A4.1 Foot-and-mouth disease infection class prevalence in a herd of African buffalo in Kruger National Park ...... 122

A4.2 Calf foot-and-mouth disease infection progression in a herd of African buffalo in Kruger National Park ...... 123

A4.3 Adult foot-and-mouth disease carrier prevalence by month in a herd of African buffalo in Kruger National Park ...... 124

LIST OF APPENDIX TABLES

Table Page

A1.1 Primer sets for TLR quantitative PCR of 7 ungulates at Wildlife Safari ...... 126

A1.2 Wildlife Safari study individuals ...... 127

A3.1 Microsatellite loci used to calculate allelic richness and heterozygosity in 11 populations of desert bighorn sheep in the Mojave Desert ...... 129

A4.1 Whole herd models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park ...... 130

A4.2 Calf models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park ...... 131

A4.3 Adult models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park ...... 133

A4.4 Tonsillar swab PCR copies of foot-and-mouth disease virus models of adult carrier buffalo in a herd of African buffalo in Kruger National Park .... 135

A4.5 Adult reinfection models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park ...... 136

A4.6 Mature female reproduction models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park ...... 137

A4.7 Models of stray African buffalo in the vicinity of Kruger National Park (1998-2013)...... 138

INTRODUCTION

The biosphere is currently undergoing its sixth mass extinction (Barnosky et al.

2011) with vertebrate extinction accelerating in the past 200 years (Ceballos et al. 2015).

The looming crisis of climate change in combination with the magnitude of disruption caused by human overpopulation and overconsumption threaten the extinction of a million described species (IPBES 2019). Moreover, 52% of Earth’s total animal population has been lost over the past 45 years, driven primarily by habitat loss and degradation and exploitation through overhunting and overfishing (WWF 2014). The severity of these extinctions, extirpations, and population declines will threaten human civilization with a break-down of ecosystem services and has been termed “biological annihilation” (Ceballos, Ehrlich, and Dirzo 2017). Despite conservation measures to protect biodiversity, 77% of land has been modified by humans (Watson et al. 2018), and

10% of the remaining wilderness has been lost in the past two decades (Watson et al.

2016).

Urbanization of land has created a diversity of wildlife-livestock-human interfaces (Hassell 2017), and a large player threatening global biodiversity is the emergence of infectious diseases, which carry dire consequences for human, domestic animal, and wildlife health (Daszak, Cunningham, and Hyatt 2000). However, in spite of the issue’s global importance the allocation of resources to disease investigation and prevention remains low (Jones et al. 2008). Even in the United States where 80% of pathogens involve wildlife, social and political will have stymied efforts to manage wildlife-livestock interface infectious diseases, and new collaborations and technology are needed for successful management (Miller, Farnsworth, and Malmberg 2013).

Furthermore, focusing veterinary medicine on the livestock-wildlife interface has the potential to better conserve biodiversity and alleviate poverty in developing countries

(Muma et al. 2014).

Studying host-parasite interactions in an ecological context in the young fields of ecoimmunology and disease ecology will help researchers understand how anthropogenic changes impact health on the wildlife-human interface (L. B. Martin, Boughton, and

Ardia 2014). Understanding how immunity and disease interacts with ecology and disturbances (e.g. population fragmentation, genetic bottlenecking, and interspecies contact) may elucidate mechanisms that can be utilized in future wildlife management

(Downs and Stewart 2014; Joseph et al. 2013; McKnight et al. 2017). Furthermore, determining what epidemiological aspects of infectious disease can be used to mitigate spillover risk or ameliorate spillover consequences between livestock and wildlife would meld these basic science questions to applied science solutions.

Here, I investigated the variation in immunity and/or disease risk in three study systems: ungulates at Wildlife Safari in Oregon, desert bighorn sheep (Ovis canadensis nelsoni) in the Mojave Desert and Peninsular Range in California, and African buffalo

(Syncerus caffer) in Kruger National Park in South Africa. The first two chapters focused on ecoimmunology in ungulates. Chapter 1 examines innate immunity across captive ungulate species through the use of hematology and laboratory experiments designed to measure innate immune response from blood samples. We expected to find differences in innate immune defense between ungulate species. We also used multiple measures of innate immunity to determine if results were correlated to help simplify what assays are required in an ecoimmunology toolkit to test the innate immune response in the field.

Chapter 2 was published in PLOS One and describes a novel method for detecting natural antibodies in desert bighorn sheep by using cell envelope proteins from the marine bacterium, Vibrio coralliilyticus. Levels of natural antibodies were validated ecologically by comparing desert bighorn sheep populations and functionally by comparing the results to a bacterial killing assay.

Chapter 3 bridges the disciplines of ecoimmunology and disease ecology by studying disease ecoimmunology in Mojave desert bighorn sheep. This study system’s metapopulation is comprised of populations confined to mountain ranges in the desert with varying levels of connectivity between populations. Genetic diversity has previously been shown to vary between populations and is driven by connectivity in this system

(Epps 2005, Epps 2018). Mycoplasma ovipneumoniae is an infectious bacterial disease that is carried by domestic sheep, has spillover to bighorn sheep via close contact, and causes respiratory pneumonia and sometimes death or a carrier state in adults with subsequent low lamb recruitment (Besser 2012). We tested how connectivity affected pathogen invasion into various populations. We then supported these findings by measuring genetic diversity at the population and individual-levels and measuring immunophenotype in individual desert bighorn sheep.

Chapter 4 describes a disease ecology study of foot-and-mouth disease in

African buffalo in Kruger National Park. Foot-and-mouth disease virus is one of the most infectious viruses known and domestic livestock outbreaks cause regionwide economic hardships (Rich and Perry 2011). African buffalo are the only known wildlife reservoir host for the virus (Weaver et al. 2013). After the initial infection in calves, juvenile and young adult buffalo become carriers for the virus, and they likely infect the new calf

cohort each year (Moonen and Schrijver 2000). We aimed to understand the temporal patterns and individual traits that are risk factors for becoming a carrier. Identifying risky animals could help prevent foot-and-mouth disease spillover to livestock on the border of the park.

CHAPTER 1

MULTIPLE INNATE ANTIBACTERIAL IMMUNE DEFENSE ELEMENTS ARE CORRELATED IN DIVERSE UNGULATE SPECIES

Brian S. Dugovich, Lucie L. Crane, Benji B. Alcantar, Brianna R. Beechler, Anna E. Jolles, and Brian P. Dolan

Manuscript submitted to PLOS One 1160 Battery Street, Koshland Building East, Suite 225, San Francisco, CA 94111

Abstract

In this study, we aimed to evaluate to what extent different assays of innate immunity reveal similar patterns of variation across ungulate species. We compared several measures of innate antibacterial immune function across seven different ungulate species using blood samples obtained from captive animals maintained in a zoological park. We measured mRNA expression of two receptors involved in innate pathogen detection, toll-like receptors 2 and 5 (TLR2 and 5), the bactericidal capacity of plasma, as well as the number of neutrophils and lymphocytes. Species examined included aoudad

(Ammotragus lervia), American bison (Bison bison bison), yak (Bos grunniens),

Roosevelt elk (Cervus canadensis roosevelti), fallow deer (Dama dama), sika deer

(Cervus nippon), and Damara zebra (Equus quagga burchellii). Innate immunity varied among ungulate species. However, we detected strong, positive correlations between the different measures of innate immunity – specifically, TLR2 and TLR5 were correlated, and neutrophil to lymphocyte ratio was positively associated with TLR2, TLR5, and bacterial killing ability. Our results suggest that ecoimmunological study results may be quite robust to the choice of assays, at least for antibacterial innate immunity, and that despite the complexity of the immune system, important sources of variation in immunity in natural populations may be discoverable with comparatively simple tools.

Introduction

Innate immune responses are important both for rapidly marshaling the body’s defenses following pathogen exposure and for initiating adaptive immune responses which can help confer long-term protection (Medzhitov and Janeway 1997). There are many components of the vertebrate innate immune system that can be targeted for study,

including recognition elements, effector proteins, and cells. Particular leukocytes, such as neutrophils and macrophages, are known to phagocytose pathogens (Silva and Correia-

Neves 2012). Neutrophils have potent antibacterial capabilities when activated. Several cell types express pathogen recognition receptors (PRRs), such as toll-like receptors

(TLRs), Nod-like receptors, and RIG-I like receptors (Kawai and Akira 2009). PRRs recognize conserved microbial components, such as proteins, lipids, or nucleic acids.

Engagement of PRRs with their appropriate ligands leads to cellular signaling events that can drive immune responses (Brubaker et al. 2015). The resulting inflammatory response can induce the production of many components, such as C-reactive protein and proteins of the complement pathway, which have antimicrobial properties and are part of the innate immune response (Gabay and Kushner 1999). Thus, multiple immune parameters are available for measuring innate immune defenses. As such, for ecoimmunological studies, it is important to evaluate to what extent the choice of immune marker is likely to alter study conclusions and to avoid redundancy between assays.

Assay choice for ecoimmunological studies is fraught with limitations related to practicality and cost. Furthermore, investigations involving rarely studied species depend on assays without species-specific reagents. Choosing the appropriate immunological measures to evaluate immune mechanisms in a study system is the most basic question to studying wildlife ecoimmunology (Demas et al. 2011). However, the ideal immunological assay strategy for a laboratory setting is often not feasible in the field due to the lack of mobile equipment and an appropriate lab space. A successful wildlife ecoimmunology study depends on efficiency and limiting the ecoimmunology toolkit to the minimum assays required to answer the research question in mind.

In this report we used multiple measures to assess innate antibacterial immune defenses from a single blood sample in various wild ungulate species. Innate immune defenses are frequently evaluated in ecoimmunological studies due to their importance as the immediate defense against pathogen invasion (Downs and Stewart 2014). Our ecoimmunology toolkit to measure innate immune defenses included expression of toll- like receptor 2 (TLR2) and toll-like receptor 5 (TLR5) in leukocytes, neutrophil and lymphocyte counts, ratio of neutrophils to lymphocytes (NLR), and measuring the bactericidal abilities of plasma using a bacterial killing assay (BKA). TLR2 is a PRR which, in combination with either TLR1 or TLR6, recognizes bacterial-derived peptidoglycan (de Oliviera Nascimento, Massari, and Wetzler 2012). TLR5 is a PRR which recognizes bacterial flagellin (Yoon et al. 2012). Neutrophils have potent antibacterial capabilities when activated. The ratio of neutrophils to lymphocytes is an important clinical measure of inflammatory status (Guthrie et al. 2013) and physiological stress (Davis, Maney, and Maerz 2008), and normal ranges vary between species (Fowler and Miller 2003). Increased NLR is an indication of inflammatory status, which may also increase those plasma components that have antibacterial properties. A bacterial killing assay (BKA) directly measures the ability of blood or plasma to kill a laboratory strain of bacteria and can be used as a proxy for innate immune competence (Beechler et al. 2012).

In whole blood, live white blood cells kill bacteria, but the mechanism that allows plasma to kill bacteria is less clear. Complement proteins interacting with antibodies are likely involved (Liebl and Martin 2009; Matson, Tieleman, and Klasing 2006), and heat inactivation of proteins showed decreased bactericidal activity (Dugovich et al. 2017).

We investigated innate, antibacterial immune mechanisms in seven wild ungulate species maintained in a zoological park. Ungulates provide a tractable, ecologically and economically important taxon to investigate questions in ecoimmunology and disease ecology (Jolles, Beechler, and Dolan 2015; Jolles and Ezenwa 2015) since ungulates commonly interface with domestic livestock resulting in infectious disease spillover

(Martin et al. 2011). Despite their ubiquity, wild ungulates are a clade of animals whose immune functions have been relatively understudied. One factor complicating the study of wild ungulate immune defenses is that immobilization drugs and capture stress can alter some blood test results (Cases-Diaz et al. 2012; Marco and Lavin 1999), while other immunological assays in other animals are robust to capture techniques (Strobel, Becker, and Encarnacao 2015). The sensitivity of some immunological assays to animal capture and handling might obscure patterns of variation due to ecological drivers that are of interest in ecoimmunological studies. However, in this study, we were not attempting to identify drivers of immune defense; instead, we focused on testing whether different assays for antibacterial innate immunity are likely to yield congruent patterns. Thus, we predicted that TLRs, cell ratios, and BKA would correlate within individual animals and across ungulate species, indicating that the choice of innate immunological assay should not affect the patterns detected in ecoimmunological studies qualitatively.

Materials and Methods

Sample collection

Blood was collected from captive adult ungulates at Wildlife Safari in Winston,

Oregon in collaboration with the park and veterinary staff. Table 1.1 lists the animal species studied. Immobilizations were performed as early in the morning as possible to

minimize the risk of hyperthermia. Each animal was immobilized with an air rifle dart containing a mixture of drugs delivered intramuscularly. The anesthesia protocol was tailored to each species. Carfentanil or thiafentanil (Zoopharm), xylazine (VetOne), and ketamine (VetOne) were used in combination to immobilize the larger ungulates (>150 kg); while, telazol (Zoetis) and xylazine were used in the smaller ungulates (<150 kg).

When the animal was immobilized, a physical exam was performed, and routine veterinary care (intravenous catheter and fluids, dart wound care, and hoof trimming) was provided. Heart rate, respiratory rate, and temperature were monitored at 5-minute intervals for the length of the sedation (approximately 20 minutes in most animals). Time to blood sampling ranged from 5 – 20 minutes. Blood samples were collected from jugular or lateral saphenous veins and placed in heparinized tubes and serum separating tubes. Time of blood draw to blood testing varied from thirty minutes to two-and-a-half hours. For reversal, naltrexone (Zoopharm) was administered intravenously and atipamezole (Zoetis) was administered intramuscularly. Animals were monitored during the time immediately following recovery and then periodically throughout the rest of the day.

Leukocyte differential analysis

Total and differential white blood cell counts were used to quantify the leukocytes present in the animal’s blood. Total white blood cell count was obtained using a

HemaTrue analyzer (Heska SN 61372) in Wildlife Safari’s veterinary laboratory. The cow setting was used for cervids and bovids while the horse setting was used for zebra.

Differential counts were performed from a blood smear slide made on the day of capture.

Absolute numbers of each leukocyte type were determined by multiplying the total white blood cell count by the differential percentages.

Bacterial killing assay

Plasma obtained from the immobilizations was stored in a -80 °C freezer. The

DH5α laboratory strain of Escherichia coli containing the gentamycin resistance plasmid pJN105 was used as a target for killing by plasma components using a previously established protocol (Dolan et al. 2016). Briefly, five colonies from an agar plate were collected using a BBL™ prompt and diluted in sterile phosphate buffered saline to approximately 20,000 CFU’s/30 μl and distributed to each well of a 96-well plate.

Frozen plasma samples were thawed for the first time in an ice water bath and were diluted 1:75 in PBS and mixed with the bacteria for 30 minutes at 37 °C. Tryptic soy broth (100 μl with 10 μg/ml gentamycin) was then added to the cells and the plate was incubated with gentle agitation at 37 °C. Starting at 5 hours, and every hour thereafter, the absorbance of each well at 600 nm was recorded and plotted as a function of time.

Each sample was run in triplicate on a different plate, and the three plates were run at the same time. The growth curve was fitted with a sigmoidal curve, using GraphPad Prism version 6.07, GraphPad Software, La Jolla, California USA, and the time to 50% growth for each sample was determined. Additional wells lacking plasma but containing increasing dilutions of the initial starting concentration of bacteria were also included on each plate to generate a standard curve. The time to 50% growth for each standard was plotted as function of percent dilution and fitted with a linear regression. The line of best fit was used to determine the percent reduction of bacteria in each sample (i.e., the longer the time to 50% growth, the more efficiently plasma was killing bacteria). The triplicate

results were averaged to calculate the percent-reduction in bacteria. If no growth was observed, then the result was deemed 100% reduction in bacterial growth.

RNA isolation and cDNA synthesis

Leukocytes from ~ 5 ml of whole blood collected in heparinized tubes were isolated by centrifuging blood collection tubes for 20 minutes at 1,500 RCF. The buffy coat was gently removed using a disposable transfer pipet, and contaminating erythrocytes were eliminated by lysing in ammonium-chloride-potassium lysing buffer.

Leukocytes were washed in PBS and pelleted. Total RNA was extracted using Macherey-

Nagel NucleoSpin RNA isolation kits following the manufacturer’s recommended guidelines. To generate cDNA, 5 μl of RNA was combined with 15 μl water and added to an RNA to cDNA EcoDry Premix reaction tube containing oligo dT primers

(Clontech) following the manufacturer's recommended guidelines.

Primer design

To design primers to work on a variety of ungulate species, most of which do not have complete genomes available, mRNA sequences from a variety of carnivores and ungulates species were aligned using Clustal W software to identify regions of homology.

Universal primers for the mammalian glyceraldehyde 3-phosphate dehydrogenase

(GAPDH), TLR2, and TLR5 genes were constructed using published mRNA sequences from the giant panda (Ailuropoda melanoleuca), horse (Equus caballus), cattle (Bos taurus), sheep (Ovis aries), dog (Canis lupus familiaris), cat (Felis catus), pig (Sus scrofa), and American bison (Bison bison). The mRNA sequences NM_001304846.1,

NM_001163856.1, NM_001034034.2, NM_001190390.1, NM_001003142.2,

NM_001009307.1, NM_001206359.1, and XM_010844969.1 were included during

development for the GAPDH primer. The mRNA sequences XM_002913846.2,

NM_001081796.1, NM_174197.2, NM_001048231.1, NM_001005264.3,

XM_003984930.3, NM_213761.1, and XM_010840725.1 were aligned for the TLR2 primer. No mRNA sequence of the TLR5 gene was available for the horse. The mRNA sequences XM_011228904, NM_001040501.1, NM_001135926.1, NM_001197176.1,

XM_011290806.1, NM_001123202.1, and XM_010852831.1 were used to construct the

TLR5 primer. These mRNA sequences were uploaded from the NCBI database and aligned using ClustalW (Lasergene Software, Megalign program). The aligned sequences were reviewed, and homologous areas were investigated as possible primer sequences.

The primer sequence needed to have no more than three degenerate bases and be within

250 base pairs of each other. The primer sequences and the expected length and average molecular weight of the replicated sequence are listed in the supporting information

(Appendix Table 1.1). Primers were purchased from IDT and validated using a random cDNA sample as a template in a polymerase chain reaction (PCR). Primers were resuspended in distilled water and 0.16 μMol of each forward and reverse primer was mixed with 1 μl cDNA, 22 μl of water, and 25 μl of 2X FideliTaq PCR Master Mix

(Affymetrix) and cycled at 95 °C for 15 seconds, 53 °C for 15 seconds, and 68 °C for 30 seconds for a total of 35 cycles. PCR products were examined by gel electrophoresis to ensure the correct size, and DNA was purified using a PCR purification kit (Macherey-

Nagel) according to the manufacturer’s instructions. The concentration of the purified

PCR product was determined using a Qubit 2.0 fluorometer (Invitrogen) and converted from ng/μl to copies/μl. The DNA solution was diluted in serial 10-fold dilutions and used to generate a standard curve in the subsequent quantitative PCR reactions. Our

primers (sequences: Appendix Table 1.1) amplified a PCR product of the predicted size in all ungulate species tested and could be used in quantitative PCR reactions.

Quantitative PCR

Quantitative PCR was completed using Fast SYBR Green (Applied Biosciences) technology and analyzed on a StepOnePlus (Applied Biosystems) thermocycler set for fast analysis. Each PCR reaction contained 0.16 μMol of each primer, 0.5 μl of cDNA,

11.5 μl of water, and 12.5 μl 2X Fast SYBR Green master mix for a total of 25 μl. The reaction was added to well of a 96-well plate (ABI) and cycled with a run method of one

10 minute 95 °C time period and then 40 cycles of 15 second 95 °C denaturing phase, 10 second 53 °C annealing phase, and 30 second 72 °C extension phase. A final melting curve analysis was performed for each plate to ensure that only one product was present.

The cycle threshold (Ct) number for each cDNA sample was calculated using StepOne

Software for StepOnePlus Real-Time PCR systems (Applied Biosystems) for all three genes. Each transcript target was run in technical triplicates and averaged. To convert Ct values to copies/μl, a standard curve was generated for each gene product and each plate run using the previously described PCR standards. TLR proteins are expressed on a variety of cell types, including leukocytes, to a varying extent. TLR2 and TLR5 copies were subsequently normalized by dividing the copies/μl for each TLR gene by the

GAPDH copies/μl and the resulting ratio used for subsequent analysis.

Statistical analysis

Animals that were severely ill were excluded from analysis. To help improve sample size and simulate wildlife sampling where few animals are perfectly healthy, animals with minor disease were kept in the study (Appendix Table 1.2). A conservative

one-way independent ANOVA power analysis showing seven groups of three samples, a large effect size of 50%, and a significance level of 0.05, showed a 23.9% chance to find significant differences confirming that we could find some strong relationships despite our small sample size. Non-normal distributions of independent variables (TLR2, TLR5, plasma BKA, NLR, and lymphocytes) were confirmed with a Shapiro-Wilk’s test and the data were log-transformed for use in linear models. An ANOVA with Tukey’s honest significant difference (HSD) test was used to test for all pair-wise differences in immune measures between species. To confirm these differences, a permutational multivariate analysis of variance (PERMANOVA) was used to compare the immune measures to species or family. To evaluate a relationship between TLR2 and TLR5 expression, a generalized linear mixed model was used. Species was used as a random effect since differences in immune measures were expected between species. Generalized linear mixed models with species as a random effect were also used to find a relationship at the individual-level between immune measures. Animals captured were predominantly female (Table 1.1). Normality testing, Tukey’s HSD ANOVA, and general linear mixed modeling were implemented using R version 3.4.3 (R Core Team, 2017). PERMANOVA used the vegan package (Oksanen et al. 2007) and used a Bonferroni correction. The nlme package (Pinheiro et al. 2014) used restricted maximum-likelihood (REML) for mixed models using non-count parameters (TLR2, TLR5, plasma BKA, and NLR). R2 for

2 mixed effect linear models was calculated using RΣ (Jaeger et al. 2017). Count data

(WBC, neutrophils, and lymphocytes) was analyzed using the lme4 package (Bates et al.

2015) to run a penalized quasi-likelihood model with a Poisson distribution and log-link.

Results

Immunological variation among ungulate species

Some aspects of innate immunity varied among ungulate species. Zebra had the highest expression of TLR2 and TLR5 (Figures 1.1A and B), and cervids tended to have a lower lymphocyte count and thus a higher NLR than the other species (Figures 1.1D and E). However, we did not detect significant differences in neutrophil titers or bacterial killing across species (Figures 1.1C and F). Permanova analysis of the overall suite of innate immune defenses showed significant differences between fallow deer and elk (p =

0.021) and fallow deer and sika (p = 0.021) and borderline significance between cervids and bovids (p = 0.057).

Correlation among immune metrics

We also found positive correlation between the following immune measures:

TLR2 and TLR5, NLR and TLR2, NLR and TLR5, and NLR and BKA (Figure 1.2).

2 TLR2 and TLR5 were positively correlated (β = 0.27, SE = 0.46, p = 0.0019, RΣ = 0.19).

2 NLR positively correlated with TLR2 (β = 0.45, SE = 0.20, p = 0.028, RΣ = 0.11), TLR5

2 (β = 0.27, SE = 0.12, p = 0.025, RΣ = 0.09), and BKA (β = 0.22, SE = 0.09, p = 0.018,

2 RΣ = 0.13). Sample sizes were too small to resolve correlations of immune measures within a species.

Discussion

In this study, we describe comparisons of immune parameters related to innate defenses against bacterial pathogens across seven wild ungulate species. We found that different parameters of innate immune defense (TLR2 and TLR5 expression, NLR, and

BKA) were correlated significantly and positively within individuals across our seven- study species and found clear differences in the innate immune mechanisms in the ungulates studied. Despite the lack of stimulation and chronic disease in our study animals, the data suggest that both TLRs tended to covary in ungulates. Direct stimulation of one TLR may also alter the expression of other TLR genes as expression of TLR5 is upregulated in neutrophils from patients with cystic fibrosis in response to

TLR2 stimulation (Koller et al. 2008). Harada, Isse, and Nakanuma (2006) noted increased expression of both TLR2 and TLR5 in human biliary epithelial cells exposed to interferon gamma (IFNγ) while Homma et al. (Homma et al. 2004) found that IFNγ in combination with other stimulants upregulated TLR2 but not TLR5 in human respiratory epithelial cells, suggesting that there may be cell type specific expression patterns. TLR2 and TLR5 expression differences have been noted in macrophage-like cells as well: infection with Borrelia burgdorferi upregulated both TLR2 and TLR5 expression in microglial cells, but only TLR2 was upregulated in monocytes (Bernardino et al. 2008;

Cabral et al. 2006). Exploring the interplay between simultaneous TLR2 and TLR5 expression would be an interesting future study.

Not only were TLR2 and TLR5 transcripts positively correlated with each other, but also with NLR. Neutrophil abundance is often used as a measure of innate immune competence in comparative studies (Demas et al. 2011), and neutrophils are known to express TLRs, so it is perhaps not surprising that TLR transcript levels were elevated in the presence of increased NLR. Previously, it has been shown that neutrophil count is positively correlated to the ability of plasma to kill bacteria (Beechler et al. 2012) or increase together in a specific treatment (Gervasi et al. 2014; Young et al. 2016), perhaps

suggesting that coordinated antibacterial innate immune responses result in both higher levels of plasma proteins (i.e. complement) and cellular effector populations (i.e. neutrophils). Though we did not directly measure it, activation of the complement system can have a direct effect on neutrophils as the cleaved C3 and C5 proteins (C3a and C5a) act as anaphylatoxins and recruit neutrophils to the sites of activated complement (Mak,

Saunders, and Jett 2013). Interestingly, while neutrophil abundance varied at the individual-level, it was relatively consistent across species, as was the ability of plasma to kill bacteria. However, the NLR did vary between species, with bovids exhibiting higher lymphocyte counts than the other species and NLR was positively correlated with BKA.

Summarizing the positive correlations of innate immune defenses, we find the TLR transcripts correlate with each other, and TLRs and BKA correlate with NLR. As such, our results document striking positive covariance between innate, antibacterial immune mechanisms.

A common objective in ecoimmunological studies is measuring an accurate and relevant snapshot of immune function in wild species. Studies must treat these often- simple measurements as proxies for the entire complex immune response (Martin, Weil, and Nelson 2006) and interpret them in the context of ecological characteristics (Prall and

Muehlenbein 2014). The assumption that observed immunological patterns are robust to the choice of immune measure is essential for interpreting and comparing trends across ecoimmunological studies but has rarely been tested. We found positive correlations between measures of innate immune surveillance (TLR2 and 5), available effector cells

(NLR), and functional antibacterial defense (BKA), validating the notion that different assays commonly used in ecoimmunological studies are likely to yield similar patterns of

interspecific variation in innate immunity. It is our hope that this work will encourage further research to compare across common ecoimmunological assays evaluating adaptive immune mechanisms and studies including a broader range of taxa.

Acknowledgements

Animal capture and sample collection was made possible by the staff of Wildlife

Safari and visiting rotating veterinary students. Funding for analyses was provided by

Oregon State University. All animal work was approved by the animal care and use committee at Oregon State University (ACUP #4450). We are grateful to Lynn Martin for guidance on the initial project proposal and findings. We would like to thank Claire

Behnke, Becca Sullivan, and Heather Broughton for fieldwork and immunological assay work.

Figures and Tables

Figure 1.1 Variation in innate immune parameters among ungulate species.

Figure 1.2 Measures of innate immunity were correlated in individual animals after accounting for species differences.

Table 1.1 Adult animals included in the study.

Figure 1.1 Variation in innate immune parameters among ungulate species. Each data point represents an individual animal with points color-coded by species at Wildlife Safari (2013 – 2014). Black points and bars represent the average and standard error for each species. Significant pairwise differences are denoted by letters above each bar. TLR2 (A), TLR5 (B), lymphocytes (C), and NLR (E) were significantly different between some species. Neutrophils (C) and the BKA (F) were not significantly different between species.

Figure 1.2. Measures of innate immunity were correlated in individual animals after accounting for species differences. Each data point represents an individual animal at Wildlife Safari (2013 – 2014) with log-transformed values and points color-coded by species. TLR2 and TLR5 (A) were positively correlated (β = 0.27, SE = 0.46, p = 0.0019, 2 2 RΣ = 0.19). NLR positively correlated with TLR2 (B; β = 0.45, SE = 0.20, p = 0.028, RΣ 2 = 0.11), TLR5 (C; β = 0.27, SE = 0.12, p = 0.025, RΣ = 0.09), and BKA (D; β = 0.22, SE 2 = 0.09, p = 0.018, RΣ = 0.13).

Table 1.1. Adult ungulate animals included in the study. The following animals were immobilized and sampled at Wildlife Safari (2013 – 2014).

Common Number in study Scientific Name Family Name (females, males) Aoudad Ammotragus lervia Bovidae 3(2,1) Bison Bison bison bison Bovidae 3(3,0) Yak Bos grunniens Bovidae 3(3,0) Roosevelt elk Cervus canadensis roosevelti Cervidae 13(8,5*) Fallow deer Dama dama Cervidae 4(4,0) Sika deer Cervus nippon Cervidae 15(14,1) Damara zebra Equus quagga burchellii Equidae 3(2,1) *includes 3 castrated males

CHAPTER 2

DETECTION OF BACTERIAL-REACTIVE NATURAL IGM ANTIBODIES IN DESERT BIGHORN SHEEP (OVIS CANADENSIS NELSONI) POPULATIONS

Brian S. Dugovich*, Melanie J. Peel*, Amy L. Palmer, Ryszard A. Zielke, Aleksandra E. Sikora, Brianna R. Beechler, Anna E. Jolles, Clinton W. Epps, and Brian P. Dolan

Published in PLOS One 1160 Battery Street, Koshland Building East, Suite 225, San Francisco, CA 94111 12 (6): e0180415. https://doi.org/10.1371/journal. pone.0180415

Abstract

Ecoimmunology is a burgeoning field of ecology which studies immune responses in wildlife by utilizing general immune assays such as the detection of natural antibody. Unlike adaptive antibodies, natural antibodies are important in innate immune responses and often recognized conserved epitopes present in pathogens. Here, we describe a procedure for measuring natural antibodies reactive to bacterial antigens that may be applicable to a variety of organisms. IgM from desert bighorn sheep plasma samples was tested for reactivity to outer membrane proteins from Vibrio coralliilyticus, a marine bacterium to which sheep would have not been exposed. Immunoblotting demonstrated bighorn sheep IgM could bind to a variety of bacterial cell envelope proteins while ELISA analysis allowed for rapid determination of natural antibody levels in hundreds of individual animals. Natural antibody levels were correlated with the ability of plasma to kill laboratory strains of E. coli bacteria. Finally, we demonstrate that natural antibody levels varied in two distinct populations of desert bighorn sheep. These data demonstrate a novel and specific measure of natural antibody function and show that this varies in ecologically relevant ways.

Introduction

Ecoimmunology seeks to explain variation in immunity within and among hosts by examining immunity in the context of the host’s ecology and life history. Because immune responses involve a complex network of protein and cellular signals and effectors, a central conundrum in the field of ecoimmunology is “what to measure”

(Brock, Murdock, and Martin 2014). Moreover, eco-immunological assays must reliably quantify immune responses in non-model species, including wildlife. In designing eco-

immunological assays, the challenge is thus two-fold: First, to identify immune components that are relevant, by demonstrating that their level is indicative of functional immune responses, and second, to design robust assays that can capture variability in the immune component among hosts and populations of non-model animal species.

Natural antibodies (nAbs) are one innate immune component that is frequently measured in ecoimmunological studies. NAbs are present early in an animal’s development and can provide innate-like immune protection. NAbs are generated by a subclass of B cells known as B-1 B cells, which are distinct from typical B cells and tend to localize to the peritoneal cavity, bone marrow, and marginal zone of the spleen (Savage and Baumgarth

2015). They are distinct from adaptive antibodies, which are highly specific to particular pathogens to which the host has previously been exposed to. Natural antibodies are mostly of the IgM subclass of antibodies and tend to have more gene rearrangements lacking N-nucleotide additions than adaptive antibodies (Yang et al. 2015; Kantor et al.

1997). While they may be considered cross-reactive, nAbs are not non-specific, rather, nAbs recognize conserved antigenic epitopes present on multiple microbial agents

(Kearney et al. 2015).

NAbs have multiple functions (Ehrenstein and Notley 2010). NAbs are known to protect against bacterial infections (Briles et al. 1981; Baumgarth, Tung, and Herzenberg

2005; Ochsenbein et al. 1999; Boes et al. 1998) presumably by binding to conserved bacterial antigenic determinants. They can also provide a first line defense against viral infections (Ochsenbein et al. 1999; Lum et al. 2013; Choi and Baumgarth 2008; Lobo et al. 2008; Jayasekera, Moseman, and Carroll 2007). In addition to their innate immune functions, nAbs can also bind to self-antigens and are necessary for clearing apoptotic

bodies (Chen et al. 2009; Wang et al. 2013; Quartier et al. 2005), which is important in limiting the development of autoimmune diseases (Nguyen, Elsner, and Baumgarth 2015;

Notley et al. 2011). Therefore nAb production is necessary for both protection and maintain physiological homeostasis (Baumgarth 2011).

Because they are present from birth and do not require vaccination or exposure to particular antigens to be generated, nAbs can be studied in a variety of wild vertebrate animals. One common technique relies on the ability of natural antibodies in serum samples to agglutinate red blood cells from a different species (Kevin D. Matson,

Ricklefs, and Klasing 2005) by binding to particular blood group glycans. However, hemagglutination assays can be hampered by factors in serum other than antibodies

(Budzko, Jelinek, and Wilcke 1981; Ludewig and Chanutin 1949; Seitanidis 1969). Other tests have been developed that examine antibody binding to a foreign protein to which study animals had not previously been exposed, such as keyhole limpet hemocyanin

(KLH) (Mayasari et al. 2016; van Knegsel et al. 2007; van der Klein et al. 2015) or hen egg ovalbumin (Sandmeier et al. 2012). While these methods provide an inexpensive and repeatable way for determining nAb levels in a variety of organisms, they do not directly address if nAbs bind to bacterial epitopes to provide initial immune protection from bacterial infections.

To address these limitations, we developed a novel method for detecting antibacterial nAb levels in desert bighorn sheep (Ovis canadensis nelsoni, DBH) that does not rely on assessing binding to allogenic blood group antigens or a single foreign protein. Bighorn sheep are an iconic wildlife species that is susceptible to several infectious diseases which historically and currently have devastated wild populations

(Wehausen, Kelley, and Ramey 2011; Besser et al. 2012; Jessup 1985), including in the desert subspecies native to the southwestern United States and northern Mexico. We tested the ability of plasma-isolated desert bighorn sheep IgM to bind to cell envelope proteins of Vibrio coralliilyticus, a marine bacterial pathogen that infects corals and oysters (Ben-Haim et al. 2003; Richards et al. 2015). Because bighorn sheep have not been exposed to this bacterium, no acquired antibody will be present to cross-react with the bacterial targets. Measuring nAbs which react to the surface antigens of a foreign bacterium is more likely to provide ecologically relevant information regarding an animal’s ability to withstand novel infections.

Importantly, we validated our assay in two ways. First, we evaluated the assay’s ecological relevance, by assessing its ability to capture immunological variation among distinct desert bighorn sheep populations. Second, we tested the assay’s functional relevance by comparing nAb levels with bactericidal activity of desert bighorn sheep plasma. Bactericidal capacity of blood or plasma is a commonly used eco-immunological measure of innate immune function. Bacterial killing assays are easily interpreted, because they directly measure variation among animals in their ability to eliminate bacterial invaders from their blood. We find that our assay detects prominent differences in nAb levels between two geographically isolated populations of desert bighorn sheep; and that nAb levels established by our assay correlate with variation in plasma killing capacity among individual bighorn sheep.

Materials and methods

Plasma sample collection

Plasma was obtained from blood samples collected from wild bighorn sheep captured in 2014 for ongoing research on recent respiratory disease outbreaks in the

Mojave Desert and Peninsular Range of California (Epps, Dekelaita, and Dugovich,

2016). Blood was collected from animals immobilized without anesthetics or sedatives into heparinized blood collection tubes via aseptic technique. Within 4 hours of collection, blood was centrifuged at 5000g for 10 minutes. The plasma was pipetted off the top and stored in micro centrifuge tubes at -80 until processing. Domestic sheep plasma was donated from Oregon State University Veterinary Teaching Hospital.

Captures were authorized under NPS research permit MOJA-2013-SCI-0030 and protocols for bighorn sheep capture were approved by the National Park Service

Institutional Animal Care and Use Committee (permit

PWR_MOJA_Epps.Powers_DesertBighorn_2013.A3).

V. coralliilyticus culture and cell envelope isolation

V. coralliilyticus strain RE22 (Elston et al. 2008) was provided by Claudia Hase

(Oregon State University) and grown in 3% NaCl LB broth at 30 °C. For isolation of cellular envelopes, 2 ml of an overnight culture of V. coralliilyticus was used to inoculate

1 L of media, and bacterial cells were grown for 4 hours until an optical density of ~0.5 was measured at 600 nm. Cells were pelleted by centrifugation for 20 minutes at 7500

RCF. After carefully decanting the supernatant, the cell pellet was frozen at -20 °C. Cell envelope proteins were isolated following previously established methods based on

French press cell lysis, followed by treatment with chaotropic reagent (sodium carbonate) and differential centrifugation (Molloy et al. 2000; Zielke, Gafken, and Sikora 2014;

Zielke and Sikora 2014). In comparison to other sub-fractionation methods this protocol

allows an enrichment of outer membrane proteins, yielding a sample compatible with mass spectrometry. Briefly, V. coralliilyticus cell pellets were re-suspended in pre-chilled

1 × phosphate buffered saline pH 7.5 (PBS), supplemented with EDTA-Free Protease

Inhibitors Tablet (Thermo-Fisher), and lysed by passage through French press cell

(Amco) at 12,000 psi. Cell debris was removed by centrifugation at 16,000 RCF. Protein concentration was measured using Bio-Rad Protein Assay, and 10 mg was mixed with 6 volumes of 0.1M sodium carbonate pH 11.0. Samples were incubated with mixing for 1 hour at 4 °C, and membrane proteins were pelleted at 200,000 × g using Type 45 rotor

(Beckman). The pellet containing cell envelope proteins was re-suspended in ice cold

PBS containing 0.1% SDS. Protein concentration was determined using a Bradford Assay

(Bio-Rad).

SDS-PAGE and western blotting

Protein samples were analyzed by SDS-PAGE using precast 4–12% Bolt® Gels and blotted using the iBlot2® dry transfer system (Thermo-Fisher). Ten μl of LDS sample buffer (Thermo-Fisher) was added to 30 μl of sample, and the mixture was boiled for 20 minutes. The gel was loaded with 25 μl of sample and run at 165 volts, until the dye fully exited the gel. Gels were then blotted onto nitrocellulose membranes using iBlot® program 3. Membranes were blocked in Tri-buffered Saline containing 0.1%

Tween-20 (TBS-T) with 5% dehydrated milk. For the analysis of IgM, plasma was diluted 1:10 in water containing 10mM DTT. Bighorn sheep IgM was then detected on blocked membranes using a rabbit polyclonal anti-sheep IgM antibody (Bio-Rad Cat#

AHP950) diluted 1:2000 in 0.5% milk/TBS-T, followed by goat-rabbit polyclonal antibody coupled to IR dye 800 (LI-COR), diluted 1:10,000 in 0.5% milk/TBS-T. For the

analysis of bighorn sheep IgM binding to V. coralliilyticus membrane proteins, protein solutions (approximately 0.7 mg/ml) were diluted 1:1 with water prior to mixing with

LDS sample buffer and loading onto the gel. After blocking membranes, bighorn sheep plasma was diluted 1:100 in TBS-T containing 0.5% milk and added to the membranes.

Membranes were gently rocked overnight, and the following day washed with TBS-T for

5 minutes. Rabbit anti-sheep IgM (diluted 1:2000, in 0.5% milk/TBS-T) was added to the membranes and incubated for 1 hour with gentle rocking. Membranes were washed in

TBS-T, and goat-rabbit polyclonal antibody coupled to IR dye 800 (1:10,000 in 0.5% milk/TBS-T) was applied to the membrane for 1 hour with gentle rocking. For all western blots, following the final incubation with detecting antibodies, membranes were washed with TBS-T for 5 minutes followed by water for 5 minutes. Antibody binding was then visualized using an Odyssey SA infrared detection system.

IgM and nAb ELISA

Bighorn sheep IgM was quantified by ELISA using a kit from Life Diagnostics

(Cat# IGM-12) following the manufacturer’s recommendations. Samples were analyzed in duplicate and converted to a concentration of IgM using standards included in the kit.

To determine nAb binding, 96-well polystyrene were coated with 350 ng of V. coralliilyticus membrane protein in 100 μl of PBS overnight at room temperature.

ELISA plates were washed twice and blocked with 300 μl of 1% bovine serum albumin in PBS for 1 hour. Plates were washed five times, and 50 μl of bighorn sheep plasma, diluted 1:2000 in PBS, was added to each well. To test specificity of the ELISA, selected plasma samples were diluted in V. coralliilyticus cellular envelope proteins prior to addition to coated and blocked ELISA plates. Plates were gently mixed on an orbital

shaker for 1 hour at room temperature, to allow nAb to bind to bacterial proteins. ELISA plates were then washed five times, and 100 μl rabbit anti-sheep IgM antibody coupled to

HRP (Bio-Rad Cat# AHP950P), diluted 1:5000 in PBS, was added. Following a one-hour incubation, ELISA plates were washed seven times and 50 μl of TMB substrate

(Amresco) was added to each well. After ~ 10 minutes, the peroxidase reaction was stopped by adding 100 μl 0.18M sulfuric acid. The wash buffer for all ELISAs was PBS containing 0.05% Tween-20. Absorbance measurements at 450 nm for all ELISA analysis were obtained with an Epoch plate reader (Biotek).

Bacterial killing assay (BKA) and growth analysis

Lyophilized ATCC 8739 E. coli (Microbiologics) cells were re-suspended at 1 pellet/10 ml PBS and diluted 10-fold in PBS to an approximate concentration of

105 cells/ml. Plate-grown V. coralliilyticus was inoculated using the BBL Prompt System

(BD) and diluting the initial kit-included inoculation tube of sterile saline 10-fold in PBS.

Bighorn sheep plasma samples were diluted 1:10 in PBS and 40 μl of plasma was mixed with 10 μl of diluted E. coli or μl diluted V. coralliilyticus in sterile 96-well plates and incubated on an orbital shaker (200 RPM) for 30 minutes at 37 °C. In a subset of the V. coralliilyticus experiments, diluted plasma was heat-inactivated by incubation at 56 °C for 30 minutes prior to exposure with bacteria. At the end of the incubation, 100 μl of

Tryptic Soy Broth was added to each well containing E. coli or 100 μl of 3% NaCl LB broth was added to wells containing V. coralliilyticus and the absorbance at 600 nm recorded. The plates were incubated at 37 °C until the conclusion of the experiment.

Starting at five hours after the start of the incubation period, the absorbance at 600 nm was recorded hourly. The experiment was concluded after 14 hours. Bacteria incubated

without plasma served as a control. Absorbance was plotted as a function of time. Growth curves had a predicted lag time, followed by a linear increase in absorbance, most often followed by a plateau. Data points within the linear range of growth were used to plot a linear regression for each individual; all individuals had at least 3 data points within the growth curve. The time to 0.3 absorbance (the approximate middle of the growth curve) was calculated from the linear regression using the regression formula for each individual and the values were natural log-transformed. In some plasma samples, bacterial growth was not detected at all over the course of the 14-hour incubation. Such samples were assigned a value of 15 hours for statistical analysis. For the purposes of graphical representation, animals were classified as “absolute killing”, “intermediate killing” or “no killing”. Plasma samples which did not inhibit bacterial growth, as defined by having a time to 0.3 absorbance less than or equal to the average time calculated from four control growth curves, were classified as “no killing.” Plasma samples which delayed the growth of E. coli, as defined by having some growth, but a time to 0.3 absorbance greater than the control samples, were labeled “intermediate killing.” Samples that never exhibited growth were classified as “absolute killing”.

Data analysis

All analyses were performed in Graphpad Prism, except for general linear mixed modeling, which was done using R version 3.4.3 (R Core Team, 2017) with the nlme package (Pinheiro et al. 2014). For ELISA results we used a non-parametric t-test with

Bonferroni correction to compare the ELISA result from an individual with maximal average staining (Sample 1) to an individual with minimal staining (Sample 2), anti-IgM alone and a negative control (BSA) (Bonferroni alpha<0.017). We then evaluated the

correlation between all samples for IgM and the V. coralliilyticus reactive IgM using a linear regression. We used a linear regression to compare the V. coralliilyticus reactive

IgM and the BKA on an individual-level and to compare bacterial growth between E. coli and V. coralliilyticus. Finally, we evaluated if the V. coralliilyticus reactive IgM and the BKA differed between the Mojave and

Peninsular sheep samples using generalized linear mixed models (Gaussian distribution, log link) with the main effect of population and a random effect of animal sex. Animal sex was used as a random effect because not all populations contained even numbers of males and females. Separate models were performed for V. coralliilyticus reactive IgM

(dependent variable: natural log concentration) and the BKA (dependent variable: natural log of time to 0.3 absorbance, which is approximately mid-log growth).

Results

Desert bighorn sheep nAb bind to cell envelope proteins of V. coralliilyticus

Some bacterial proteins, such as ribosomal proteins, are well-conserved across the prokaryotic kingdom, and adaptive antibodies generated against such antigens may cross react with proteins from V. coralliilyticus. To minimize this possibility, cell envelope proteins enriched in outer membrane protein fractions were isolated from V. coralliilyticus. As expected, the proteome of the cell-surface fraction was distinct from the whole cell lysate (Figure 2.1A), and many proteins present in the total cell lysate were absent in the membrane fraction. In addition to avoiding cross-reactivity with conserved intracellular proteins, the use of the bacterial-surface antigens ensured that we measured nAbs that bind to epitopes relevant for elimination of the pathogen, either by

activating the classical complement system or enhancing opsonization by phagocytic cells.

To determine the extent of nAb binding, we utilized a commercially available rabbit anti-sheep IgM polyclonal antibody raised against domestic sheep IgM. When applied to bighorn plasma proteins separated by SDS-PAGE and blotted onto nitrocellulose, the anti-sheep IgM detected a band of approximately 75 kDa via western blot analysis, matching the predicted size of the bighorn sheep IgM heavy chain (Figure

2.1B) in four plasma samples. The heavy chain IgM band in bighorn sheep was the same size as domestic sheep (Figure 2.1B). Therefore, anti-sheep IgM polyclonal antibodies can be used to detect bighorn sheep IgM.

To determine if bighorn nAb could bind to bacterial proteins, we exposed V. coralliilyticus membrane proteins (separated by SDS-PAGE and blotted onto a nitrocellulose membrane) to bighorn sheep plasma, and subsequently detected bighorn

IgM bound to bacterial proteins with rabbit anti-sheep IgM antibody. Plasma IgM from each animal showed some conserved and some unique protein bands (Figure 2.1C). The blue arrows in Figure 2.1C show bands which are detected in almost every animal tested, with varying levels of intensity. Orange arrows indicate bands which are only apparent in some individuals and not others. These data indicate that 1.) bighorn sheep IgM molecules can recognize cell-surface proteins from bacteria which do not infect bighorn sheep and 2.) slight variations in IgM binding, evident as variation in intensity of signal at bands recognized by all individuals or unique bacterial proteins recognized, exist in different individual animals.

Quantification of desert bighorn sheep nAbs using ELISA

While immunoblotting provides visual confirmation of natural IgM binding to bacterial membrane proteins, the technique is notoriously difficult to quantify and not a cost-effective way to measure natural IgM levels in large populations of animals. An

ELISA was therefore developed which utilized V. coralliilyticus cell envelope proteins as an antigenic target for IgM. A reliable signal could be detected and distinguished from all negative controls (Figure 2.2A). Depicted in Figure 2.2A are levels of bound IgM from two different animals representing the maximal and minimal signal detected in our initial assay, which included samples from 16 different individual bighorn sheep. The ELISA signal detected was specific, as pre-incubation of bighorn sheep plasma with V. coralliilyticus cell envelope proteins prior to addition to the ELISA plate reduced signal by 93% (n = 6, standard deviation = 12%). Variation in the ELISA signal from different animals might be due to differing overall levels of IgM in the plasma samples, rather than reflecting levels of nAb that bind specifically to V. coralliilyticus surface proteins. We therefore directly compared the V. coralliilyticus reactive IgM to total IgM determined using a commercially available kit (Figure 2.2B). Linear regression suggested that 75% of the variance in V. coralliilyticus reactive IgM is unexplained by total plasma

IgM (r2 = 0.25, p<0.05). Therefore, it is unlikely that natural IgM levels are directly related to total IgM levels in the plasma.

Assessing ecological and functional relevance: nAb levels differ between bighorn sheep populations, and correlate with bactericidal capacity

V. coralliilyticus-reactive nAb was determined by ELISA in 93 individual desert bighorn sheep from populations in two geographically distinct regions. Natural antibody levels were higher in bighorn sheep sampled from California’s Peninsular Mountains

when compared to bighorn sheep sampled from the Mojave Desert mountain ranges

(Figure 2.3A, p<0.001), after accounting for sex as a random effect in a general linear mixed model.

Increased levels of nAb may enhance the ability of complement to kill bacteria, via the classical pathway. We therefore incubated plasma with E. coli bacteria and monitored bacterial growth over time. Some plasma samples were able to completely inhibit bacterial growth (Figure 2.3B Sample #1), while others delayed growth by several hours (Figure 2.3B Samples 2 and 3). There were also some samples which did not appreciably halt bacterial growth to any measurable extent (Figure 2.3B, Sample #4), at the dilutions used in this experiment. Heat-inactivation of plasma returned bacterial growth to control levels (Figure 2.3B), demonstrating that a heat-labile component, most likely complement proteins, is responsible for killing bacteria. We then compared the plasma killing of E. coli between sheep captured in the Peninsular and Mojave ranges.

Bactericidal activity of Peninsular plasma samples was higher than samples taken from the Mojave (Figure 2.3C, p<0.001), after accounting for sex as a random effect in a general linear mixed model. In fact, over 50% of the plasma samples (23 out of 43 animals) collected from the Mojave range did not kill E. coli cells to a measurable extent, while only one plasma sample completely prevented bacterial growth (absolute killing) and the remaining animals (19 out of 43) showed intermediate killing (Figure 2.3D). In sharp contrast, ~25% of DBH plasma samples from the Peninsular range completely prevented bacterial growth while another 49% showed intermediate killing of bacteria, and the remaining 26% of individuals showed no inhibition of bacterial growth (Figure

2.3D). Because bactericidal activity of plasma is mediated by complement, we would

hypothesize that plasma bacterial killing would be proportional to nAb levels, which could serve to activate the classical complement pathway. At the individual animal level, we found that the bactericidal capacity of bighorn plasma showed a positive correlation with the amount of V. coralliilyticus-reactive nAb (Figure 2.3E, P<0.05). Finally, we determined that the ability of plasma to kill E. coli cells was highly correlated to the ability of plasma to kill V. coralliilyticus cells (Figure 2.3F, P<0.05).

Discussion

The field of ecoimmunology relies on a limited number of immune assays to document variation in immunity within and among wild animals. Experiments conducted with laboratory animals can utilize many different techniques to study natural antibody biology, including but not limited to, tracking specific B cell populations, analyzing germ-free or transgenic animals, expanding clonal B cells, and experimentally infecting animals. In wild animals, however, these techniques are not possible to implement. Two methods have commonly been used to measure nAb levels in a variety of wild animal systems: assessing the ability of nAbs to agglutinate red blood cells or to bind to a specific foreign protein, such as KHL or ovalbumin. However, these methods may not specifically measure nAb activity against bacterial antigens. For agglutination, it is widely held that nAbs are responsible for binding to foreign blood groups, however, not all nAbs may recognize foreign blood groups to the same extent that they recognize pathogenic bacteria signatures. In a study from Baxendale et al. (2008), natural antibody clones bearing non-mutated yet distinct heavy chains were able to bind to pneumococcal capsular polysaccharides and ABO blood antigens with different affinities. Therefore, measuring erythrocyte agglutination may reflect the presence of natural antibodies with

specificity for blood group antigens as opposed to bacterial products. In the foreign protein methods, KLH and ovalbumin are both glycoproteins and nAbs reactive to particular carbohydrate structures may bind to such targets. A vast heterogeneity of glycans is present on KLH (Geyer et al. 2005), although ovalbumin tends to have fewer such structures (Harvey et al. 2000). However, without exact knowledge about which moieties are specifically recognized by nAbs, it is difficult to know the degree to which these glycans will be present on the foreign protein chosen for the assay. Furthermore, bacterial glycan structures are very diverse, using many monosaccharides with a variety of branch-structures (Herget et al. 2008), suggesting that bacterial-specific nAbs which react to specific polysaccharide epitopes may not react with KLH or ovalbumin.

Additionally, nAbs can also react to lipid moieties, such as phosphocholine (Briles et al.

1981), which will not be detected by measuring antibody interactions with glycoproteins.

We therefore chose to isolate bacterial cell envelopes, which would contain several components that could serve as potential targets for nAbs including bacterial- glycoproteins and lipids as the antigen for nAb-binding in our assays to more accurately assess nAb function in a wild animal.

Hunter et al. (2008) demonstrated that western blot could be a powerful tool for discriminating between natural and acquired antibodies by examining the pattern of antibody staining to Mycoplasma antigens in naïve and infected desert tortoises. The binding of “naïve” bighorn sheep IgM to a pathogen also displays a particular pattern

(Figure 2.1C) with several protein bands consistently appearing by western blot, though variations in intensity are apparent. Additionally, some animals appear to recognize unique bacterial proteins not recognized by other individuals. Whether these bands are

the result of adaptive IgM cross-reacting to a specific bacterial protein, or the presence of a particular natural IgM antibody present in certain individuals, cannot be determined. It is also important to note that not all bacterial proteins present react with natural antibodies. The most prominent band in the membrane proteome (Figure 2.1A, marked with an asterisk) did not react with IgM antibody, nor did a group of low molecular weight proteins which migrated in the 15–25 kDa range on the gel. These data demonstrate that nAb IgM binding has specificity for particular bacterial epitopes and is not simply binding to proteins non-specifically.

Western blot data provided valuable information about the consistency with which particular proteins are recognized, but this is an impractical (and expensive) method for researchers who need to test hundreds of samples. For large sample sizes important in ecological studies, ELISAs are far more effective. Our data demonstrate that

IgM antibodies can react to V. coralliilyticus cell envelopes bound to an ELISA plate.

Importantly, the ELISA signal for natural antibody binding is not entirely explained by the total amount of IgM, with 75% of the variance left unexplained after accounting for total IgM. This finding also supports the conclusion that natural antibody reactivity to bacterial proteins is specific.

The bactericidal activity of plasma is most-likely mediated by the complement proteins present in serum, as heat-inactivation of plasma removed the bactericidal components (Figure 2.3B), however activation of the complement cascade requires recognition of the bacteria by other proteins such as antibody (either IgM or IgG) or lectins to be bound to the surface of the bacteria. We therefore anticipated that bacterial killing would be related to natural antibody levels. As shown in Figure 2.3E, levels of

natural antibody and bacterial killing by plasma are correlated. Despite being correlated, the goodness of fit is relatively low (R2 = 0.06), which suggests that alterations in natural antibody levels can explain some, but not all of the variation in bactericidal activity between individuals. Other factors such as glycosylation of antibodies, adaptive IgG antibodies that react with bacterial antigens, or even variations with the numerous complement proteins also likely contribute to the observed bactericidal capability of plasma.

The data presented here indicate that nAb levels differ between populations of bighorn sheep, and that these differences may manifest in the ability of individuals to kill potentially infectious bacteria. Why these populations differ in the innate immune parameters we measured here is unknown but could be the subject of future study, especially if susceptibility to infectious agents differs between Peninsular and Mojave

DBH. One possible explanation could be the different environmental conditions in the bighorn sheep populations. Elevation, precipitation, and access to water are known to have impacts on bighorn sheep populations, and differ between the two regions assessed in this study (Epps et al. 2004). Differences in forage quality among mountain ranges could account for differences in innate immune responses, as antibody responses are known to be influenced by food intake (L. B. Martin 2nd et al. 2007). Moreover, the

Peninsular and Mojave metapopulations are genetically distinct (Buchalski et al. 2016), and have at times been considered separate subspecies (Ian McTaggart Cowan 1940;

Subspecies 1993). Mojave Desert populations exhibit varying degrees of habitat fragmentation and habitat conditions, both of which influence neutral and adaptive genetic diversity (Creech et al. 2016; Epps et al. 2006, 2005) which could alter

susceptibility of these populations to infectious disease. Thus, genetic factors could be responsible for the variation in nAb levels between bighorn sheep in the two regions.

Different disease exposures could also lead to alterations in nAb levels as IgM secreted from B1 B cells rapidly increases following exposure to bacteria (F. Martin, Oliver, and

Kearney 2001). The Peninsular and Mojave bighorn sheep populations are known to have different disease histories (Elliott et al. 1994); higher prevalence of diseases in the

Peninsular regions could predispose individuals there to produce more nAbs. Combining our approach to measuring relevant antibacterial natural antibody levels in wild animals with epidemiologic histories, monitoring, or natural experiments could clarify these hypotheses in this and other systems.

In conclusion, we developed an assay for measuring natural antibodies reactive to bacterial antigens that may be applicable to a variety of organisms. Our assay detects nAb with affinity for a broad range of cell surface proteins from a marine pathogen, Vibrio coralliilyticus, to which terrestrial animals are unlikely to have been exposed. We validated our assay by demonstrating that it captures prominent differences in nAb levels among bighorn sheep populations, and that this variation is associated with differences in actual innate immune function. For this iconic wildlife species, our assay thus provides an ecologically and functionally relevant measure of nAbs with affinity for bacterial epitopes. We hope that this novel assay for pathogen-specific nAb will serve as a useful tool in studying immunity in this, and other non-model species.

Acknowledgments

Bighorn sheep captures within the Mojave National Preserve were conducted under NPS research permit MOJA-2013-SCI-0030. We would like to thank Brian

Henrichs and Lindsay Hunter for assistance in fieldwork and sample collection and

Claudia Hase for providing V. coralliilyticus.

Figures

Figure 2.1 Desert bighorn sheep IgM binds to cell-surface proteins from the marine pathogen V. coralliilyticus.

Figure 2.2 ELISA reliably detects desert bighorn sheep IgM binding to V. coralliilyticus cell envelope proteins.

Figure 2.3 Natural IgM levels differ between geographically distinct desert bighorn sheep populations and correlate with bacteria killing capability of plasma.

Figure 2.1. Desert bighorn sheep IgM binds to cell-surface proteins from the marine pathogen V. coralliilyticus. A. Cell envelopes (CE) and bacterial whole cell lysates (WCL) were separated by SDS-PAGE and visualized by staining with commassie blue dye. Molecular weight ladders flank the cell lysates and the weight of the protein (kDa) is denoted. Several distinct bands which are enriched in the CE fraction are marked orange arrows while bands absent from the membrane but present in the WCL are marked with a blue arrow. B. Plasma from domestic sheep or four randomly selected desert bighorn sheep were resolved by SDS-PAGE and transferred onto nitrocellulose. Rabbit anti-sheep IgM polyclonal antibodies were added to the membrane and detected with an anti-rabbit polyclonal antibodies coupled to a fluorescent dye. C. V. coralliilyticus cell envelope proteins were resolved by SDS-PAGE and blotted into a nitrocellulose membrane. Desert bighorn sheep plasma was added to the membranes, followed by rabbit-anti sheep IgM and anti-rabbit polyclonal antibodies coupled to a fluorescent dye. The image marked with an asterisk is a negative control, where the nitrocellulose membrane was not exposed to plasma. The asterisk also denotes a background band which is detected by the secondary antibodies. Several bands present in all animals tested are marked with blue arrows, whereas orange arrows denote bands which are only detectable in a subset of animals.

Figure 2.2. ELISA reliably detects desert bighorn sheep IgM binding to V. coralliilyticus cell envelope proteins. A. An ELISA for detecting desert bighorn sheep IgM binding to the membrane fractions of V. coralliilyticus was conducted with 17 random DBH sheep plasma samples. Samples were analyzed in triplicate and the individual with maximal average staining (Sample 1) and the minimal staining (Sample 2) are plotted with standard error bars. The background staining of anti-sheep IgM alone and BSA control wells is also shown (*, P<0.016, **, P<0.001). B. Average ELISA readings for the 17 plasma samples are plotted as a function of the total IgM determined by a separate ELISA analysis. The regression line is plotted showing 95% confidence intervals (orange dotted line) and the slope is statistically different from zero (P<0.05).

Figure 2.3. Natural IgM levels differ between geographically distinct desert bighorn sheep populations and correlate with bacteria killing capability of plasma. A. Average V. coralliilyticus ELISA values for each animal were averaged together by population location B. The ability of four bighorn sheep samples to kill laboratory grown E. coli for each population is depicted. Bacterial growth was monitored over a 14- hour period following exposure to plasma and the time to mid-exponential growth was calculated and technical replicates averaged. Here the inverse of time to mid-exponential growth is plotted. Plasma was either left untreated (black bars) or heat-inactivated prior to incubation with bacteria (white bars). C. The ability of plasma to prevent bacterial

growth for each sample was determined and was then averaged for each population. D. Same as in (C) but each animal was classified as no-killing if bacterial growth was not delayed compared to untreated controls, intermediate killing if growth did occur, but was delayed compared to controls, or absolute killing if plasma prevented bacteria from growing at all. E. The natural log of the time to mid-exponential phase growth was compared to ELISA values for IgM binding to V. coralliilyticus and the regression line is plotted with 95% confidence levels (orange dotted lines). The slope of the regression line is statistically different from zero (P<0.05). F. Nine bighorn sheep plasma samples were tested for the ability to kill both E. coli and V. coralliilyticus bacteria and the time to mid- exponential growth of each bacterial culture was compared. The slope of the regression line is statistically different from zero (P<0.05).

CHAPTER 3

POPULATION CONNECTIVITY PROTECTS DESERT BIGHORN SHEEP (OVIS CANADENSIS NELSONI) FROM INFECTIOUS PNEUMONIA

Brian S. Dugovich, Brianna R. Beechler, Rachel S. Crowhurst, Ben Gonzalez, Brian P. Dolan, Clinton W. Epps**, and Anna E. Jolles**

Abstract

Habitat fragmentation is an important driver of biodiversity loss and can be remediated by increasing connectivity in a species’ metapopulation. However, increased connectivity may also increase the potential for infectious disease spread across vulnerable populations. Alternatively, connectivity increasing immunogenetic diversity may underlie disease resistance and protect hosts from adverse infectious challenges. In

2013, a novel strain of Mycoplasma ovipneumoniae invaded the metapopulation of desert bighorn sheep in the Mojave Desert, causing pneumonia. We tracked the spread of this epizootic across the metapopulation to investigate how variation in connectivity among populations influenced disease outcomes. Among exposed populations, we found that connectivity reduced disease risk and increased immunogenetic diversity at the population level. Immunophenotype predicted diversity of a potential pneumonia defense gene linked to interferon-gamma (BL4) and conferred protection from infection at the individual-level. Our findings suggest that the genetic benefits of population connectivity can outweigh the potential risk of infectious disease spread once spillover occurs and support management to increase connectivity in metapopulations.

Introduction

Two processes that are central to driving anthropogenic loss of biodiversity are habitat fragmentation and the accelerating emergence of infectious diseases (Daszak,

Cunningham, and Hyatt 2000). Fragmentation contributes to population declines and extinctions by eroding the genetic basis for adaptation to changing environments

(Keyghobadi 2007), whereas infectious diseases can reduce population resilience through their effects on individual fitness and population growth rates (McCallum 2008). Habitat

fragmentation can, in principle, be partly remediated through management decisions that increase connectivity among fragmented populations. However, wildlife management views connectivity as a double-edged sword with equally sharp sides (Hess 1994): connectivity increases the risk of infectious disease transmission among populations

(Hess 1996), but is essential to maintaining genetic diversity in fragmented populations

(Cushman et al. 2013); genetic diversity, in turn, may mediate the susceptibility or resistance to infectious diseases (McKnight et al. 2017). As such, the relative magnitude of effects on disease transmission versus disease susceptibility determines the net impact of population connectivity on infectious disease risk. Despite abundant theoretical models on the topic (Brearley et al. 2013; Plowright et al. 2011; Heard et al. 2015), the role of population connectivity in mediating or mitigating infectious disease risk remains controversial, as these costs and benefits have not been quantified in any natural wildlife host-parasite system.

Desert bighorn sheep (Ovis canadensis nelsoni, DBH) are uniquely adapted to the extreme environments they inhabit and are the only large herbivore in much of the

Mojave Desert in southeastern California (Wehausen 2005). In this region, DBH inhabit small island-like mountain ranges separated by desert flats that are largely avoided; each range functions as a relatively independent population (Bleich, Wehausen, and Holl

1990). Gene flow and thus movement between ranges is predicted by distance, anthropogenic barriers (e.g. highways), and topography (Epps et al. 2005; Epps et al.

2007). Current Mojave DBH populations are small (Abella et al. 2011) and genetic drift is high (Epps et al. 2005), which has worrying conservation implications if their fitness and resilience are compromised when challenged by infectious disease.

DBH are vulnerable to pneumonia epizootics transmitted from domestic sheep

(Foreyt and Jessup 1982). Mycoplasma ovipneumoniae is a key respiratory pathogen for

DBH (Besser et al. 2008, 2012, 2013). Elsewhere in North America, respiratory disease in bighorn sheep has resulted in high mortality in adults and lambs at initial infection, with high lamb mortality but little adult mortality in subsequent years (Cassirer and

Sinclair 2007; Wehausen, Kelley, and Ramey 2011; McClintock and White 2007). In extremely arid habitats, as in the Mojave Desert of California, domestic sheep are rarely present (Epps et al. 2004), limiting the exposure risk of wild DBH populations. However, in 2013 an epizootic of ovine pneumonia was detected in one population resulting in extensive DBH adult mortality (Dekelaita et al. unpublished data). This strain of

Mycoplasma ovipneumoniae then spread throughout the metapopulation, missing only a single, very isolated population, but further die-offs were not observed. We investigated effects of population connectivity on disease risk in the Mojave DBH metapopulation during this deadly and fast-moving outbreak of ovine pneumonia in 2013 – 14 (Figure

3.1). For the first time in a natural mammalian host metapopulation, we assessed links between connectivity, genetic diversity, immune function, and susceptibility to disease to understand to the balance of risks and benefits of population connectivity in the face of exposure to an invading infectious disease.

Materials and Methods

Capture and sample collection

113 unique DBH were captured by the California Department of Fish and

Wildlife in November of 2013 and October through November of 2014 (n = 72, n = 41, respectively) for disease testing and satellite collaring of 11 populations (Cady, Clipper,

Granite, Hackberry, Marble, Newberry, North Bristol, Old Dad, Orocopia, Soda, and

South Bristol; Figure 3.1). DBH were caught by helicopter net-gunning and then were blindfolded and hobbled before sampling. Rectal temperatures were taken before sample collection and animals were monitored throughout to prevent hyperthermia. In the field, blood samples were drawn from the jugular vein into sodium heparin, EDTA, and non- coated blood tubes. Blood samples were either processed immediately, as was the case for immunoassays, or stored in liquid nitrogen for later use. Animals were captured and processed under the National Park Service Institutional Animal and Use Committee

(ACUP #PWR_MOJA_Epps.Powers_DesertBighorn_2013.A3, 2013 – 2015).

Disease testing

The California Department of Fish and Wildlife collected blood and nasal swabs for Mycoplasma ovipneumoniae testing, and these samples were stored at -20 oC.

Samples were sent to the Washington Animal Disease Diagnostic Laboratory (Pullman,

WA). Nasal swabs were used for polymerase chain reaction (PCR) by detecting

Mycoplasma ovipneumoniae specific DNA sequences, while serum was used for detection of antibodies against Mycoplasma ovipneumoniae by competitive enzyme- linked immunosorbent assays. PCR positive animals were considered actively infected and PCR and/or serology positive were designated as previously exposed animals.

Indeterminate results were classified as negative. Four animals were missing either a

PCR or serology test and were excluded from analysis involving disease status.

Connectivity

A previous study of gene flow between these DBH populations was used to estimate connectivity by network theory using demographic weighted closeness (Creech

et al. 2014). This metric modeled the potential of patch recolonization from neighboring patches and was limited by female dispersal range. That study used DNA samples from

392 DBH from 26 populations to measure sharing of mitochondrial DNA and observations of radio-collared females moving between patches to calculate their maximum effective dispersal distance. Slope and barrier resistance models were utilized to calculate least-cost paths between patches in the metapopulation.

Genetic diversity

Expected heterozygosity and allelic richness estimations used 10 – 25 DBH samples per population (Cady – 20 , Clipper – 19, Granite – 17, Hackberry – 18, Marble

– 25, Newberry – 20, North Bristol – 21, Old Dad – 20, Orocopia – 10, South Bristol –

23, and South Soda – 15), including those sampled for disease diagnostics and additional samples collected non-invasively from fecal pellets, for 208 total samples. Eight microsatellite loci near known immune function genes (adaptive-linked, Nickerson

2014) and thirteen putatively neutral microsatellite loci (neutral-linked, Epps,

Crowhurst, and Nickerson 2018) were used (Appendix Table 3.1). At the population- level, adaptive-linked and neutral-linked allelic richness and heterozygosity were calculated. BL4 is an adaptive-linked gene near the locus for interferon gamma, which upregulates the adaptive immune response (Coltman et al. 2001). Heterozygosity of this gene was calculated at the individual level. Adaptive-linked heterozygosity was also calculated at the individual level.

Hematology

Total white blood cells were counted with a Leuko-TIC test kit (Medix #BLT-

4013) according to the manufacturer's protocol. Differential blood cell counts were

performed on blood smears. Absolute numbers of neutrophils and lymphocytes were used to measure the neutrophil to lymphocyte ratio (NLR). One animal had missing hematology results and was excluded from individual analysis.

Bacterial killing assay

For the bacterial killing assay, a previously established spectrometer method was employed (French and Neuman-Lee 2012). Frozen plasma was thawed in an ice bath and was diluted 1:10 with phosphate-buffered saline to a final volume of 150 µl, and 40 µl of each diluted sample was placed in duplicate on 96-well plates. An E. coli solution was created from freeze-dried commercial pellets (ATCC 8739) and diluted to 105 colony forming units/ml and 10 µl was added to the wells with plasma and four wells of 40 µl of

PBS for controls. A pair of blank wells at the end of the plate only contained 50 µl of

PBS. The plate was incubated on a shaker at 37 oC at 200RPM for 30 minutes. 125 µl of tryptic soy agar broth (Sigma-Aldrich #22092) was added to all of the wells and absorbance was measured at 300 nm. The plate was incubated on a shaker for 12 hours, and the absorbance was checked again at 12 hours. The bacterial killing proportion was calculated by:

(control absorbance – average experimental absorbance) / control absorbance

Five animals had missing frozen plasma samples and were excluded from individual- level analysis.

Lymphocyte proliferation assay

A lymphocyte proliferation assay (LPA) assessed the lymphocyte proliferative response to a mitogenic challenge, requiring fresh lymphocytes. The assay was therefore run in a mobile lab in the field. Lymphocytes were separated from the heparin blood

tubes using UNI-SEP lymphoseparation tubes (Novamed #U-05). 10 μl of this cell suspension was added to 90 μl of Trypan Blue (Sigma-Aldrich #T8154), and the viable lymphocytes were counted on a hemocytometer. After the concentration of lymphocytes per μl was determined, the solution was diluted to 10 million cells/ml using AIM-V complete media (Life Technologies #12055091). Lipopolysaccharide (LPS, Sigma-

Aldrich #L3012) was used as a mitogen for B cells. The samples were plated on a 96- well plate with 1 μg LPS mitogen and the appropriate controls. The plates were incubated at 37 oC in an incubator with increased carbon dioxide. This was accomplished in the field by placing a burning candle inside an incubator sealed with duct tape; the flame extinguished itself after converting some of the oxygen in the incubator to carbon dioxide. A modified version colorimetric LPA was used (Cory et al. 1991): At 72 hours,

20ul of Alamar Blue was added to the wells. At 96 hours, the LPS well absorbance was measured. To calculate absorbance, the optical density at 600 nm was subtracted from the optical density at 570 nm. Negative values were set to zero. Cell proliferation after mitogen stimulation was calculated as:

(control absorbance – average experimental absorbance) / control absorbance

32 animals had missing LPA results and were excluded from individual analysis.

Data analysis

At the population level, connectivity was expected to be driving previous disease exposure either by facilitating Mycoplasma ovipneumoniae spread or providing immunogenetic protection. Linear regression was used to examine this relationship between connectivity and Mycoplasma ovipneumoniae exposed animals in 10 populations comprising 107 DBH sampled during the outbreak and connectivity

generated from a greater metapopulation of 392 DBH. Exposed animals were PCR positive and/or seropositive. Due to isolation from the disease outbreak, the Newberry population was excluded from disease analyses. Also, at the population level, increased connectivity would be expected to increase genetic diversity. Linear regression was performed between connectivity and neutral-linked heterozygosity, neutral-linked allelic richness, adaptive-linked heterozygosity, and adaptive-linked allelic richness in 11 populations of DBH comprised of genetic data from 208 DBH in these populations.

Finally, would increasing genetic diversity decrease Mycoplasma ovipneumoniae exposure? Linear regression models between disease exposure and neutral-linked heterozygosity, neutral-linked allelic richness, adaptive-linked heterozygosity, and adaptive-linked allelic richness were created using 10 populations comprised of 107 DBH sampled in the field.

Variation in individual bighorn sheep immune responses would be expected to influence Mycoplasma ovipneumoniae susceptibility and resistance. However, capturing comprehensible information from the assorted opposing and overlapping immune responses during an infectious disease outbreak is challenging (Downs and Stewart

2014). Therefore, to summarize the disparate effectors of the immune response, an individual-level immunophenotype was generated by using principal component analysis

(PCA) on the neutrophil to lymphocyte ratio, plasma bacterial killing assay, and lymphocyte proliferation assay to generate two principal component axes (PC1 and PC2) describing the majority of the variation in each individual’s immune response. No transformation was performed since these values were inherently scaled 0 – 1. With

missing immunoassay samples or results (34 individuals), the immunophenotype PCA encompassed 79 individual DBH.

If immunogenetic diversity at the population level were to detect changes in

Mycoplasma ovipneumoniae exposure in the past, then an individual DBH’s immunophenotype summarizing the current immune response at the time of capture would be expected to be influencing active infection (Mycoplasma ovipneumoniae PCR status). To test this relationship of active disease and immune response, a general linear model with a logit link was used. The binomial response variable was the positive or negative PCR test result, and the explanatory variables were either PC1 or 2 and sex. If higher immunogenetic diversity does improve the immune response, an association with the individual immunophenotype should be detectable. To test for an association between immunophenotype and genetic diversity, multiple linear regression was used with PC1 or

2 as the response variable and BL4 heterozygosity or adaptive-linked heterozygosity, the

Mycoplasma ovipneumoniae PCR test result (to account for immune response changes with active infection), and sex as explanatory variables.

Linear regression, general linear modeling, and principal component analysis were implemented using R version 3.4.3 (R Core Team, 2017). The lme4 package was used for general linear mixed models (Bates et al. 2015).

Results and Discussion

Mycoplasma ovipneumoniae exposure was lower in well-connected populations

Initially, we examined the role of connectivity on Mycoplasma ovipneumoniae invasion into the metapopulation. Using serology and PCR to detect Mycoplasma

ovipneumoniae, we determined the prevalence of disease in each population, sampling

113 unique DBH from eleven populations. We also used previous work that quantified connectivity between populations (Creech et al. 2014). The Newberry population was completely isolated from the rest of the study area and was not exposed to Mycoplasma ovipneumoniae; thus, we excluded it from disease-related models but included it as a de- facto control group in other analyses. Within Mycoplasma ovipneumoniae exposed populations, we found that more connected populations had lower exposure (p < 0.001,

Figure 3.2, Table 3.1), suggesting that protective effects of connectivity via disease susceptibility may have outweighed more rapid transmission in well-connected populations, limiting Mycoplasma ovipneumoniae infection.

Well-connected populations had higher neutral and adaptive-linked genetic diversity

To investigate mechanisms that might underlie this putative protective effect of connectivity against Mycoplasma ovipneumoniae infection, we assessed links between connectivity, genetic diversity, immune function, and disease status in the sheep. To investigate associations of connectivity with genetic diversity, we used samples from 10-

25 DBH per population, including those sampled for disease diagnostics and additional samples collected non-invasively from fecal pellets, to estimate allelic richness and expected heterozygosity in each population. We included eight microsatellite loci near known immune function genes (adaptive-linked, Nickerson 2014), and thirteen neutral microsatellite loci (neutral-linked, Epps, Crowhurst, and Nickerson 2018). Strong correlations were found between connectivity and adaptive-linked heterozygosity (p =

0.005), neutral-linked allelic richness (p = 0.003, Figure 3.3A), and adaptive-linked

allelic richness (p < 0.001, Figure 3.3B, Table 3.1) similar to previous findings

(Nickerson 2014).

Adaptive-linked genetic diversity was associated with lower Mycoplasma ovipneumoniae prevalence in populations

Importantly, exposure to Mycoplasma ovipneumoniae was negatively correlated with adaptive-linked heterozygosity (p = 0.025) and adaptive-linked allelic richness (p =

0.018, Figure 3.3D), but not with measures of neutral genetic diversity (neutral-linked heterozygosity [p = 0.45] and neutral-linked allelic richness [p = 0.075, Figure 3.3C,

Table 3.1]), suggesting that it was specifically immunogenetic variation (as opposed to the more general effects of lower genetic drift) that drove differences in disease exposure among populations.

Individual immunophenotype represented immune response strength and the innate and adaptive immunity axis

To further investigate immunity variation, we measured the immune response in

DBH and asked whether immunogenetic diversity was associated with variation in immune function among individual sheep. On a spectrum from innate to adaptive immunity, the neutrophil to lymphocyte ratio, bacterial killing ability of plasma, and lymphocyte proliferation ability were used to study the DBH immune response. A higher neutrophil to lymphocyte ratio indicates a stronger innate immune response (Davis,

Maney, and Maerz 2008). Plasma in the bacterial killing assay has a mixture of innate and adaptive immune factors, such as complement being involved in innate immunity

(Liebl and Martin 2009) and natural antibodies (French and Neuman-Lee 2012) combining both innate and adaptive immune defenses (Panda and Ding 2015). Finally, the lymphocyte proliferation assay measured the adaptive immune response of B cells

(antibody producing cells) when stimulated with lipopolysaccharide (Lewis et al. 2013).

Using principal component analysis (PCA), these factors were collapsed into a composite measure of immunophenotype with the first principal component (PC1) accounting for

43.1% of variation and indicating the strength of the immune response on the x-axis

(driven mostly by the innate immune response) and the second principal component

(PC2) accounting for 32.7% of the variation and representing an innate to adaptive immune response on the y-axis (Figure 3.4). Interestingly, a PCA of immune measures in cheetahs and leopards shared similar principal component axes (Heinrich et al. 2017).

Infected individuals had upregulated innate immune responses, but stronger adaptive responses were associated with lower Mycoplasma ovipneumoniae prevalence

We tested for associations between the immune function PCA axes and individual heterozygosity across the adaptive-linked microsatellite loci included in the study.

Actively infected sheep showed elevated immune responses, as indicated by a positive correlation with PC1 (p = 0.005); however, active infection was negatively associated with PC2 (p = 0.045, Table 3.2). During active infection, a higher PC1 indicated that the innate immune response was upregulated, and this was driven mostly by the bacterial killing ability of blood as one might expect during a bacterial infection. Despite using plasma, this immunoassay still likely measures innate immunity through complement proteins (Dugovich et al. 2017). The negative association of PC2 with infection suggests that individuals with more robust adaptive immune responses (i.e. antibody-mediated defenses) may be less susceptible to Mycoplasma ovipneumoniae infection. Respiratory epithelial cells trigger adaptive immune responses by dendritic cells, B cells, and T cells

that can eliminate the pathogen at the site of invasion if innate defenses fail (Schleimer et al. 2007).

Individual variation in immune function was associated with immunogenetic variation at the BL4 locus

We also examined associations with heterozygosity at a specific adaptive-linked microsatellite (BL4) that was linked to the interferon gamma gene in Soay Sheep that upregulates adaptive immunity (Coltman et al. 2001). Thus, increased heterozygosity at the BL4 locus would be expected to decrease Mycoplasma ovipneumoniae pathogenesis susceptibility. Individual immunophenotypes showed correlation with the BL4 gene

(Table 3.2). BL4 heterozygosity was positively correlated with PC2 (p = 0.008), but not

PC1 (p = 0.849) signifying the importance of the BL4 gene to adaptive immunity and resistance to Mycoplasma ovipneumoniae infection. In a laboratory study, interferon gamma knock-out mice had increased susceptibility to bacterial respiratory pneumonia caused by Legionella pneumophila (Shinozawa et al. 2002). More pertinently, the heat shock protein 70 antigen from Mycoplasma ovipneumoniae was demonstrated in mice to induce increased interferon gamma production, and this antigen could be a Mycoplasma ovipneumoniae vaccine candidate in sheep (Jiang et al. 2016). Additionally, the heterozygosity of the BL4 locus in our study demonstrated differences in immune function between individuals. Interestingly, this locus showed a signal of being under positive selection in the Mojave DBH system (Clinton W. Epps, Crowhurst, and

Nickerson 2018) possibly demonstrating its ability to increase fitness in this system.

Adaptive heterozygosity as a whole was not found to be correlated with PC1 (p = 0.25) or

PC2 (p = 0.385, Table 3.2), which was not surprising due to these adaptive-linked loci

being a miscellany of loci near known immune genes and their competing actions likely nullified a detectable signal on our defined axes.

Implications

In this study system, adaptive-linked genes were more correlated with connectivity compared to neutral-linked genes, which suggests positive selection pressure for adaptive-linked genotypes. Indeed, genetic diversity has changed over two generations and was associated with new connections in the metapopulation (Clinton W.

Epps, Crowhurst, and Nickerson 2018). However, gene flow simulations in this system showed that adaptive-linked genes were slow to spread throughout this metapopulation, and adaptation to a new stressor would be likely far too slow to be effective (Creech et al.

2017). Therefore, standing genetic variation in adaptive-linked genes may be of primary importance as a first line of defense in limiting novel pathogen invasion into a metapopulation, and understanding long-term processes underlying the observed variation in genetic diversity among populations may be fundamental to predicting patterns of disease in wildlife.

Beyond the risk of spreading disease through physical contact, connectivity maintains genetic diversity in systems with strong genetic drift. If a metapopulation is already connected, the sharper edge of connectivity is maintaining genetic diversity and immunity through increased gene flow. For example, a fungal-plant system in Finland found more connected patches had lower pathogen colonization and higher pathogen extinction (Jousimo et al. 2014). Although an isolated population like the Newberry herd

(Figure 3.1) did not suffer from Mycoplasma ovipneumoniae infection during this epizootic, its dwindling genetic diversity has hypothetically left it less likely to resist

pathogen invasion and possible to suffer higher mortality, an infection debt. Currently, isolation has protected the population, and these DBH appear healthy from lack of circulating infectious disease, but if a spillover event were to happen in this area, the infection debt would come due and the results could be devastating.

This study examined disease invasion into a fragmented metapopulation, but the persistence of the pathogen would be a critical future study. Isolation was not sufficient to prevent disease invasion in all but the Newberry population, and susceptibility

(mediated by immune function and underlying immunogenetic variation) was lower in well-connected populations. However, for a pathogen like Mycoplasma ovipneumoniae, which tends to persist endemically in its host populations once it has invaded, the other relevant question is how does metapopulation connectivity structure affect persistence?

For example, the Orocopia population is not directly connected to the other populations and was not affected by the 2013 Mycoplasma ovipneumoniae outbreak but had a different strain detected by California Department of Fish and Wildlife in 2014 (personal communication). Despite this, the Orocopia population data still reflected the patterns found in the other populations. If the Orocopia strain was from an earlier Mycoplasma ovipneumoniae outbreak, persistent infections may abide by the same patterns in this system long after the initial outbreak.

While the results of this study are exciting for conservation management strategies for increasing connectivity, care must be taken for applying decisions elsewhere. While connectivity decreased Mycoplasma ovipneumoniae in our study system, negative effects of connectivity have been emphasized in the Salmon River canyon where a study of a Rocky Mountain bighorn sheep metapopulation implicated

rams at risk of spreading infectious pneumonia (Borg et al. 2017). In this Mojave system, these bighorn sheep are native to this habitat and connectivity drives the genetic framework of the metapopulation. Other bighorn metapopulations are the amalgamation of translocation efforts (Singer, Papouchis, and Symonds 2000) and different genetic structures are likely at work (Malaney et al. 2015). Even DBH populations in Nevada had lasting genetic signatures from translocations that started 50 years ago (Jahner et al.

2019). Our study of Mojave DBH have less livestock spillover risk than Sierra and Rocky

Mountain bighorn sheep that share the landscape with grazing domestic sheep (Cahn et al. 2011; Carpenter et al. 2014), and management decisions hinge on limiting contact on this interface and attempting to eliminate Mycoplasma ovipneumoniae seems unlikely at the metapopulation level (Cassirer et al. 2018). While it is too early to see the effects of persistent infection after our study system’s Mycoplasma ovipneumoniae epizootic, persistent infections in Hells Canyon have been documented to cause juvenile mortality and have compromised long-term population growth (Manlove et al. 2016) in Rocky

Mountain bighorn sheep. To further complicate matters, lack of cross-strain immunity and high spillover pressure in this system leads to new deadly Mycoplasma ovipneumoniae outbreaks among chronically infected bighorn sheep (Cassirer et al.

2017).

Conclusion

To our knowledge, this observational wildlife study is the first to link population- level connectivity, genetic diversity, and infectious disease and individual-level immunophenotype with genetic diversity and infectious disease. In a world with increasing habitat fragmentation rife with emerging infectious diseases, wildlife

management is becoming staggeringly complex. Here, we parsed out the effects of disease spread in a fragmented metapopulation to show the net benefit that connectivity gives to population health. Connectivity may be a double-edged sword, but the benefits cut sharper than the risks.

Figures and Tables

Figure 3.1 Map of desert bighorn sheep populations in the study area in the Mojave Desert.

Figure 3.2 Mycoplasma ovipneumoniae exposure was lower in well-connected populations of desert bighorn sheep in the Mojave Desert.

Figure 3.3 Well-connected populations of desert bighorn sheep in the Mojave Desert had higher neutral and adaptive-linked genetic diversity, and adaptive-linked genetic diversity was associated with lower Mycoplasma ovipneumoniae prevalence in populations.

Figure 3.4 Principal component analysis of individual immunophenotype represented immune response strength and the innate and adaptive immunity axis in desert bighorn sheep in the Mojave Desert.

Table 3.1 Population-level relationships of connectivity, Mycoplasma ovipneumoniae exposure, and genetic diversity in desert bighorn sheep in the Mojave Desert by linear regression.

Table 3.2 Infected individual desert bighorn sheep in the Mojave Desert had upregulated innate immune responses, but stronger adaptive responses were associated with lower active Mycoplasma ovipneumoniae prevalence; individual variation in immune function was associated with immunogenetic variation at the BL4 locus but not adaptive-linked heterozygosity.

Figure 3.1. Map of desert bighorn sheep populations in the study area in the Mojave Desert. These animals are mostly confined to these mountain ranges. Low elevation desert flats and highways act as barriers for movement. The Newberry population was isolated from the rest of the study area. The Orocopia population is not shown.

Figure 3.2. Mycoplasma ovipneumoniae exposure was lower in well-connected populations of desert bighorn sheep in the Mojave Desert. Animals were exposed if they were PCR positive and/or seropositive for Mycoplasma ovipneumoniae and sampling occurred 2013 – 2014. Due to isolation from the disease outbreak, the Newberry population was excluded from disease analyses. Connectivity measures were defined as demographic weighted closeness from another study in this metapopulation (Creech et al. 2014). Higher values on the x-axis indicate that population is more connected to other populations. Higher connectivity was strongly associated with decreased Mycoplasma ovipneumoniae exposure (β = -0.019, SE = 0.004, Adj. R2 = 0.74, p = 0.001).

Figure 3.3. Well-connected populations of desert bighorn sheep in the Mojave Desert had higher neutral and adaptive-linked genetic diversity, and adaptive- linked genetic diversity was associated with lower Mycoplasma ovipneumoniae prevalence in populations. Animals were exposed if they were PCR positive and/or seropositive for Mycoplasma ovipneumoniae and sampling occurred 2013 – 2014. Due to isolation from the disease outbreak, the Newberry population was excluded from disease analyses. Connectivity measures were defined as demographic weighted closeness from another study in this metapopulation (Creech et al. 2014). Higher values on the x-axis indicate that population is more connected to other populations. Connectivity formed strong linear relationships with both neutral-linked (A, β = 0.038, SE = 0.009, Adj. R2 = 0.61, p = 0.003) and adaptive-linked allelic richness (B, β = 0.039, SE = 0.007, Adj. R2 = 0.75, p < 0.001). Neutral-linked allelic richness was not significantly associated with

Mycoplasma ovipneumoniae exposure (C, β = -1.144, SE = 1.44, Adj. R2 = -0.04, p = 0.45), but adaptive-linked allelic richness appeared to have a protective effect (D, β = - 1.941, SE = 0.7, Adj. R2 = 0.42, p = 0.025). Connectivity had a stronger relationship with adaptive-linked rather neutral-linked genes indicating their importance in disease resistance.

Figure 3.4. Principal component analysis of individual immunophenotype represented immune response strength and the innate and adaptive immunity axis in desert bighorn sheep in the Mojave Desert. The neutrophil to lymphocyte ratio (NLR), plasma bacterial killing assay (BKA), and lymphocyte proliferation assay (LPA) were used to measure individual immunocompetence and sampling occurred 2013 – 2014. PC1 represented the strength of the immune response and was driven mainly by the BKA, while PC2 represented an axis of innate to adaptive immunity with adaptive immunity being a positive correlation and innate immunity being a negative correlation.

Table 3.1. Population-level relationships of connectivity, Mycoplasma ovipneumoniae exposure, and genetic diversity in desert bighorn sheep in the Mojave Desert by linear regression. Animals were exposed if they were PCR positive and/or seropositive for Mycoplasma ovipneumoniae and sampling occurred in 11 populations from 2013 – 2014. Due to isolation from the disease outbreak, the Newberry population was excluded from disease analyses.

Relationship β Std. Error R2 p-value Connectivity and disease

Disease exposure -0.019 0.004 0.74 0.00077*

Connectivity and genetic diversity

Neutral heterozygosity 0.0032 0.0015 0.26 0.062

Adaptive heterozygosity 0.007 0.0019 0.56 0.0049*

Neutral allelic richness 0.0376 0.0092 0.61 0.0028*

Adaptive allelic richness 0.0387 0.0068 0.75 0.00029*

Disease exposure and genetic diversity

Neutral heterozygosity -1.144 1.44 -0.04 0.450

Adaptive heterozygosity -1.941 0.70 0.42 0.025*

Neutral allelic richness -0.295 0.145 0.26 0.075

Adaptive allelic richness -0.488 0.165 0.46 0.018*

Table 3.2. Infected individual desert bighorn sheep in the Mojave Desert had upregulated innate immune responses, but stronger adaptive responses were associated with lower active Mycoplasma ovipneumoniae prevalence; individual variation in immune function was associated with immunogenetic variation at the BL4 locus but not adaptive-linked heterozygosity. Active disease indicates a PCR positive animal and sampling occurred in 11 populations from 2013 – 2014. Due to isolation from the disease outbreak, the Newberry population was excluded from disease analyses. The BL4 locus was linked to the interferon gamma gene, and adaptive-linked heterozygosity is the summation of eight different immune genes. General linear modeling was performed for active disease models, and linear regression was performed for immune-gene heterozygosity models.

Relationship β Std. Error Adj. R2 p-value Active disease and immunophenotype

Immune

response 0.759 0.268 0.13 0.0046* strength

Adaptive immune -0.58 0.289 0.07 0.0446* response

BL4 heterozygosity and immunophenotype

Immune

response -0.047 0.239 0.14 0.849 strength

Adaptive immune 0.552 0.203 0.12 0.0081 response

Adaptive-linked heterozygosity and immunophenotype

Immune

response -0.750 0.648 0.15 0.25 strength

Adaptive immune 0.49 0.56 0.04 0.385 response

CHAPTER 4

FOOT-AND-MOUTH DISEASE EPIDEMIOLOGY IN A HERD OF ITS NATURAL RESERVOIR HOST, AFRICAN BUFFALO (SYNCERUS CAFFER)

Brian Dugovich, Bryan Charleston, Francois Maree, Eva Perez-Martin, Katherine Scott, Brianna Beechler, and Anna Jolles

Abstract

Foot-and-mouth disease virus (FMDV) is endemic in African buffalo (Syncerus caffer) in Southern Africa and has spillover potential to cattle on the wildlife-livestock interface causing region-wide socioeconomic hardships. Newly infected calves and carrier buffalo are the likely host individuals to be involved in FMDV transmission, which may put cattle at risk. In Kruger National Park, South Africa, we aimed to determine when the carrier state occurred and which adult buffalo were more likely to be carriers. We captured a herd of 60 – 70 buffalo in a 900 hectare enclosure within the park every 2 – 3 months for 3 years. Using FMDV PCR, virus isolation, and serology assays, we identified the infection classes of each buffalo at each time point: maternally protected, susceptible, infected, carrier, and recovered. In adult buffalo, exposure to infected calves increased the carrier state. We found that adult carrier animals were more likely to have a lower body condition score and were younger and male. Analyzing reports of stray buffalo revealed a peak in the dry season. According to the demographic patterns we observed, thin and young male buffalo may pose the highest risk for FMDV transmission to livestock. Basic epidemiological information describing disease transmission on an individual-level can help manage spillover risk in the face of variable environments and shrinking conservation budgets.

Introduction

Foot-and-mouth disease virus (FMDV) is a highly contagious pathogen that affects cloven hoofed animals and is associated with decreased productivity in livestock and severe economic losses globally (Rich and Perry 2011). Miniscule amounts of virus are sufficient to cause infection in a naïve host (Gibson and Donaldson 1986) and this,

combined with the large quantities of virus excreted by affected animals and multiple routes of infection, allows for a fast rate of viral spread (Alexandersen et al. 2003). This capacity for explosive spread of foot-and-mouth disease (FMD) in cattle (R0 = 21; Ster,

Dodd, and Ferguson 2012) makes epidemics in naïve host populations difficult to control and exerts large economic costs for control and management (Knight-Jones and Rushton

2013).

African buffalo are the only known wildlife reservoir host for FMDV (Weaver et al. 2013) despite its ubiquity throughout the world. Serological surveys indicate that over

98% of buffalo in Kruger National Park (KNP), South Africa have been exposed to all three local serotypes (Southern African Territories [SAT] 1, 2 and 3) of FMDV by the time they are 2-years-old (Thomson et al. 1992). In KNP the majority of calves are born between January and April with a peak in January and February that coincides with the optimum period for high protein grass growth (Ryan, Knechtel, and Getz 2007). FMDV’s basic reproductive number (R0) has been estimated in buffalo as 15.8, 95% CI [4.1, 65.6]

(Jolles et al. unpublished data). Carrier animals likely drive FMDV persistence (Moonen and Schrijver 2000), at least for SAT 1 and SAT 3 (Jolles et al. unpublished data), by carrier-to-calf and then calf-to-calf transmission among a newly susceptible calf cohort of

African buffalo. Other mechanisms for persistence may include antigenic shift and fluctuations in immune protection (Scott et al. unpublished data, Grimaldo et al. unpublished data). The likely mechanism that allows for persistent carriers is that FMD virions are maintained on follicular dendritic cells (FDC) in lymphoid tissue following primary infection in cattle, sheep, and African buffalo (Juleff et al. 2008; Juleff et al.

2012). Virions held on the surface of FDCs can then infect lymphoid cells that come into

contact with the FDC network (El Shikh and Pitzalis 2012), and thus, active virus can recrudesce and transmit to other animals.

Persistence of FMDV in buffalo causing spillover to cattle on the livestock- wildlife interface is the main challenge in controlling FMD in Southern Africa (Arzt et al.

2011), and the long-term viability of conservation areas depends on mitigating conflicts with surrounding human populations (Cumming 2011). A border fence on the western boundary of KNP was first erected in 1958 to help keep cattle and wildlife separate and also helped prevent FMD outbreaks, but despite upgrades, it is permeable to elephants, flood damage, and increasingly, human tampering (Jori et al. 2009). Between 2000 –

2008, stray buffalo caused 5 of 6 cattle FMD outbreaks outside of the park (van

Schalkwyk et al. 2016). Farmers in the FMD control area on the KNP border view wildlife as an impediment to cattle farming and rank FMD as the most concerning disease

(Chaminuka, McCrindle, and Udo 2012). Furthermore, disease transmission has been one of the main obstacles in the formation of the Great Limpopo Transfrontier Park (Wolmer

2003).

In this longitudinal study of a herd of African buffalo, we aimed to characterize the temporal patterns of primary and persistent FMDV infection, and to identify individual characteristics of acutely and persistently infected animals, which drive transmission of FMDV among buffalo and spillover risk to livestock. We also used park records to determine when buffalo are more likely to leave the park. Our study is intended to provide essential information to help focus limited resources for disease prevention to periods during the year and specific animals that pose the greatest risk for

FMDV spillover from buffalo to livestock.

Materials and Methods

Buffalo capture

An established FMDV-positive herd of 60 – 70 buffalo in a 900 hectare enclosure in the Satara section of Kruger National Park was utilized in this study. The founding animals of the herd were captured in 2001 in the northern section of Kruger National

Park, where they had been exposed to all three local serotypes of FMDV (Southern

African Territories [SAT] 1, 2, and 3). A double fence around the enclosure excluded large predators and helped prevent outside disease transmission. Within, the buffalo herd was largely unmanaged and experienced natural breeding, births, and deaths, grazed on the vegetation present, and had access to water from natural water pans and man-made water troughs. During the dry season, supplemental grass and alfalfa hay were provided when food was scarce. The entire herd was captured every 2 – 3 months for three years; data presented here span the first 13 captures (February 2014 to August 2016). This study followed three calf cohorts (2013 – 14, n = 16; 2014 – 15, n = 14; and 2015 – 16, n = 4), and included a total of N individuals.

For each capture, a spring-loaded gate captured the herd in a fenced area surrounding a water source. The buffalo were then moved into a boma where they were sedated in groups of 4 – 10. During the wet season, some animals were darted from a helicopter and processed out in the enclosure proper. Buffalo were sedated with 7 – 10 ug etorphine hydrochloride and 0.04 – 0.07 mg azaperone per kilogram body weight.

Buffalo age was estimated using incisor length (Jolles 2007) and were body condition scored (BCS) as a proxy for nutritional status (Ezenwa, Jolles, and O’Brien 2009). The width between the outside curve of the horns was measured for adult buffalo (Ezenwa

2007), while the widest section of the boss (horn base) was also measured in adult males

(Gandy 2004). In calves, the horn length was determined by measuring from the base of the horn to the tip. In all males, the testes were manipulated into the scrotum, and then the circumference was measured around the widest section of the scrotum. Pregnancy was determined by rectal palpation (Carmichael 1977), and adult females were milked for lactation status (Jolles, Cooper, and Levin 2005). This animal work was approved by the institutional animal care and use committee at Oregon State University, ACUP #4478 and at Kruger National Park.

FMDV testing

Blood was collected from the jugular vein with an 18-gauge needle and transferred on ice for testing at the Animal Research Council – Onderstepoort Veterinary

Institute in South Africa. The samples were used to quantify FMDV-specific antibodies for each individual (SAT 1, SAT 2, and SAT 3) via a liquid-phase blocking enzyme- linked immunosorbent assay (Ferris 1990). Esophageal-pharyngeal live virus was quantified by collecting mucus and epithelial cell samples using the established probang sample method (Sutmolle and Gaggero 1965) and a palatine tonsillar swab method (Juleff et al. 2012). Samples were frozen for transport by plane to the Pirbright Institute in the

UK. Live FMDV was detected and quantified by virus isolation (VI; Reid et al. 2002) and serotyped by enzyme-linked immunosorbent assay (Ferris and Dawson 1988). Real- time reverse-transcription polymerase chain reaction (RT-qPCR, Kasanga et al. 2014) was used to detect the FMDV RNA-dependent RNA polymerase 3D(pol) gene. The PCR we used was not FMDV serotype-specific, so we did not differentiate between viral serotypes in PCR positive animals. Sequences from virus that we isolated from the herd

indicate that the majority of circulating virus was SAT 1, but antibodies to all three serotypes were present in the herd. FMDV infection classes were defined as maternally protected (PCR/VI: negative; Serology: positive), susceptible (PCR/VI: negative;

Serology: negative), infected (PCR/VI: positive; Serology: negative), recovered (PCR/VI: negative; Serology: positive), and carrier (PCR/VI: positive; Serology: positive). PCR positive results were much more common than VI-positives (28% PCR + were also VI

+); as such, for most of the carrier and infected animals, which are of primary interest in this study, we had no information on viral serotype. Maternally protected animals were distinguished from recovered animals by identifying when calves seroconverted to susceptible or infected status and designating FMDV seropositivity as maternally protected before this period. Serology and PCR data were available for the first 13 captures, resulting in a study period of 2.5 years. Eliminating missing samples and unsuccessful tests resulted in 666 data points describing FMDV infection status of 98 individual animals during up to 13 time points.

Data analysis

Model selection of general linear mixed effects models with a logit-link was used to determine the strongest relationships between infection classes and the following explanatory variables: age, sex, temporal patterns, BCS, horn measurements, scrotum circumference, and pregnancy and lactation statuses. Animal ID was used as a random effect in all models to avoid pseudoreplication due to the longitudinal study design.

Interaction terms frequently caused non-convergence of models due to limited sample size, so they were removed from analysis. Calves were classified as younger than 2- years-old since they should be positive to all three SAT serotypes by then (Thomson et al.

1992). Three temporal models were run: “real” age, season, and capture. Age at first capture was used to avoid temporal autocorrelation with time measures except in models where the real age was used with the absence of temporal factors. The wet and dry seasons were attributed to captures based on rainfall patterns in the park (SANParks

2019): wet season (captures in October, December, and March) and dry season (captures in June and August). Capture number was used to look for long-term trends over the course of the study. Due to high correlation with age, residuals generated from the entire study of horn width and length, boss width, and scrotal circumference as linear regressions on age (Ezenwa 2007) were used during model selection. To account for sexual maturity in female buffalo occurring from 4.5 to 5.5 years of age (Jolles 2007) and lactation occurring after gestation, model selection was performed on a subset of adult females 6 years of age and older. Parameters were centered and rescaled by subtracting the mean and dividing by two standard deviations using data from the entire study to help improve model accuracy (Gelman 2008). Models were backwards selected using the lowest Akaike information criteria (AIC) score. Due to small sample size, infected models were forward selected to avoid model non-convergence. Each model best fit set contained a real age, season, and capture model, and the best of these using the lowest

AIC was chosen for the final model. Model selection of general linear mixed models was performed for the whole herd (Table 4.2), calves (Table 4.3), and adults (Tables 4.4 – 5).

Appendix tables 4.1 – 7 contain the full selection of models. To test the idea that primary infection in calves might replenish virus reservoirs in carriers, we included the number of infected calves for the three prior captures as temporal explanatory variables in a model of carrier status as the response variable (Table 4.5). Currently infected calves were also

tested (Appendix Table 4.5). Similar to the general linear mixed models, linear mixed model selection was performed on Log(FMDV PCR copies + 1) of carrier adults as the response variable (Tables 4.4 – 5, Appendix Tables 4.4 – 5).

A database of stray buffalo reports was provided by Mpumalanga State

Veterinarian Louis van Schalkwyk. The stray buffalo log had data from 1998 – 2003 and included information from ranger reports including fence patrols and following up on cases from the public observing stray buffalo. Unconfirmed observations were excluded except in cases where a ranger saw fresh buffalo tracks. Sex, age, and disease information for the stray buffalo were sparse. Data were compiled and linear regression was used to find relationships between fence breaks and season, sex, and age group.

Linear regression, linear mixed modeling, and general linear mixed modeling were implemented using R version 3.4.3 (R Core Team, 2017). Rescaling was performed by the arm package (Gelman and Hill 2007). The lme4 package was used for general linear mixed models (Bates et al. 2015). The nlme package was used for linear mixed models (Pinheiro et al. 2014).

Results

FMDV infection status was assessed for 37 calves (163 data points) and 61 adults

(503 data points, Table 4.1). Some calves aged into adults during the course of the study.

The frequency of different FMDV infection classes in the population varied temporally

(seasonal and long-term variation) and by animal characteristics (age class, sex, body condition) and capture date (Appendix Figure 4.1).

Temporal Patterns of Infection

Overall, the force of infection decreased throughout the course of the study.

Infection was more commonly observed during earlier captures (p = 0.004, Table 4.2), and recovered animals were more prevalent later in the study (p = 0.006, Table 4.2).

Adult carriers did not significantly decrease throughout the study period (Appendix Table

4.3). Maternally derived antibodies were present in most calves during June (dry season) captures, when calves were between 3 – 6 months old; but by August (still dry season), many calves lost maternal protection and were observed as susceptible to FMDV.

Primary infections peaked during the December captures (wet season, Appendix Figure

4.2). Infected calves increased the prevalence of carriers at the subsequent capture time point (p = 0.002, Figure 4.1, Table 4.5), resulting in a peak in the frequency of carrier animals in October (wet season, p = 0.04, Table 4.2, Appendix Figure 4.3). Thus, loss of maternally derived immunity drove the timing of primary infections in our study herd; and exposure to infectious calves replenished viral reservoirs in carriers.

Body Condition Score

Low body condition score was consistently associated with the presence of

FMDV, both in calves with primary infection, and in carriers: Calves with a low BCS (p

= 0.02, Table 4.3) were more likely to be infected, and adults with a poor BCS (p =

0.002, Table 4.4) were more likely to be carriers. Poor BCS appeared to have a stronger signal and larger effect size (Figure 4.2B) in males (β = -1.249, p = 0.025, Appendix

Table 4.3) vs. females (β = -0.663, p = 0.058, Appendix Table 4.3). Adult carrier buffalo with a poor BCS also had higher tonsillar swab PCR copies (p < 0.001, Figure 4.2C,

Table 4.4).

Age

Overall, susceptible, infected, and carrier status were associated with young age

(p < 0.001, p = 0.004, and p = 0.02, respectively; Figure 4.3A, Table 4.2), and this result was robust across age groups and sexes. Moreover, within the carrier group, younger animals had higher tonsillar swab FMDV PCR copies (p = 0.005, Table 4.4). In contrast, recovered buffalo, including both males and females, tended to be older (p < 0.001, p =

0.047, p = 0.012, respectively; Table 4.4, Appendix Table 4.3). Taken together, these results suggest that FMDV infected younger animals are more likely to be involved in

FMDV transmission than older individuals.

Sex

Adult carriers were more likely to be male (p = 0.052, Figure 4.2B, Table 4.5), while recovered animals were more often female (p = 0.028, Table 4.2). Lactating females were less likely to be carriers (p = 0.047, Appendix Table 4.6).

Stray Buffalo

Stray buffalo events (n = 628) involved 2,288 animals. Demographic information was very incomplete, but 259 males, 163 females, and 13 female-calf pair stray events were recorded. The stray buffalo event data suggests that the likelihood of encountering buffalo outside the park was higher in the dry season (p < 0.001, Figure 4.4, Appendix

Table 4.7). Thus, buffalo were observed outside the park predominantly during the time of year when they were less likely to be infectious for foot-and-mouth disease.

Discussion

FMDV infection in this herd waned throughout the course of the study. No calves were detected as infected for the study’s last capture in August 2016 when infections

would be expected to be increasing. Infection and carrier classes in calves were higher in earlier captures. Furthermore, recovered females were more likely to be identified in later captures. This enclosed buffalo herd was artificially isolated from other buffalo herds. In a more natural setting, not only does dispersal of young adults to different herds occur

(Spaan et al. 2019), but large fusion-fission events would expose multiple herds to more transmission events (Cross, Lloyd-Smith, and Getz 2005). In addition, the smaller size of this herd may have predisposed FMDV to stochastically fade out. Indeed, between the first and second captures there was a spike in FMDV seroprevalence after the buffalo were first concentrated into the bomas suggesting that we may inadvertently have increased transmission in a herd that was operating under a low force of infection.

Despite this, FMDV transmission declined during our study, in part probably due to weak calf recruitment in the last year, and viral isolation data suggest that at least one of the serotypes, SAT 2, may have been extirpated from the herd during this period (Jolles et al. unpublished data).

Taken together, our data on FMDV dynamics in this buffalo herd suggest that

FMDV primary infection occurs annually when the year's new calves lose maternal immunity, 4 – 6 months after the birthing period (which in KNP peaks from December –

February). Primary infections beginning in August in calves re-expose the rest of the herd to FMDV, which leads to a peak in carrier adult buffalo (both calves and young adults) in

October (wet season). Viral carriage declines over 6 – 9 months in buffalo (Jolles et al. unpublished data), leading to a reduced prevalence of carriers in the herd between

February and August, and then infections in the new calf cohort once again boost the frequency of carriers in the herd and the titer of FMDV detected in tonsillar swab

samples. Older animals were less often observed to carry FMDV, suggesting that the efficacy of the immune response to FMDV increases after several rounds of exposure, concordant with findings from FMDV vaccine studies (Lazarus et al. 2018). Our findings imply that exposure of young adults, without stable immune protection against FMDV to infectious calves in the wet season is a primary driver of frequency and viral load of carriers in the herd. This dynamic may change when dispersing individuals or small groups of buffalo avoid being exposed to a high force of infection in a herd setting.

Interestingly, low body condition was consistently associated with FMDV infection: primary infections were detected more often in calves with poor body condition, carriers tended to have worse body condition than recovered animals, and the viral titers were higher in carriers with poor body condition. These patterns suggest that clearance of virus may be slower or more incomplete in animals in poor body condition.

Epithelial tissues are effectively defended by mucosal antibodies, but mucosal antibody titers can be suppressed during protein restriction (R. R. Watson et al. 1985), such as free-ranging buffalo during the dry season in association with decreased nutrition (Ryan,

Knechtel, and Getz 2007; Leader-Williams 1996). In addition, recovered animals had a higher body condition score suggesting better suppression of virus shedding or eventually full viral clearance and resistance to the annual calf infection cycle.

The alternative explanation for low BCS in carrier animals is that FMDV infection may be energetically costly in buffalo. However, carriers maintain FMDV passively on follicular dendritic cells in the palatine tonsils (Juleff 2012), do not mount increased immune responses (antibody or inflammatory) compared to recovered animals

(Perez-Martin et al. unpublished data), and as such should not incur appreciable clinical

signs to inhibit food intake. It seems unlikely that this would result in detectable body condition losses during infection. This lack of a negative trade-off can be seen with horn length being longer in carrier calves indicating better growth despite carrying active virus. However, longer horn length could indicate that these calves were healthier and survived to the carrier stage where infirm calves did not survive the first few months of life.

Finally, while both sexes tended to be younger as carriers and older as recovered animals, adult male and female buffalo also had differences in their infection classes. A decreased BCS was more likely in carriers of both sexes, but this was more pronounced in males. Temporally, males were more often carriers in the wet season. Male buffalo bachelor herds rejoin the main herd to breed in the wet season which would expose them to newly infected calves. Mating is energetically costly due to increased time mating and fighting and decreased time eating (Turner, Jolles, and Owen-Smith 2005). This could explain why males were more likely to lose weight in the wet season compared to females. The study enclosure size was large enough to allow male buffalo to segregate themselves from the rest of the herd outside the breeding season to maintain this natural behavior. Alternatively, young adult male buffalo may have lost body condition at the end of the dry season, and their low rank in the herd restricted their access to food and they regained condition slower than females and older male buffalo who had the highest body condition in the herd. Also, lactation was negatively associated with carrier females.

Lactation requires adequate energy stores (Cook et al. 2013), and we found that high BCS is associated with being recovered likely due to a stronger immunity.

Matching heightened times of FMDV infectiousness to temporal patterns of stray buffalo to formulate a risk assessment of spillover is challenging. Carrier status in adults peaked in the wet season after FMDV re-exposure from infected calves, but stray buffalo were more likely during the dry season. Movement outside the park through a permeable fence is likely driven by a search for food during the dry season. However, buffalo are easier to see in the bushveld during the dry season due to less vegetation, so an observational bias cannot be discounted. A more thorough investigation of stray buffalo in southern KNP using the same stray buffalo data, and workshops with local stakeholders, found that although more stray buffalo were reported in the dry season, it was not different from the wet season (van Schalkwyk et al. 2016). This study also found that stray buffalo during the dry season remained closer to the fence compared to the wet season indicating that the fewer wet season strays made further excursions into farmland further complicating seasonal risk comparisons. Brahmbhatt et al. (2012) did not find any seasonal patterns between cattle and wildlife contact on the western border of KNP. A study in Northern KNP found that cattle FMD outbreaks caused by stray buffalo varied by season depending on site (Miguel et al. 2013). Buffalo contact with cattle peaked in the dry season when both species congregated around limited water sources.

Furthermore, rivers can damage fencing such as wet season flooding in 2000 – 2001 which allowed entire herds of buffalo to leave KNP (van Schalkwyk et al. 2016).

Conversely, during the dry season in northern KNP, falling water levels on the Limpopo

River permitted buffalo to cross the international border between South Africa and

Zimbabwe (Jori et al. 2016). Therefore, temporal patterns of buffalo movement outside of

conservation areas will likely depend on local characteristics of the wildlife-livestock interface, which could lead to numerous regional differences in a park as large as KNP.

The long-term health of KNP depends on mitigating the threat of FMDV spillover from buffalo to cattle. The precise mechanism of spillover is not well-understood (Maree et al. 2014). Buffalo shed virus in respiratory aerosols, nasal secretions, and saliva for longer than cattle (Bengis and Erasmus 1988). Male buffalo mating with female cattle has even been implicated as a possible route of transmission (Bastos et al. 1999), but evidence that this is a common transmission pathway is lacking. Stray buffalo were the highest perceived risk (92%) for FMD outbreaks by the communal livestock farmers in the FMD protection zone adjoining KNP (Lazarus et al. 2017). Stakeholder groups in

South Africa rank fencing as the highest preference for FMD control and vaccination as the lowest preference (Roberts and Fosgate 2018). However, a fencing regime brings its own drawbacks, including potential negative impacts on wildlife (Ferguson et al. 2013).

Nevertheless, reduction in vaccination and fencing increases the risk of a cattle FMD outbreak three-fold (Jori and Etter 2016). Even the creation of the Great Limpopo

Transfrontier Park is threatened as FMDV has been shown to circulate between buffalo and cattle across transboundary lines (Jori et al. 2016).

In conclusion, this observational study of a herd of African buffalo revealed that newly infected calves reinfect young adults with active virus primarily in the wet season.

Males and buffalo in poor body condition also increase the risk of carrier status, whereas older females rarely carry the virus. Temporal patterns of wildlife movement across conservation area boundaries may vary spatially, depending on local characteristics of the wildlife-livestock interface, and temporally, according to opportunities (fence breaks) and

motivation to leave (resource shortages within the conservation area). Basic epidemiological information on the timing of disease transmission and characteristics of animals most likely to be infectious can help manage spillover risk in the face of variable environments and shrinking conservation budgets.

Figures and Tables

Figure 4.1 Foot-and-mouth disease virus infection class prevalence in a herd of African buffalo over the course of the study by capture number.

Figure 4.2 A low body condition score in adult African buffalo is associated with carrier status that includes seasonal and sex patterns and increased numbers of foot-and-mouth disease virions recovered.

Figure 4.3 Age and sex influence foot-and-mouth disease virus infection class in a herd of African buffalo in Kruger National Park.

Figure 4.4 Stray African buffalo events in the vicinity of Kruger National Park (1998 – 2013).

Table 4.1 Summary of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park.

Table 4.2 Whole herd models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park.

Table 4.3 Calf models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park.

Table 4.4 Adult models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park.

Table 4.5 Adult reinfection models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park.

Figure 4.1. Foot-and-mouth disease virus infection class prevalence in a herd of African buffalo over the course of the study by capture number. This study in Kruger National Park took place from 2014 – 2016 with captures every 2 – 3 months. A. Initial calf FMDV infections are halted in the new calf cohort by their maternal antibodies. Adult carrier animals likely infect the new calf cohort. B. Adult carriers increase after initial foot-and-mouth disease virus outbreaks in calves, which suggests infected calves reinfect adults with the virus.

Figure 4.2. A low body condition score in adult African buffalo is associated with carrier status that includes seasonal and sex patterns and increased numbers of foot-and-mouth disease virions recovered. This study in Kruger National Park took place from 2014 – 2016 with captures every 2 – 3 months. Captures were not performed in March. The body condition score (BCS) was performed by manual palpation of the ribs, spine, hips, and tail base and scored on a scale of 1 – 5 for each category and averaged. A. In the adult buffalo models, low carrier BCS was more pronounced at the beginning of the wet season (β = -0.902, SE = 0.29, p = 0.002). B. In the adult male buffalo models, males had a lower body condition score if they were a carrier (β = -1.25, SE = 0.557, p = 0.025). C. In the adult reinfection buffalo models, foot-and-mouth disease virus copies were negatively correlated with BCS (β = -2.269, SE = 0.74, p = 0.003).

Figure 4.3. Age and sex influence foot-and-mouth disease virus infection class in a herd of African buffalo in Kruger National Park. This study took place from 2014 – 2016 with captures every 2 – 3 months. A. Susceptible (mean: 1.9 years, SE: 0.42 years) and infected (mean: 2.5 years, SE: 0.5 years) classes occur in younger animals. Carriers were young adults at an average of 4.9 years old (SE: 0.23 years). Recovered animals tended to be older adults (mean: 8.5 years, SE: 0.27 years). B. In the adult reinfection models, males were more likely to be carriers (β = -1.325, SE = 0.673, p = 0.049).

Figure 4.4. Stray African buffalo events in the vicinity of Kruger National Park (1998 – 2013). Stray buffalo events significantly peaked (p < 0.001, β = 25.5, SE = 8.0, R2 = 0.51) in the dry season (May – September) opposed to the wet season (October – April), but observational bias of more visible animals in the bare vegetation of the dry season cannot be discounted.

Table 4.1. Summary of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park. Longitudinal data (2014 – 2016) from 37 calves (163 data points) and 61 adults (503 data points) were used to generate the following 666 infection classes. In adult animals, there were 17 males (154 data points) and 44 females (349 data points). Infection Class Calves Adults Adult Males Adult Females Maternally protected 15 ------Susceptible 54 7 1 6 Infected 17 8 2 6 Carrier 46 207 89 118 Recovered 31 281 62 219

Table 4.2. Whole herd models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park. Longitudinal data (2014 – 2016) were used in generalized linear mixed models with logit links. The temporal dependent variable (none [real age], wet season, or capture) was the best model found during model selection using Akaike information criteria (see Appendix Table 4.1 for values). The estimates of dependent variables are displayed with standard error below in parentheses. Asterisks designate significance. Dependent variable: Maternal Susceptible Infected Carrier Recovered (1) (2) (3) (4) (5) Male 2.481 -0.466 0.741* -0.914** (1.727) (0.618) (0.438) (0.406) Real Age -77.996* -4.049*** (45.574) (0.819) 1st Capture Age -2.712*** -1.035** 3.043*** (0.949) (0.449) (0.446) BCS 4.206 0.034 -0.769 -0.619** 0.644** (2.567) (0.432) (0.470) (0.246) (0.263) Capture -0.227*** 0.102*** (0.078) (0.031) Wet Season 0.419** (0.207) Constant -50.822* -3.792*** -2.675*** -1.452*** -0.857** (29.339) (0.498) (0.568) (0.288) (0.337) Observations 666 666 666 666 666 Log Likelihood -21.850 -149.843 -89.451 -364.695 -333.307 Note: *p<0.1; **p<0.05; ***p<0.01

Table 4.3. Calf models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park. Longitudinal data (2014 – 2016) were used in generalized linear mixed models with logit links. The temporal dependent variable (none [real age], wet season, or capture) was the best model found during model selection using Akaike information criteria (see Appendix Table 4.2 for values). The estimates of dependent variables are displayed with standard error below in parentheses. Asterisks designate significance. Dependent variable: Maternal Susceptible Infected Carrier Recovered (1) (2) (3) (4) (5) Male 2.481 -0.660 (1.727) (1.024) Real Age -77.994* 38.400* (45.607) (23.317) BCS 4.206 -1.643** 0.318 (2.567) (0.714) (0.738) 1st Capture Age -26.373 (17.851) Horn Length 0.861 0.529 1.986** -2.394 (0.629) (0.799) (1.008) (2.283) Wet Season -1.510*** (0.558) Capture -0.520*** (0.166) Constant -50.821* -10.992 -2.961*** 1.226 19.740 (29.360) (7.341) (0.682) (0.947) (12.965) Observations 163 163 163 163 163 Log Likelihood -21.850 -67.933 -33.611 -51.952 -48.776 Note: *p<0.1; **p<0.05; ***p<0.01

Table 4.4. Adult models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park. Longitudinal data (2014 – 2016) were used in generalized linear mixed models with logit links. The temporal dependent variable (none [real age], wet season, or capture) was the best model found during model selection using Akaike information criteria (see Appendix Table 4.3 for values). The estimates of dependent variables are displayed with standard error below in parentheses. Asterisks designate significance. Dependent variable: Carrier Log (Swab PCR Carrier Recovered Copies) generalized linear generalized linear linear mixed effects mixed effects mixed effects (1) (2) (3) Real Age -3.943*** (1.377) Male 1.138* -0.897 -0.645 (0.586) (0.575) (1.028) 1st Capture Age -2.656*** 2.662*** (0.687) (0.670) BCS -0.903*** 0.791*** -2.822*** (0.290) (0.300) (0.556) Horn Width 0.025 -0.140 1.919* (0.604) (0.594) (1.070) Wet Season 0.445* (0.244) Capture 0.102*** (0.037) Constant -0.707** -0.630 3.854*** (0.342) (0.452) (0.588) Observations 503 503 183 Log Likelihood -254.710 -259.613 -502.927 Note: *p<0.1; **p<0.05; ***p<0.01

Table 4.5. Adult reinfection models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park. Longitudinal data (2014 – 2016) were used in generalized linear mixed models with logit links for carriers and linear mixed effect models for tonsillar swab viral PCR copies in carriers. The estimates of dependent variables are displayed with standard error below in parentheses. Asterisks designate significance. Dependent variable: Carrier Log (Swab PCR Adult Carriers Copies) generalized linear linear mixed effects mixed effects (1) (2) # calves 1 capture prior 0.817** 1.573** (0.333) (0.608) # calves 2 captures prior 1.124*** 0.378 (0.363) (0.704) # calves 3 captures prior -1.246*** -0.855 (0.338) (0.711) Male 1.325** (0.673) 1st Capture Age -3.610*** -2.940** (0.791) (1.471) BCS -0.832** -2.245*** (0.381) (0.737) Horn Width 1.677* (1.004) Constant -0.555 3.368*** (0.401) (0.495) Observations 503 161 Log Likelihood -187.804 -435.210 Note: *p<0.1; **p<0.05; ***p<0.01

CONCLUSION

The dawning of the Anthropocene brings incredible conservation challenges to maintaining biodiversity as humans are not only causing mass extinction and biological annihilation, but the evolutionary trajectories of the remaining life have been altered to adapt to a less wild world (Otto 2018). Daszak et al. (2000) synthesized the risks posed by emerging infectious diseases in wildlife but little has been done policy-wise to mitigate these threats (Cunningham, Daszak, and Wood 2017). Wildlife-livestock interface diseases are often complex systems that rely on reservoir hosts that cannot be eliminated, and understanding immune function and pathogen dynamics in wildlife is critical for effective disease management (Downs and Stewart 2014; Miller, Farnsworth, and Malmberg 2013).

Refining and improving the immunoassay toolkit has been a main focus of ecoimmunology (Downs and Stewart 2014). The challenge lies in refining methods that are employable in the field and give enough information to try to make sense of what is often only a snapshot look at ecoimmunological patterns in wildlife. At least for innate immunity, researchers can be confident that different methods of measuring the immune response yield similar results, so resources can be spent on other tests. In Chapter 1, we found positive correlations between measures of innate immunity: TLR2, TLR5, neutrophil to lymphocyte ratio, and a bacterial killing assay. In Chapter 2, we further validated this notion by developing an assay that detects natural antibodies in desert bighorn sheep. This assay was both ecologically and functionally relevant. Peninsular desert bighorn sheep had a stronger innate immune response than the Mojave metapopulation. These results were then validated against a bacterial killing assay, which

100

again showed that different methods of measuring innate immunity are correlated.

Natural antibodies hold promise for future ecoimmunological studies to measure the innate immune response in a variety of wildlife.

Combining landscape genetics and ecoimmunology will provide new insights for infectious disease investigation and management on the wildlife-livestock interface

(White, Forester, and Craft 2018). In Chapter 3, the Mycoplasma ovipneumoniae outbreak occurring at the same time we launched the desert bighorn sheep ecoimmunology project was a rare chance to combine disease ecology and ecoimmunology. Measures of connectivity, genetic diversity, pathogen exposure, and immunophenotype were combined to form a compelling picture of connectivity’s protective effect on deadly infectious disease spillover in the Mojave system. Increasing connectivity by providing wildlife crossings could lead to improved health of the metapopulation. Ongoing work in the system is assessing lamb survival post-outbreak. A longitudinal study with recaptures would offer a more complete understanding of

Mycoplasma ovipneumoniae’s persistence in populations and long-term effects.

Most wildlife disease research is descriptive and lacks understanding of pathogen dynamics and risk factor identification (Gortázar, Ruiz-Fons, and Höfle 2016). In

Chapter 4, an observational study of an African buffalo herd identified temporal and demographic patterns of FMDV dynamics. Newly infected calves shedding FMDV during the wet season reinfect young adult buffalo causing them to become carriers.

These carriers are then the likely mechanism to infect the new calf cohort. Risk factors for carrier status were identified. Having a poor body condition score, younger age, and being male increased the chances of being a carrier buffalo. Although more stray buffalo

101

were recorded in the dry season from park records, overall temporal patterns of stray buffalo are often equivocal and may depend on local characteristics of the wildlife- livestock interface. While this study found exciting relationships in FMDV dynamics in

African buffalo, the infection class categorization was a rough approximation of FMDV disease ecology and actual transmission events were not described. Ongoing research in the buffalo study by my colleagues will use viral genetics to pinpoint the transmission and spread of the virus throughout the herd.

102

BIBLIOGRAPHY

Abella, Regina K., Vernon C. Bleich, Randall A. Botta, Ben J. Gonzales, Thomas R. Stephenson, Steven G. Torres, and John D. Wehausen. 2011. “Status of Bighorn Sheep in California22010.” Desert Bighorn Council Transactions 51 (54): 68. Alexandersen, S., Z. Zhang, A. I. Donaldson, and A. J. M. Garland. 2003. “The Pathogenesis and Diagnosis of Foot-and-Mouth Disease.” Journal of Comparative Pathology 129 (1): 1–36.

Arzt, J., B. Baxt, M. J. Grubman, T. Jackson, N. Juleff, J. Rhyan, E. Rieder, R. Waters, and L. L. Rodriguez. 2011. “The Pathogenesis of Foot-and-Mouth Disease II: Viral Pathways in Swine, Small Ruminants, and Wildlife; Myotropism, Chronic Syndromes, and Molecular Virus-Host Interactions.” Transboundary and Emerging Diseases 58 (4): 305–26.

Barnosky, A. D., N. Matzke, S. Tomiya, G. O. U. Wogan, B. Swartz, T. B. Quental, C. Marshall, et al. 2011. “Has the Earth’s Sixth Mass Extinction Already Arrived?” Nature 471 (7336): 51–57.

Bastos, A. D., H. J. Bertschinger, C. Cordel, C. D. van Vuuren, D. Keet, R. G. Bengis, D. G. Grobler, and G. R. Thomson. 1999. “Possibility of Sexual Transmission of Foot- and-Mouth Disease from African Buffalo to Cattle.” The Veterinary Record 145 (3): 77–79.

Bates, Douglas, Martin Mächler, Ben Bolker, and Steve Walker. 2015. “Fitting Linear Mixed-Effects Models Using lme4.” Journal of Statistical Software, Articles 67 (1): 1–48.

Baumgarth, Nicole. 2011. “The Double Life of a B-1 Cell: Self-Reactivity Selects for Protective Effector Functions.” Nature Reviews. Immunology 11 (1): 34–46.

Baumgarth, Nicole, James W. Tung, and Leonore A. Herzenberg. 2005. “Inherent Specificities in Natural Antibodies: A Key to Immune Defense against Pathogen Invasion.” Springer Seminars in Immunopathology 26 (4): 347–62.

Baxendale, H. E., M. Johnson, R. C. M. Stephens, J. Yuste, N. Klein, J. S. Brown, and D. Goldblatt. 2008. “Natural Human Antibodies to Pneumococcus Have Distinctive Molecular Characteristics and Protect against Pneumococcal Disease.” Clinical and Experimental Immunology 151 (1): 51–60.

Beechler, B. R., H. Broughton, A. Bell, V. O. Ezenwa, and A. E. Jolles. 2012. “Innate Immunity in Free-Ranging African Buffalo (Syncerus Caffer): Associations with Parasite Infection and White Blood Cell Counts.” Physiological and Biochemical Zoology: PBZ 85 (3): 255–64.

103

Bengis, R. G., and J. M. Erasmus. 1988. “Wildlife Diseases in South Africa: A Review.” Revue Scienfifique et Technique de l’OIE 7 (4): 807–21.

Bengis, R. G., G. R. Thomson, R. S. Hedger, V. Devos, and A. Pini. 1986. “Foot-and- mouth-disease and African buffalo (Syncerus caffer). 1. Carriers as a source of infection for cattle.” The Onderstepoort Journal of Veterinary Research 53 (2): 69– 73.

Ben-Haim, Y., F. L. Thompson, C. C. Thompson, M. C. Cnockaert, B. Hoste, J. Swings, and E. Rosenberg. 2003. “Vibrio Coralliilyticus Sp. Nov., a Temperature-Dependent Pathogen of the Coral Pocillopora Damicornis.” International Journal of Systematic and Evolutionary Microbiology 53 (Pt 1): 309–15.

Bernardino, A. L. F., T. A. Myers, X. Alvarez, A. Hasegawa, and M. T. Philipp. 2008. “Toll-like Receptors: Insights into Their Possible Role in the Pathogenesis of Lyme Neuroborreliosis.” Infection and Immunity 76 (10): 4385–95.

Besser, Thomas E., E. Frances Cassirer, Margaret A. Highland, Peregrine Wolff, Anne Justice-Allen, Kristin Mansfield, Margaret A. Davis, and William Foreyt. 2013. “Bighorn Sheep Pneumonia: Sorting out the Cause of a Polymicrobial Disease (vol 108, Pg 85, 2013).” Preventive Veterinary Medicine 109 (3-4): 185–185.

Besser, Thomas E., Margaret A. Highland, Katherine Baker, E. Frances Cassirer, Neil J. Anderson, Jennifer M. Ramsey, Kristin Mansfield, et al. 2012. “Causes of Pneumonia Epizootics among Bighorn Sheep, Western United States, 2008-2010.” Emerging Infectious Diseases 18 (3): 406–14.

Besser, Thomas E., E. Frances Cassirer, Catherine Yamada, Kathleen A. Potter, Caroline Herndon, William J. Foreyt, Donald P. Knowles, and Subramaniam Srikumaran. 2012. “Survival of Bighorn Sheep (Ovis Canadensis) Commingled with Domestic Sheep (Ovis Aries) in the Absence of Mycoplasma Ovipneumoniae.” Journal of Wildlife Diseases 48 (1): 168–72.

Besser, Thomas E., E. Frances Cassirer, Kathleen A. Potter, John VanderSchalie, Allison Fischer, Donald P. Knowles, David R. Herndon, Fred R. Rurangirwa, Glen C. Weiser, and Subramaniam Srikumaran. 2008. “Association of Mycoplasma Ovipneumoniae Infection with Population-Limiting Respiratory Disease in Free- Ranging Rocky Mountain Bighorn Sheep (Ovis Canadensis Canadensis).” Journal of Clinical Microbiology 46 (2): 423–30.

Bleich, V. C., J. D. Wehausen, and S. A. Holl. 1990. “Desert-dwelling mountain sheep – conservation implications of a naturally fragmented distribution.” Conservation Biology: The Journal of the Society for Conservation Biology 4 (4): 383–90.

Boes, M., A. P. Prodeus, T. Schmidt, M. C. Carroll, and J. Chen. 1998. “A Critical Role

104

of Natural Immunoglobulin M in Immediate Defense against Systemic Bacterial Infection.” The Journal of Experimental Medicine 188 (12): 2381–86.

Borg, Nathan J., Michael S. Mitchell, Paul M. Lukacs, Curt M. Mack, Lisette P. Waits, and Paul R. Krausman. 2017. “Behavioral Connectivity among Bighorn Sheep Suggests Potential for Disease Spread.” The Journal of Wildlife Management 81 (1): 38–45.

Brahmbhatt, Dipa P., Geoffrey T. Fosgate, Edwin Dyason, Christine M. Budke, Bruce Gummow, Ferran Jori, Michael P. Ward, and R. Srinivasan. 2012. “Contacts between Domestic Livestock and Wildlife at the Kruger National Park Interface of the Republic of South Africa.” Preventive Veterinary Medicine 103 (1): 16–21.

Brearley, Grant, Jonathan Rhodes, Adrian Bradley, Greg Baxter, Leonie Seabrook, Daniel Lunney, Yan Liu, and Clive McAlpine. 2013. “Wildlife Disease Prevalence in Human-Modified Landscapes.” Biological Reviews of the Cambridge Philosophical Society 88 (2): 427–42.

Briles, D. E., M. Nahm, K. Schroer, J. Davie, P. Baker, J. Kearney, and R. Barletta. 1981. “Antiphosphocholine Antibodies Found in Normal Mouse Serum Are Protective against Intravenous Infection with Type 3 Streptococcus Pneumoniae.” The Journal of Experimental Medicine 153 (3): 694–705.

Brock, Patrick M., Courtney C. Murdock, and Lynn B. Martin. 2014. “The History of Ecoimmunology and Its Integration with Disease Ecology.” Integrative and Comparative Biology 54 (3): 353–62.

Brubaker, S. W., K. S. Bonham, I. Zanoni, and J. C. Kagan. 2015. “Innate Immune Pattern Recognition: A Cell Biological Perspective.” Annual Review of Immunology Vol 33 33: 257–90.

Buchalski, Michael R., Benjamin N. Sacks, Daphne A. Gille, Maria Cecilia T. Penedo, Holly B. Ernest, Scott A. Morrison, and Walter M. Boyce. 2016. “Phylogeographic and Population Genetic Structure of Bighorn Sheep ( Ovis canadensis ) in North American Deserts.” Journal of Mammalogy 97 (3): 823–38.

Budzko, D. B., D. F. Jelinek, and B. W. Wilcke Jr. 1981. “Nonspecific Reactions in the Hemagglutination Inhibition Test for Detection of Rubella Antibodies.” Journal of Clinical Microbiology 13 (5): 818–23.

Cabral, E. S., H. Gelderblom, R. L. Hornung, P. J. Munson, R. Martin, and A. R. Marques. 2006. “Borrelia Burgdorferi Lipoprotein-Mediated TLR2 Stimulation Causes the down-Regulation of TLR5 in Human Monocytes.” The Journal of Infectious Diseases 193 (6): 849–59.

105

Cahn, Maya L., Mary M. Conner, Oswald J. Schmitz, Thomas R. Stephenson, John D. Wehausen, and Heather E. Johnson. 2011. “Disease, Population Viability, and Recovery of Endangered Sierra Nevada Bighorn Sheep.” The Journal of Wildlife Management 75 (8): 1753–66.

Carpenter, Tim E., Victor L. Coggins, Clinton McCarthy, Chans S. O’Brien, Joshua M. O’Brien, and Timothy J. Schommer. 2014. “A Spatial Risk Assessment of Bighorn Sheep Extirpation by Grazing Domestic Sheep on Public Lands.” Preventive Veterinary Medicine 114 (1): 3–10.

Cases-Diaz, E., I. Marco, J. R. Lopez-Olvera, G. Mentaberre, E. Serrano, and S. Lavin. 2012. “Effect of Acepromazine and Haloperidol in Male Iberian Ibex (Capra Pyrenaica) Captured by Box-Trap.” Journal of Wildlife Diseases 48 (3): 763–67.

Cassirer, E. Frances, Kezia R. Manlove, Emily S. Almberg, Pauline L. Kamath, Mike Cox, Peregrine Wolff, Annette Roug, et al. 2018. “Pneumonia in Bighorn Sheep: Risk and Resilience.” The Journal of Wildlife Management 82 (1): 32–45.

Cassirer, E. Frances, Kezia R. Manlove, Raina K. Plowright, and Thomas E. Besser. 2017. “Evidence for Strain-Specific Immunity to Pneumonia in Bighorn Sheep.” The Journal of Wildlife Management 81 (1): 133–43.

Cassirer, E. F., and A. R. E. Sinclair. 2007. “Dynamics of Pneumonia in a Bighorn Sheep Metapopulation.” The Journal of Wildlife Management 71 (4): 1080–88.

Ceballos, Gerardo, Paul R. Ehrlich, Anthony D. Barnosky, Andrés García, Robert M. Pringle, and Todd M. Palmer. 2015. “Accelerated Modern Human-Induced Species Losses: Entering the Sixth Mass Extinction.” Science Advances 1 (5): e1400253.

Ceballos, Gerardo, Paul R. Ehrlich, and Rodolfo Dirzo. 2017. “Biological Annihilation via the Ongoing Sixth Mass Extinction Signaled by Vertebrate Population Losses and Declines.” Proceedings of the National Academy of Sciences of the United States of America 114 (30): E6089–96.

Chaminuka, Petronella, Cheryl M. E. McCrindle, and Henk M. J. Udo. 2012. “Cattle Farming at the Wildlife/Livestock Interface: Assessment of Costs and Benefits Adjacent to Kruger National Park, South Africa.” Society & Natural Resources 25 (1-3): 235–50.

Chen, Yifang, Yong-Beom Park, Ekta Patel, and Gregg J. Silverman. 2009. “IgM Antibodies to Apoptosis-Associated Determinants Recruit C1q and Enhance Dendritic Cell Phagocytosis of Apoptotic Cells.” Journal of Immunology 182 (10): 6031–43.

Choi, Youn Soo, and Nicole Baumgarth. 2008. “Dual Role for B-1a Cells in Immunity to Influenza Virus Infection.” The Journal of Experimental Medicine 205 (13): 3053–

106

64.

Coltman, D. W., K. Wilson, J. G. Pilkington, M. J. Stear, and J. M. Pemberton. 2001. “A Microsatellite Polymorphism in the Gamma Interferon Gene Is Associated with Resistance to Gastrointestinal Nematodes in a Naturally-Parasitized Population of Soay Sheep.” Parasitology 122 (Pt 5): 571–82.

Condy, J. B., and R. S. Hedger. 1974. “Survival of foot-and-mouth-disease virus in African buffalo with nontransference of infection to domestic cattle.” Research in Veterinary Science 16 (2): 182–85.

Cook, Rachel C., John G. Cook, David J. Vales, Bruce K. Johnson, Scott M. Mccorquodale, Lisa A. Shipley, Robert A. Riggs, et al. 2013. “Regional and Seasonal Patterns of Nutritional Condition and Reproduction in Elk.” Wildlife Monographs 184 (1). https://doi.org/10.1002/wmon.1008.

Cory, A. H., T. C. Owen, J. A. Barltrop, and J. G. Cory. 1991. “Use of an aqueous soluble tetrazolium formazan assay for cell-growth assays in culture.” Cancer Communications 3 (7): 207–12.

Creech, Tyler G., Clinton W. Epps, Erin L. Landguth, John D. Wehausen, Rachel S. Crowhurst, Brandon Holton, and Ryan J. Monello. 2017. “Simulating the Spread of Selection-Driven Genotypes Using Landscape Resistance Models for Desert Bighorn Sheep.” PloS One 12 (5): e0176960.

Creech, Tyler G., Clinton W. Epps, Ryan J. Monello, and John D. Wehausen. 2014. “Using Network Theory to Prioritize Management in a Desert Bighorn Sheep Metapopulation.” Landscape Ecology 29 (4): 605–19.

Creech, Tyler G.; Clinton W. Epps; Ryan J. Monello; John D. Wehausen. 2016. “Predicting Diet Quality and Genetic Diversity of a Desert-Adapted Ungulate with NDVI.” Journal of Arid Environments 127 (April): 160–70.

Cross, P. C., J. O. Lloyd-Smith, and W. M. Getz. 2005. “Disentangling Association Patterns in Fission-Fusion Societies Using African Buffalo as an Example.” Animal Behaviour 69: 499–506.

Cumming, D. H. M. 2011. “Constraints to Conservation and Development Success at the Wildlife-Livestock-Human Interface in Southern African Transfrontier Conservation Areas: A Preliminary Review.” Wildlife Conservation Society. https://www.wcs- ahead.org/kaza/rpt_constraints_to_cons_&_dev_cumming_ltr_final.pdf.

Cunningham, Andrew A., Peter Daszak, and James L. N. Wood. 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 (1725). https://doi.org/10.1098/rstb.2016.0167.

107

Cushman, Samuel A., Brad McRae, Frank Adriaensen, Paul Beier, Mark Shirley, and Kathy Zeller. 2013. “Biological Corridors and Connectivity.” Key Topics in Conservation Biology 2 (384-404). https://esajournals.onlinelibrary.wiley.com/doi/pdf/10.1002/9781118520178#page=3 95.

Daszak, P., A. A. Cunningham, and A. D. Hyatt. 2000. “Wildlife Ecology - Emerging Infectious Diseases of Wildlife - Threats to Biodiversity and Human Health.” Science 287 (5452): 443–49.

Davis, A. K., D. L. Maney, and J. C. Maerz. 2008. “The Use of Leukocyte Profiles to Measure Stress in Vertebrates: A Review for Ecologists.” Functional Ecology 22 (5): 760–72.

Demas, G. E., D. A. Zysling, B. R. Beechler, M. P. Muehlenbein, and S. S. French. 2011. “Beyond Phytohaemagglutinin: Assessing Vertebrate Immune Function across Ecological Contexts.” The Journal of Animal Ecology 80 (4): 710–30.

Dolan, B. P., K. M. Fisher, M. E. Colvin, S. E. Benda, J. T. Peterson, M. L. Kent, and C. B. Schreck. 2016. “Innate and Adaptive Immune Responses in Migrating Spring-Run Adult Chinook Salmon, Oncorhynchus Tshawytscha.” Fish & Shellfish Immunology 48: 136–44.

Downs, C. J., and K. M. Stewart. 2014. “A Primer in Ecoimmunology and Immunology for Wildlife Research and Management.” California Fish and Game 100 (3): 371–95.

Dugovich, B. S., M. J. Peel, A. L. Palmer, R. A. Zielke, A. E. Sikora, B. R. Beechler, A. E. Jolles, C. W. Epps, and B. P. Dolan. 2017. “Detection of Bacterial-Reactive Natural IgM Antibodies in Desert Bighorn Sheep Populations.” PloS One 12 (6): 15.

Ehrenstein, Michael R., and Clare A. Notley. 2010. “The Importance of Natural IgM: Scavenger, Protector and Regulator.” Nature Reviews. Immunology 10 (11): 778–86.

Elliott, L. F., W. M. Boyce, R. K. Clark, and D. A. Jessup. 1994. “Geographic Analysis of Pathogen Exposure in Bighorn Sheep (Ovis Canadensis).” Journal of Wildlife Diseases 30 (3): 315–18.

El Shikh, M. E., and C. Pitzalis. 2012. “Follicular Dendritic Cells in Health and Disease.” Frontiers in Immunology 3: 292.

Elston, Ralph A., Hiroaki Hasegawa, Karen L. Humphrey, Ildiko K. Polyak, and Claudia C. Häse. 2008. “Re-Emergence of Vibrio Tubiashii in Bivalve Shellfish Aquaculture: Severity, Environmental Drivers, Geographic Extent and Management.” Diseases of Aquatic Organisms 82 (2): 119–34.

108

Epps, Clinton W., Rachel S. Crowhurst, and Brandon S. Nickerson. 2018. “Assessing Changes in Functional Connectivity in a Desert Bighorn Sheep Metapopulation after Two Generations.” Molecular Ecology 27 (10): 2334–46.

Epps, Clinton W., Daniella Dekelaita, and Brian Dugovich. 2016. “Updates on respiratory disease affecting desert bighorn sheep in and near Mojave National Preserve.” Science Newsletter.

Epps, Clinton W., John D. Wehausen, Vernon C. Bleich, Steven G. Torres, and Justin S. Brashares. 2007. “Optimizing Dispersal and Corridor Models Using Landscape Genetics.” The Journal of Applied Ecology 44 (4): 714–24.

Epps, Clinton W., Per J. Palsbøll, John D. Wehausen, George K. Roderick, and Dale R. McCullough. 2006. “Elevation and Connectivity Define Genetic Refugia for Mountain Sheep as Climate Warms.” Molecular Ecology 15 (14): 4295–4302.

Epps, Clinton W., Per J. Palsbøll, John D. Wehausen, George K. Roderick, Rob R. Ramey, and Dale R. McCullough. 2005. “Highways Block Gene Flow and Cause a Rapid Decline in Genetic Diversity of Desert Bighorn Sheep.” Ecology Letters 8 (10): 1029–38.

Epps, Clinton W., Dale R. McCullough, John D. Wehausen, Vernon C. Bleich, and Jennifer L. Rechel. 2004. “Effects of Climate Change on Population Persistence of Desert-Dwelling Mountain Sheep in California.” Conservation Biology: The Journal of the Society for Conservation Biology 18 (1): 102–13.

Ezenwa, Vanessa O., Anna E. Jolles, and Michael P. O’Brien. 2009. “A Reliable Body Condition Scoring Technique for Estimating Condition in African Buffalo.” African Journal of Ecology 47 (4): 476–81.

Ferguson, Kenneth J., Sarah Cleaveland, Daniel Thomas Haydon, Alexandre Caron, Richard A. Kock, Tiziana Lembo, J. Grant C. Hopcraft, et al. 2013. “Evaluating the Potential for the Environmentally Sustainable Control of Foot and Mouth Disease in Sub-Saharan Africa.” EcoHealth 10 (3): 314–22.

Ferris, N. P., and M. Dawson. 1988. “Routine Application of Enzyme-Linked Immunosorbent Assay in Comparison with Complement Fixation for the Diagnosis of Foot-and-Mouth and Swine Vesicular Diseases.” Veterinary Microbiology 16 (3): 201–9.

Fowler, Murray E., and R. Eric Miller. 2003. Zoo and Wild Animal Medicine. 5th ed. St. Louis, Mo.: St. Louis, Mo.: Saunders.

French, Susannah S., and Lorin A. Neuman-Lee. 2012. “Improved Ex Vivo Method for Microbiocidal Activity across Vertebrate Species.” Biology Open 1 (5): 482–87.

109

Foreyt, W. J., and D. A. Jessup. 1982. “Fatal Pneumonia of Bighorn Sheep Following Association with Domestic Sheep.” Journal of Wildlife Diseases 18 (2): 163–68.

French, Susannah S., and Lorin A. Neuman-Lee. 2012. “Improved Ex Vivo Method for Microbiocidal Activity across Vertebrate Species.” Biology Open 1 (5): 482–87.

Gabay, C., and I. Kushner. 1999. “Mechanisms of Disease: Acute-Phase Proteins and Other Systemic Responses to Inflammation.” The New England Journal of Medicine 340 (6): 448–54.

Gelman, Andrew, and Jennifer Hill. 2007. Data Analysis Using Regression and Multilevel/Hierarchical Models. Cambridge University Press.

Gelman, Andrew. 2008. “Scaling Regression Inputs by Dividing by Two Standard Deviations.” Statistics in Medicine 27 (15): 2865–73.

Gervasi, S. S., E. G. Hunt, M. Lowry, and A. R. Blaustein. 2014. “Temporal Patterns in Immunity, Infection Load and Disease Susceptibility: Understanding the Drivers of Host Responses in the Amphibian-Chytrid Fungus System.” Functional Ecology 28 (3): 569–78.

Geyer, Hildegard, Manfred Wuhrer, Anja Resemann, and Rudolf Geyer. 2005. “Identification and Characterization of Keyhole Limpet Hemocyanin N-Glycans Mediating Cross-Reactivity with Schistosoma Mansoni.” The Journal of Biological Chemistry 280 (49): 40731–48.

Gibson, C. F., and A. I. Donaldson. 1986. “Exposure of sheep to natural aerosols of foot- and-mouth-disease virus.” Research in Veterinary Science 41 (1): 45–49.

Gortázar, C., J. F. Ruiz-Fons, and U. Höfle. 2016. “Infections Shared with Wildlife: An Updated Perspective.” European Journal of Wildlife Research 62 (5): 511–25.

Guthrie, G. J. K., K. A. Charles, C. S. D. Roxburgh, P. G. Horgan, D. C. McMillan, and S. J. Clarke. 2013. “The Systemic Inflammation-Based Neutrophil-Lymphocyte Ratio: Experience in Patients with Cancer.” Critical Reviews in Oncology Hematology 88 (1): 218–30.

Harada, K., K. Isse, and Y. Nakanuma. 2006. “Interferon Gamma Accelerates NF-Kappa B Activation of Biliary Epithelial Cells Induced by Toll-like Receptor and Ligand Interaction.” Journal of Clinical Pathology 59 (2): 184–90.

Harvey, D. J., D. R. Wing, B. Küster, and I. B. Wilson. 2000. “Composition of N-Linked Carbohydrates from Ovalbumin and Co-Purified Glycoproteins.” Journal of the American Society for Mass Spectrometry 11 (6): 564–71.

110

Hassell, James M., Michael Begon, Melissa J. Ward, and Eric M. Fèvre. 2017. “Urbanization and Disease Emergence: Dynamics at the Wildlife-Livestock-Human Interface.” Trends in Ecology & Evolution 32 (1): 55–67.

Heard, Geoffrey W., Chris D. Thomas, Jenny A. Hodgson, Michael P. Scroggie, David S. L. Ramsey, and Nick Clemann. 2015. “Refugia and Connectivity Sustain Amphibian Metapopulations Afflicted by Disease.” Ecology Letters 18 (8): 853–63.

Heinrich, Sonja K., Heribert Hofer, Alexandre Courtiol, Jörg Melzheimer, Martin Dehnhard, Gábor Á. Czirják, and Bettina Wachter. 2017. “Cheetahs Have a Stronger Constitutive Innate Immunity than Leopards.” Scientific Reports 7 (March): 44837.

Herget, Stephan, Philip V. Toukach, René Ranzinger, William E. Hull, Yuriy A. Knirel, and Claus-Wilhelm von der Lieth. 2008. “Statistical Analysis of the Bacterial Carbohydrate Structure Data Base (BCSDB): Characteristics and Diversity of Bacterial Carbohydrates in Comparison with Mammalian Glycans.” BMC Structural Biology 8 (August): 35.

Hess, George. 1996. “Disease in Metapopulation Models: Implications for Conservation.” Ecology 77 (5): 1617–32.

Hess, G. R. 1994. “Conservation corridors and contagious-disease – a cautionary note.” Conservation Biology: The Journal of the Society for Conservation Biology 8 (1): 256–62.

Homma, T., A. Kato, N. Hashimoto, J. Batchelor, M. Yoshikawa, S. Imai, H. Wakiguchi, H. Saito, and K. Matsumoto. 2004. “Corticosteroid and Cytokines Synergistically Enhance Toll-like Receptor 2 Expression in Respiratory Epithelial Cells.” American Journal of Respiratory Cell and Molecular Biology 31 (4): 463–69.

Hunter, Kenneth W., Jr, Sally A. Dupré, Tiffanny Sharp, Franziska C. Sandmeier, and C. Richard Tracy. 2008. “Western Blot Can Distinguish Natural and Acquired Antibodies to Mycoplasma Agassizii in the Desert Tortoise (Gopherus Agassizii).” Journal of Microbiological Methods 75 (3): 464–71.

Ian McTaggart Cowan. 1940. “Distribution and Variation in the Native Sheep of North America.” The American Midland Naturalist 24 (3): 505–80.

IPBES. 2019. “Global assessment report on biodiversity and ecosystem services of the Intergovernmental Science- Policy Platform on Biodiversity and Ecosystem Services.” E. S. Brondizio, J. Settele, S. Díaz, and H. T. Ngo (editors). IPBES Secretariat, Bonn, Germany.

Jaeger, Byron C., Lloyd J. Edwards, Kalyan Das, and Pranab K. Sen. 2017. “An R2 Statistic for Fixed Effects in the Generalized Linear Mixed Model.” Journal of

111

Applied Statistics 44 (6): 1086–1105.

Jahner, Joshua P., Marjorie D. Matocq, Jason L. Malaney, Mike Cox, Peregrine Wolff, Mitchell A. Gritts, and Thomas L. Parchman. 2019. “The Genetic Legacy of 50 Years of Desert Bighorn Sheep Translocations.” Evolutionary Applications 12 (2): 198–213.

Jayasekera, Jerome P., E. Ashley Moseman, and Michael C. Carroll. 2007. “Natural Antibody and Complement Mediate Neutralization of Influenza Virus in the Absence of Prior Immunity.” Journal of Virology 81 (7): 3487–94.

Jessup, David A. 1985. “Diseases of Domestic Livestock Which Threaten Bighorn Sheep Populations.” Desert Bighorn Council Transactions 29: 29–33.

Jiang, Fei, Jinyan He, Nalu Navarro-Alvarez, Jian Xu, Xia Li, Peng Li, and Wenxue Wu. 2016. “Elongation Factor Tu and Heat Shock Protein 70 Are Membrane-Associated Proteins from Mycoplasma Ovipneumoniae Capable of Inducing Strong Immune Response in Mice.” PloS One 11 (8): e0161170.

Jolles, A. E., B. R. Beechler, and B. P. Dolan. 2015. “Beyond Mice and Men: Environmental Change, Immunity and Infections in Wild Ungulates.” Parasite Immunology 37 (5): 255–66.

Jolles, A. E., D. V. Cooper, and S. A. Levin. 2005. “Hidden Effects of Chronic Tuberculosis in African Buffalo.” Ecology 86 (9): 2358–64.

Jolles, A. E., and V. O. Ezenwa. 2015. “Ungulates as Model Systems for the Study of Disease Processes in Natural Populations.” Journal of Mammalogy 96 (1): 4–15.

Jolles, Anna E. 2007. “Population Biology of African Buffalo (Syncerus Caffer) at Hluhluwe-iMfolozi Park, South Africa.” African Journal of Ecology 45 (3): 398–406.

Jones, K. E., N. G. Patel, M. A. Levy, A. Storeygard, D. Balk, J. L. Gittleman, and P. Daszak. 2008. “Global Trends in Emerging Infectious Diseases.” Nature 451 (7181): 990–U4.

Jori, F., A. Caron, P. N. Thompson, R. Dwarka, C. Foggin, M. de Garine-Wichatitsky, M. Hofmeyr, J. Van Heerden, and L. Heath. 2016. “Characteristics of Foot-and-Mouth Disease Viral Strains Circulating at the Wildlife/livestock Interface of the Great Limpopo Transfrontier Conservation Area.” Transboundary and Emerging Diseases 63 (1): e58–70.

Jori, Ferran, and Eric Etter. 2016. “Transmission of Foot and Mouth Disease at the Wildlife/livestock Interface of the Kruger National Park, South Africa: Can the Risk Be Mitigated?” Preventive Veterinary Medicine 126 (April): 19–29.

112

Jori, F., W. Vosloo, B. Du Plessis, R. Bengis, D. Brahmbhatt, B. Gummow, and G. R. Thomson. 2009. “A Qualitative Risk Assessment of Factors Contributing to Foot and Mouth Disease Outbreaks in Cattle along the Western Boundary of the Kruger National Park.” Revue Scientifique Et Technique-Office International Des Epizooties 28 (3): 917–31.

Joseph, Maxwell B., Joseph R. Mihaljevic, Ana Lisette Arellano, Jordan G. Kueneman, Daniel L. Preston, Paul C. Cross, and Pieter T. J. Johnson. 2013. “Taming Wildlife Disease: Bridging the Gap between Science and Management.” The Journal of Applied Ecology 50 (3): 702–12.

Jousimo, Jussi, Ayco J. M. Tack, Otso Ovaskainen, Tommi Mononen, Hanna Susi, Charlotte Tollenaere, and Anna-Liisa Laine. 2014. “Disease Ecology. Ecological and Evolutionary Effects of Fragmentation on Infectious Disease Dynamics.” Science 344 (6189): 1289–93.

Juleff, N. D., F. F. Maree, R. Waters, R. G. Bengis, and B. Charleston. 2012. “The Importance of FMDV Localisation in Lymphoid Tissue.” Veterinary Immunology and Immunopathology 148 (1-2): 145–48.

Juleff, Nicholas, Miriam Windsor, Elizabeth Reid, Julian Seago, Zhidong Zhang, Paul Monaghan, Ivan W. Morrison, and Bryan Charleston. 2008. “Foot-and-Mouth Disease Virus Persists in the Light Zone of Germinal Centres.” PloS One 3 (10). https://doi.org/10.1371/journal.pone.0003434.

Kantor, A. B., C. E. Merrill, L. A. Herzenberg, and J. L. Hillson. 1997. “An Unbiased Analysis of V(H)-D-J(H) Sequences from B-1a, B-1b, and Conventional B Cells.” Journal of Immunology 158 (3): 1175–86.

Kasanga, Christopher J., Wataru Yamazaki, Valerie Mioulet, Donald P. King, Misheck Mulumba, Ezekia Ranga, Jimis Deve, et al. 2014. “Rapid, Sensitive and Effective Diagnostic Tools for Foot-and-Mouth Disease Virus in Africa.” The Onderstepoort Journal of Veterinary Research 81 (2): E1–5.

Kawai, T., and S. Akira. 2009. “The Roles of TLRs, RLRs and NLRs in Pathogen Recognition.” International Immunology 21 (4): 317–37.

Kearney, John F., Preeyam Patel, Emily K. Stefanov, and R. Glenn King. 2015. “Natural Antibody Repertoires: Development and Functional Role in Inhibiting Allergic Airway Disease.” Annual Review of Immunology 33 (January): 475–504.

Keyghobadi, Nusha. 2007. “The Genetic Implications of Habitat Fragmentation for Animals.” Canadian Journal of Zoology 85 (10): 1049–64.

Klein, S. A. S. van der, T. V. L. Berghof, J. A. J. Arts, H. K. Parmentier, J. J. van der

113

Poel, and H. Bovenhuis. 2015. “Genetic Relations between Natural Antibodies Binding Keyhole Limpet Hemocyanin and Production Traits in a Purebred Layer Chicken Line.” Poultry Science 94 (5): 875–82.

Knegsel, A. T. M. van, G. de Vries Reilingh, S. Meulenberg, H. van den Brand, J. Dijkstra, B. Kemp, and H. K. Parmentier. 2007. “Natural Antibodies Related to Energy Balance in Early Lactation Dairy Cows.” Journal of Dairy Science 90 (12): 5490–98.

Knight-Jones, T. J. D., and J. Rushton. 2013. “The Economic Impacts of Foot and Mouth Disease - What Are They, How Big Are They and Where Do They Occur?” Preventive Veterinary Medicine 112 (3-4): 161–73.

Koller, B., M. Kappler, P. Latzin, A. Gaggar, M. Schreiner, S. Takyar, M. Kormann, et al. 2008. “TLR Expression on Neutrophils at the Pulmonary Site of Infection: TLR1/TLR2-Mediated up-Regulation of TLR5 Expression in Cystic Fibrosis Lung Disease.” Mian Yi Xue Za Zhi = Journal of Immunology 181 (4): 2753–63.

Lazarus, D. D., G. T. Fosgate, O. L. van Schalkwyk, R. E. J. Burroughs, L. Heath, F. F. Maree, B. Blignaut, et al. 2017. “Serological Evidence of Vaccination and Perceptions Concerning Foot-and-Mouth Disease Control in Cattle at the Wildlife- Livestock Interface of the Kruger National Park, South Africa.” Preventive Veterinary Medicine 147 (November): 17–25.

Lazarus, D. D., O. L. van Schalkwyk, R. E. J. Burroughs, A. Mpehle, B. Reininghaus, O. Rikhotso, L. Heath, F. F. Maree, B. Blignaut, and G. T. Fosgate. 2018. “Serological Responses of Cattle Inoculated with Inactivated Trivalent Foot-and-Mouth Disease Vaccine at the Wildlife-Livestock Interface of the Kruger National Park, South Africa.” Preventive Veterinary Medicine 158 (October): 89–96.

Leader-Williams, N. 1996. “Serengeti II: Dynamics, Management, and Conservation of an Ecosystem: Edited by A.R.E. Sinclair and Peter Arcese University of Chicago Press, 1995. 90.00/£34.00 Hbk, 34.00/£27.95 Pbk (xii + 665 Pages) ISBN 0 226 76031 6.” Trends in Ecology & Evolution 11 (4): 184.

Lewis, Catherine A., Daniel A. Cristol, John P. Swaddle, Claire W. Varian-Ramos, and Patty Zwollo. 2013. “Decreased Immune Response in Zebra Finches Exposed to Sublethal Doses of Mercury.” Archives of Environmental Contamination and Toxicology 64 (2): 327–36.

Liebl, A. L., and L. B. Martin. 2009. “Simple Quantification of Blood and Plasma Antimicrobial Capacity Using Spectrophotometry.” Functional Ecology 23 (6): 1091–96.

Lobo, Peter I., Kailo H. Schlegel, Wen Yuan, Gregory C. Townsend, and Jennifer A.

114

White. 2008. “Inhibition of HIV-1 Infectivity through an Innate Mechanism Involving Naturally Occurring IgM Anti-Leukocyte Autoantibodies.” Journal of Immunology 180 (3): 1769–79.

Ludewig, S., and A. Chanutin. 1949. “Factors Influencing the Agglutination of Red Blood Cells, Red Blood Cell Stroma, and Lymphocytes.” The Journal of Biological Chemistry 179 (1): 271–78.

Lum, Fok-Moon, Teck-Hui Teo, Wendy W. L. Lee, Yiu-Wing Kam, Laurent Rénia, and Lisa F. P. Ng. 2013. “An Essential Role of Antibodies in the Control of Chikungunya Virus Infection.” Journal of Immunology 190 (12): 6295–6302.

Malaney, Jason L., Chris R. Feldman, Michael Cox, Peregrine Wolff, John D. Wehausen, and Marjorie D. Matocq. 2015. “Translocated to the Fringe: Genetic and Niche Variation in Bighorn Sheep of the Great Basin and Northern Mojave Deserts.” Diversity and Distributions 21 (9): 1063–74.

Mak, T. W., M. E. Saunders, and B. D. Jett. 2013. Primer to the Immune Response. Elsevier Science.

Manlove, Kezia, E. Frances Cassirer, Paul C. Cross, Raina K. Plowright, and Peter J. Hudson. 2016. “Disease Introduction Is Associated with a Phase Transition in Bighorn Sheep Demographics.” Ecology 97 (10): 2593–2602.

Marco, I., and S. Lavin. 1999. “Effect of the Method of Capture on the Haematology and Blood Chemistry of Red Deer (Cervus Elaphus).” Research in Veterinary Science 66 (2): 81–84.

Maree, Francois F., Christopher J. Kasanga, Katherine A. Scott, Pamela A. Opperman, Melanie Chitray, Abraham K. Sangula, Raphael Sallu, et al. 2014. “Challenges and Prospects for the Control of Foot-and-Mouth Disease: An African Perspective.” Veterinary Medicine-Research and Reports 5: 119–38.

Martin, Claire, Paul-Pierre Pastoret, Bernard Brochier, Marie-France Humblet, and Claude Saegerman. 2011. “A Survey of the Transmission of Infectious Diseases/infections between Wild and Domestic Ungulates in Europe.” Veterinary Research 42 (June): 70.

Martin, F., A. M. Oliver, and J. F. Kearney. 2001. “Marginal Zone and B1 B Cells Unite in the Early Response against T-Independent Blood-Borne Particulate Antigens.” Immunity 14 (5): 617–29.

Martin, Lynn B., 2nd, Kristen J. Navara, Zachary M. Weil, and Randy J. Nelson. 2007. “Immunological Memory Is Compromised by Food Restriction in Deer Mice Peromyscus Maniculatus.” American Journal of Physiology. Regulatory, Integrative

115

and Comparative Physiology 292 (1): R316–20.

Martin, Lynn B., Raoul K. Boughton, and Daniel R. Ardia. 2014. “A New Division of Ecoimmunology and Disease Ecology.” Integrative and Comparative Biology 54 (3): 338–39.

Martin, Lynn B., II, Zachary M. Weil, and Randy J. Nelson. 2006. “Refining Approaches and Diversifying Directions in Ecoimmunology.” Integrative and Comparative Biology 46 (6): 1030–39.

Matson, K. D., B. I. Tieleman, and K. C. Klasing. 2006. “Capture Stress and the Bactericidal Competence of Blood and Plasma in Five Species of Tropical Birds.” Physiological and Biochemical Zoology: PBZ 79 (3): 556–64.

Matson, Kevin D., Robert E. Ricklefs, and Kirk C. Klasing. 2005. “A Hemolysis- Hemagglutination Assay for Characterizing Constitutive Innate Humoral Immunity in Wild and Domestic Birds.” Developmental and Comparative Immunology 29 (3): 275–86.

Mayasari, N., W. Rijks, G. de Vries Reilingh, G. J. Remmelink, B. Ducro, B. Kemp, H. K. Parmentier, and A. T. M. Van Knegsel. 2016. “The Effects of Dry Period Length and Dietary Energy Source on Natural Antibody Titers and Mammary Health in Dairy Cows.” Preventive Veterinary Medicine 127 (May): 1–9.

McCallum, Hamish. 2008. “Tasmanian Devil Facial Tumour Disease: Lessons for Conservation Biology.” Trends in Ecology & Evolution 23 (11): 631–37.

McClintock, Brett T., and Gary C. White. 2007. “Bighorn Sheep Abundance Following a Suspected Pneumonia Epidemic in Rocky Mountain National Park.” The Journal of Wildlife Management 71 (1): 183–89.

McKnight, Donald T., Lin Schwarzkopf, Ross A. Alford, Deborah S. Bower, and Kyall R. Zenger. 2017. “Effects of Emerging Infectious Diseases on Host Population Genetics: A Review.” Conservation Genetics 18 (6): 1235–45.

Medzhitov, R., and C. A. Janeway. 1997. “Innate Immunity: Impact on the Adaptive Immune Response.” Current Opinion in Immunology 9 (1): 4–9.

Michel, Anita L., and Roy G. Bengis. 2012. “The African Buffalo: A Villain for Inter- Species Spread of Infectious Diseases in Southern Africa.” The Onderstepoort Journal of Veterinary Research 79 (2): 453.

Miguel, Eve, Vladimir Grosbois, Alexandre Caron, Thierry Boulinier, Herve Fritz, Daniel Cornelis, Chris Foggin, Pious V. Makaya, Priscillia T. Tshabalala, and Michel de Garine-Wichatitsky. 2013. “Contacts and Foot and Mouth Disease Transmission from Wild to Domestic Bovines in Africa.” Ecosphere 4 (4).

116

https://doi.org/10.1890/ES12-00239.1.

Miller, Ryan S., Matthew L. Farnsworth, and Jennifer L. Malmberg. 2013. “Diseases at the Livestock-Wildlife Interface: Status, Challenges, and Opportunities in the United States.” Preventive Veterinary Medicine 110 (2): 119–32.

Molloy, M. P., B. R. Herbert, M. B. Slade, T. Rabilloud, A. S. Nouwens, K. L. Williams, and A. A. Gooley. 2000. “Proteomic Analysis of the Escherichia Coli Outer Membrane.” European Journal of Biochemistry / FEBS 267 (10): 2871–81. Moonen, P., and R. Schrijver. 2000. “Carriers of Foot-and-Mouth Disease Virus: A Review.” The Veterinary Quarterly 22 (4): 193–97.

Muma, John B., Kennedy K. Mwacalimba, Hetron M. Munang’andu, Gift Matope, Akinbowale Jenkins, Victor Siamudaala, Aaron S. Mweene, and Tanguy Marcotty. 2014. “The Contribution of Veterinary Medicine to Public Health and Poverty Reduction in Developing Countries.” Veterinaria Italiana 50 (2): 117–29.

Nguyen, Trang T. T., Rebecca A. Elsner, and Nicole Baumgarth. 2015. “Natural IgM Prevents Autoimmunity by Enforcing B Cell Central Tolerance Induction.” Journal of Immunology 194 (4): 1489–1502.

Nickerson, Brandon Scott. 2014. “Effects of Genetic Drift, Natural Selection, and Population Connectivity on Adaptive-Linked Genetic Diversity of Desert Bighorn Sheep,” September. https://ir.library.oregonstate.edu/concern/graduate_thesis_or_dissertations/ 5q47rs554.

Notley, Clare A., Mark A. Brown, Graham P. Wright, and Michael R. Ehrenstein. 2011. “Natural IgM Is Required for Suppression of Inflammatory Arthritis by Apoptotic Cells.” Journal of Immunology 186 (8): 4967–72.

Ochsenbein, A. F., T. Fehr, C. Lutz, M. Suter, F. Brombacher, H. Hengartner, and R. M. Zinkernagel. 1999. “Control of Early Viral and Bacterial Distribution and Disease by Natural Antibodies.” Science 286 (5447): 2156–59.

Oksanen, J., R. Kindt, P. Legendre, and B. O’Hara. 2007. “The Vegan Package.” Researchgate.net. https://www.researchgate.net/profile/Gavin_Simpson/publication/228339454_The_v egan_Package/links/0912f50be86bc29a7f000000/The-vegan-Package.pdf.

Oliviera Nascimento, Laura de, Paola Massari, and Lee Wetzler. 2012. “The Role of TLR2 in Infection and Immunity.” Frontiers in Immunology 3 (79). https://doi.org/10.3389/fimmu.2012.00079.

Otto, Sarah P. 2018. “Adaptation, Speciation and Extinction in the Anthropocene.”

117

Proceedings. Biological Sciences / The Royal Society 285 (1891). https://doi.org/10.1098/rspb.2018.2047.

Panda, Saswati, and Jeak L. Ding. 2015. “Natural Antibodies Bridge Innate and Adaptive Immunity.” Journal of Immunology 194 (1): 13–20.

Pinheiro, Jose, Douglas Bates, Saikat DebRoy, Deepayan Sarkar, and R Team, Core. 2014. Linear and Nonlinear Mixed Effects Models. Vol. 3.

Plowright, Raina K., Patrick Foley, Hume E. Field, Andy P. Dobson, Janet E. Foley, Peggy Eby, and Peter Daszak. 2011. “Urban Habituation, Ecological Connectivity and Epidemic Dampening: The Emergence of Hendra Virus from Flying Foxes (Pteropus Spp.).” Proceedings. Biological Sciences / The Royal Society 278 (1725): 3703–12.

Prall, S. P., and M. P. Muehlenbein. 2014. “Testosterone and Immune Function in Primates: A Brief Summary with Methodological Considerations.” International Journal of Primatology 35 (3-4): 805–24.

Quartier, Pierre, Paul K. Potter, Michael R. Ehrenstein, Mark J. Walport, and Marina Botto. 2005. “Predominant Role of IgM-Dependent Activation of the Classical Pathway in the Clearance of Dying Cells by Murine Bone Marrow-Derived Macrophages in Vitro.” European Journal of Immunology 35 (1): 252–60.

R Core Team. 2017. “R: A language and environment for statistical computing.” R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R- project.org/.

Reid, S. M., N. P. Ferris, G. H. Hutchings, Z. D. Zhang, G. J. Belsham, and S. Alexandersen. 2002. “Detection of All Seven Serotypes of Foot-and-Mouth Disease Virus by Real-Time, Fluorogenic Reverse Transcription Polymerase Chain Reaction Assay.” Journal of Virological Methods 105 (1): 67–80.

Richards, Gary P., Michael A. Watson, David S. Needleman, Karlee M. Church, and Claudia C. Häse. 2015. “Mortalities of Eastern and Pacific Oyster Larvae Caused by the Pathogens Vibrio Coralliilyticus and Vibrio Tubiashii.” Applied and Environmental Microbiology 81 (1): 292–97.

Rich, K. M., and B. D. Perry. 2011. “The Economic and Poverty Impacts of Animal Diseases in Developing Countries: New Roles, New Demands for Economics and Epidemiology.” Preventive Veterinary Medicine 101 (3-4): 133–47.

Roberts, Laura C., and Geoffrey T. Fosgate. 2018. “Stakeholder Perceptions of Foot-and- Mouth Disease Control in South Africa.” Preventive Veterinary Medicine 156 (August): 38–48.

118

Ryan, S. J., C. U. Knechtel, and W. M. Getz. 2007. “Ecological Cues, Gestation Length, and Birth Timing in African Buffalo (Syncerus Caffer).” Behavioral Ecology: Official Journal of the International Society for Behavioral Ecology 18 (4): 635–44.

Sandmeier, Franziska C., C. Richard Tracy, Sally Dupré, and Kenneth Hunter. 2012. “A Trade-off between Natural and Acquired Antibody Production in a Reptile: Implications for Long-Term Resistance to Disease.” Biology Open 1 (11): 1078–82.

SANParks. 2019. "South African National Parks - SANParks - Official Website - Accommodation, Activities, Prices, Reservations." SANParks. Accessed June 19, 2019. https://www.sanparks.org/conservation/scientific_new/savannah_arid/data_resources/ weather.php.

Savage, Hannah P., and Nicole Baumgarth. 2015. “Characteristics of Natural Antibody- Secreting Cells.” Annals of the New York Academy of Sciences 1362 (December): 132–42.

Schalkwyk, O. L. van, D. L. Knobel, E. M. De Clercq, C. De Pus, G. Hendrickx, and P. Van den Bossche. 2016. “Description of Events Where African Buffaloes (Syncerus Caffer) Strayed from the Endemic Foot-and-Mouth Disease Zone in South Africa, 1998-2008.” Transboundary and Emerging Diseases 63 (3): 333–47.

Schleimer, Robert P., Atsushi Kato, Robert Kern, Douglas Kuperman, and Pedro C. Avila. 2007. “Epithelium: At the Interface of Innate and Adaptive Immune Responses.” The Journal of Allergy and Clinical Immunology 120 (6): 1279–84.

Seitanidis, B. 1969. “A Comparison of the Monospot with the Paul-Bunnell Test in Infectious Mononucleosis and Other Diseases.” Journal of Clinical Pathology 22 (3): 321–23.

Shinozawa, Yoko, Tetsuya Matsumoto, Kou Uchida, Shiro Tsujimoto, Yoichiro Iwakura, and Keizo Yamaguchi. 2002. “Role of Interferon-Gamma in Inflammatory Responses in Murine Respiratory Infection with Legionella Pneumophila.” Journal of Medical Microbiology 51 (3): 225–30.

Silva, M. T., and M. Correia-Neves. 2012. “Neutrophils and Macrophages: The Main Partners of Phagocyte Cell Systems.” Frontiers in Immunology 3. https://doi.org/10.3389/fimmu.2012.00174.

Singer, F. J., C. M. Papouchis, and K. K. Symonds. 2000. “Translocations as a Tool for Restoring Populations of Bighorn Sheep.” Restoration Ecology 8 (4): 6–13.

Ster, Irina Chis, Peter J. Dodd, and Neil M. Ferguson. 2012. “Within-Farm Transmission Dynamics of Foot and Mouth Disease as Revealed by the 2001 Epidemic in Great

119

Britain.” Epidemics 4 (3): 158–69.

Strobel, S., N. I. Becker, and J. A. Encarnacao. 2015. “No Short-Term Effect of Handling and Capture Stress on Immune Responses of Bats Assessed by Bacterial Killing Assay.” Mammalian Biology = Zeitschrift Fur Saugetierkunde 80 (4): 312–15.

Subspecies, Bighorn. 1993. “A morphometric reevaluation of the peninsular bighorn subspecies.” Desert Bighorn Council Transactions 37: 1–10.

Sutmolle.P, and A. Gaggero. 1965. “Foot-and-mouth disease carriers.” The Veterinary Record 77 (33): 968 – &.

Thomson, G. R., W. Vosloo, J. J. Esterhuysen, and R. G. Bengis. 1992. “Maintenance of Foot and Mouth Disease Viruses in Buffalo (Syncerus Caffer Sparrman, 1779) in Southern Africa.” Revue Scientifique et Technique 11 (4): 1097–1107.

Turner, W. C., A. E. Jolles, and N. Owen-Smith. 2005. “Alternating Sexual Segregation during the Mating Season by Male African Buffalo (Syncerus Caffer).” Journal of Zoology 267: 291–99.

Wang, Chunguang, S. Pauliina Turunen, Outi Kummu, Marja Veneskoski, Jaakko Lehtimäki, Antti E. Nissinen, and Sohvi Hörkkö. 2013. “Natural Antibodies of Newborns Recognize Oxidative Stress-Related Malondialdehyde Acetaldehyde Adducts on Apoptotic Cells and Atherosclerotic Plaques.” International Immunology 25 (10): 575–87.

Watson, James E. M., Danielle F. Shanahan, Moreno Di Marco, James Allan, William F. Laurance, Eric W. Sanderson, Brendan Mackey, and Oscar Venter. 2016. “Catastrophic Declines in Wilderness Areas Undermine Global Environment Targets.” Current Biology: CB 26 (21): 2929–34.

Watson, James E. M., Oscar Venter, Jasmine Lee, Kendall R. Jones, John G. Robinson, Hugh P. Possingham, and James R. Allan. 2018. “Protect the Last of the Wild.” Nature 563 (7729): 27–30.

Watson, R. R., D. N. McMurray, P. Martin, and M. A. Reyes. 1985. “Effect of age, malnutrition and renutrition on free secretory component and IgA in secretions.” The American Journal of Clinical Nutrition 42 (2): 281–88.

Weaver, Genevieve V., Joseph Domenech, Alex R. Thiermann, and William B. Karesh. 2013. “Foot and Mouth Disease: A Look from the Wild Side.” Journal of Wildlife Diseases 49 (4): 759–85.

Wehausen, J. D., S. T. Kelley, and R. R. Ramey. 2011. “Domestic Sheep, Bighorn Sheep, and Respiratory Disease: A Review of the Experimental Evidence.” California Fish

120

and Game 97 (1): 7–24.

Wehausen, John D. 2005. “Nutrient Predictability, Birthing Seasons, and Lamb Recruitment for Desert Bighorn Sheep.” In Symposium Proceedings for the Sweeney Granite Mountains Desert Research Center 1978--2003; A Quarter Century of Research and Teaching. University of California Natural Reserve Program. http://www.academia.edu/download/42877162/Wehausen_Granite_Mtn_Reserve_Sy mposium_2005.pdf.

White, Lauren A., James D. Forester, and Meggan E. Craft. 2018. “Dynamic, Spatial Models of Parasite Transmission in Wildlife: Their Structure, Applications and Remaining Challenges.” The Journal of Animal Ecology 87 (3): 559–80.

Wolmer, W. 2003. “Transboundary Conservation: The Politics of Ecological Integrity in the Great Limpopo Transfrontier Park.” Journal of Southern African Studies 29 (1): 261–78. WWF. 2014. “Living Planet Report 2014: Species and spaces, people and places.” WWF International, Gland, Switzerland.

Yang, Yang, Chunlin Wang, Qunying Yang, Aaron B. Kantor, Hiutung Chu, Eliver Eb Ghosn, Guang Qin, Sarkis K. Mazmanian, Jian Han, and Leonore A. Herzenberg. 2015. “Distinct Mechanisms Define Murine B Cell Lineage Immunoglobulin Heavy Chain (IgH) Repertoires.” eLife 4 (September): e09083. Yoon, S. I., O. Kurnasov, V. Natarajan, M. S. Hong, A. V. Gudkov, A. L. Osterman, and I. A. Wilson. 2012. “Structural Basis of TLR5-Flagellin Recognition and Signaling.” Science 335 (6070): 859–64.

Young, H. S., R. Dirzo, K. M. Helgen, D. J. McCauley, C. L. Nunn, P. Snyder, K. E. Veblen, S. Zhao, and V. O. Ezenwa. 2016. “Large Wildlife Removal Drives Immune Defence Increases in Rodents.” Functional Ecology 30 (5): 799–807.

Zielke, Ryszard A., Philip R. Gafken, and Aleksandra E. Sikora. 2014. “Quantitative Proteomic Analysis of the Cell Envelopes and Native Membrane Vesicles Derived from Gram-Negative Bacteria.” Current Protocols in Microbiology 34 (August): 1F.3.1–16.

Zielke, Ryszard A., and Aleksandra E. Sikora. 2014. “Isolation of Cell Envelopes and Naturally Released Membrane Vesicles of Neisseria Gonorrhoeae.” Current Protocols in Microbiology 34 (August): 4A.3.1–17.

121

APPENDIX FIGURES

A4.1 Foot-and-mouth disease infection class prevalence in a herd of African buffalo in Kruger National Park.

A4.2 Calf foot-and-mouth disease infection progression in a herd of African buffalo in Kruger National Park.

A4.3 Adult foot-and-mouth disease carrier prevalence by month in a herd of African buffalo in Kruger National Park.

122

Appendix Figure 4.1. Foot-and-mouth disease infection class prevalence in a herd of African buffalo in Kruger National Park. This study took place from 2014 – 2016 with captures every 2 – 3 months. The whole herd, calves, adults, adult males, and adult females varied with infection class over the course of the study. Infections appeared to fade out over the course of the study.

123

Appendix Figure 4.2. Calf foot-and-mouth disease infection progression in a herd of African buffalo in Kruger National Park. This study took place from 2014 – 2016 with captures every 2 – 3 months. Infection class prevalence was averaged by month. No captures took place in March. Calves predictably moved from maternal protection early in the year to susceptibility peaking in the dry season (August) to infection peaking in the wet season (December).

124

Appendix Figure 4.3. Adult foot-and-mouth disease carrier prevalence by month in a herd of African buffalo in Kruger National Park. This study took place from 2014 – 2016 with captures every 2 – 3 months. Carrier prevalence was averaged by month. No captures took place in March. Carrier prevalence in adult buffalo peaked at the beginning of the wet season in October.

125

APPENDIX TABLES

A1.1 Primer sets for TLR quantitative PCR of 7 ungulates at Wildlife Safari.

A1.2 Wildlife Safari study individuals.

A3.1 Microsatellite loci used to calculate allelic richness and heterozygosity in 11 populations of desert bighorn sheep in the Mojave Desert.

A4.1 Whole herd models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park.

A4.2 Calf models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park.

A4.3 Adult models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park.

A4.4 Tonsillar swab PCR copies of foot-and-mouth disease virus models of adult carrier buffalo in a herd of African buffalo in Kruger National Park.

A4.5 Adult reinfection models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park.

A4.6 Mature female reproduction models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park.

A4.7 Models of stray African buffalo in the vicinity of Kruger National Park (1998-2013).

126

Appendix Table 1.1. Primer sets for TLR quantitative PCR of 7 ungulates at Wildlife Safari. These ungulates were sampled in 2013 – 2014. Target Gene Primer Sequence (5’ – 3’) Size of PCR product GAPDH Forward - ATCAAGAAGGTRGTGAAGCAGG 175 base pairs Reverse - TGTCRTACCAGGAAATGAGCTT TLR2 Forward - ACKCTMCCRGATGCCTCCTT 158 base pairs Reverse - GACAGGAABTCACAGGAGCA TLR5 Forward - GACGCSTGGTGCCTSGAA 170 base pairs Reverse - TCYTCRGGCCACCTCAARTAC

127

Appendix Table 4.2. Wildlife Safari study individuals. These ungulates were sampled 2013 – 2014. Health status indicates minor health problems of some animals included in the study. Individual Species Sex* Age (months) Health Status WSP13-003 elk cm 156.5 ulcerated cornea WSP13-004 elk cm 144.7 epistaxis WSP13-005 elk f 157.1 normal WSP13-006 sika f 144.4 normal WSP13-013 sika f 181.3 normal WSP13-015 sika f 146 hoof ulcer WSP13-016 aoudad m 147.1 pododermatitis WSP13-017 sika f 133 normal WSP13-018 fallow f 145.3 abnormal eye WSP13-020 sika f 241.2 normal WSP13-027 bison f 127.2 normal WSP13-028 zebra m 19.6 normal WSP13-029 elk f 165.4 normal WSP13-031 sika f 133.6 normal WSP13-033 yak f 143.5 normal WSP13-034 elk f 146.9 pododermatitis WSP13-035 aoudad f 87.7 normal WSP13-036 bison f 43.8 normal WSP13-037 zebra f 36.9 capture hyperthermia WSP13-038 yak f 160.7 normal WSP13-039 sika f 147 normal WSP13-040 sika m 153.7 overgrown hooves WSP13-041 aoudad f 74.7 normal WSP13-042 zebra f 26.3 normal WSP13-043 bison f 104.7 normal WSP13-044 yak f 313.8 lethargic WSP14-01 elk f 168.7 normal WSP14-02 sika f 176.9 normal WSP14-03 elk f 156.8 normal WSP14-04 sika f 133 normal WSP14-05 sika f 168 normal WSP14-06 fallow f 190.5 multiple masses WSP14-07 sika f 217.5 normal WSP14-08 fallow f 157.1 multiple masses WSP14-09 elk f 178.6 normal WSP14-10 sika f 167.6 normal WSP14-11 sika f 181.7 normal WSP14-12 fallow f 156.3 normal WSP14-13 sika f 192.9 normal

128

WSP14-14 elk f 167.9 normal WSP14-15 elk cm 204.9 minor laceration WSP14-16 elk f 181.6 normal WSP14-18 elk m 158.2 normal WSP14-19 elk m 30.9 normal *cm refers to castrated males

129

Appendix Table 3.1. Microsatellite loci used to calculate allelic richness and heterozygosity in 11 populations of desert bighorn sheep in the Mojave Desert. Heterozygosity of the adaptive-linked gene, BL4, was also used in individual immunophenotype analyses. Locus Type ADCYAP Adaptive AE129 Neutral AE16 Neutral BL4 Adaptive FCB11 Neutral FCB193 Neutral FCB266 Neutral FCB304 Neutral HH62 Neutral JMP29 Neutral KP6 Adaptive MAF209 Neutral MAF33 Neutral MAF36 Neutral MAF48 Neutral MAF65 Neutral MHC1 Adaptive MMP9 Adaptive RA01 Adaptive TCRBV6 Adaptive TGLA38 Adaptive

130

Appendix Table 4.1. Whole herd models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park. Longitudinal data (2014 – 2016) were used in generalized linear mixed models with logit links. The estimates of dependent variables are displayed with standard error below in parentheses. Asterisks designate significance. Dependent variable: Maternal Susceptible Infected Carrier Recovered (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) Male 2.481 2.552 -0.466 0.088 0.799* 0.741* 0.742* -0.899** -0.892** -0.914** (1.727) (2.034) (0.618) (0.474) (0.439) (0.438) (0.436) (0.409) (0.403) (0.406) Real Age -77.996* -4.049*** -2.848*** -0.685* 3.065*** (45.574) (0.819) (0.861) (0.408) (0.428) 1st Capture Age -40.787 -72.648 -3.556*** -3.729*** -2.689*** -2.712*** -1.035** -1.003** 3.052*** 3.043*** (25.274) (70.517) (0.833) (0.907) (0.889) (0.949) (0.449) (0.447) (0.446) (0.446) Capture -1.650 -0.143** -0.227*** -0.001 0.102*** (2.765) (0.062) (0.078) (0.031) (0.031) BCS 4.206 4.748** 7.696 0.034 -1.067** -1.258** -0.769 -0.686*** -0.619** -0.723*** 0.756*** 0.886*** 0.644** (2.567) (2.214) (6.494) (0.432) (0.461) (0.516) (0.470) (0.241) (0.246) (0.256) (0.250) (0.255) (0.263) Wet Season 0.015 -0.710* -0.211 0.419** -0.105 (0.935) (0.364) (0.472) (0.207) (0.212) Constant -50.822* -22.815** -34.796 -3.792*** -3.140*** -2.630*** -4.401*** -3.927*** -2.675*** -1.379*** -1.452*** -1.273*** 0.216 -0.040 -0.857** (29.339) (11.102) (27.983) (0.498) (0.419) (0.527) (0.561) (0.478) (0.568) (0.273) (0.288) (0.365) (0.230) (0.245) (0.337)

Observations 666 666 666 666 666 666 666 666 666 666 666 666 666 666 666 Log Likelihood -21.850 -37.670 -32.337 -149.843 -153.193 -152.142 -92.939 -94.480 -89.451 -367.860 -364.695 -366.748 -333.652 -338.585 -333.307 Akaike Inf. Crit. 53.701 87.339 74.674 309.686 314.385 312.284 195.878 198.959 188.903 745.720 741.391 745.496 677.304 689.171 678.614 Bayesian Inf. 76.185 114.320 97.158 332.170 332.390 330.289 218.362 221.443 211.387 768.204 768.371 772.477 699.788 716.151 705.595 Crit. Note: *p<0.1; **p<0.05; ***p<0.01

131

Appendix Table 4.2. Calf models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park. Longitudinal data (2014 – 2016) were used in generalized linear mixed models with logit links. The estimates of dependent variables are displayed with standard error below in parentheses. Asterisks designate significance. Dependent variable: Maternal Susceptible Infected Carrier Recovered (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) Male 2.481 2.292 5.333 -0.498 0.602 -0.277 -0.660 -1.274 -0.861 (1.727) (1.766) (4.720) (0.633) (0.704) (1.188) (1.024) (0.866) (0.848) Real Age -77.994* -5.823 6.398 38.400* (45.607) (4.175) (18.380) (23.317) 1st Capture -29.621 -98.640 -26.373 -13.754 2.252 26.541 4.373 Age (25.784) (77.804) (17.851) (15.425) (15.013) (23.620) (9.776) BCS 4.206 4.698** 7.525* -1.643** -0.949 -0.854 -0.995 0.318 0.034 (2.567) (2.156) (4.442) (0.714) (0.593) (0.712) (0.762) (0.738) (0.674) Horn Length 0.861 -0.201 0.529 -0.055 -0.157 0.324 1.986** -2.394 0.842 (0.629) (0.633) (0.799) (0.833) (1.854) (0.789) (1.008) (2.283) (0.750) Wet Season -0.183 -1.510*** 2.068** -0.303 0.382 (0.934) (0.558) (0.830) (0.666) (0.546) Capture -1.151 0.193** -0.084 -0.520*** 0.118 (0.710) (0.098) (0.085) (0.166) (0.123) Constant -50.821* -17.245 -46.876 -4.107* -10.992 -7.415 -2.961*** -2.937 -1.878*** 1.214 8.530 1.226 19.740 -0.442 -2.709** (29.360) (11.563) (32.971) (2.360) (7.341) (6.119) (0.682) (6.169) (0.693) (10.814) (9.638) (0.947) (12.965) (3.789) (1.155)

Observations 163 163 163 163 163 163 163 163 163 163 163 163 163 163 163 Log Likelihood -21.850 -35.977 -31.563 -96.296 -67.933 -70.270 -33.611 -32.105 -51.976 -56.732 -56.172 -51.952 -48.776 -66.535 -51.266 Akaike Inf. 53.701 83.954 75.125 198.591 145.867 152.541 75.222 76.210 111.953 125.465 124.345 111.904 109.551 145.069 112.532 Crit.

132

Bayesian Inf. 69.108 102.442 93.614 207.873 159.846 169.316 86.372 92.985 124.278 142.190 141.069 123.087 126.276 163.558 126.511 Crit. Note: *p<0.1; **p<0.05; ***p<0.01

133

Appendix Table 4.3. Adult models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park. Longitudinal data (2014 – 2016) were used in generalized linear mixed models with logit links. The estimates of dependent variables are displayed with standard error below in parentheses. Asterisks designate significance. Dependent variable:

Adult Adult Adult Male Adult Male Adult Female Adult Female Carrier Recovered Carrier Recovered Carrier Recovered (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18)

Male 1.141** 1.138* 1.067* -0.911 -1.024* -0.897 (0.571) (0.586) (0.579) (0.573) (0.581) (0.575) Real Age -2.623*** 2.745*** -2.890** 2.577** -2.141** 2.226*** (0.648) (0.654) (1.147) (1.297) (0.836) (0.824) 1st Capture -2.656*** -2.734*** 2.414*** 2.662*** -3.351** -3.332** 2.700* 2.736** -1.928*** -1.915*** 1.853** 2.149** Age (0.687) (0.686) (0.663) (0.670) (1.308) (1.314) (1.400) (1.395) (0.690) (0.684) (0.821) (0.852) BCS -0.829*** -0.903*** -0.873*** 0.923*** 1.010*** 0.791*** -0.971* -1.249** -1.484** 1.080** 1.287** 1.358** -0.663* -0.651* -0.654* 0.751** 0.811** 0.567 (0.283) (0.290) (0.303) (0.283) (0.288) (0.300) (0.506) (0.557) (0.638) (0.523) (0.561) (0.633) (0.349) (0.355) (0.364) (0.344) (0.349) (0.358) Pregnant -0.337 -0.323 -0.335 0.296 0.313 0.291 (0.342) (0.339) (0.339) (0.332) (0.330) (0.336) Lactating -1.115** -1.067** -1.074** 0.961** 1.033** 0.966** (0.479) (0.464) (0.461) (0.446) (0.443) (0.444) Horn Width 0.061 0.025 0.128 -0.152 0.069 -0.140 0.780 -0.436 -0.082 -0.627 (0.591) (0.604) (0.602) (0.589) (0.602) (0.594) (1.064) (1.029) (0.998) (1.047) Wet Season 0.445* -0.262 0.530 -0.378 0.308 -0.128 (0.244) (0.239) (0.442) (0.442) (0.308) (0.302) Capture -0.041 0.102*** 0.035 0.0001 -0.027 0.126*** (0.037) (0.037) (0.070) (0.070) (0.046) (0.048) Boss Width 0.305 0.526 0.529 -0.253 -0.388 -0.364 (0.532) (0.553) (0.577) (0.543) (0.553) (0.569) Scrotum -2.293* -2.125 -1.892 1.965 1.956 1.778 Circum.

134

(1.210) (1.306) (1.321) (1.336) (1.412) (1.400) Constant -0.694** -0.707** -0.139 0.386 0.389 -0.630 0.905** 0.801** 0.720 -0.985** -0.970** -1.105 -0.313 -0.395 -0.040 0.024 -0.015 -1.122** (0.296) (0.342) (0.454) (0.296) (0.337) (0.452) (0.385) (0.403) (0.676) (0.436) (0.438) (0.704) (0.368) (0.396) (0.499) (0.361) (0.418) (0.560)

Observations 503 503 503 503 503 503 154 154 154 154 154 154 349 349 349 349 349 349 Log -255.717 -254.710 -255.776 -260.722 -262.936 -259.613 -69.092 -67.809 -68.414 -71.183 -70.785 -71.152 -161.134 -164.278 -164.615 -166.543 -167.989 -164.409 Likelihood Akaike Inf. 523.433 523.421 525.553 533.443 539.871 533.225 150.183 149.617 150.828 154.366 155.571 156.304 336.268 342.557 343.229 347.085 351.978 344.817 Crit. Bayesian Inf. 548.624 552.810 554.942 558.634 569.261 562.615 167.833 170.209 171.419 172.016 176.162 176.896 362.559 369.043 369.716 373.376 382.024 374.863 Crit.

Note: *p<0.1; **p<0.05; ***p<0.01

135

Appendix Table 4.4. Tonsillar swab PCR copies of foot-and-mouth disease virus models of adult carrier buffalo in a herd of African buffalo in Kruger National Park. Longitudinal data (2014 – 2016) were used in generalized linear mixed models with logit links. The estimates of dependent variables are displayed with standard error below in parentheses. Asterisks designate significance. Dependent variable: Log (Swab PCR Copies) in Carriers (1) (2) (3) Real Age -3.943*** (1.377) Male -0.645 -0.514 (1.028) (1.042) Age 1st Capture -2.782** -3.737*** (1.285) (1.425) BCS -2.822*** -2.770*** -2.242*** (0.556) (0.598) (0.635) Horn Width 1.919* 1.278 1.912* (1.070) (0.970) (1.076) Wet Season 0.766 (0.570) Capture -0.242*** (0.089) Constant 3.854*** 3.446*** 6.030*** (0.588) (0.540) (0.967) Observations 183 183 183 Log Likelihood -502.927 -504.402 -502.514 Akaike Inf. Crit. 1,019.854 1,022.804 1,021.027 Bayesian Inf. Crit. 1,042.126 1,045.076 1,046.436 Note: *p<0.1; **p<0.05; ***p<0.01

136

Appendix Table 4.5. Adult reinfection models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park. Longitudinal data (2014 – 2016) were used in generalized linear mixed models with logit links for carriers and linear mixed effects models for tonsillar swab viral PCR copies in carriers. The estimates of dependent variables are displayed with standard error below in parentheses. Asterisks designate significance. Dependent variable: Carrier Log(FMDV PCR Copies) generalized linear linear mixed effects mixed effects (1) (2) (3) (4) # infected calves current capture -0.598 -0.171 (0.391) (0.858) # infected calves 1 capture prior 0.773** 0.817** 1.476** 1.573** (0.324) (0.333) (0.613) (0.608) # infected calves 2 captures prior 0.947*** 1.124*** 0.195 0.378 (0.302) (0.363) (0.623) (0.704) # infected calves 3 captures prior -1.246*** -0.855 (0.338) (0.711) Male 1.186* 1.325** -0.396 (0.606) (0.673) (1.071) 1st Capture Age -3.269*** -3.610*** -3.438** -2.940** (0.685) (0.791) (1.488) (1.471) BCS -1.206*** -0.832** -2.660*** -2.245*** (0.429) (0.381) (0.826) (0.737) Horn Width 1.968* 1.677* (1.075) (1.004) Constant -0.567 -0.555 3.593*** 3.368*** (0.365) (0.401) (0.655) (0.495)

Observations 503 503 166 161 Log Likelihood -212.903 -187.804 -448.204 -435.210 Akaike Inf. Crit. 441.806 391.609 916.409 888.420 Bayesian Inf. Crit. 474.733 423.796 947.035 915.753 Note: *p<0.1; **p<0.05; ***p<0.01

137

Appendix Table 4.6. Mature female reproduction models of foot-and-mouth disease virus infection classes in a herd of African buffalo in Kruger National Park. Female buffalo 6 years of age and older were examined. Longitudinal data (2014 – 2016) were used in generalized linear mixed models with logit links. The estimates of dependent variables are displayed with standard error below in parentheses. Asterisks designate significance. Dependent variable: Recovered Carrier (1) (2) (3) (4) (5) (6) Real Age 0.777 -0.486 (1.029) (1.312) 1st Capture Age 0.437 0.511 -0.598 -0.493 (1.075) (1.190) (1.316) (1.451) BCS 0.935* 1.028** 0.736 -1.004* -1.002* -0.950* (0.495) (0.510) (0.503) (0.563) (0.571) (0.577) Pregnant 0.286 0.323 0.231 -0.222 -0.311 -0.217 (0.422) (0.424) (0.427) (0.463) (0.463) (0.463) Lactating 0.782 0.840* 0.835 -1.178** -1.107* -1.176** (0.494) (0.493) (0.509) (0.594) (0.568) (0.592) Wet Season 0.339 -0.055 (0.409) (0.444) Horn Width -0.059 0.295 0.309 (1.627) (1.989) (2.031) Capture 0.134** -0.037 (0.068) (0.074) Constant 1.194* 1.127 0.139 -1.849** -1.741* -1.504 (0.673) (0.783) (0.966) (0.855) (0.980) (1.161) Observations 226 226 226 226 226 226 Log Likelihood -100.671 -100.539 -96.685 -86.773 -88.819 -86.676 Akaike Inf. Crit. 213.342 215.078 209.370 187.546 191.637 189.353 Bayesian Inf. Crit. 233.704 238.833 236.260 211.075 215.393 216.243 Note: *p<0.1; **p<0.05; ***p<0.01

138

Appendix Table 4.7. Models of stray African buffalo in the vicinity of Kruger National Park (1998-2013). Linear regression was performed with season and sex as dependent variables. Interactions of sex and season were not significant. Only a seasonal model was performed for calves due to their low sample size. Dependent variable: Stray Buffalo Events Calf (1) (2) (3) (4) Wet Season -21.178*** -22.091*** -21.935** -0.500 (0.800) (6.264) (8.542) (0.342) Male -2.205 (3.633) Wet Season:Male -0.029 (4.479) Female 3.538 (5.028) Wet Season:Female -2.389 (5.314) Constant 68.137*** 69.671*** 64.769*** 1.500*** (0.545) (4.883) (7.713) (0.224)

Observations 628 68 56 7 R2 0.528 0.518 0.481 0.300 Adjusted R2 0.528 0.496 0.452 0.160 Residual Std. Error 9.996 (df = 626) 11.149 (df = 64) 12.437 (df = 52) 0.447 (df = 5) F Statistic 701.314*** (df = 1; 626) 22.952*** (df = 3; 64) 16.092*** (df = 3; 52) 2.143 (df = 1; 5) Note: *p<0.1; **p<0.05; ***p<0.01