INVESTIGATING SPATIA L DYNAMICS OF ZOONOS ES BETWEEN ANIMAL AN D HUMAN POPULATIONS: A ONE HEALTH PERSPECT IVE By SHELDON GEORGE BRUNO WAUGH A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 201 8
201 8 Sheldon George Bruno Waugh
4 ACKNOWLEDGMENTS I thank the chair and members of my supervisory committee for their mentoring and I thank my parents and my wife for my unwavering support. I would also like to thank my friends and peers for their support, friendship and genuine care of myself and my goals of obtaining a PhD.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 LIST OF ABBREVIATIONS ................................ ................................ ........................... 10 ABSTRA CT ................................ ................................ ................................ ................... 11 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 13 Overview of the Dissertation ................................ ................................ ................... 13 Burden of Zoonoses ................................ ................................ ............................... 13 One Health Methodology and Zoonotic Disease Ecology ................................ ....... 14 Zoonotic Disease Surveillance and Control ................................ ............................ 15 Incorporating Spatial Dynamics ................................ ................................ .............. 16 Zoonotic Surveillance Efforts of Companion Animals Within the United States Military ................................ ................................ ................................ ................. 18 Brucella S urveillance and Control Efforts in Southern Kazakhstan ......................... 23 Objectives and Chapters ................................ ................................ ......................... 29 2 VISULAIZING THE OCCURRENCE OF ZOONOTIC DISEASES AMONG MILITARY ASSOCIATED CANINES ................................ ................................ ...... 30 Background ................................ ................................ ................................ ............. 30 Methods ................................ ................................ ................................ .................. 35 ROVR ................................ ................................ ................................ ............... 35 Study Area ................................ ................................ ................................ ........ 36 Secondary Validati on ................................ ................................ ....................... 37 Regional and GOA/POA Comparisons ................................ ............................. 37 Results ................................ ................................ ................................ .................... 38 Total Population and Encounter Comparison ................................ ................... 38 Secondary Validation ................................ ................................ ....................... 38 Regional and GOA/POA Comparisons ................................ ............................. 39 Discussion ................................ ................................ ................................ .............. 40 Summary ................................ ................................ ................................ ................ 43 3 BRUCELLOSIS TRANSMISSION BETWEEN HUMANS AND DOMESTICATED LIVESTOCK IN SOUTHERN KAZAKHSTAN: INFERENCES THROUGH MLVA TYPING ................................ ................................ ................................ .................. 52
6 Background ................................ ................................ ................................ ............. 52 Methods ................................ ................................ ................................ .................. 54 Study Area and Data Collection ................................ ................................ ....... 54 MLVA ................................ ................................ ................................ ................ 55 Visualizing and Comparin g Genetic Diversity Between Discrete Geography and Other Characteristics ................................ ................................ .............. 55 Determining Source of Human Infection Through MST Netw ork Creation ....... 56 Results ................................ ................................ ................................ .................... 57 Discussion ................................ ................................ ................................ .............. 58 Summary ................................ ................................ ................................ ................ 64 4 SPATIAL GENOMIC ASSOCIATION OF CO CIRCULATING BRUCELLA STRAINS IN SOUTHERN KAZAKHSTAN: PHYLOGENETIC INFERENCES USING MLVA DATA ................................ ................................ ............................... 71 Background ................................ ................................ ................................ ............. 71 Methods ................................ ................................ ................................ .................. 75 Study Area and Data Collection ................................ ................................ ....... 75 MLVA ................................ ................................ ................................ ................ 76 Phylogenetic Reconstruction and Comparing Spatial Genomic Association by Phase Collection and Isolate Group ................................ ......................... 77 Phylogenetic reconstruction ................................ ................................ ....... 77 Genotype designati on ................................ ................................ ................ 78 Comparing spatial genomic association by phase collection and isolate group ................................ ................................ ................................ ...... 78 Outcomes of Interest ................................ ................................ ........................ 80 Results ................................ ................................ ................................ .................... 81 Phylogenetic Reconstruction ................................ ................................ ............ 81 Genotype Designation ................................ ................................ ...................... 81 Comparing Spatial Genomic Association by Phase Collection and Isolate Group ................................ ................................ ................................ ............ 82 Village comparisons ................................ ................................ ................... 82 Tau estimation and comparison ................................ ................................ 82 Mantel tests ................................ ................................ ................................ 82 Discussion ................................ ................................ ................................ .............. 83 Summary ................................ ................................ ................................ ................ 85 5 CONCLUSIONS AND FU TURE DIRECTIONS ................................ ...................... 93 Overall Conclusions ................................ ................................ ................................ 93 Future Directions ................................ ................................ ................................ .... 95 APPENDIX A SUPPLEMENTARY TABLES AND FIGURES ................................ ........................ 96 LIST OF REFERENCES ................................ ................................ ............................. 101
7 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 109
8 LIST OF TABLES Table P age 2 1 Encounter querying accuracy stratified by zoonotic VRME. .............................. 49 2 2 Occurrence of zoonotic VRMEs, during 2016, stratified by military associated designation. ................................ ................................ ................................ ........ 50 2 3 Occurrence of zoonotic VRMEs, during 2016, stratified by region. .................... 51 3 1 Pairwise PERMANOVA results ................................ ................................ .......... 68 4 1 Descriptive table of KZ2 surveillance data ................................ ......................... 88 4 2 Stratification of 42 designated clusters by isolate group ................................ .... 90 A 1 List of 34 veterinary reportable medical events (VRMEs) used in analyses ...... 96
9 LIST OF FIGURES Figure P age 2 1 Spatial population breakdown of all veterinary facilities. ................................ ..... 44 2 2 Encounter flowchart from estimated popula tion of US dogs to ROVR queries .. 45 2 3 Heatmaps of 2016 zoonotic VRMEs Counts ................................ ...................... 46 2 4 Prevalence of the available zoonotic VRMEs ................................ ..................... 48 3 1 Methodological flowchart for this study. ................................ .............................. 66 3 2 Non Metric Multidimensional Scaling (NMDS) plots of 522 provided KZ2 isolates ................................ ................................ ................................ ............... 67 3 3 Analysis of Brucella spp. MLVA strain type clustering and MST network construction using PhyloVizv2.0. ................................ ................................ ........ 69 4 1 Graphical representation of study area and isolate distribution. ......................... 87 4 2 Methodological flowchart for this study. ................................ .............................. 88 4 3 The phylogenetic and compositional depiction of 42 designated genotypes. ..... 89 4 4 Within/Between village designations violin plots. ................................ ................ 91 4 5 Tau estimation stratified by isolate groups. ................................ ........................ 92 A 1 Phylogenetic reconstruction using alternative tree building methods. ................ 97 A 2 Phylogenetic reconstruction using a maximum likelihood approach ................... 98 A 3 Mantel spatial correlograms ................................ ................................ ................ 99 A 4 Bray Curtis dissimilarity violin plots ................................ ................................ .. 100
10 LIST OF ABBREVIATIONS EID Emerging Infectious Disease GOA Government Owned Animals MLVA Multiple Locus Variable Number Tandem Repeat (VNTR) Analysis MWD Military Working Dogs POA Privately Owned Animals VNTR Variable Number Tandem Repeat VTF Veterinary Treatment Facility
11 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy INVESTIGATING SPATIA L DYNAMICS OF ZOONOS ES BETWEEN ANIMAL AND HUMAN POPULATIONS: A ONE HEALTH P ERSPECTIVE By Sheldon George Bruno Waugh May 2018 Chair: Robert L Cook Major: Epidemiology With an increasing emerging zoonotic disease burden leading to novel problem solving, the One Health approach is a burgeoning approach to ensure the well being of people, animals, and the environment through collaborative problem solving locally, nationa lly, and globally. The ability to solve complex infectious disease related questions has significantly improved due to the development of this approach that intersects numerous paradigms of health. Zoonotic disease surveillance can be used to evaluate the current trends of disease burden, to equip organizations with precise information, enacting concentrated interventions, or to focus on factors with the highest payoff. By incorporating spatial aspects in surveillance and control programs, we increase the c ollaborative power that a One Health approach brings to the table. We investigated the spatial dynamics, by utilizing two separate zoonotic surveillance efforts to associate zoonotic diseases in animals with the human disease incidence. The studies investi gated the spatial dynamics of two distinct populations with a high propensity of zoonotic transmission dynamics between animals and humans. The first study investigated the zoonotic disease occurrence of military associated animals within health infrastructure. The second and third study investigated the
12 occurrence, Spatial Genomic association, and transmission and of co circulating Brucella spp among domesticated livestock and humans.
13 CHAPTER 1 INTRODUCTION Overview of the Dissertation Zoonotic disease surveillance is used to evaluate the current trends of disease burden, to equip organizations with precise information, enacting concen trated interventions, or focusing on factors with the highest payoff. Spillover is defined as the transmission of infectious agents from reservoir animal populations (often domesticated species) to sympatric (occurring in the sa me geographical area) wildlife The dynamics of zoonotic spillover are related to disease ecology in that they a re r elated to the pathogen, host and the environmental factors. Zoonotic disease surveillance and spillover as critical features to understand and prevent/eliminate the spread of infectious diseases from animals to humans. An importan t One Health related contribution to zoonotic disease surveillance and control is the addition of spatial dynamics using spillover I will discuss the main aspects of zoonotic disease and disease burden, zoonotic disease surveillance and control programs, spillover and spatial dynamics and describe two examples of how the incorporation of spillover and spatial dynamics can yield a One Health approach, improving these two distinct surveillance initiatives. Burden of Zoonoses Zoonotic diseases continue to pl ace a substantial burden on global health, through human and animal populations. An updated literature survey identified out of 1,407 recog nized species of human pathogens, 58% of them are zoonotic 1 Over one billion cases of human zoonotic disease are estimated to occur annu ally, resulting in hundreds of m illions of dollars in economic losses 2,3 Most of emerging infectious disease events (EIDs) have wildlife origins and require the need for a deeper
14 understanding of transmission between human and animal populations Jones et al. analyzed a database of 335 EID events betw een 1940 and 2004 and demonstrated temporal and environmental patterns associated with increasing occurrence 4 They witnessed global trends showing an increasing contribution of zoonotic diseases (both wildlife and non wildlife related) present in the overall increasing trend of EIDs by decade 4 One Health Methodology and Zoonotic Disease Ecology The One Health approach is a methodology to ensure the well being of people, animals, and the environment t hrough collaborative problem solving locally, nationally, and globally 5,6 This approach interse cts anthropogenic, e nvironme ntal and biological/v eterinary dimensions to observe disease outbreaks and spillover in human and animal populations. The ability to solve complex infectious disease related questions has significantly improved due to the develo pment of this interdisciplinary and integrative approach that intersects various paradigms of health ecology, and biology. The complexity of zoonotic disease transmission and spread necessitates a collaborative approach such as the One Health approach, among professionals from multiple disciplines such as veterinary sciences and disease ecology While these dimensions remain stationary the con tributions to each dimension differ depending on each environment and/or population For example, the conceptual model for the zoonotic disease dynamics in military associated animals will include significant contributions from the environment with exposure to disease carrying mosquito and tick species, like Lyme disease, and behavioral contributions in te rms of occupational roles of military working dogs, such as exposure to zoonoses associated with deployed environments or close living situations, like Giardiasis.
15 Zoonotic Disease Surveillance and Control Ecological approaches are methods that address the complex transactions between people and their environment They are used in conjunction with zoonotic disease surveillance to observe disease burden in populations and locate possible factors and behaviors that could contribute to the distribution and spread of disease Surveillance measures c an be used to evaluate the current trends of disease burden, to equip organizations with precise information, enacting concentrated interventions, or focusing on factors with the highest payoff. Zoonotic disease s urveillance is defined as t he ongoing systematic and timely collection, analysis, interpretation, and dissemination of information about the occurrence, distribution, and determinants of diseases transmitted between humans and animals. 7 Zoonotic disease surveillance primarily focuses on clinical diagnosis followed by a confirmatory test or assay. It i s optimal when used to plan, implement, and evaluate responses to reduce infectious disease morbidity and mortality in human and animal populations through a functionally integrated human and animal health system similar to a One Health methodology 7 Additionally, the identification of risk factors leads to modeling approaches that estimate outbreak trajectories or predict the effectiveness of possible interventions to advise public health professionals on how to best re allocate resources to mitigate the impact of an outbreak or to determine the best cost effective approaches to control potential en demic disease spread. Forecasts and estimates of intervention effectiveness using surveillance data are useful in they provide preliminary guidance to public health authorities on risk factors and possible control efforts. Furthermore, the se results can se rve to generate hypotheses and improve further efforts on data collection, increase the focus on
16 transmission dynamics and control mechanisms of zoonotic diseases in populations. For example, locating the zoonotic disease distribution and burden in popula tions with limited appropriated funds, such as the Army Veterinary Service or in a Low Middle Income Country (LMIC) such as Kazakhstan, generates essential data for the generation of useful hypotheses and the eventual concentration of interventional and fi nancial efforts in order efficiently distribute resources related to disease control and prevention. Disease control programs are often established with the aim of eventual eradication of agents at a country, zone or an other administrative level While thi s approach is desirable, the needs of stakeholders may require a broader range of outcomes 8 such as the correct disease prevalence, in the affected area. Additionally, the program should provide detailed information on: The disea se situation Disease impacts (animal and public health, food safety, food security, biodiversity and socioeconomic impact) and how these are distributed among stakeholders Identity level of interest and involvement of stakeholders 8 One of the main goals of zoonotic disease surveillance and control programs is to implement interventions that will control the size and geographic scope of the outbreak and to minimize morbidity, mortality, and economic losses in both human and animal populations 7 T o control the geographic scope of possible outbreaks, it is necessary to incorporate spatial dynamics and analyses to locate hotspots and identify significant areas of disease prevalence and/or incidence. Incorporating Spatial Dynamics Spatial Epidemiology is defined as the description and analysis of geographic variations in disease concerning demographic, environmental, behavioral,
17 socioeconomic, genetic, and infectious risk factors 9 The use of spatial epidemiology allows for the inclusion of complex epidemiological factors, normally related to zoonotic disease s such as multiple hosts, routes of transmission and other wildlife factors One of the main contributors of zoonotic transmission spatial dynamics involve the concepts of spillover and spillback events. Spillover is defined as the transmission of infectious agents from reservoir animal populations (often domesticated species) to sympatric (occurring in the same geograph ical area) wildlife with spillback defined as the reverse process of spillover The dynamics of zoonotic spillover are related to disease ecology in that they a re r elated to pathogen, host and the environment al factors Spillover events vary among host sp ecies due to differences in pathogen abundance, the opportunity for contact with humans and other behavioral or ecological differences. Previous studies have determined that the spillover events are best explained by a significant relationship between geographical overlap and pathogen sharin g 10,11 The dynamics between pathogen and h ost requir e an overlapping shift to account for the increased interaction required for zoonotic spillover. Spillover directionality is an aspect primarily determined by geographic overlap and pathogen sharing between host species and populations. This significant co ntributor can be calculated by the association between genomic and spatial data. Spatial Genomic associations are correlation s or interactions between genomic data or diversity and geographic factors such as spatial distances or discrete geography. Associa tion s can be quantified as simply as plotting on the X and Y axis 12 subset into discrete geographic categories 12 or utilize correlograms 12,13 or statistical tests 12,14 to compare geographic and genomic matrices. Additionally, interpolation 13 and predictive
18 modeling methods such as Ecological niche modeling (ENM) 15,16 can be used to predict spatial patterns of specific datasets and integrate genotypic data. T o control the geographic scope of possible outbreaks, it is necessary to incorporate spatial dynamics and spatial genomic association analyses to locate hotspots and identify significant areas of disease prevalence and/or incidence The incorporation of sp atial dynamics to zoonotic disease surveillance and control is an approach that can be applied in numerous environments and populations. With this as our main objective, we demonstrate this with two distinct surveillance efforts with a high propensity of z oonotic spillover dynamics between animals and humans Zoonotic Surveillance Efforts of Companion Animals W ith in the United States Military Dogs, cats and other companion animals have played an integral role in ma ny aspects of human life. Human and compan ion animal interactions have a wide range of benefits to human health 17 19 The threat of zoonotic spillover, among companion animals and humans, is compounded by proximity and the numerous amount of diseases o f companion animals share with humans. Many of these zoonoses are spread by direct contact between the species or are vector transmitted (e.g., fleas, ticks, flies, and mosquitoes) diseases for which companion animals might act as reservoirs for the pathog en. From a One Health approach companion animals can serve as sources of zoonotic infections, as intermediate hosts between wildlife reservoirs and humans, or as possibl e as a sentinel or proxy species for emerging disease surveillance 20 As compared to human and livestock diseases, there is a lack of international health agencies coordinating surveillance of diseases in companion animals with surveillance systems exist ing for only a few zoonoses
19 The like lihood of companion animals providing zoonotic diseases a route of transmission to their human handler/ owners is large enough for the Department of Defense (DoD) to devote critical time and resources to investigate the disease transmission dynamic between humans and companion animals. With certain zoonoses exhibiting a differential effect on certain populations, it is crucial to divert appropriate resources and capital for the appropriate diseases in the appropriate populations. For example, Giardiasis is a n intestinal infection that affects both animals and humans, with at risk populations residing in areas with close quarters, such as dog kennels. Giardia reproduces and spreads through the shedding of parasites through the stool and can remain infective in the environment for long periods of time. Infection occurs through the ingestion of parasites in contaminated water, stool, plant material or food. Observational prevalence studies and meta analyses have associated closer living conditions and kennels wit h significantly higher risks of giardia infections 21 24 With Government owned anim al ( GOA ) populations, canines, such as Military Working Dogs ( MWDs ) will have exposure to larger amount s of foreign bodies, via deployments to foreign countries or a higher likelihood of exposure to zoonotic outbreaks due to close living quarters (kennels). With such a differential lifestyle between GOAs and POAs, it is necessary to identify the expected di fferences in zoonotic disease occurrence and burden. Large scale efforts to monitor clinical syndromes and zoonotic diseases in companion animals have been made in the past, in collaboration with private laboratories, veterinary clinics, and government ent ities 25,26 The CALLISTO project, which evaluated the zoonotic disease burden between the compan ion animals and
20 humans, in Europe, utilizing expert opinions and extensive literature reviews to identify and rank zoonotic diseases with the highest risk of spillover events and other ecological characteristics. These collaborative databases not only allo w investigators to increase knowledge about non human reservoirs and inter animal transmission, across a diverse environmental landscape but to characterize and display existing spatiotemporal clusters of disease, in real time. Although the Centers for Dis ease Control and Prevention (CDC) and Banfield Hospital funded a country wide surveillance effort, utilizing animal hospitals and laboratories, named The National Companion Animal Surveillance Program (NCASP) 25 there have been no efforts in the last several years to conduct real time surveillance for syndromes or diseases in companion animals on a national scale in the United States. The surveillance effort became a publically funded effort since 2007 and was limited by confide ntiality issues, delayed dissemination of results and difficulties in managing such large volume s of data 25 leading to its discontinuation 27 blishes a yearly report that states the burden and occurrence of zoonotic diseases of roughly 2.5 million dogs and 505 thousand cats nationwide. The 2016 report ranked Giardia, Kennel Cough, Lyme disease, Canine Parvovirus, Rabies, Canine Distemper, Canin e Influenza and Leptospirosis as the top 8 infectious diseases i n dogs. Of these top eight, six of them are zoonotic with prevalence data (Giardia Infection: 36 cases per 10,000, Kennel Cough : 119 cases per 10,000 and Lyme disease : 65 cases per 10,000) wit h a proclivity for spillover events between dogs and humans. The DoD sought to establish a centralized veterinary zoonotic surveillance system to provide Senior Staff and Commanders with a clear picture of disease burden
21 and highlight areas of increased ri sk. Under Department of Defense Directive 6400.04E, it is the primary responsibility of Army Veterinary Service to establish services to 28 With this responsibility, Army Veterinary Service seeks to centralize and enhance surveillance efforts through the Remote Online Veterinary Record (ROVR), to accurately establish an epidemiological baseline for zoonotic disease burden in military associa ted animal populations. However, prior to 2014, most of records for Government Owned Animals (GOAs: including MWDs and equine animals) encounters were still on paper, including Privately Owned Animals (POAs: Animals owned by active and retired service mem bers) records stored separately using commercial software, limiting the ability of a real time and accurate data repository. The utilization of physical paper records leads to numerous issues such as a potential loss of files during moving, and difficult c ollection and transfer of data Additionally, the amount of time required to read, annotate and analyze records, through physical paper data mining, would restrict the amount of information obtained from these databases Finally, a lack of standardization in disease reporting efforts leads to a systematic differential bias that could skew the conclusions made by analysts, uninformed of the differences in initial data collection. Before the directive and push for an online record, veterinary epidemiologist s primarily investigated retrospective measures of disease burden, including mortality and morbidity trends among provided animal population data, mainly in MWD population. These retrospective analyses are suitable for determining potential risk factors as sociated with certain diseases; however certain trends may not always temporally
22 correlate to current trends leading to the problem of predicting future outbreaks based on variables not temporally associated with the outcome, decreasing the explanato ry pow er of any created predictive model The apparent solution to these pas t limitations was to utilize a Veterinary Online Record System, to standardize the way diseases are reported and annotated and to decrease the time and effort needed to bring the surveil lance effort closer to a real time system. By moving to a centralized online system, The Army Veterinary Service significantly improves the centralization of data allowing for improved estimates of zoonotic disease burden in GOA and POA populations. An on line system would also increase the ability for investigators to locate and establish trends closer to real time; to stay ahead of any potential zoonotic disease outbreak. The Remote Online Veterinary Record (ROVR) database was developed by HEALTHeWV, in 2014, to serve as the primary Electronic Health Record (EHR) for all veterinary services for the US Department of Defense. The database was made for all 150 DoD Veterinary Treatment F acilities (VTFs) worldwide, across three continents 29 The overall purpose and goal for ROVR are to serve GOAs and POAs as the official Ele ctronic Health Record (EHR) for the Army Veterinary Service Additionally, a secondary goal is to act as a real time disease data repository for all GOA and POA EHR data. With military personnel experiencing apparent increased rates of job reducing ailments such as diarrheal and viral disease 30 32 it is essential that the US military should focus on maximiz ing thei r operational potential by minimizing the amount of time personnel are sick to transmissible diseases. By observing the zoonotic disease burden
23 in POAs and GOAs, public health investigators can hone in on what diseases are at greatest risk of being spread from animal to human between these two populations. A n online record system allows investigators to answer new questions, primarily if companion animals, owned by military personnel could serve as possible sentinels for following outbreaks of human disease from animal transmission. Sentinel animals can serve as precursor snapshot of a population closely associated with the population of interest, which are humans in our case. Additionally, regional differences in disease occurrence allow for the integratio n of geographic data into the overall surveillance effort to investigate spatial variation in disease occurrence. Following this, we take a look at a zoonotic surveillance effort that primarily focuses on a molecular epidemiological ( investigates the contr ibution of genetic and environmental risk factors based on molecular biology) approach. Brucella S urveillance and C ontrol E fforts in Southern Kazakhstan Brucellosis is a chronic granulomatous infection caused by intracellular bacteria 33 The Brucella genera are best identified by species and subspecies associated with different hosts. The organism is known for environmental persistence which includes resistance to drying, temperature, pH, humidity 33 The disease is responsible for a considerable amount of clinical morbidity as well as a substantial loss of productivity in domesticated livestock within developing world 34 The most common clinical signs of Brucella infection in humans, are a chronic, low grade fever, followed by articular, muscular, and back pain resulting from pathogen induced inflammation in the joints and vertebrae 35 The Central Asian countries of Kazakh stan, Uzbekistan, Kyrgyzstan, Tajikistan, Turkmenistan, and Afghanistan all are former republics of the Soviet Union (USSR)
24 These countries make up a regional hotspot that pose a significant public health and a veterinary threat to the involved and surrou nding countries 36 Brucellosis control and surveillance are vastly different in these countries, compared to advanced countries, primarily due to the developing nature of the nations. This developing nature of these countries le ads to a weakened central government and loss of institutional command and control. After the fall of the Soviet Union in 1991, collective farms run by the USSR were privatized and broken up, leading to fragmentation of the state agricultural system 37 This decentralization of government along with the fall of the Soviet Union lead to an overall lack of government oversight regarding control and eradication programs, resulting from a lack of monetary control to sustain vaccina tion and eradication programs 38,39 Kazakhstan, for example, has an uncontrolled trade system of livestock, to and from the country, directly contributing to the widespread incidence of brucellosis found, in the country. From 2006 to 2013 the incidence of human Brucellosis has decreased from 17.5 to 8.49 cases per 100,000 persons respectively 40 Even though the decrease in incidence is significant, the country still poses a significant public health and veterinary threat to the involved and surrounding countries. Additionally, Kazakhstan and Central Asia ha ve multiple Brucella spp. circulating s imultaneously, which adds additional uncertainty to the overall picture of outbreaks and Brucella spp spatiotemporal distribution The lifestyles will be more likely to include more pastoral or agricultural related work and occupations. With 60% of the po pulation living in rural areas and agriculture accounting for over 45% of people employed, Central Asian citizens are more likely to be in contact with infected animals (or animals or wildlife in general),
25 increasing the overall risk of animal to human tra nsmission 39 Due to this uncertainty, Central Asia observes an expanded group of potential transmitters to livestock and humans, increasing the risk of potential spillover events in Brucella spp. and increasing overall burden of Brucella in livestock and humans. One of the main ecological factors contributing to the spillover and transmission of Brucella is a spatiotemporal overlap of agricultural areas, human populations, and the grazing or migratory patterns of wildlife known to be potential hosts for Brucella infection. Spatiotemporal overlap is considered to be an event in which feature For livestock owners and producers in Kazakhstan, potential spatiotemporal overlap, such as the seasonal grazing patterns of potential wild hosts, such as maral deer, mountain sheep, mountain goats, roe deer, and Saiga antelope 41 is taken into consideration Potential interventions such as delayed migration could lea d to an overall decrease in the spatiotemporal overlap between livestock grazing and Kazakh wildlife migratory, su mmering or overwintering areas 41 These spillover events have implications in human populations, in that the major ity of human related Brucella spread occurs from exposure to infected livestock 42 One of the strongest tools to control spatiotemporal overlap is to util ize a zoonotic surveillance program to obtain a clear picture of diseas e occurrence within the study area. For countries that track zoonotic diseases that have reached endemic stability, li ke Kazakhstan, the requirements and considerations for surveillance shifts to account for the relatively stable epidemiological outlook. For many developing countries with endemic zoonotic diseases, the majority of surveillance programs primarily lean on serological testing. The isolation and identification of zoonotic diseases from infected
26 animals are critical if specific interventions mu st be implemented Indeed, besides providing material for performing molecular epidemiology and documenting transmission patterns, it is essential to investigate the possibility of undiscovered spillover in reservoir hosts. In these situations, transmissio n investigations are especially important two fold by locating origins of transmission and incorporating a One Health approach by integrating animal and human data to discover the directionality of disease transmission. To identify sources of transmission brucellosis surveillance and control efforts in Central Asia, require s the detection, identification and eventual quarantine or culling of infected livestock 43,44 For Lower Middle Income Countries (LMICs), like countries in Central Asia, the selection of an appropriate surveillance system requires multiple considerations, the weighing of multiple advantages and disadvantages (tools, software, 45 For K azakhstan, Brucellosis surveillance and cont rol efforts focus on investigating livestock and human populations to identify sources of transmission, and to stop the eventual spread and possible spillover events. A crucial factor is the ability to correctl y identify, classify and group Brucella isolates, collected from livestock and humans. For an area like Kazakhstan with multiple co circulating Brucella spp. 46 48 it is essential to correctly determine the pr oper Brucella spp, for outbreak tracing Primarily, this approach leans on the use of Whole Genome Sequencing to identify Single Nucleotide Polymorphisms (SNPs), create phylogenetic trees, observe genetic diversity, and illuminate possible discrete separat ion, by country or city for example. WGS, while on the overall financial decrease, is still a considerable
27 financial investment in necessary reagents and computing power for any bioinformatics 49 Additionally, the genetic similarity between Brucella spp is exceptionally high, where an approach such as WGS may not lead to clear taxonomic classification. Therefore, a cheaper and more parsimonious alternative to genotypic identification is necessa ry for this situation. The use of Multiple Locus Variable N umber Tandem Repeat (VNTR) Analysis (MLVA), serves an optimal methodological intermediary for taxonomic classification, classification, and comparison 50 VNTRs are shor t nucleotide sequences that vary in re pea t numbers in bacterial genomes. They are thought to arise through DNA strand slippage during replication and are of unknown function 51 MLVA targets tandemly repeated DNA regions that h ave high speed molecular clocks, like certain areas of Brucella spp. genomes The addition or deletion of repeat units reflecting either slipped strand nucleotide mispairing during replication or unequal crossover events results in a high rate of mutation at these loci 52 Effective MLVA arrays group samples into meaningful genetic groups that reflect evolutionary relationships (more stable, lower diversity markers), while simultaneously permitting high levels of strain resolution (high diversity markers). Numerous MLVA assays have b een developed for numerous bacteria l genera and species including B. anthracis 53 and Brucella spp 52,54 56 For epidemiological tracing markers that accurately reflect broad evolutionary relationships are useful for comparing the genetic similarity of an isolate to other isolates, on a regional or global s cale, whereas high resolution markers are valuable for detailed tracing of smaller. As a result, this approach has proven particularly useful as a tool for strain discrimination in bacterial species with little genomic variation.
28 These systems capitalize o n diversity at specific places on the genome, or loci that are prone to mutations which result in varying alleles (variants on a locus). Some alleles may appear at random in the population and are perpetuated without selection because they do not affect th e function or appearance of the organism. Low computational cost tree building algorithms like Neighbor Joining or Unweighted Pair Group Method with Arithmetic Mean (UPGMA) hierarchical clustering allows for researchers to quickly delineate Brucella spp. a nd observe finer scaled genetic diversity within spp Until recently, it has been common knowledge that specific species of the Brucella genera only affected (symptomatically) certain species of animal hosts which in turn establish specific chains of trans mission. For example, B. melitensis is a species primarily found in sheep and goats due to the symptomatic effects, but can spread asymptomatically through cattle, and propagated back to goat and sheep populations. The origin and eventual management of br ucellosis outbreaks remain uncertain due to a lack of detailed data on Brucella transmission dynamics in Southern Kazakhstan. This lack of detailed data is primarily driven by the lack of sensitivity in current molecular diagnostic tools and uneven samplin g in current surveillance programs. The KZ surveillance program represents one of the largest passive surveillance efforts in Kazakhstan, using isolates collected by the Kazak Republican Sanitary Epidemiological Station (RSES) and the Kazak National Refer ence Veterinary Center (NRVC) over a period of 4 years, split up between two separate phases. Phase one was from 2007 2008, and phase two was from 2012 to 2013.
29 Objectives and Chapters An important One Health related contribution to zoonotic disease survei llance and control is the addition of spatial data and dynamics. By incorporating spatial aspects in surveillance and control programs, we increase the collaborative power that a One Health approach brings to the table. The three studies presented utilize two separate zoonotic surveillance efforts to associate zoonotic diseases in animals with the human disease incidence. The studies investigated the spati al dynamics of two distinct populations with a high propensity of zoonotic transmission dynami cs between animals and humans. First w e intend to compare the current occurrence of zoonotic disease between POAs and GOAs in 2016. An analysis never conducted before in this population with in the United States Military. Additionally, we will discuss the shortfalls and limitations of the ROVR database, discuss potential improvements, and propose future analyses, collaborative projects, and comparative analyses of the database system to improve the surveillance of zoonotic diseases, in military associated animals. Second w e intend to fill the gaps by investigating the spatial g enomic association, occurrence, and potential trans m ission directionality of co circulating Brucella spp, among human and domesticated livestock, in Southern Kazakhstan.
30 CHAPT ER 2 VISULAIZING THE OCCURRENCE OF ZOONOTIC DISEASES AMONG MILITARY ASSOCIATED CANINES Background There exists a burgeoning awareness among veterinary and human health professionals that disease events in companion animal populations have direct relevance to human health 1 4 Zoonotic Diseases (also known as zoonoses) are caused by infections that are shared between animals and people. T he possibility of animals serving as agents of zoonotic spillover justifies an enhanced focus on establishing a framework to facilitate improved surveillance efforts among animals and humans. Util izing a One Health approach, the intersection between animal and human health requires collaboration, between the two disciplines, to observe measurable associations. Large scale surveillance efforts to monitor clinical syndromes and zoonotic diseases in c ompanion animals have been made in the past, in collaboration with private laboratories, veterinary clinics, and government entities 5,6 These collaborative efforts not only allow investigators to increase knowledge about non human reservoirs and transmission dynamics, across a diverse environmental landscape but to characterize and display spatiotempora l clusters of disease. The Department of Defense (DoD) sought to establish a centralized veterinary zoonotic surveillance system to provide Senior Staff and Commanders with a clear picture of companion animal disease burden and highlight geographic (large and small scale) areas of increased risk of infection and/or possible spillover to humans Under Department of Defense Directive 6400.04E, it is the primary responsibility of the Army emiology,
31 7 With this responsibility, Army Veterinary Service seeks to centralize and enhance surveillance efforts through the Remote Online Veterinary Record (ROVR), to accurately establish an epidemiological baseline for zoonotic disease burden in a military associated animal population including animals that primarily work for the government and animals privately owned by military service members. Government Owned Animals (GOAs: including Mi litary Working Dogs [MWD] and equine animals) have numerous roles within the military and civilian government from Scout/Patrol and Explosive Detection Dogs directly attached with military units, to Sentry, Casualty and Drug detection Dogs that are attache d to law enforcement and civilian divisions of the US Government. The majority of MWDs reside in permanent community kennels either located on military installations or specialized federal buildings. Due to the differences in environmental exposure and dif ferential behavior, it is likely that the zoonotic disease burden among MWDs will differ significantly from that of animals privately owned by active or retired service members (Privately Owned Animals: POAs). For instance POAs might observe higher incide nces of diseases involving mosquito and tick species, like Lyme disease, due to POAs living in more open areas as compared to GOAs that live in kennels Additionally, GOAs (MWDs) adhere to stricter disease preventative measures, such as vaccination or other medicinal methods, leading to a decreased burden of diseases with effective prevent iv e measures. Previous studies have observed exposure to zoonoses of military working dogs possibly associated with deployed environments or close livi ng situations, such as Leishmaniasis 8,9 and Giardiasis 10,11 respectively
32 P rior to 2014, the majority of records for GOA encounters were sti ll primarily utilizing physical records (paper) including POA records stored separately using commercial software, limiting the ability of a real time and readily accessible data repository The utilization of physical paper records leads to numerous issues such as a potential loss of files during mov ing, and a difficulty of transferring certain types of data. Additionally, the amount of time required to read, annotate and analyze records, through physical paper data mining, restricts the amount of information obtained from these databases in an effici ent manner Finally, a lack of standardization in disease reporting efforts from providers using free text diagnoses to undeveloped case definitions, lead ing to a systematic bias based on the difference in installation, state or regional reporting polici es that could skew the conclusions made by analysts, uninformed of the differences in data collection. Before the directive and push for an online record, veterinary epidemiologists primarily investigated retrospective measures of disease burden, includin g mortality and morbidity trends among provided population data, mainly in the MWD population. For MWDs, the primary objective of these retrospective studies w as to describe the prevalence of diseases, in informative environments such as combat zones and d eployed environments. Previous studies, conduct these analyses with encounter data with estimated time differences of 3 4 years from data to publication 8,9,12,13 These retrospective analyses are suitable for determining potential risk factors associated with certain diseases; however certain trends may not always temporally correlate to cu rrent trends This lea ds to the problem of predicting future outbreaks based on variables not
3 3 temporally associated with the outcome, decreasing the explanatory power of any resultant predictive model. The perceived solution to these historical limitations was to use the new Veterinary Online Record System to standardize the way diseases are reported and annotated and to decrease the time and effort needed to bring the surveillance effort closer to a real time system evaluating outbreaks with c urrent data By moving to a centralized online system, the Army Veterinary Service sought to improve the centralization of data allowing for improved estimates of zoonotic disease burden in GOA and POA populations. Banfield Pet Hospital a US based chain of veterinary clinics with over 900 locations produces a yearly report detailing the current burden of infectious diseases in common pet species, including Cats and Dogs. Banfield uses its proprietary data/electronic medical records system, to collect da ta from every pet cared for in Banfield hospitals. Information is updated daily to the medical database and common and medically important diagnoses affecting dogs and c ats in the United States, according to their age, breed and geographical location. Additionally, this report also contains details on the most common (or medically important to the burden on a The l ist represents the Top 8 diseases found in Dogs, with the diseases highlighted in bold being zoonoses with an elevated risk of spillover to humans 14 Giardia Infection Kennel Cough Lyme Disease Canine Parvovirus Rabies Canin e Distemper
34 Canine Influenza Leptospirosis Additionally, The Companion Animal Parasite Council (CAPC) an independent council of veterinarians, veterinary parasitologists, and other animal health care professionals develops monthly and yearly US and Cana dian maps of the estimated prevalence of parasitic, viral, tick borne, and intestinal parasitic diseases 15 The Army Veterinary Service lacks summar ies like this, investigating if military associated animals are a novel population Additionally, we want to compute surveillance efforts of observing zoonotic diseases, specifically. Additionally, a n online system would also increase the ability for investigators to locate and e stablish trends with a n easily accessible database, closer to real time; potentially stay ing ahead of any zoonotic disease outbreak. A n online record system allows investigators to answer new questions, for example, if companion animals owned by military p ersonnel could serve as possible sentinels for later outbreaks of human disease from animal transmission. By utilizing summarized data or quickly analyzing real time information, the ROVR possible setbacks and restrictions. Geographical scope and differentiation are vital for zoonotic surveillance programs to identify large and small scale hotspots, to concentrate interventions to areas in greatest need. In addition to locating areas of greatest risk and burden, it is also important to obtain an overall scope of zoonoses to determine whether the diseases are differ the difference in installation, state or regional reporting p olicies, we expect to find significant
35 differences between regions and possible spatial correlations based on states and military installation. It is essential that military personnel maximize their operational potential by minimizing the amount of time p ersonnel are sick from transmissible diseases. By observing the zoonotic disease burden in POAs and GOAs, public health professionals can hone in on what diseases are at greatest risk of being spread from animal to human and where they are or likely to sp read. Additionally, regional differences in policy and disease burden necessitate the integration of geographic data into the overall surveillance effort to investigate spatial variation in disease occurrence across the US operations The primary objective of this study is to describe the current occurrence (or frequency) of specific zoonotic diseases in POAs and GOAs, and spatiall y compare the differences of zoonotic Veterinary Reportable Medical Event ( VRME ) occurrence by military installation, State, and Region, in 2016. Additionally, we will discuss the shortfalls and l imitations of the ROVR EHR database and disc uss potential improvements to the EHR system to improve the surveillance of zoonotic diseases, in canines associated with the DoD This study is to describe the current occurrence of specific zoonotic diseases in POAs and GOAs, spatially, in 2016 Methods ROVR The Remote Online Veterinary Record (ROVR) EHR was developed from HEALTHeWV, in 2014, to s erve as the primary Electronic Health Record (EHR) for all veterinary services for the US Department of Defense. The EHR was made for all 1 38 DoD Veterinary facilities worldwide, across three continents. ROVR serves as the
36 official Electronic Health Record (EHR) for GOAs and POAs and acts as a disease data repository for all GOA and POA EHR data. ROVR is an enterprise web based application for worldwide use to support the United States Army Veterinary Services (VS). ROVR supports GOA and POA that use milita ry Veter i nary Treatment Facilities (VTFs) and clinics. ROVR has a single point of entry and fully accessible data repository within a military network and requires a DoD Common Access Card with an authorized certificate to access the system. Encounter D ata Collection Encounters were queried in the ROVR database for the zoonotic diseases listed as Veterinary Reportable Medical Events ( VRMEs ), as specified by the US Army Veterinary Service. The total list of 33 zoonotic diseases (See Appendix) represents the VRMEs seen in GOAs and POAs. This query covered all of 2016 (January 1 st 2016 December 31 st 2016) including information about address, military installation and VTF). Encounters were queried by searching for each relating to animal zoonosis 16 The data were spatially aggregated to the county level and above for HIPPA compliance. Even though the data primarily deals with animal health, HIPPA was complied animal ROVR records. Study Area The sc ope of ROVR includes all military Veterinary Treatment Facilities located within four regions located worldwide. The Atlantic, Central, Europe and Pacific regions
37 encompass numerous countries including The United States, The United Kingdom, Germany Japan, Korea, Italy, Spain, Middle East (Bahrain), Guam and, Turkey. Secondary Validation A secondary validation was necessary due to the unknown reliability of the databases querying system, and to gauge the overall accuracy of the EHR for future improvements and reports. We validated all encounters within individual veterinary medical records, inspecting the encounter date and determining if the d iagnosis date resides in 2016. We represented the correct querying percentage with tables stratified by VRMEs. A co rrect querying percentage will be calculated by dividing the correct number of encounters, after secondary validation by the total amount of encounters pulled from the initial query, by VRME. Regional and GOA/POA Comparisons Comparisons of zoonotic encoun ter occurrence between GOAs and POAs within the ROVR database were assessed by the X 2 test of homogeneity Additionally, comparisons were made between permanent regions. The analysis allowed for the discovery of significant differences between the two mili tary associated populations (GOAs and POAs) and the observation of large scale spatial heterogeneity between Veterinary Reportable Medical Events ( VRMEs ) Prevalence estimates were created using population estimates obtained from the ROVR database, by utilizing the following equation. ROVR yearly total unique canine visits will be used as proxies for true GOA and POA populations, for 2016. Unique canine visi ts were defined as unique visits made by canines in and logged in the ROVR system.
38 To compare the occurrence of VRMEs by military installation and state heatmaps depicting the yearly occurrence of VRMEs are provided utilizing occurrence data and clustered using Ward Hierarchical Clustering to observe similar burdens by State and Military Installation. Clustering techniques are used to observe installations and states that act similarly regarding standardized occurrence (frequency) of zoonotic VRMEs. Results Total Population and Encounter Comparison Figure 2 1 shows the geographical location breakdown of all DoD VTFs The majority of the veterinary facilities are in the Atlantic permanent region with a high density of facilities located in the Northeastern and Southeastern United States. Secondary Validation Of the 33 reportable zoonotic diseases, 12 were present after querying the corresponding SNOMED codes. Of these 12 diseases, only nine were present in the database, after secondary validation for 2016. The diseases were Ancylostomiasis (Hookworm), Coccidioidomycosis, Dermatophytosis (Ringworm), Giardiasis, Le ishmaniasis, Leptospirosis, Lyme Bor re liosis Rocky Mountain Spotted Fever (RMSF) and Toxocariasis (Roundworm). Ancylostomiasis was the most occurring (largest frequency) zoonotic disease, making up roughly 46 percent of the total observed zoonotic disease burden in 2016 dogs found to be infected with a zoonotic V RME The total flowchart of our study size is depicted in Figure 2 2. Of the 166,227 total dogs who visited a VTF, 1,511 encounters met our query requirements. Of those 1,511 encounters, 515 encounters remained for further analysis, after secondary validation, where 471of these were from canines.
39 Table 2 1 d epicts the querying and record verification accuracy stratified by VRME. Overall, the total correct querying percentage after secondary validation was 35 percent, with three VRMEs being eliminated, from further analysis, due to incorrect data. The 65 perce nt of the incorrect data consisted of encounter dates not occurring during 2016 or incomplete dates. This inconsistency in diagnosis dates is primarily due to the querying system of the ROVR database collecting encounters outside of the time related search filters. Regional and GOA/ POA Comparisons Table 2 2 depicts the occurrence (frequency) of all zoonotic VRMEs, during 2016, stratified by canine military association. We observe a significant differential occurrence of Giardia (p > 0.001) and Ancylostomi asis (p > 0.001) between GOAs and POAs (57.6% vs 7.5% and 9.1% vs 50.5%, respectfully ) Table 2 3 depicts the occurrence of all zoonotic VRMEs, during 2016, stratified by Region. An expected spatial observation in this table is the significant amount of Lyme disease in the Atlantic Region with Lyme making up 28.2 percent of all Atlantic Region VRME encounters. Additionally, Toxocar i asis and Hookworm ( Ancylostomiasis ) had significant differential occurrence between region with Toxocar i asis occurring most i n the Pacific region making up 32 percent of all regional zoonotic VRMEs and hookworm occurring in roughly half of all cases in the Central and Atlantic regions. Figure 2 3 is a heatmap of states and military installations with a case breakdown of the n ine zoonotic diseases found in the 2016 breakdown along with a Permanent Regional comparison of counts. The states and installations were ordered using a hierarchal clustering technique grouping like total counts together with other like states or installa tions. Lyme disease, considered to be a regional disease mostly contained in
40 the Northeastern United States, demonstrated spatially related clustering among the northeastern states in the heatmap Figure 2 4 depicts the prevalence of the zoonotic VRMEs, utilizing the occurrence and total population data from the ROVR database, stratified by animal military association. Similar to Table 2 2 w e observe a significant difference in Giardia prevalence between the GOA and POA populations observing a large burden in GOAs. Discussion This study has compared the frequency of reportable zoonotic diseases, across military associated animals and demonstrated the effectiveness of how the implementation and use of a centr al online record could allow investigators, veterinary and public health professionals to effectively track the occurrence of zoonotic diseases among the military associated animal population across the United States. Additionally, we identified pitfalls a nd proposed improvements to the EHR system to progress the surveillance of zoonotic diseases, in military associated animals. Finally we recorded the zoonotic spatial frequency and disease burden providing professionals and investigators the ability to c ompare the occurrence of zoonotic diseases of humans and military associated animals to potentially observe areas of the increased animal to human spillover risk The significantly differing occurrence of zoonotic diseases between GOAs and POAs represent a revealing dynamic of how differing factors such as environment and lifestyle differences affects the overall disease burden, between these populations. For P OAs, the strong occurrence of Hookworm (Ancylostomiasis) is significantly connected and a lack of a systemic preventive intervention against hookworm, present in GOA populations
41 Future analyses could compare global clustering s tructures related to increased risk of transmission among GOAs and POAs confirming the hypothesis of kennels, directly relating to use of kennels for MWDs. Novel methods allow for this risk clustering calculation without knowledge of the underlying populat ion distribution 17 Additionally, we observed significant amounts of Hookworm (Ancylostomiasis) cases in all POAs and GOAs recorded populations. According to national data available by the CAPC diagnostic service, hookworm prevalence emulate the occurrence in the ROVR database with higher risks of infection present in southern states 15 With Ancylostomiasis known to be pathogenic in human populations, military associated animals may serve as a li nk for potential transmission to human populations. The widespread occurrence in POA populations (~ 50.5 % of all zoonotic disease cases in 2016), is primarily attributed to the exposure of hookworm infected soil, primarily enriched in areas frequented by do gs and cats (dog parks for example). Additional fine scaled spatial analyses, over time, would allow investigators, veterinary, and public health professionals to effectively locate hotspots, within space and time, where increased cases of hookworm are pre sent. There are multiple limitations with this study and the ROVR database that substantially limited the amount of data used and the conclusions made. The low secondary querying percentage may yield an inaccurate burden in our observed populations. We an ticipated the low percentage due to a poor querying software, but the percentage will have little to no effect on the actual encounters from 2016. Additionally, list which leads the querying software to select encounters, not in the correct period.
42 Finally, the SNOMED code query method excluded any cases where the provider entered the disease in free text instead of using the SNOMED codes/drop down menu This will lea d to an automatic underreporting of cases due to missing cases with free text diagnoses. Advanced analyses, required to answer more advanced surveillance research questions, requires multiple years of confirmed data, which unfortunately under the current p rotocols of querying and confirming data, could lead to months to years of data collection and quality control. Additionally, providers not utilizing SNOMED codes will also increase the time for a complete dataset with additional time needed for diagnosis list mining. This lag time will drastically increase the wait time for military leaders and public health professionals to receive time sensitive information about possible zoonotic disease outbreaks and decrease the significance of any investigative quest ions asked. A reduction in wait time, due to the improved database and querying structures would improve the operational picture for commanders and senior military personnel and allow investigators to establish a zoonotic animal disease surveillance progr am closer to real time acuity. Currently, the ROVR database is developing an control measures and stronger querying structures, within the ROVR database. Additionally, ser vice wide policies have been drafted to improve provider data entry, to standardize the way encounters are recorded in ROVR. Finally, we utilized ROVR yearly totals as proxies for true totals for GOA and POA populations. This skew the calculated prevalence estimates higher due to the underreporting of military associated dogs. However, the study still has merits in that it i s the first of its kind looking at diagnoses from a central repository on a global scale.
43 Summary In this paper, we have described the current occurrence of specific zoonotic diseases in POAs and GOAs, spatially, in 2016. We have discussed the shortfalls and EHR system imp ro ving the surveillance of zoonotic di seases in military associated animals
44 Tables and Figures F igure 2 1 Spatial population b r eakdown of all veterinary f acilities.
45 A B C Figure 2 2. Encounter f lowchart from estimated population of US dogs to ROVR queries. A) Estimated US military associated dog population to validated ROVR encounters. B) ROVR encounters undergoes secondary validation stratified by GOA or POA. C) ROVR encounters undergoes seconda ry val idation stratified by region
46 A 1 Ancylostomiasis 2 Coccidioidomycosis 3 Dermatophytosis 4 Giardiasis 5 Leishmaniasis 6 Leptospirosis 7 Lyme 8 RMSF 9 Toxocariasis Figure 2 3 H eatmaps of 2016 z oonotic VRME s Counts by A) Military Installation and B) State Military Installations and States were clustered using the Ward agglomerative algorithm on Euclidean distances demonstrated with dendograms.
47 B Figure 2 3. Continued
48 Figure 2 4. P revalence of th e available zoonotic VRMEs, utilizing the occurrence and total population data from the ROVR database, stratified by animal military association and Region.
49 Table 2 1. Encounter querying a cc uracy stratified by z oonotic VRME. Zoonotic VRME (2016) Total Encounters from Query Encounters not within 2016 Correct Encounters Unknown Dates Correct Querying Percentage Ancylostomiasis 506 217 246 43 49% Campylobacteriosis 2 1 0 1 0% Coccidioidomycosis 5 1 2 2 40% Cryptosporidiosis 2 1 0 1 0% Dermatophytosis 152 75 67 10 44% Giardiasis 348 280 57 11 16% Leishmaniasis 8 3 5 0 63% Leptospirosis 7 3 1 3 14% Lyme 311 178 95 38 31% Rabies 3 3 0 0 0% Rocky Mountain Spotted Fever (RMSF) 16 11 3 2 19% Toxocariasis 151 75 60 16 40% TOTAL 1511 848 536 127 35%
50 Table 2 2. Occurrence of zoonotic VRMEs, during 2016, stratified by military associated d esignation. GOA POA p n= 33 438 Permanent Region (%) <0.001 Atlantic 6 (18.2) 256 (58.4) Central 15 (45.5) 124 (28.3) Europe 2 (6.1) 18 (4.1) Pacific 10 (30.3) 40 (9.1) Age (mean (sd)) 4.09 (2.42) 2.86 (3.32) 0.037 Sex = F/M (%) 10/23 (30.3/69.7) 214/224 (48.9/51.1) 0.06 Zoonotic VRMEs Ancylostomiasis (%) 3 (9.1) 221 (50.5) <0.001 Coccidioidmycosis (%) 2 (6.1) 0 (0.0) <0.001 Dermatophytosis (%) 3 (9.1) 34 (7.8) 1 Giardiasis (%) 19 (57.6) 33 (7.5) <0.001 Leishmaniasis (%) 0 (0.0) 5 (1.1) 1 Leptospirosis (%) 0 (0.0) 1 (0.2) 1 Lyme (%) 3 (9.1) 92 (21.0) 0.156 RMSF (%) 2 (6.1) 1 (0.2) 0.003 Toxocariasis (%) 1 (3.0) 51 (11.6) 0.217
51 Table 2 3. O ccurrence of zoonotic VRM Es, during 2016, stratified by r egion. Atlantic Central Europe Pacific p n 262 139 20 50 Permanent Region (%) <0.001 Atlantic 262 (100.0) 0 (0.0) 0 (0.0) 0 (0.0) Central 0 (0.0) 139 (100.0) 0 (0.0) 0 (0.0) Europe 0 (0.0) 0 (0.0) 20 (100.0) 0 (0.0) Pacific 0 (0.0) 0 (0.0) 0 (0.0) 50 (100.0) Age(mean (sd)) 3.14 (3.41) 2.25 (2.69) 3.55 (3.79) 3.56 (3.62) 0.023 Sex = F/M (%) 127/135 (48.5/51.5) 71/68 (51.1/48.9) 7/13 (35.0/65.0) 19/31 (38.0/62.0) 0.275 Zoonotic VRMEs Ancylostomiasis (%) 144 (55.0) 65 (46.8) 2 (10.0) 13 (26.0) <0.001 Coccidioidmycosis (%) 0 (0.0) 1 (0.7) 0 (0.0) 1 (2.0) 0.219 Dermatophytosis (%) 14 (5.3) 16 (11.5) 0 (0.0) 7 (14.0) 0.027 Giardiasis (%) 16 (6.1) 22 (15.8) 5 (25.0) 9 (18.0) 0.001 Leishmaniasis (%) 2 (0.8) 0 (0.0) 3 (15.0) 0 (0.0) <0.001 Leptospirosis (%) 0 (0.0) 1 (0.7) 0 (0.0) 0 (0.0) 0.495 Lyme (%) 74 (28.2) 10 (7.2) 7 (35.0) 4 (8.0) <0.001 RMSF (%) 2 (0.8) 1 (0.7) 0 (0.0) 0 (0.0) 0.912 Toxocariasis (%) 10 (3.8) 23 (16.5) 3 (15.0) 16 (32.0) <0.001
52 CHAPTER 3 BRUCELLOSIS TRANSMISSION BETWEEN HUMANS AND DOMESTICATED LIVESTOCK IN SOUTHERN KAZAKHSTAN: INFERENCES THROUGH MLVA TYPING Background Brucellosis, one of the most common zoonotic bacterial diseases worldwide, is a major source of concern due to substantial worldwide economic losses, chronic comorbidities, and severe health risks 36,42,92 The principal causative agent of human brucellosis is the bacterium B. melitensis Kazakhstan represents a significant hotspot posing a public health and veterinary threat to the involved and surrounding countries. From 2006 to 2013, the incidence of human Bru cellosis decreased from 17.5 to 8.49 cases per 100,000, respectively, still representing some of the high est rates in the world. Additionally, within the country exists substantial spatial differentiation between regions, with Almaty, South Kazakhstan and Zhambyl exhibiting particularly high incidence rates 40 Brucellosis control measures commonly associated with vaccination measures (RB51 and S19 for B. abortus and Rev 1 for B. melitensis ), test and culling/quarantine progr ams and movement control. To effectively spread an eradicate h uman b rucellosis l ivestock control is an essential factor due to the lack of an effective human vaccine & a lack of vaccination effectiveness Control measures, such as this, have been found to create abrupt and permanent reductions in brucellosis human infection risk in surrounding lower middle income countries (LMIC ) and confirmed through agent based and mathematical modeling 93 95 (Kracalik et al 2018) Kazakh b rucellosis control and surveillance strategies are vastly different, compared to Western countries, primarily due to
53 Kazakhstan, according to the latest FAO report on Brucellosis Control i n Central Asia has no defined control program and primarily bases its interventions on culling and eradication 43 With the fall of the Soviet Union in 1991, collective farms run by the USSR were privatized and broken up, leading to fragmentation o f the state agricultural system 39,74 leading a current picture of 60% of the population living i n rural areas and agriculture This pastoral environment increases the likelihood to be in contact with infected animals (or animals or wildlife in general), potentially increasing the overall risk of animal to human transmission For an area like Kazakhst an with multiple co circulating strains 96 to control human brucellosis, it is essential to correctly determine the human source of origin as well as determining the overall spatial structure, for effective epidemiological tra cing The ability to investigate and identify transmission d irectionality, in Brucella spp ., is something that requires the use of genomic data to properly relate each sample genotypically Kamath et al. utilize d whole genome sequencing and Bayesian phyl ogenetic reconstruction to relate genetic factors with discrete and continuous geographic characteristics and factors 84 They additionally sought to identify where B. abortus initially came from in the Greater Yellowstone Area (GYE), both in space and time and investigated how the B. abortus changed and diversified, genetically, from initial int roduction to the GYE, to present day. The researchers identified genetically distinct groups of B. abortus samples primarily related to a specific animal group, either Bison, Elk or Domestic Livestock and relate d these distinct groups to discrete geographi c position 84
54 brucellosis outbreaks using a VNTR 9 assay by creating and identifying multiple haplotypes into a median joining network integrating di screte geography to evaluate the spatial burden of the created haplotypes and to locate the area and species of origin 97 al we aim to utilize a Multiple Locus Variable number tandem repeat Analysis (MLVA) ass ay to epidemiologically trace and d etermine the human source of infection and discrete geographic patterns for Southern Kazakh brucellosis as a part of two separate surveillance phases. Methods Study Area and Data Collection The study area includes a maj ority of Southern Kazakhstan and a small part of Northern Kyrgyzstan Kazakhstan has borders with Russia, China, and the Central Asian countries of Kyrgyzstan, Uzbekistan, and Turkmenistan. It is the world's ninth biggest country by size, and it is more th an twice the size of the other Central Asian states combined. Kaza khstan is also the largest landlocked country, worldwide, with the majority of the country exhibiting a low population density, including urban agglomerations in the far northern and southern portions of the country The terrain of Kazakhstan is dominated by steppes and deserts, with mountainous terrain occupying the majority of southeastern Kazakhstan. A total of 517 animal and human isolates were collected by the Kazak Republican Sanitary Epidemiological Station (RSES) and the Kazak National Reference Veterinary Center (NRVC) over a period of 4 years, split up between two separate phases. Phase one was from 2007 2008, and phase two was from 2012 2013. Serum samples were collected along with demographic, year collected, and spatial position determined by a handheld Global Positioning System unit. Figure 3 1 is a visual
55 representation of the isolate spatial distribution across Southern Kazakhstan. The dataset includes co circulating spp. of Brucella residing in Southern Kazakhstan, including B. melitensis and B. a bortus in roughly si x percent of the total isolates, identified using a novel Bruce ladder approach. MLVA We utilized a MLVA 15 assay based o n the pr otocol from Huynh et al. 85 using a modified MLVA 15 protocol utilizing a Beckman Coulter CEQ 8800 Genetic Analysis System The 15 marker primer pairs were divided into two groups: Panel 1 (MLVA 8): eight markers including bruce0 3, bruce07, bruce14, bruce16, bruce20, bruce21, bruce25, and bruce28, and Panel 2 (MLVA 7): seven loci including bruce01, bruce02, bruce27, bruce29, bruce30, bruce31, and bruce33. VNTR locus PCR amplifications aining 0.25 L of forward primer (2.5 M), 0.25 L of reverse primer (2.5 M), 7.5 L Platinum PCR SuperMix, and 1.0 L genomic DNA (10ng/L). Cycling conditions consisted of five minutes of initial denaturation at 95C, followed by 35 cycles of 30 secon ds of denaturation at 95C, 30 seconds of annealing at 60C, and 60 seconds of extension at 72C. PCR was completed with a final extension at 72C for five minutes. VNTR amplicons were sized using the Beckman Coulter CEQ8000 or CEQ8800 capillary electrop horesis instruments, and CEQ fragment analysis software. Visualizing a nd Comparing Genetic Diversity Between Discrete Geography and O ther Characteristics The repeat data for the 517 total isolates, used for the analysis were converted into Bray Curtis Dissimilarity indices. Bray Curtis Dissimilarity i s a statistic used to
56 quantify the compositional dissimilarity between two different sites isolates or samples based on counts at each site depicted by Equation 3 1. (3 1) For our analysis, our VNTR loci represent our sites (n) and our isolates representing the samples (i and i `). Non metric multidim ensional scaling (NMDS), a multivariate non parametric data r eduction technique, was used on the created dissimilarity matrices to visualize and delineate geographical regions of Southern Kazakhstan and other categories such as phase collection and isolate groups Two dimensional NMDS models were fit to the MLVA dissimilarity datasets using the metaMDS algorithm in R 3.3.3 with the R package vegan Additionally, Violin and box plots were utilized to categorize the data and visualize the distribution of gene tic dissimilarity present in each group. The plots were made using R 3.3.3 with the R package vegan and ggplot2 To test for differences in community structure among discrete geography and other categorized groups we used permutational multivariate analys is of variance (PERMANOVA) using the Adonis function in the R package vegan, using R 3.3.3 The procedure tested for differentiation in community structure among our study areas, with additional stratifications such as phase collection, isolate group and o blasts. Determi ning Source of Human Infection T hrough MST Network Creation Network construction of the MLVA isolates was performed with PhyloVizv2.0 and minimum spanning trees (MST) were built with the goeBURST full MST algorithm using global optimal eBURST (goeBURST) and Euclidean distances. MSTs display the
57 genetic diversity within the collections of isolates and relationships between isolates though network connections or relatedness due to discrete characteristics Results Figure 3 1 depicts the NMDS plots stratified by phase collection, isolate group and Oblast. Through the plots, comparing the human and veterinary isolate groups, we observe a smaller dissimilarity present in the human isolates, yielding a smaller average distance in the NMDS an d apparent clustering among the isolate group. Additionally, through visual inspection, it is clearer to observe less of a delineation between oblasts when comparing the human and veterinary isolates groups. This lack of separation is primarily clear and e vident when observing the violin/box plots, depicted in Figure A 2, utilizing the Bray Curtis indices. Observing the dissimilarity within oblasts, we observe a tighter clustering pattern in the Almaty Oblast category in both isolate groups and an overall l ower average dissimilarity when comparing to the other study area oblasts. Additionally, the B. abortus isolates provide a much larger dissimilarity compared to all other isolate categories. Table 3 1 depicts the p airwise PERMANOVA results stratified by C ollection Phase, Isolate Gr oup and Oblast, categorized by a ll isolates, Veterinary Isolates and Human Isolates The results back up within the violin plot and NMDS plot results, showing visual agreeance with our significant results All pairwise compariso ns within the isolate group category were significantly dissimilar from each other (all pairwise comparisons with an adjusted p << 0.001). Among all isolates, there is significant dissimilarity between collection phases (R 2 = 0.016, p<<0.001) and between t he regions of Almaty and Zhambyl (R 2 = 0.043 p<<0.001), Almaty and Kyrgyzstan (R 2 = 0.029
58 p<<0.001), Zhambyl and Kyrgyzstan (R 2 = 0.024 p<<0.001) and South Kazakhstan and Kyrgyzstan (R 2 = 0.143 p<<0.016). Comparing the veterinary and human Isolates gro ups, we observe similar significant pairwise combinations in phase collection (Veterinary: R 2 = 0.02 3 p<<0.001, Human: R 2 = 0.036, p = 0.001) and between the oblasts of Almaty and Zhambyl (Veterinary: R 2 = 0.046 p<<0.001, Human: R 2 = 0.046 p = 0.004). Figure 3 2 depicts the created MST networks. By observing the isolate group categorization, we observe the veterinary isolates have a larger network degree centrality than human isolates (average degree centrality: 2.06 vs 1.86, for veterinary and human g roups, respectively). The increased degree centrality along with clusters of human isolates originating from veterinary isolates supports the directionality of Brucella transmission from animals to humans. A deeper look into the veterinary isolates, depict ed by Figure 3 2C, show that the majority of isolates originated from sheep (96% of all veterinary isolates), supporting the hypothesis that human cases originate from sheep to human spillover. A dditionally the MST identified 58 clonal complex e s (CCs) at the SLV level. All CCs were homogenous concerning isolate group and the five largest CCs (CC3, CC8, CC1, CC2, and CC5) primarily belonging to the Zhambyl oblast (85% percent of top 5 CCs isolates). At the DLV level, we encountered 49 CCs, with the top five CCs with respect to total isolate counts, primarily containing veterinary isolates from Zhambyl (78% of isolates in top 5 CCs isolates). Discussion We characterized the geographic and species dependent structure of Co circulating Brucella spp. isolates co llected as part of a large scale surveillance program in Southern Kazakhstan. The NMDS plots yielded v isually apparent ecological and
59 genetic dissimilarities between phase collec tion, oblast and isolate groups. This result demonstrates a differential effec t in surveillance phases, discrete geography, and host species. PERMANOVA analysis show s a significant dissimilarity between Almaty and Zhambyl Oblasts vs. all Kazakh oblasts were significantly dissimilar with isolates collected from Kyrgyzstan. PERMANOVA results backed up by the violin plot results, showing visual agreeance with significant results. Our MST networks highlight the directionality of human brucellosis spillover being from sheep to human transmission. This transmission follows the majority of research demonstrating a transmission from animals to humans, due to humans being incidental or dead end hosts and the instance of both vertical and horizontal human to human transmission being rare 34,98 The relationships of our identified B. abo rtus strains and other isolates associate the strains with isolates from cattle. This relationship may indicate that the mechanisms driving the evolution of B. a bortus (an isolate common to cattle) and the B. melitensis strains, isolated in cattle may have a common evolutionary thread in the cattle s pecies. Additionally, veterinary isolate groups have highest network degree; revealing the majority of human isolates stem from animal isolates, phylogenetically. Visual clustering within isolate groups also indicate s that the similarity persists between h uman and veterinary samples, indicating either differing evolutionary mechanisms or the human outbreaks stemming from few common isolates. The latter hypothesis is supported by the increased diversity of the veterinary isolates and estimated spillover dire ctionality of animal to human transmission. From our results we observe human brucellosis is primarily driven by B. melitensis solidifying the crucial aspect of controlling human brucellosis is controlling B.
60 melitensis We also see human isolates exhibi t ing lower diversity as compared to our veterinary Limited diversity in human strains demonstrate s that not all human related B. melitensis strains are transmitted equally Possible contributors to this decreased diversity could be external to a common source of infection or e nvironmental pressure s or internal with shifting pathogen fitness with in humans Additionally, populations with decreased biodiversity frequently increase disease transmission 90 Temporally, w e observe a significant decrease in diversity in human B. melitensis isolates. From this, we can say that the spillover events occurring from 2007 2013 are originating from more and more common sources, and since we know from our results that they originat e from sheep, we should anticipate dramatic decreases in human Brucellosis incidence with sheep vaccination and proper culling programs. With a s imilar diversity by region demonstrating a homogeneous diversity, a region wide interventional control effort o n B. melitensis could lead to a far reaching and significant impacts in human brucellosis control decreasing the risk of infection over multiple areas over a brief period In contrast, the amount of h igher diversity present in our veterinary isolates yiel ds a h igher risk of spillover events initially due to a larger source pool 90 With an increased likelihood of spillover event comes an increased likelihood of novel strain development and spillover to humans, due to evolutionary pressures affecting B. melitensis in animals, increasing the increasing pathogen fitness in humans, complicating control measures. With these two varying populations in mind, the effort on controlling the animals must be seen as the approach with the g reatest likelihood to eradicate human is going after sheep and goats via vaccination and culling programs
61 A deeper investigation into transmission capabilities between human and animal population depends on the assumption that brucellosis acts as a pathog en with a density dependent disease transmission structure. With B. melitensis present in sheep, we acknowledge the spread of the pathogen from sheep to sheep through abortion or other infected tissues. A sheep population with increased density will inevit ably lead to increased exposure and transmission. This dynamic works for humans, with respect to increased density, leads to increased exposure to infected sheep and sheep byproducts. Dobson, in 2004, utilized a mathematical model framework to investigate how diversity in species affects diseases with density dependent transmission. Dobson concluded that the density dependent population could amplify transmission and outbreaks when the host species diversity increases 99 In ou r dataset, we did not encounter B. melitensis strains present in other species besides sheep and humans, however, we did observe B. abortus strains present in cattle and sheep. Observing B. increased its species diversity and recent outbreaks in the United States, we should use this as a warning and responding to keep B. melitensis in check, to reduce the risk of further spread. Additionally, Dobson recommended to that to eradicate a disease with this type of transmission, eliminating the host sp ecies (sheep) that dominate the terms of transmission, would lead to the greatest likelihood of pathogen eradication 99 This holds true to our brucellosis environment as well, by hypothesizing our within sheep transmission is far greater than our sheep to human transmission. Additionally, we have observed the directionality of B. melitensis transmission moving from sheep to humans. By identifying sheep as the main culprit of human
62 brucellosis transmission, public health interve ntionists can further concentrate their control efforts to sheep to yield a maximum effect in future human infections. Focusing on discrete geography, we observed Zhambyl represent ed most of the network backbone (oblast isolates with the highest network de gree) possibly revealing the oblast of origin for most B. melitensis isolates. Additionally, Zhambyl was revealed to have most diversity among its veterinary isolates as detailed in the NMDS plots and violin plots. This along with our network results, confirms Zhambyl as a possible region of origin for B. melitensis This combination of info will concentrate control i nterventions focus ing on sheep in Zhambyl for a maximum effect in controlling B. melitensis transmission to humans. Previous research ha s observed three separate transmission dynamics involving B. abortus and B. melitensis in wildlife, livestock and humans. Widespread circulation of B. abortus in wildlife and livestock, with multiple spillover events occurring from bison to elk and elk to livestock (Greater Yellowstone Area) 84 ; a widespread circulation of B. melitensis in sheep, recorded spillover from sheep to cattle and cattle to humans (Spain) 100 102 ; and a combination of these environments with co circulating strains with all these populations occurring simultaneously (Kazakhstan) 46,48,96 While we do have a n overall sample that contains subsets of B. melitensis in humans and sheep and B. abortus in sheep and cattle confirmed via BruceLadder we acknowledge that B. abortus is co circulating on a large scale. Shevtsov et al (2015) already investigated the same area with a large scale approach and observed large amoun ts of B. abortus in cattle around southern Kazakhstan 96 With this combo environment occurring in Kazakhstan with co circulating spp., the likelihood of spillover of B. melitensis and B. abortus into
63 incidental hosts increases dramatically, demonstrated in our research with sheep with B. abortus and cattle with B. melitensis This scenario has already been noted and observed in Kazakhstan, within conventional farming operations with sheep and cattle residing together 103 This increased likelihood is important due to successful infection of incidental hosts could be a sign of possible shifts in pathogen fitness, affecting the way the pathogen evolves and drives the brucellosis virulence in the host possibly leading to future novel host transmission. With all of this known, potential human brucellosis control interventions should primarily focus on sheep Rev 1 vaccination and culling campaigns in Zhambyl O blast Interventions should also include edu cational campaigns focusing on the proper handling, control, cleaning and use of sheep and sheep byproducts (dairy and dairy byproducts), due to the increased risk of transmission through these possible routes and collaboration with farmers to control move ments of sheep during migration, due to the increased risk of spillover due to spatial overlap. This study confirms and recommends with genomic data, what previous studies h ave concluded through time series analysis, agent based and mathematical modeling techniques 93 95 that control programs aimed at livestock leads to significant decreases in human risk of brucellosis infection Strengths of this study include the use one of the largest dat a sets on human and animal isolates present in Southern Kazakhstan. The surveillance effort also represents the largest effort in the area over the four year period. To the best of our knowledge, this study represents the most comprehensive spatial/genetic association analyse s, utilizing three separate methods to locate associations between spatial and genetic
64 distances. The ability to ascertain spatial/genetic association and phylogenetic reconstruction using a MLVA 15 assay allows for a more economical method compared to Bay esian Phylogenetic reconstruction using WGS. Additionally, this inve stgation reprsents one of the first studies to have B. abortus identified in sheep and to investigate the role of these isolates within more traditional B. abortus isolates from cattle. Th ere were several limitations to our study. The main weakness of this study is that the passive surveillance efforts yield a biased isolate distribution, which could be unrepresentative of the true genetic burden of isolates present in Southern Kazakhstan. Additionally, the h igh F scores could be due to uneven sampling ; however PERMANOVA performs the best out of all the dissimilarity statistical tests, due to its robustness 104 MLVA used in outbreak investigations have shown e xceptional merit in tracing pathogens. MLVA assays are primarily designed to have high discriminatory power and proposed as an alternative for genotyping of highly clonal groups of bacteria such as the Brucella genus For future analyses, t he use of WGS could increase the discriminatory power, involving SNP data and a larger part of the genome for analysis However, WGS, while on the overall financial decrease, is still a considerable financial investment in necessary reagents and computing power for any bioinformatics 49 For a n LMIC like Kazakhstan, the widespread use of WGS is a significant financial limit, and the use of MLVA serves as an informative intermediary for outbreak tracing and identification. Summary In conclusion, we utilize d a Multiple Locus Variable number tandem repeat Analysis (MLVA) assay to epidemiologically trace and determine the human source of
65 infection and discrete geographic patterns for Southern Kazakh brucellosis as a part of two separate surveillance phases Our findings concluded and demonstrate d the human health benefits of livestock vaccination and culling programs (primarily in sheep) that reduce spillover of brucellosis to multiple species Additionally, we have successfully integrated discrete g eography identifying Zhambyl as a possible source region of human brucellosis. Controlling brucellosis requires the use of a One Health approach that emphasizes cooperation between human, veterinary, and environmental health practitioners. By integrating s patial data with molecular epidemiological data, we can further concentrate interventions and incorporate novel education campaigns to effectively control human brucellosis We find MLVA data to be informative to successfully integrate discrete geographic and demographic variables with enough definition to associate with genetic diversity.
66 Figure 3 1. Methodological flowchart for this s tudy
67 Figure 3 2. Non Metric Multidimensional Scaling (NMDS) plots of 522 provided KZ2 i solates categorized by all isolates further stratified by A) Isolate Group B ) Collection Phase and C) Region by h uman isolates D ) Collection Phase and E) Region and by Animal isolates F) Collection Phase and G) Region
68 Table 3 1. Pairwise P ERMANOVA results stratified by collection phase, isolate group and region, categorized by all isolates, v eterinary Isolates and human i solates Pairs F. Model R 2 p (Adjusted) sig All Isolates Phase 1 vs 2 8.118 0.016 0.0001 *** Isolate Group Veterinary vs Human 31.494 0.059 0.0002 ** Veterinary vs Veterinary B. a bortus 8.502 0.021 0.0002 ** Human vs Veterinary B. a bortus 14.863 0.105 0.0002 ** Oblast Almaty vs Zhambyl 22.147 0.043 0.0003 ** Almaty vs South Kazakhstan 1.088 0.005 0.335 Almaty vs Kyrgyzstan 6.710 0.029 0.0003 ** Zhambyl vs South Kazakhstan 2.383 0.008 0.06 Zhambyl vs Kyrgyzstan 7.570 0.024 0.0006 ** South Kazakhstan vs Kyrgyzstan 3.989 0.143 0.016 Veterinary Isolates Phase 1 vs 2 9.375 0.023 0.0001 *** Oblast Almaty vs Zhambyl 18.983 0.046 0.0001 *** Human Isolates Phase 1 vs 2 4.371 0.036 0.001 Oblast Almaty vs Zhambyl 4.315 0.046 0.004 Almaty vs South Kazakhstan 1.619 0.028 0.197 Almaty vs Kyrgyzstan 4.413 0.054 0.004 Zhambyl vs South Kazakhstan 1.102 0.029 0.355 Zhambyl vs Kyrgyzstan 5.113 0.082 0.004 South Kazakhstan vs Kyrgyzstan 3.989 0.143 0.017
69 A B Figure 3 3. Analysis of Brucella spp. MLVA strain type clustering and MST network construction using PhyloVizv2.0. Size of circle represents number of isolates of each MLVA type. The shading indicates isolates of A) Isolate Group, B) Oblast, C ) Host Species and D) Dou ble Loci Variant Clonal Complexes.
70 C D Figure 3 3. Continued
71 CHAPTER 4 SPATIAL GENOMIC ASSOCIATION OF CO CIRCULATING BRUCELLA STRAINS IN SOUTHERN KAZAKHSTAN: PHYLOGENETIC INFERENCES USING MLVA DATA Background most prevalent zoonotic disease with a worldwide incidence of almost 500,000 new human cases a year 36 Brucellosis is caused by a chronic granulomatous infection caused by intracellular bacteria. The organism is known for environ mental persistence which includes resistance to drying, temperature, pH, humidity. Significantly higher incidence rates and overall burden in livestock and humans occur in the continent of Asia, specifically Central Asia 36 The disease causes significant clinical morbidity, due to persistent and reoccurring symptoms, as well as a considerable loss of productivity in animal husbandry in the developing world, due to eradication and culling programs 33,70 Transmission of human brucellosis is primarily spread through occupational exposure to infected tissue and dairy products. Spillover events between reservoirs and human hosts are primarily driven by domesticated livestock (cattle, sheep, and goats) through infected meat and unpasteurized dairy products. The Central Asian country of Kazakhstan has faced significant issues prim arily due to its history as a former republic of the Soviet Union. The country represents a significant locus for human Brucellosis incidence as a hyperendemic nation within a global disease hotspot 36 From 2006 to 2013, the inci dence of human Brucellosis has decreased from 17.5 to 8.49 cases per 100,000, respectively 46 Even though we observe a decrease in incidence, the incidence rate and burden represents a national average that is astronomically h igh, compared to other countries. Additionally, we
72 observe substantial spatial differentiation between regions, with Almaty, South Kazakhstan and Zhambyl exhibiting particularly high incidence rates. As a whole, livestock control is essential to brucellos is control in humans due to the lack of an effective human vaccine, a lack of vaccination effectiveness in domesticated livestock 71,72 and the persistence of natural wildlife reservoirs 73 For a country like Kazakhstan, brucellosis control and surveillance measures are vastly different from countries, like the United State nature. Kazakhstan, according to the latest FAO report on Brucellosis Control in Central Asia, has no defined control program and primarily bases its interventions on culling and eradication 43 After the fall of the Soviet Union in 1991, collective farms run by the USSR were privatized and broken up, leading to fragmentation of the state agricultural system and eventual loss of centralized livestock control 39,74 This developing nature of Kazakhstan led to a weakened central government and loss of institutional command and con trol. This decentralization of government along with the fall of the Soviet Union lead to an overall lack of government oversight regarding control and eradication programs, resulting from a lack of monetary control to sustain vaccination and eradication p rograms. Additionally, Kazakhstan has multiple Brucella spp circulating simultaneously, which adds uncertainty and confusion to the overall picture of outbreaks and Brucella spp. spatiotemporal distribution 46 48 The control programs of greatest effectiveness require the use of surveillance to correctly identify and track humans, domesticated livestock and wildlife reservoirs infected with Brucella Properly identifying Brucella isolates allows for researchers and
73 stakeholders to properly trace outbreaks and correctly identify spatiotemporal overlap (areas of increased risk of spillover) between wildlife, potential livestock hosts and human populations. The ability to correctly i dentify, classify, and group isolates Brucella isolates, collected from livestock and humans, primarily requires the use of taxonomic assays and/or genomic sequencing data developed to delineate specific genotypes and identify fine scale changes in genetic diversity. The use of Multiple Locus Variable number Tandem Repeat (VNTR) Analysis (MLVA), serves an optimal methodological intermediary for taxonomic classification, classification, and comparison 50 VNTRs are short nucleotid e sequences that vary in repeat numbers in bacterial genomes. They are thought to arise through DNA strand slippage during replication and are of unknown function 75 MLVA targets tandemly repeated DNA regions that have high sp eed molecular clocks. The addition or deletion of repeat units reflecting either slipped strand nucleotide mispairing during replication or unequal crossover events results in a high rate of mutation at these loci 52 Effective MLVA arrays group samples into meaningful genetic groups that reflect evolutionary relationships (more stable, lower diversity markers), while simultaneously permitting high levels of strain resolution (high diversity markers). The use of MLVA data within phylogenetic analyses is fairly common, producing clusters and genotypes utilizing distance based clustering algorithms, such as Neighbor Joining or Unweighted Pair Group Method with Arithmetic Mean (UPGMA). These traditional methods, in MLVA phylogenetics excel at determining broad clusters among isolates utilizing distance matrices created from a generalized maximum parsimony algorithm 76,77 Novel methods utilize more advanced mathematical models
74 that estimate the evolutionary history using Maximum Likelihood 78 and Bayesian 75,79 phylogenetic reconstruction, that allow for a more detailed integration of metadata, such as demographic or geographical data. Pre vious research has identified geographic variation in Brucella species and genera 46,47,80 84 Spatial Genomic associations are considered to be correlations or interactions between genomic data or diversi ty and geographic factors such as spatial distances or discrete geography. Associations can be quantified as simply as plotting on the X and Y axis 12 subset into discrete geographic categories 12 or utilize correlograms 12,13 or statistical tests 12,14 to compare geographic and genomic matrices. Additionally, interpolation 13 and predictive modeling methods such as Ecological Niche Modeling (ENM) 15,16 can be used to measure and estimate genetic variation o ver various spatial scales. A recent study conducted in the same study area, utilized co circulating B. melitensis and B. abortus strains collected primarily from sheep and cattle and observed differences phylogenetically and used discrete geography to co mpare Kazakhstan isolates with isolates from collected from other outbreaks across the world 47 The analysis revealed genetic homogeneity in seemingly unrelated outbreaks in China, Turkey, and Kazakhstan 47 The studies however, lacked a systematic approach to evaluate the genomic spatial associations in Brucella Additionally, Kazakhstan has lacked a study investigating the spatial dynamics of Brucella genetic diversity, utilizing human and veterin ary samples. This study utilized one of the largest passive surveillance efforts in Kazakhstan, using MLVA and isolate metadata collected by a two phase collaborative human and veterinary surveillance effort. Using a MLVA 15 assay 85
75 previously developed and validated, we will genotypically identify animal and human isolates from two phases of a surveillance effort in Southern Kazakhstan using novel phylogenetic approaches and identify the relationship between spatial and genomi c data. This study aims to identify, characterize and compare the spatial genomic associations of co circulating Brucella spp. on a regional scale, in Southern Kazakhstan, utilizing an established MLVA 15 genomic typing protocol, between collection phase a nd isolate group. Methods Study Area and Data Collection The study area includes a majority of Southern Kazakhstan and a small part of Northern Kyrgyzstan Kazakhstan has borders with Russia, China, and the Central Asian countries of Kyrgyzstan, Uzbekistan, and Turkmenistan. It is the world's ninth biggest country by size, and it is more than twice the size of the other Central Asian states combined. Kaz akhst an is also the largest landlocked country, worldwide, with the majority of the country exhibiting a low population density, including urban agglomerations in the far northern and southern portions of the country The terrain of Kazakhstan is dominated by s teppes and deserts, with mountainous terrain occupying the majority of southeastern Kazakhstan. A total of 584 animal and human isolates were collected by the Kazak Republican Sanitary Epidemiological Station (RSES) and the Kazak National Reference Veterin ary Center (NRVC) over a period of 4 years, split up between two separate phases. Phase one was from 2007 2008, and phase two was from 2012 2013. Serum samples were collected along with demographic, year collected, and spatial position determined by a hand held Global Positioning System unit. Figure 4 1 is a visual
76 representation of the isolate spatial distribution across Southern Kazakhstan. Table 4 1 is a demographic breakdown of our surveillance data. The dataset includes co circulating spp. of Brucella r esiding in Southern Kazakhstan, including B. melitensis and B. a bortus in ro ughly si x percent of the total isolates, identified using a no vel Bruce ladder approach Villages were identified as being groups of points occurring at the same geographic coordin ates. Due to the isolated nature of the country and collection protocol, it can be reasonably assumed that different geographic coordinates represented different villages. MLVA We utilized a MLVA 15 assay based o n the protocol from Huynh et al. 85 using a modified MLVA 15 protocol utilizing a Beckman Coulter CEQ 8800 Genetic Analysis System The 15 marker primer pairs were divided into two groups: Panel 1 (MLVA 8): eight markers including bruce03, bruce07, bruce14, bru ce16, bruce20, bruce21, bruce25, and bruce28, and Panel 2 (MLVA 7): seven loci including bruce01, bruce02, bruce27, bruce29, bruce30, bruce31, and bruce33. VNTR locus PCR amplifications d primer (2.5 M), 0.25 L of reverse primer (2.5 M), 7.5 L Platinum PCR SuperMix, and 1.0 L genomic DNA (10ng/L). Cycling conditions consisted of five minutes of initial denaturation at 95C, followed by 35 cycles of 30 seconds of denaturation at 95 C, 30 seconds of annealing at 60C, and 60 seconds of extension at 72C. PCR was completed with a final extension at 72C for five minutes. VNTR amplicons were sized using the Beckman Coulter CEQ8000 or CEQ8800 capillary electrophoresis instruments, and CEQ fragment analysis software.
77 Phylogenetic Reconstruction and Comparing Spatial Genomic Association by Phase C ollection a nd Isolate Group Figure 4 2 is a flowchart to depict the m ethodological structure of this study To comp are the genetic diversity of Brucella isolate groups with geographic distances genetic distance matrices will be generated from the MLVA repeat data and compared observing the Spatial genomic association Phylogenetic r econstruction We will construct phylogenetic trees using three different phylogeny building approaches: A maximum likelihood phylogenetic tree using a MLVA assay with 15 separate loci 85 with numbers representing repeats. A maximum likelihood phylogeny was under the Lewis MK model for d iscrete morphological character data including the B_melentesis_M16, B_abortus_S19, B_melitensis_biovar_Abortus_2308, B_melitensis_Rev1 and B_abortus_RB51 reference isolates as an outgroup and evaluating statistical support for individual nodes based on 10 0 non parametric bootstraps. A Neighbor Joining tree building Approach A UPGMA tree building and hierarchical clustering approach *Provided in supplementary sections not discussed in dissertation The maximal conditional probability of the data (sequ ences) given a hypothesis and branch lengths). The following equation depicts the likelihood equation used for phylogenetic reconstruction. We will utilize the tree and set of parameters that has the maximum likelihood calculation. The model of evolution equations 4 1 and 4 2, utilized for this study will be the Lewis MK model for discrete morphological character data a generalized version of the Juke and Cantor model. ( 4 1) ( 4 2)
78 The phyl ogenetic reconstruction will be conducted using R 3.3.2 86 and the R package phangorn 87 Genotype d esignation The genotyp es were determined utilizing a methodology based on the partitioning of large scale phylogenies and the inference of transmission clusters. This approach utilizes a depth first search algorithm that automates the partition of phylogenetic trees based on clade reliability and the nth percentile of patristic phylogenetic dis tances Based on referenced article and software recommendations, we utilized a percentile value of 0.05. Each isolate will then be categorized based on the The analysis wa s analyzed with the Java based program, PhyloPart 88 Comparing spatial genomic association by phase collection and isolate g roup We will utilize three distinct methods to compare the spatial genomic association by phase colle ction and isolate group; Within/Between village comparisons of patristic distances, Tau estimation and comparisons, and bootstrapped Mantel tests and correlograms. First, the patristic distances ( the sum of the branch length on a path between a pair of ta xa ) were aggregated by phase collection, village and isolate species and then compared to using violin plots and bootstrapped non parametric t tests, due to the non parametric distribution of the resultant distance matrices, with 1000 permutations. With th e spatial surveillance collection, villages were categorized by isolates with matching GPS coordinates. Second, t o fully explore the spatial patterns of Brucella based on genotypes, we will utilize the Tau Statistic, developed by Lessler et al., to dete rmine the risk of infection
79 statistic is a measure of the relative risk of someone at a particular spatial distance from a case also being a case, versus the risk of anyone in the population being a case. The nove l aspect of this statistic is that the underlying population does not need to be known to effectively obtain a risk estimate This assumption is key to our dataset, due to the differential phased surveillance efforts and the unknown true prevalence of co c irculating Brucella spp. in Southern Kazakhstan. statistic to use information on the infecting pathogen (such as serotype/genotype) to distinguish between pairs of cases that are consistent with coming from the same transmission chain and pairs that are in consistent with coming from the same chain. Equation 4 3 is the adapted formula for tau when the underlying population is unknown; where is an estimator of the odds that a case within the distance range of a case is homotypic (or with the same type or genotype) ( 4 3) statistic can allow for multiple genotypes of the same genotypes. The statistic and risk estimates will be obtained using the geocoded isolates and create d genotypes determined by the phylogenetic trees created previously. We statistic of Brucella isolates categorized by our previously designated statistic was calculated within households (i.e., at 0m) and in a sliding 8km w indow up to the average distance between community households, 347km The analysis will be analyzed with R 3.3.2 86 and the R package IDSpatialStats 68
80 Finally, u sing the grouped genetic and geographic distance matrices, we will observe the associations between the two mediums. Matrix observations are primarily non independent and based on pairwise comparisons. Additionally, geographic sampling points are usually not independently spread and exhibit som e form of spatial autocorrelation. To account for this, we will utilize Mantel testing. Mantel tests are used to test for correlation between distance matrices when the underlying probability distribution of the test statistic is unknown/non normal and whe n dependence is present. Bootstrapped Mantel tests are utilized to overcome the lack of independence by randomly shuffling the values c hosen matrices multiple times, and calculating correlations between the shuffled and original matrices. For the geographic distances, distance matrix. We will calculate the geographic distance in meters between each isolate using the great circle distance measure, tak ing the surface into account. Bootstrapped Mantel tests, with 10,000 permutations, will be conducted stratified by phase and species comparing genetic and geographic distance matrices. Additionally, Mantel correlograms will be created, o bserving the gradual spatial patterns, by the provided great circle distances. All analyses will be analyzed with R 3.3.2 86 and the R package vegan 89 Comparisons between phase collection (phase 1 vs. phase 2) and isolate species (Animal vs. Human) will be made using the Mantel and Tau methods. Outcomes of Interest Our overall outcome of interest will be the measured spatial association between the phylogenetic distances from our constructed trees and the geographic distances obtained from our geocoded isolates. These measures will include violin plots, stratified
81 by village and phas e collection/isolate group, our tau statistic estimations measuring the risk of cases being observed with a given spa tial interval, and the bootstrapped Mantel test with a recorded correlation and significance test value for the dataset at all distances with a Mantel correlogram that will record the correlation at provided distances (every 8km until 350km). Results Phylo genetic Reconstruction Of the 584 isolates, 517 were properly annotated with GPS coordinates. Therefore, the 517 properly annotated isolates with the five reference isolates were used in the phyloge netic reconstructions. Figure A 1 is a phylogenetic tree reconstruction utilizing a n ML approach using the 517 isolates with geographic coordinates. According to the tree, there is distinctive clustering among the human and animal isolates with the isolates identified as B. abortus using the Bruce ladder method. This effect was present in all three tree building approaches (UPGMA and NJ trees in Appendix). Among the collection phase, we observe no distinct grouping between phase 1 and phase 2. Genotype Designation Figure 4 3 represents the phylogenetic and compositional breakdown of the 4 2 d esignated genotypes designated by PhyloPart Table 4 1 stratifies the 4 2 genotypes by isolate groups composition. Figure 4 3 B is a p hylogenetic depiction of the 4 2 genotyp es Isolates were aggregated int o genotypes highlighted in red with unclustered isolates categorized by phase collection and isolate group
82 Comparing Spatial Genomic Association by Phase C ollection a nd Isolate Group Village c omparisons Figure 4 4 demonstrates the comparison of genetic d istances between or within villages stratified by phase collection and isolate species. Violin plots were used to compare and observe the differences in mean genetic distances and the overall distribution of each stratified dataset. Among the between and within village groups, a marked decrease of genetic diversity is present in the between village group, demonstrating a marked similarity in isolate strains within a village. We saw a similar marked decrease when stratified for phase collection. When compar ing between/within village diversity between species, we observed the significant decrease diminish substantially among human isolates and remain in the animal samples Tau e stimation and c omparison Figure 4 5 illustrates the comparisons between the smoothed Tau estimates based on the same spatial scale, between human and animal isolates, using a sliding 8km window, from 0km to the average dist ance between isolates, 347 km. Comparing human to animal isolates, w e found that animal isolates located < 8 km apart were two times more likely (Humans: 3.4 ( 2. 4 4.3 ); Animals: 1. 53 (1.2 9 1 .72 )) to encounter an isolate of the same genotype within the same spatial window, as compared to animal isolates. Additionally, spatial clustering was observed at distances up to 15km. Mantel t ests Figure A 1 depicts the Mantel c orrelograms, stratified by isolate group, located in the Supplementary section that confirms diversity results we found from the tau estimations
83 Discussion We characterized the geographic and species dependent structure of Co circulating Brucella spp. isolates collected as part of a large scale surveillance program in Southern Kazakhstan. Our results illustrate the relationships between space and genetics among Kazakh Brucella spp. in domesticated livestock (primarily sheep) and humans, which highlights and extends novel findings linking geography (scaled and discrete), with genetic distances. A positive, statistically significant relationship between geographic and genetic distance exists utilizing a MLVA 15 assay and across all four years of analysis and two different surveillance phases. Since the Brucella genus as a whole has a low evolutionary rate and the collection of only four years of data, we chose to on ly investigate the discrete differences between isolate species and surveillance phases. From this stratification we found spatial associations at both the species and phase collection categorization. We conducted multiple analyses to investigate the asso ciation between spatial and genetic distances from multiple angles and approaches. Stronger and significant associations were present within the domesticated ani mal isolates, dictated from the within/between genetic comparisons (violin plots) and tau estimates analyses. We observed a weaker association among the human isolates according to the violin plot analysis and a stronger tau clustering effect concluding a decreased amount of genetic diversity present in the human isolates, as compared to the veterinary isolates. The increased diversity present in veterinary isolates is evidence of potential spillover/spillback of wildlife reservoirs to domesticated livestock. We observe equivalent res ults within other free ranging reservoirs of B. melitensis and B. abortus such as Bison in the Greater Yellowstone Area, Buffalo in South east Africa, and Ibex in the French Alps. In contrast, the decreased genetic diversity present in
84 human isolates is p rimarily due to the outbreak dynamics of human brucellosis infection and the limited possibilities of transmission (contact with infected tissue and ingestion of infected unpasteurized dairy products). The limited sources and similar livestock exposure yie lds isolate s very similar regardless of time or space. With the majority of our dataset included B. melitensis isolates collected from sheep and past research showing sheep accounting for the majority of livestock B. melitensis cases, w e estimate the direc tionality of Brucella transmission to follow animals (sheep) to humans, due to genetic diversity differences in populations Research has shown that the amount of h igher diversity present in our veterinary isolates yields a h igher risk of spillover events initially due to a larger source pool, creating novel strains spreading to humans or other livestock species 90 Additionally, the l imited diversity in our human isolates, demonstrate the possibility that not all Brucella str ains are transmitted equally and the high diversity in veterinary B. melitensis strains is lost during spillover due possibly to a founder effect reducing the genetic variation of B. melitensis from the original population Other possible factors include external environmental pressures and evolutionary fitness in humans and animals augmenting pathogen fitness. This study represents one of the largest analyses on human and animal isolates present in Southern Kazakhstan. The surveillance effort also represents the largest efforts in the area over the eight year period. The data collection represents a four year longitudinal surveillance study effort of cohabitating humans and livestock populations. To the best of our knowledg e, this study represents the most comprehensive spatial/genetic association analyses, utilizing three separate methods to locate associations between spatial and genetic distances. The ability to ascertain
85 spatial/genetic association and phylogenetic recon struction using a MLVA 15 assay allows for a more economic al method compared to Bayesian p hylogenetic reconstruction using WGS. The main weakness of this study is that the passive surveillance efforts yield a biased isolate distribution, which could be unr epresentative of the true genetic burden of isolates present in Southern Kazakhstan. For our Tau section, we observed increasing tail end significantly high tau estimates when approaching the spatial cut off values of our analysis. This effect is primarily due to an increased likelihood of a false spatial clustering effect at very large distances comparable to our cut off values and has been replicated in similar studies 91 The total dataset had about 8 percent of GPS coordinat e data missing, resulting in the exclusion of 64 isolates in the analysis. Finally, the MLVA assay has l ower resolution data as compared to informative Single Nucleotide Polymorphism ( SNP ) or Single Nucleotide Repeat ( SNR ) data For future analyses, t he use of WGS could increase the discriminatory power, involving SNP data and a larger part of the genome for analysis This higher definition would allow for finer genotype designations and allow for more finely scaled spatial analyses. Summary In conclu sion, we identif ied, characterized and compared the spatial genomic associations of co circulating Brucella spp. on a regional scale, in Southern Kazakhstan, utilizing an established MLVA 15 genomic typing protocol between collection phase and isolate gro up. Our findings concluded and demonstrate d the human health benefits of livestock vaccination and culling programs (primarily in sheep) that reduce spillover of brucellosis to multiple species Additionally, we have successfully identified differential sp atial genomic association patterns between human and domesticated livestock
86 brucellosis cases. This crucial differential effect yielded a possible spillover directionality in B. melitensis cases between veterinary (sheep) and humans. Controlling brucellosi s requires the use of a One Health approach that emphasizes cooperation between human, veterinary, and environmental health practitioners. By integrating spatial data with molecular epidemiological data, we can further concentrate interventions and incorpo rate novel education campaigns to effectively control human brucellosis We find MLVA data to be informative to successfully integrate discrete geographic and demographic variables with enough definition to associate with genetic diversity.
87 Figure 4 1. Graphical r epresentation of study area and isolate d istribution. Isolate distributions were stratified by the two surveillance collection phases (A) Phase 1 and B) Phase 2) and individual isolate species (C) Veterinary related isolates and D) H uman isolates
88 Table 4 1. Descriptive table of KZ2 surveillance data stratified by collection p hase Phase 1 (2007 2008) Phase 2 (2012 2013) n= 11 9 398 Year (%) 2007 64 (53.8) 2008 55 (46.2) 2012 269 (67.6) 2013 129 (32.4) Brucella Species = B. abortus/B. melitensis (%) 0/119 (0.0/100.0) 11/387 (2.8/97.2) Region (%) Almaty 66 (55.5) 137 (34.4) Kyrgyzstan 2 (1.7) 21 (5.3) South Kazakhstan 3 (2.5) 0 (0.0) Zhambyl 48 (40.3) 240 (60.3) Host Species (%) Cattle 0 (0.0) 4 (1.0) Goat 0 (0.0) 10 (2.5) Human 58 (48.7) 60 (15.1) Sheep 61 (51.3) 324 (81.4) Figure 4 2. Methodological flowchart for this s tudy
89 A B Figure 4 3 The p hylogenetic and compositional depiction of 4 2 designated genotypes. A) The p hylogenetic depiction of the 4 2 genotype clusters obtained from the software PhyloPart, designed to automate the partition of phylogenetic trees based on clade reliability and a n nth percentile patristic distance algorithm. Isolates were aggregated into ge notypes highlighted in red. B) Represents the compositional breakdown of the 4 2 designated genotypes ordered by decreasing cluster count
90 Table 4 2 Stratification of 4 2 designated clusters by isolate group Isolate Group Cluster Composition Count Human 17 Human/Veterinary 9 Veterinary 16 Total 42
91 A B Figure 4 4 Within/Between village d esignations v iolin plots stratified by A) collection phase and B) isolate group. V iolin plots are similar to box plots, except that they also show the kernel probability density of the data at different values.
92 Figure 4 5 statistic of Brucella isolates categorized by genotypes designated by the software PhyloPart. The statistic is calculated within households ( i.e at 0m) and in a sliding 8km window up to the average distance between community households, 347km.
93 CHAPTER 5 CONCLUSIONS AND FUTURE DIRECTIONS Overall Conclusions Zoonotic disease surveillance and spillover as critical features to understand and prevent/eliminate the spread of infectious diseases from animals to humans. An important One Health related contribution to zoonotic disease surveillance and control is the addition of spatial dynamics, by means of spillover. In this dissertation, I discuss ed the main aspects of zoonotic disease and disease burden, zoonotic disease surveillance and control programs, spillover and spatial dynamics and describe d three study examples of how the incorporation of spillover and spatial dynamics can yield a One Health approach, improving these two distinct surveillance initiatives. In the first study, we described the current occurrence of specific zoonotic diseases in privately o wned animals (POAs) and government owned animals (GOAs), and spatially compare the differences of zoonotic Veterinary Reportable Medical Events ( VRME ) occurrence by military installation, State, and Region, in 201 6. Additionally, we discussed the shortfall s and limitations of the ROVR database and discuss potential improvements of the database system to improve the surveillance of zoonotic diseases, in military associated canines. The second study, identified characterized and compared the spatial genomic a ssociations of co circulating Brucella spp on a regional scale, in Southern Kazakhstan, utilizing an established MLVA 15 genomic typing protocol, between collection phase and isolate group. The third study utilized a MLVA 15 assay to epidemiologically tra ce and determine the human source of infection and discrete geographic patterns for Southern Kazakh brucellosis as a part of two separate surveillance phases. Within our ROVR database, we observed significantly differing
94 occurrences of zoonotic disease bet ween GOAs and POAs. Visualizing a novel revealing dynamic of how differing factors such as environment and lifestyle differences affects the overall disease burden, between these populations. For GOAs, the strong occurrence of Giardiasis is significantly c include close quarters kennels for all MWDs, an optimal transmission dynamic for Giardia. For both, GOAs and POAs we observed elevated occurrence of Ancylostomiasis, known to be pathogenic in human populations. Ad ditionally, we observed spatial definition with Lyme disease, considered to be a regional disease mostly contained to the Northeastern United States, and Hookworm showing elevated occurrence in the southeastern and island states. For our second study, we s uccessfully incorporated spatial dynamics, to evaluate comparisons made between human and veterinary isolates. We observed a weaker spatial genomic association among the human isolates according to the violin plot analysis and a stronger tau effect conclud ing a decreased amount of genetic diversity present in the human isolates, as compared to the animal isolates and a strong spatial clustering risk related to the same genotype, in humans, concluding a similar source of infection. Our third study successful ly integrated discrete geography identifying Zhambyl as a possible source region of human brucellosis, in addition to, concluding and demonstrating the human health benefits of livestock vaccination and culling programs (primarily in sheep) that reduce spi llover of brucellosis to multiple species. All three studies demonstrated the necessity of incorporating spatial data and dynamics in surveillance and control programs, in order to apply a One health approach and install collaborative ideas and interventio ns, improving the programs as a whole.
95 The main strength of these studies include the use of novel datasets, investigating rarely investigated populations. The conclusions made in these studies will allow public health professionals to actively concentrate and improve their surveillance and control programs, in order to control disease spread with the smallest financial footprint. The weaknesses of these studies, are consistent with any novel surveillance system, including missing data and differential/unev en sampling schemes. Future Directions For our ROVR database, f uture analyses could compare global clustering structures related to increased risk of transmission among GOAs and POAs to confirm the hypothesis of kennels, directly relating to use of kennels for MWDs. Novel methods allow for this risk clustering calculation without knowledge of the underlying population distribution Future analyses should include the use of Whole Genome Sequencing to allow for a higher definition of genetic detail. T he use o f WGS could increase the discriminatory power, involving of SNP data and a larger part of the genome for analysis. However, WGS, while on the overall financial decrease, is still a considerable financial investment in necessary reagents and computing power for any bioinformatics For a LMIC like Kazakhstan, the widespread use of WGS is a significant financial limit and the use of MLVA serves as an informative intermediary for outbreak tracing and identification A dditional ly, research is needed to investig ate the health seeking behavior of Kyrgyz people and why they travel very long distances and rugged terrain to obtain healthcare in Kazakhstan. We also observed s imilar dis similarity by region demonstrating a homogeneous diversity, possibly leading to a po ssible region wide success with interventional efforts on B. melitensis
96 APPENDIX A SUPPLEMENTARY TABLES AND FIGURE S Table A 1. List of 34 veterinary reportable medical e vents (VRMEs) used in analyses Zoonotic Disease Anaplasmosis Ancylostomiasis (Hookworm) Anthrax Babesiosis Botulism Brucellosis Campylobacteriosis CCHF Crimean Congo Hemorrhagic Fever Coccidioidomycosis Colibacillosis (STEC) Cryptosporidiosis Ebola EEE Eastern Equine Encephalitis Dermatophytosis Giardiasis Hemorrhagic fevers Leishmaniasis Leptospirosis Lyme Borreliosis Multi drug resistant bacteria Meliodiosis Plague Q fever (coxiella burnetti) Rabies Rickettsial disease (tick borne) Rocky Mountain Spotted Fever Salmonellosis Shigellosis Tick borne encephalitis Tuberculosis Toxocariasis Tularemia West Nile fever
97 A B Figure A 1. Phylogenetic reconstruction using alternative tree building methods. A) Neighbor Joining and B) Unweighted Pair Group Method with Arithmetic Mean (UPGMA) tree building approach.
98 Figure A 2 P hylogenetic reconstruction using a maximum l ikelihood approach categorized by phase collection and isolate g roup
99 A B Figure A 3 Mantel spatial correlograms, stratified by isolate group, A) Human and B) Veterinary. Correlograms show the relationship between geographic distance (x axis) and the Mantel r correlation score (y axis ). Points above the zero line indicate lower genetic distance between case pairs. Points below the zero line indicate greater genetic distance between case pairs. S olid symbols are statistically significant; hollow symbols are not.
100 Figure A 4 Bray Curtis d issimilarity v iolin plots grouped by all isolates further stratified by A) Isolate Group B) Collection Phase and C) Oblast, by Human Isolates stratified by D) Co llection Phase and E) Oblast and by Animal isolates stratified by F) Collection Phase and G) Oblast Violin plots are similar to box plots, except that they also show the kernel probability density of the data at different values.
101 LIST OF REFERENCES 1. Gowtage Sequeria, M. W. and S. Host Range and Emerging and Reemerging Pathogens Volume 11, Number 12 December 2005 Emerging Infectious Disease journal CDC. doi:10.3201/eid1112.050997 2. Torgerso n, P. R. & Macpherson, C. N. L. The socioeconomic burden of parasitic zoonoses: Global trends. Vet. Parasitol. 182, 79 95 (2011). 3. Narrod, C., Zinsstag, J. & Tiongco, M. A One Health Framework for Estimating the Economic Costs of Zoonotic Diseases on So ciety. EcoHealth 9, 150 162 (2012). 4. Jones, K. E. et al. Global trends in emerging infectious diseases. Nature 451, 990 (2008). 5. Gibbs, E. P. J. The evolution of One Health: a decade of progress and challenges for the future. Vet. Rec. 174, 85 91 (20 14). 6. Conrad, P. A., Mazet, J. A., Clifford, D., Scott, C. & Wilkes, M. Evolution of a the University of California, Davis. Prev. Vet. Med. 92, 268 274 (2009). 7. Institute of Medicine (U.S.), Committee on Achieving Sustainable Global Capacity for Surveillance and Response to Emerging Diseases of Zoonotic Origin & Keusch, G. Sustaining global surveillance and response to emerging zoonotic diseases (National Academi es Press, 2009). 8. World Organisation for Animal Health (OIE). Guidelines for Animal Disease Control 9. Elliott, P. & Wartenberg, D. Spatial Epidemiology: Current Approaches and Future Challenges. Environ. Health Perspect. 112, 998 1006 (2004). 10. Davies, T. J. & Pedersen, A. B. Phylogeny and geography predict pathogen community similarity in wild primates and humans. Proc. R. Soc. Lond. B Biol. Sci. 275, 1695 1701 (2008). 11. Olival, K. J. et al. Host and viral traits predict zoonotic spillover fr om mammals. Nature 546, 646 (2017). 12. Carrel, M. A., Emch, M., Jobe, R. T., Moody, A. & Wan, X. F. Spatiotemporal Structure of Molecular Evolution of H5N1 Highly Pathogenic Avian Influenza Viruses in Vietnam. PLoS ONE 5, e8631 (2010). 13. Fisher, M. C. et al. Low Effective Dispersal of Asexual Genotypes in Heterogeneous Landscapes by the Endemic Pathogen Penicillium marneffei. PLOS Pathog. 1, e20 (2005).
102 14. Smith, K. L. et al. Meso scale ecology of anthrax in southern Africa: a pilot study of diversit y and clustering. J. Appl. Microbiol. 87, 204 207 (1999). 15. Mullins, J. C. et al. Ecological Niche Modeling of Bacillus anthracis on Three Continents: Evidence for Genetic Ecological Divergence? PLoS ONE 8, e72451 (2013). 16. Jia, P. & Joyner, A. Human brucellosis occurrences in inner mongolia, China: a spatio temporal distribution and ecological niche modeling approach. BMC Infect. Dis. 15, 36 (2015). 17. Edney, A. T. Companion animals and human health: an overview. J. R. Soc. Med. 88, 704p 708p (1995 ). 18. Wells, D. L. The Effects of Animals on Human Health and Well Being. J. Soc. Issues 65, 523 543 (2009). 19. road ahead. J. Vet. Behav. Clin. Appl. Res. 5, 226 234 (2010). 20. Day, M. J. et al. Surveillance of Zoonotic Infectious Disease Transmitted by Small Companion Animals. Emerg. Infect. Dis. 18, e1 (2012). 21. Itoh, N. et al. Prevalence of intestinal parasites in breeding kennel dogs in Japan. Parasitol. Res. 114, 1221 1224 (2015). 22. Itoh, N. et al. Giardia and other intestinal parasites in dogs from veterinary clinics in Japan. Parasitol. Res. 109, 253 256 (2011). 23. Upjohn, M. et al. Prevalence, molecular typing and risk factor analysis for Giardia duodenalis infections in dogs in a central London rescue shelter. Vet. Parasitol. 172, 341 346 (2010). 24. Bouzid, M., Coln Gonzlez, F. J., Lung, T., Lake, I. R. & Hunter, P. R. Climate change and the emergence of vector borne diseases in Europe: case study of dengue fever. BMC Public Health 14, 781 (2014). 25. Glickman, L. T. et al. Purdue University Banfield National Companion Animal Surveillance Program for emerging and zoonotic disea ses. Vector Borne Zoonotic Dis. Larchmt. N 6, 14 23 (2006). 26. Day, M. J. The CALLISTO Project: A Summary. J. Comp. Pathol. 155, S1 S7 (2016). 27. Brady, S., Norris, J. M., Kelman, M. & Ward, M. P. Canine parvovirus in Australia: The role of socio econo mic factors in disease clusters. Vet. J. 193, 522 528 (2012).
103 28. Richardson, T. R. DoD Directive 6400.04E: DoD Veterinary Public and Animal Health Services. (2000). 29. Nelson, M. C. & Burke, R. L. Fielding the remote online veterinary record, a veterina ry electronic health record to improve patient care and practice management. US Army Med. Dep. J. 61 67 (2015). 30. Cook, G. C. Influence of diarrhoeal disease on military and naval campaigns. J. R. Soc. Med. 94, 95 97 (2001). 31. Sanchez, J. L., Gelnett J., Petruccelli, B. P., Defraites, R. F. & Taylor, D. N. Diarrheal disease incidence and morbidity among United States military personnel during short term missions overseas. Am. J. Trop. Med. Hyg. 58, 299 304 (1998). 32. Russell, K. L. et al. Vaccine p reventable adenoviral respiratory illness in US military recruits, 1999 2004. Vaccine 24, 2835 2842 (2006). 33. Pappas, G., Akritidis, N., Bosilkovski, M. & Tsianos, E. Brucellosis. N. Engl. J. Med. 352, 2325 2336 (2005). 34. Atluri, V. L., Xavier, M. N. Jong, M. F. de, Hartigh, A. B. den & Tsolis, R. M. Interactions of the Human Pathogenic Brucella Species with Their Hosts. Annu. Rev. Microbiol. 65, 523 541 (2011). 35. Dean, A. S. et al. Clinical Manifestations of Human Brucellosis: A Systematic Review and Meta Analysis. PLoS Negl. Trop. Dis. 6, e1929 (2012). 36. Pappas, G., Papadimitriou, P., Akritidis, N., Christou, L. & Tsianos, E. V. The new global map of human brucellosis. Lancet Infect. Dis. 6, 91 99 (2006). 37. Dean, A. S., Crump, L., Greter, H., Schelling, E. & Zinsstag, J. Global Burden of Human Brucellosis: A Systematic Review of Disease Frequency. PLoS Negl. Trop. Dis. 6, e1865 (2012). 38. Golunov, S. Border security in Central Asia: before and after September 11. in Geneva Center for the Democratic Control of Armed Forces DCAF Conference Papers (2005). 39. Djalalov, S. & Babu, S. C. Policy reforms and agriculture development in Central Asi a 28, (Springer Science & Business Media, 2006). 40. Grushina, T. et al. Universal indirect enzyme linked immunosorbent assay for monitoring of human and animal brucellosis in Kazakhstan. Vaccine 28, F46 F48 (2010).
104 41. Walker, R. & Blackburn, J. Biothr eat Reduction and Economic Development: The Case of Animal Husbandry in Central Asia. Front. Public Health 3, (2015). 42. Pappas, G. The changing Brucella ecology: novel reservoirs, new threats. Int. J. Antimicrob. Agents 36, S8 S11 (2010). 43. FAO. Regi onal workshop on brucellosis control in Central Asia and Eastern Europe 44. Brucellosis in humans and animals (WHO, 2006). 45. Remais, J. V., Zeng, G., Li, G., Tian, L. & Engelgau, M. M. Convergence of non communicable and infectious diseases in low a nd middle income countries. Int. J. Epidemiol. 42, 221 227 (2013). 46. Shevtsova, E. et al. Epidemiology of Brucellosis and Genetic Diversity of Brucella abortus in Kazakhstan. PLOS ONE 11, e0167496 (2016). 47. Shevtsov, A. et al. Genetic diversity of Br ucella abortus and Brucella melitensis in Kazakhstan using MLVA 16. Infect. Genet. Evol. 34, 173 180 (2015). 48. et al. Genotyping of Brucella melitensis and Brucella abortus strains in Kazakhstan using MLVA 15. Infect. Genet. Evol. 58, 13 5 144 (2018). 49. Land, M. et al. Insights from 20 years of bacterial genome sequencing. Funct. Integr. Genomics 15, 141 161 (2015). 50. Van Belkum, A. Tracing isolates of bacterial species by multilocus variable number of tandem repeat analysis (MLVA). FEMS Immunol. Med. Microbiol. 49, 22 27 (2007). 51. Pitt, T. L. & Barer, M. R. 3 Classification, identification and typing of micro organisms. in Medical Microbiology (Eighteenth Edition) 24 38 (Churchill Livingstone, 2012). 52. Whatmore, A. M. et al. Identification and Characterization of Variable Number Tandem Repeat Markers for Typing of Brucella spp. J. Clin. Microbiol. 44, 1982 1993 (2006). 53. Ert, M. N. V. et al. Global Genetic Population Structure of Bacillus anthracis. PLOS ONE 2, e461 (2007) 54. Adams, A. T. et al. Two stage Genome wide Methylation Profiling in Childhood HLA Loci: Inflamm. Bowel Dis. 20, 1784 1793 (2014).
105 55. Tiller, R. V. et al. Comparison of Tw o Multiple Locus Variable Number Tandem Repeat Analysis Methods for Molecular Strain Typing of Human Brucella melitensis Isolates from the Middle East. J. Clin. Microbiol. 47, 2226 2231 (2009). 56. Allen, A. et al. Molecular Epidemiology of Brucella abort us in Northern Ireland 1991 to 2012. PLOS ONE 10, e0136721 (2015). 57. Damborg, P. et al. Bacterial Zoonoses Transmitted by Household Pets: State of the Art and Future Perspectives for Targeted Research and Policy Actions. J. Comp. Pathol. 155, S27 S40 (2 016). 58. Cito, F. et al. Prioritization of Companion Animal Transmissible Diseases for Policy Intervention in Europe. J. Comp. Pathol. 155, S18 S26 (2016). 59. Rijks, J. M., Cito, F., Cunningham, A. A., Rantsios, A. T. & Giovannini, A. Disease Risk Asse ssments Involving Companion Animals: an Overview for 15 Selected Pathogens Taking a European Perspective. J. Comp. Pathol. 155, S75 S97 (2016). 60. Reperant, L. A. et al. Companion Animals as a Source of Viruses for Human Beings and Food Production Animal s. J. Comp. Pathol. 155, S41 S53 (2016). 61. Burkman, K. D. Incidence of zoonotic diseases in military working dogs serving in operations Desert Shield and Desert Storm. Mil. Med. Bethesda 166, 108 11 (2001). 62. Takara, M. S. & Harrell, K. Noncombat rel ated injuries or illnesses incurred by military working dogs in a combat zone. J. Am. Vet. Med. Assoc. 245, 1124 1128 (2014). 63. Ballweber, L. R., Xiao, L., Bowman, D. D., Kahn, G. & Cama, V. A. Giardiasis in dogs and cats: update on epidemiology and public health significance. Trends Parasitol. 26, 180 189 (2010). 64. Mohamed, A. S., Glickman, L. T., Camp, J. W., Lund, E. & Moore, G. E. Prevalence and risk factors for Giardia spp. infection in a large national sample of pet dogs visiting veterinary hospitals in the United States (2003 2009). Vet. Parasitol. 195, 35 41 (2013). 65. Evans, R. I., Herbold, J. R., Bradshaw, B. S. & Moore, G. E. Causes for discharge of military working dogs from service: 268 cases (2000 2004). J. Am. Vet. Med. Assoc. 231, 1215 1220 (2007). 66. Moore, G. E., Burkman, K. D., Carter, M. N. & Peterson, M. R Causes of death or reasons for euthanasia in military working dogs: 927 cases (1993 1996). J. Am. Vet. Med. Assoc. 219, 209 214 (2001).
106 67. Zimmerman, K. L. et al. SNOMED representation of explanatory knowledge in veterinary clinical pathology. Vet. Clin Pathol. 34, 7 16 (2005). 68. Lessler, J., Salje, H., Grabowski, M. K. & Cummings, D. A. T. Measuring Spatial Dependence for Infectious Disease Epidemiology. PLOS ONE 11, e0155249 (2016). 69. Companion Animal Parasite Council. Parasite prevalence maps. (2016). 70. Franco, M. P., Mulder, M., Gilman, R. H. & Smits, H. L. Human brucellosis. Lancet Infect. Dis. 7, 775 786 (2007). 71. Kreeger, T. J., Miller, M. W., Wild, M. A., Elzer, P. H. & Olsen, S. C. SAFETY AND Efficacy Of Brucella Abortus Strain Rb51 Vaccine In Captive Pregnant Elk J. Wildl. Dis. 36, 477 483 (2000). 72. Olsen, S. C. & Johnson, C. S. Efficacy of Dart or Booster Vaccination with Strain RB51 in Protecting Bison against Experimental Brucella abortus Challenge. Clin. Vaccine Immunol. 19, 886 890 (2012). 73. Yang, X. et al. Progress in Brucella vaccine development. Front. Biol. 8, 60 77 (2013). 74. Beauvais, W., Coker, R., Nurtazina, G. & Guitian, J. Policies and Livestock Systems Driving Brucellosis Re emergence in Kazakhstan. EcoHealth 1 9 (2015). doi:10.1007/s10393 015 1030 7 75. Sainudiin, R., Durrett, R. T., Aquadro, C. F. & Nielsen, R. Microsatellite Mutation Models: Insights From a Comparison of Humans and Chimpanzees. Genetics 168, 383 395 (2004). 76. Aikembayev, A. M. et al. His torical Distribution and Molecular Diversity of Bacillus anthracis Kazakhstan. Emerg. Infect. Dis. 16, 789 796 (2010). 77. Swofford, D. PAUP*: phylogenetic analysis using parsimony, version 4.0b10. (2003). 78. Lewis, P. O. A Likelihood Approach to Estimating Phylogeny from Discrete Morphological Character Data. Syst. Biol. 50, 913 925 (2001). 79. Wu, C. H. & Drummond, A. J. Joint Inference of Microsatellite Mutation Models, Population History and Genealogies Using Transdimensional Markov Chain Monte Carlo. Genetics 188, 151 164 (2011). 80. Garofolo, G. Multiple locus variable number tandem repeat (VNTR) analysis (MLVA) using multiplex PCR and multicolor capillary electrophoresis: application to the genotyping of Brucella species. Methods Mol. Biol. Clifton NJ 1247, 335 347 (2015).
107 81. Her, M. et al. Application and evaluation of the MLVA typing as say for the Brucella abortus strains isolated in Korea. BMC Microbiol. 9, 230 (2009). 82. De Massis, F. et al. Genetic characterization of Brucella melitensisand Brucella abortusgeographical clusters in Italy. Vet. Ital. 51, 225 233 (2015). 83. De Massis F. et al. MLVA as an Epidemiological Tool To Trace Back Brucella melitensis Biovar 1 Re Emergence in Italy. Transbound. Emerg. Dis. 62, 463 469 (2015). 84. Kamath, P. L. et al. Genomics reveals historic and contemporary transmission dynamics of a bacter ial disease among wildlife and livestock. Nat. Commun. 7, 11448 (2016). 85. Huynh, L. Y. et al. Multiple Locus Variable Number Tandem Repeat (VNTR) Analysis (MLVA) of Brucella spp. Identifies Species Specific Markers and Insights into Phylogenetic Relatio nships. in National Institute of Allergy and Infectious Diseases, NIH 47 54 (Humana Press, 2008). 86. Ihaka, R. & Gentleman, R. R: A Language for Data Analysis and Graphics. J. Comput. Graph. Stat. 5, 299 314 (1996). 87. Schliep, K. P. phangorn: phylogen etic analysis in R. Bioinformatics 27, 592 593 (2011). 88. Prosperi, M. C. F. et al. A novel methodology for large scale phylogeny partition. Nat. Commun. 2, 321 (2011). 89. Dixon, P. & Palmer, M. W. VEGAN, a package of R functions for community ecology. J. Veg. Sci. 14, 927 930 (2003). 90. Keesing, F. et al. Impacts of biodiversity on the emergence and transmission of infectious diseases. Nature 468, 647 652 (2010). 91. Barro, A. S. et al. Identifying hotspots of human anthrax transmission using three local clustering techniques. Appl. Geogr. 60, 29 36 (2015). 92. Godfroid, J. Brucellosis in livestock and wildlife: zoonotic diseases without pandemic potential in need of innovative one heal th approaches. Arch. Public Health 75, 34 (2017). 93. Kracalik, I. T. et al. Human Brucellosis Trends: Re emergence and Prospects for Control Using a One Health Approach in Azerbaijan (1983 2009). Zoonoses Public Health 63, 294 302 (2016).
108 94. Hou, Q. et al. Modeling the transmission dynamics of sheep brucellosis in Inner Mongolia Autonomous Region, China. Math. Biosci. 242, 51 58 (2013). 95. Havas, K. A., Boone, R. B., Hill, A. E. & Salman, M. D. A Brucellosis Disease Control Strategy for the Kakheti Re gion of the Country of Georgia: An Agent Based Model. Zoonoses Public Health 61, 260 270 (2014). 96. Shevtsov, A. et al. Genetic diversity of Brucella abortus and Brucella melitensis in Kazakhstan using MLVA 16. Infect. Genet. Evol. J. Mol. Epidemiol. Evo l. Genet. Infect. Dis. 34, 173 180 (2015). 97. et al. Brucellosis Transmission between Wildlife and Livestock in the Greater Yellowstone Ecosystem: Inferences from DNA Genotyping. J. Wildl. Dis. 53, 339 343 (2017). 98. Aydin, B. et al. Pre maturity due to maternal brucella infection and review of the literature. Turk. J. Pediatr. 55, 433 (2013). 99. Dobson, A. Population Dynamics of Pathogens with Multiple Host Species. Am. Nat. 164, S64 S78 (2004). 100. Reviriego, F. J., Moreno, M. A. & D om nguez, L. Risk factors for brucellosis seroprevalence of sheep and goat flocks in Spain. Prev. Vet. Med. 44, 167 173 (2000). 101. Sanz, C. et al. Mass vaccination as a complementary tool in the control of a severe outbreak of bovine brucellosis due to Brucella abortus in Extremadura, Spain. Prev. Vet. Med. 97, 119 125 (2010). 102. lvarez, J. et al. Management of an outbreak of brucellosis due to B. melitensis in dairy cattle in Spain. Res. Vet. Sci. 90, 208 211 (2011). 103. Grushina, T. et al. Diagnostics of human brucellosis caused by Brucella melitensis. in International Congress on Anthrax, Brucellosis, CBPP, Mycobacterial Disease 26 (2003). 104. Anderson, M. J. & Walsh, D. C. I. PERMANOVA, ANOSIM, and the Mantel test in the face of heterog eneous dispersions: What null hypothesis are you testing? Ecol. Monogr. 83, 557 574 (2013).
109 BIOGRAPHICAL SKETCH Sheldon Waugh was born in Gloucester, UK in May of 1989. Sheldon then moved to Naples, Florida in 1996, where his hometown currently resides. Sheldon received his undergraduate education at the University of Florida, completing his Bachelor of Science and Master of Science in g eography in 2011 and 2014, respectively Sheldon has also completed on a Certificate of Public Health with a concentration in Epidemiology. Sheldon has received his Ph D in e pidemiology from the University of Florida in the spring of 2018 His subject areas include the study of vector borne diseases, ecological infectious disease modeling public health and bioinformatics bioinformatician and epidemiologist at the Army Public Health Center, located in Aberdeen Proving Ground MD, continuing his work on mapping and studying infectious diseases for the One Health Division.