Improving National Surveillance of Wildlife and Vector‑Borne Diseases Using Spatial Data‑Science

Abstract

This thesis investigates how spatial models can be used to map infectious disease pathogens in wildlife and vectors, identify factors associated with observed patterns, and assess potential impacts on wildlife populations. Models are simplified representations of reality that help identify patterns and make complex systems more understandable. The focus is on spatial models using data from the Netherlands. The main findings are summarized below by chapter. Chapter 1 introduces the thesis and provides an overview of data-, process-, and decision-driven spatial epidemiology, as well as the research context. Chapters 2–4 use models to describe observed pathogen occurrences by quantifying infection rates, mapping spatial distributions, and identifying environmental, ecological, climatic, and host-related factors. Chapter 2 examines roe deer as hosts for vector-borne pathogens. Of nine pathogens studied, Anaplasma phagocytophilum, Bartonella species, and Babesia species were regularly detected. Their prevalence was primarily associated with age and nutritional or health status, but not roe deer density. A few Rickettsia species and one Borrelia burgdorferi s.l. case were also detected. These results suggest that roe deer primarily contribute to maintaining A. phagocytophilum, Bartonella, and Babesia species. Chapter 3 examines pathogens in Dutch brown hares, associated environmental and climatic factors, and changes in disease occurrence. Hares hosted multiple pathogens, including zoonotic variants. Yersinia species were most common, with higher prevalence associated with intensive agriculture and lower minimum temperatures. The chapter highlights the importance of long-term, systematic, standardized surveillance of pathogens and populations, as retrospective data have limitations for investigating disease and population trends. Chapter 4 investigates Toxoplasma gondii type II in squirrels collected by the DWHC between 2014 and 2024. Just under one-third were infected, and in over half of infected squirrels, severe inflammation caused by infection was the cause of death. Rainfall and the number of hot days were identified as potential risk factors. These findings suggest substantial environmental oocyst contamination and an important role for squirrels as intermediate hosts in urban ecosystems. Some squirrels were also infected with Hammondia hammondi, which has a similar life cycle but cannot infect cats through environmental oocysts. Chapter 5 uses camera-trap data to examine rats and mice in private gardens. The absence of domestic cats had the greatest effect on the presence of both species. For mice, greener environments were additionally associated with higher occupancy, while other predators had weaker and opposite effects. Across Chapters 2–5, data-driven models examine associations between environmental, ecological, or socioeconomic variables and pathogen, vector, or host presence. When observations are limited, process-based models can complement these approaches. Chapter 6 applies a process-based model to assess habitat suitability for tick-borne encephalitis virus. Suitability was primarily determined by the sheep tick and small mammals, while larger mammals had minimal effect. Combined with surveillance data, the model can identify areas where the virus may circulate and enable more targeted monitoring. Chapter 7 applies a trait-based vulnerability assessment to examine wildlife vulnerability to climate change and evaluates representation in the DWHC database. Vulnerability depends on exposure, sensitivity, and adaptive capacity. Twenty-five percent of mammal species had high exposure, 24% high sensitivity, and 22% limited adaptive capacity. The whiskered bat and garden dormouse were most vulnerable, yet were poorly represented in surveillance data. Chapter 8 summarizes what can be learned from wildlife and vector-borne disease surveillance data while considering their limitations. Two themes emerge: what can be learned from existing data using spatial modelling, and which data are insufficiently collected. Spatial modelling can provide valuable insights and guide surveillance, but its effectiveness depends fundamentally on the availability, quality, and design of the underlying data. Collaboration between spatial modelers, host/pathogen/vector experts, and surveillance practitioners is therefore essential.