The value of visually apparent diseases to advance landscape epidemiology in wildlife

Understanding the patterns and processes driving landscape-scale dynamics of wildlife pathogens is challenging because the logistics of sampling pathogens from hosts is generally difficult and resource intensive. As a result, dynamics across large spatial scales are generally understudied in wildlife-host systems, impeding the progress of landscape epidemiology. However, pathogens which present as visually apparent diseases offer an opportunity to study landscape epidemiology, especially when combined with remote-surveillance technology, such as camera trapping. Camera trapping is widely used for statistical techniques such as community composition, abundance, and occupancy of wildlife species, however, few examples of its use for inferences in disease ecology exist.
In this thesis, I demonstrate that pathogens which cause visual signs of disease in wildlife can be used to inform pathogen occurrence, severity, and ascertain epidemiological drivers across large spatial areas. I begin with a systematic review of one visually apparent disease in wildlife (dermatitis) to establish patterns and causes in disease prevalence at a biogeographical scale (Chapter 2). Then, I present three empirical visually apparent disease case studies using observational camera trapping data, landscape variables, and spatial modelling methods: sarcoptic mange in bare-nosed wombats (Vombatus ursinus) (Chapter 3), and rumpwear in common ringtail possums (Pseudocheirus peregrinus) (Chapter 4) and common brushtail possums (Trichosurus vulpecula) (Chapter 5).
In Chapter 2, I reveal the overall trends in causes and reports of dermatitis in wildlife. The highest proportion of dermatitis cases reported in the literature were of an unknown cause, and free-living wildlife were 2-6 times more likely to be reported with dermatitis than those in captivity or laboratories. I also determined that environmental, genetic, and social factors play into the development of skin conditions in threatened species, particularly those in captivity and outside of their endemic home range. This research was the first to systematically review clinical signs of wildlife disease and determine conditions that make them more likely to be documented globally. The outcomes of this chapter emphasize the need for cross-disciplinary research between animal health and disease ecology to better manage causes of skin diseases in wildlife.

In Chapter 3, I used the visual clinical signs of sarcoptic mange (aetiologic agent Sarcoptes scabiei) in bare-nosed wombats to determine the factors driving both host and pathogen occurrence. Using over 53,000 images from 3,000 camera traps, landscape variables (like climate, land use, and vegetation type) and species distribution modelling, I found that i) bare-nosed wombats were well?suited to the Tasmanian landscape, with only high annual precipitation reducing habitat suitability; ii) sarcoptic mange in bare-nosed wombats had a more restricted distribution; and iii) disease was more likely to or occur in landscapes that were highly suitable for the host, in combination with proximity to fresh water, and in areas with gentle topography, such as agricultural areas. The findings have illuminated areas where wombats are at higher risk for disease occurrence, where disease management interventions are best targeted, and can help guide disease risk assessments for proposed translocations.
In Chapter 5, I assessed patterns and drivers of rumpwear (a poorly understood disease characterised by hair damage on the lumbosacral region) in brushtail possums (Trichosurus vulpecula) at regional and landscape scales. Using over 40,000 camera trapping images, landscape variables, generalised additive models, and deep learning neural networks, I assessed factors driving rumpwear in possums. I found that the time of year (month) and vegetation type were significant contributing factors to individual possums presenting with the clinical signs of the disease. Specifically, the prevalence of rumpwear increasing during summer months and decreasing during autumn, and there was a higher prevalence or rumpwear in rainforest vegetation. The findings suggest the putative etiological agent may have a cyclic lifecycle, and seasonal activities of possums, or preferred habitat, may also influence the disease agent. Prior to my research, rumpwear had only been described in brushtail possums (Trichosurus genus), but through the extensive network of camera trap images available in Tasmania, I showed this disease to also occur in ringtail possums (P. peregrinus) (Chapter 4). I also revealed that rumpwear in P. peregrinus occurred across Tasmania at comparable rates (7-14%) to that observed in brushtail possums (8-16%) at a regional level and similar to rumpwear in brushtail possums across the landscape (18.6%). The wildlife-pathogen case studies in this thesis represent some of the largest spatial assessments of wildlife diseases, globally. This thesis exemplifies that using a combination of visually apparent diseases, technology (both camera trapping and computational) and spatial modelling, technical limitations of sampling diseases at large spatial scales in wildlife can be overcome. This thesis provides insight into a pathogen of global relevance (sarcoptic mange), uncovering disease clinical signs in a new genus, and providing foundational detail into a disease with poorly understood aetiology. Overall, my thesis contributes to epidemiological assessments for each of these host-pathogen systems, advances the field of landscape epidemiology for wildlife, and has applied implications for the spatial allocation of resources were pathogens cause conservation and welfare issues.