PhD Thesis

Spatial Risk Assessment of Mosquito-Borne Viral Diseases – Research at the Intersection of Ecology and Epidemiology.

Nils Tjaden (01/2013-06/2020)

Support: Carl Beierkuhnlein, Cyrus Samimi, Heike Feldhaar

Mosquito-borne viral diseases pose an increasing threat to human and ani-mal health on a global level. Over the past few decades, competent vector spe-cies like the Asian tiger mosquito (Aedes albopictus) or the Asian bush mosquito (Aedes japonicus) have spread vigorously across the globe and far beyond their native distribution. During the same time, large outbreaks of diseases that are being transmitted by these and other mosquito species (such as chikungunya, Zika, West-Nile fever and Usutu) have been recorded. Diseases that were formerly considered purely tropical by many, such as dengue and chikungunya, showed repeated outbreaks along the coast of the Mediterranean Sea – far away from the tropics. Usutu virus (which was largely neglected in the past as long as it was spatially limited to Africa) emerged in Europe, causing mass extinction events among blackbird populations. Evidence suggests that increasing temperatures due to climate change will facilitate future spread. Clearly, there is an increasing need for spatial risk assessment of these diseases. In this thesis, I use two established approaches, Ecological Niche Models and Epidemiological Models, to assess the spatial risk arising from different mosquito-borne viral diseases. Building models for chikungunya and Usutu viruses as well as the mosquito vector Ae. albopictus, I produce risk maps at global, continental, national and local scales. I explore the strengths and weaknesses of the different approaches and make suggestions for future improvements. All models in this thesis suggest a potential for a continued increase in mosquito-borne viral disease occurrence in large parts of the respective study area. On a global scale, chikungunya is expected to increase its presence on all continents except for Antarctica as well as some areas in Australia and northern India (where climate change will lead to conditions that may prohibit vector survival). On a continental scale, two fundamentally different models for Usutu suggest that large parts of Europe offer favorable environmental conditions for transmission of the disease. However, they differ considerably at the local scale. At the national scale, large parts of western Germany are projected to become climatically suitable for the establishment of Ae. albopictus in the near future due to climate change. Most of these areas (including those that are already highly suitable today) also showed elevated incidence rates of travel-related dengue and chikungunya infections, suggesting an elevated risk for virus transmission. Risk maps are an important tool that can be used by field entomologists and epidemiologists for more targeted surveillance and monitoring. And they can help to communicate essential information to politicians and decision makers in order to facilitate the establishment of the infrastructure that is necessary for these endeavors. Both Epidemiological Models and Ecological Niche Models suffer from a lack of essential data. For Epidemiological Models, laboratory studies and field data about the underlying mechanisms of transmission are severely lacking for many diseases. This is demonstrated in this thesis using the extrinsic incubation period (EIP) of dengue as an example. It has long been known that the duration of the EIP inside the mosquito vector highly depends on ambient temperature. However, among the few experimental works that investigate that relationship, several are based on flawed methodology or otherwise outdated. For many less-studied diseases (such as Usutu) the gaps in knowledge are still much larger. The need for more fundamental research in this area is high. For Ecological Niche Models, the availability of high-quality occurrence records of vectors and diseases is a major problem. International and interdisciplinary efforts towards a centralized, open data repository need to be intensified. The centralized climate data repository of the Earth System Grid Foundation (ESGF, https://esgf.llnl.gov) and the data base of species occurrence records at the Global Biodiversity Information Facility (GBIF, http://www.gbif.org) could serve as inspiration for this. Transferability of model results across different climate zones is another issue that warrants further investigation. Finally, different models have different pros and cons, and different ques-tions require different approaches. Ecological Niche Models require only a lim-ited amount of a-priori knowledge about the environmental parameters governing a species’ spatial distribution. Even with relatively low numbers of occurrence records, they can be very useful for rapid, coarse scale risk assessment. Epidemiological Models are built upon a much more detailed theoretical background, and if they are parameterized thoroughly, they can add valuable information on fine spatio-temporal scales. While Ecological Niche Models have always been intended for spatial applications, the adaption of Epidemiological Models for the creation of spatial risk maps involves some unresolved hurdles that will be addressed in future works.