Context and objectives
Bluetongue is a viral disease of ruminants which is transmitted by Culicoides midges. Since the late 1990’s a series of bluetongue virus (BTV) serotypes have invaded Mediterranean Europe. In 2006 other BTV serotypes also invaded temperate Europe. This twofold invasion pattern has resulted in large economic loses mainly for sheep farmers. In Mediterranean Europe bluetongue is mainly spread by Culicoides imicola which is also the main tropical vector in Northern Africa. In temperate Europe other indigenous midge species (mainly from the C. obsoletus group) are involved. The occurrence of arthropod vectors of disease in general and biting midges in particular is determined by a large number of (a)biotic factors. Amongst others, soil moisture is one of the variables that is considered to be an explanatory factor of the presence or absence of the insect vector. However, until now, epidemiological research focusing on the prediction of the spread of Culicoides, only partially integrates knowledge on soil moisture within its spatial models. This knowledge is included in these models through the Normalized Difference Vegetation Index (NDVI), even though the correlation between NDVI and soil moisture was not validated at most sites. Therefore this project will investigate alternative proxies for soil moisture to apply in spatial epidemiological models.
The main goal of the proposal is to improve presence/absence modelling of Culicoides through (1) the integration of soil moisture proxies other than the currently used NDVI proxy and (2) the use of machine learning models instead of the popular and frequently used statistical models.
More accurate predictions of the presence of Culicoides imicola will contribute to a better understanding of the factors affecting the spread of BTV in the Mediterranean, which may reflect in improved control measures. In addition this may also improve our ability assess the risk of introduction of African horse sickness, another prime candidate of the 60+ viruses which are transmitted by Culicoides, and which may invade Europe in the near future.
Project outcome
Expected scientific results
The algorithms developed are implemented within the Avia‐GIS software suite for VECMAP. The soil moisture imagery will be offered as a service to all VECMAP clients.
- A spatial and temporal analysis of soil moisture in the topsoil was contrasted from sites of Culicoides imicola presence and absence. Although the temporal soil moisture pattern was not significantly different between these sites, the volumetric soil moisture content was found to differ substantially, with consistently wetter topsoil conditions at the C. imicola presence sites which are more favourable for larval development;
- The potential of remote sensing for soil moisture estimation was assessed. Apparent thermal inertia derived from multitemporal Terra and Aqua MODIS products showed promising results, but also indicated the limitations of this approach for vegetated terrain.. Another approach to estimate soil moisture from optical remote sensing explores the negative relationship between vegetation index and surface temperature. A topographic normalization methodology was developed using stratified linear regression. The normalization resulted in better soil moisture (soil dryness) estimates based on the vegetation index and soil moisture relationship in comparison with the existing state‐of‐the‐art methodologies. Although soil moisture estimates derived using this method seem more accurate than those derived using optical data, the temporal resolution of SAR data is possibly too low to include this data source in epidemiological models;
- The use of Random Forests was found to improve C. imicola distribution models compared to the established models using linear discriminant analysis or logistic regression, especially when species dispersal is taken into account.