Big data - modelling of midges in Europa using machine learning techniques and satellite imagery

Ana Carolina Cuellar, Lene Jung Kjær, Henrik Skovgaard, Søren Archim Nielsen, Anders Stockmarr, G. Anderson, Anders Lindström, J. Chirico, T. Lilja, R. Lühken, S. Steinke, E. Kiel, Magdalena Larska, S. I. Hamnes, S. Sviland, Petter Hopp, K. Brugger, F. Rubel, T. Balenghien, C. GarrosI. Rakotoarivony, X. Allene, J. Lhoir, J. C. Delecolle, B. Mathieu, D. Delecolle, M. L. Setier-Rio, R. Venail, B. Scheid, M. A. Miranda Chueca, C. Barcelo Segui, J. Lucientes, R. Estrada, Tack Wesley, A. Mathis, Rene Bødker

Research output: Chapter in Book/Report/Conference proceedingConference abstract in proceedingsResearchpeer-review

227 Downloads (Pure)

Abstract

Biting midges (Diptera, Ceratopogonidae) of the genus Culicoides are important vectors of pathogens causing diseases in free living and production animals and can lead to large economic losses in many European countries. In Europe, Culicoides imicola and the Obsoletus group are considered to be the main vectors of bluetongue virus that mostly affects ruminants such as cattle and sheep. Spatio-temporal modelling of vector distribution and abundance allows us to identify high risk areas for virus transmission and can aid in applying effective surveillance and control measures.

We used presence-absence and monthly abundance data of Culicoides from 1005 sites across 9 countries (Spain, France, Denmark, Poland, Switzerland, Austria, Poland, Sweden, Norway) collected between the years 2007 and 2013. The dataset included information on the vector species abundance (number of specimens caught per night), GPS coordinates of each trap, start and end dates of trapping. We used 120 environmental predictor variables together with Random Forest machine learning algorithms to predict the overall species distribution (probability of occurrence) and monthly abundance in Europe. We generated maps for every month of the year, to visualize the abundance of C. imicola and Obsoletus group in Europe as well as distribution maps showing the probability of occurrence.

We were able to create predictive maps of both Culicoides sp. occurrence and abundance using Random Forest models, and although the variance was large, the predicted abundance values for each site had a positive correlation with the observed abundance. We found relatively large spatial variations in probability of occurrence and abundance for both C. imicola and the Obsoletus group. For C. imicola probability of occurrence and abundance was higher in southern Spain, where as the Obsoletus group had higher probability of occurrence and abundance in central and northern Europe such as France and Germany. Temporal variation was also observed with higher abundance occurring during summer months and low or no abundance during winter months for both C. imicula and the Obsoletus group, although abundance was generally higher for a longer period of time for C. imicula than for the Obsoletus group.

Using machine learning techniques, we were able to model the spatial distribution in Europe for C. imicola and the Obsoletus group in terms of abundance and suitability (probability of occurrence). Our maps corresponded well with the previously reported distribution for C. imicola and the Obsoletus group. The observed seasonal variation was also consistent with reported population dynamics for Culicoides, as it depends on environmental factors such as temperature and rainfall. Longer seasonal abundance for C. imicula compared to the Obsoletus group can be explained by the species distribution, as C. imicula is limited to the southern parts of Europe where the warm season lasts longer, whereas the Obsoletus group is found further north. The outputs obtained here will be used as input for epidemiological models and can be helpful for determining high risk areas for disease transmission.
Original languageEnglish
Title of host publicationNKVet Symposium 2017 - abstract book
Number of pages2
Place of PublicationOslo, Norway
Publication date2017
Publication statusPublished - 2017
EventNKVet Symposium 2017 - Oslo, Norway
Duration: 11 Jan 201712 Jan 2017

Conference

ConferenceNKVet Symposium 2017
CountryNorway
CityOslo
Period11/01/201712/01/2017

Fingerprint Dive into the research topics of 'Big data - modelling of midges in Europa using machine learning techniques and satellite imagery'. Together they form a unique fingerprint.

Cite this