1. Overview

Introduction/Study Description

VBORNET [1] was an initiative of the European Centre for Disease Prevention and Control (ECDC), which ran from 2009 to 2014. The project established a European network of entomological and public health specialists in order to assist ECDC in its preparedness activities on vector-borne diseases. As part of this work a database collating validated records of key vector species distributions was commissioned. In this data paper we describe work done on Aedes vexans and Culex modestus, both vectors of West Nile fever virus, and Anopheles plumbeus, a potential vector of malaria parasite.

VectorNet [2] is continuing this work and builds upon VBORNET by supporting the collection of data on vectors and pathogens in vectors, related to both animal and human health. VectorNet is a joint initiative of the European Food Safety Authority (EFSA) and the European Centre for Disease Prevention and Control (ECDC), which started in May 2014.

Whilst VBORNET and VectorNet have made substantial progress collating European data on key vector species, the coverage is still incomplete. The ‘Gap Analysis’ work within these projects aims to identify those areas of likely species distribution within the project extent where there are no current data. These estimates produced by spatial modelling techniques are intended to meet two objectives: firstly to help direct extensive VectorNet sampling efforts in the field; and secondly to provide first indications of the current likely extent and distribution of key vector species within continental Europe and its surrounding regions. It is hoped that publishing these models will aid the VectorNet network of experts to engage the wider research and professional community in the drive to expand and validate the VectorNet database. Readers are encouraged to visit the VectorNet website [2] or to directly contact the authors to report complementary data.

For each species, probability of presence maps at the resolution of 1 km were generated using a variety of well-established spatial modelling techniques available through the VECMAP system [3]. Both the input data and the resulting models were iteratively assessed by project experts and the best performing are included in this data package.

2. Context

Spatial coverage

Description: Continental Europe

Northern boundary: 71.8

Southern boundary: 33.5

Eastern boundary: 62.3

Western boundary: –19.2

Temporal coverage

Known presence up to 31/01/2013.


The inland floodwater mosquito, Aedes vexans, vector of West Nile fever and Rift Valley fever viruses.

Anopheles plumbeus, a potential vector of Plasmodium falciparum.

Culex modestus, a vector of West Nile fever virus.

3. Methods

For each of the species the following method was followed.


Identifying presence and absence training data

The reported distributions of each of the three mosquito species by VBORNET were used as the basis for species present training data. Data reported from the VBORNET map published January 2013 were utilised for Aedes vexans and Culex modestus. Data reported from January 2014 were used for Anopheles plumbeus. Maps of the recorded known distributions at that time are presented in Appendix 1 available within this data package. These reported distributions were recorded in VBORNET at a coarse NUTS 3 polygon scale. The data originates from a combination of aggregated data contributed by the authors and listed contributors, as well as a literature review completed by the VBORNET vector group leaders. The full data set and sources are available to contributors of VBORNET and VectorNet.

Habitat suitability and environmental limits

The recorded distribution at a NUTS 3 scale was too coarse to be utilised by the model framework. In addition, the selected modelling methods required information on both presence and absence to calibrate the modelling process. It was therefore necessary to identify areas of absence within NUTS 3 regions assigned as present. To do this a suitability mask at 1 km resolution was compiled by requesting experts within the network (see the Data Creators section) to identify primary, secondary and unsuitable land cover classes. Where available environmental limiting factors such as altitude or precipitation limits which are derived from remotely sensed imagery. Land cover masks were defined using the 100 m Corine land cover dataset [4] and the 300 m GLOBCOVER [5] product where no Corine data was available. Definitions of land class suitability for each species as defined by experts can be found in Tables 1 and 2.

Table 1

Reclassed values defining the Corine suitability layers 1 = suitable and 0 = unsuitable. AEVE = Aedes vexans, ANPL = Anopheles plumbeus and CUMO = Culex modestus.


Continuous urban fabric 0 0 0
Discontinuous urban fabric 0 0 0
Industrial or commercial units 0 0 0
Road and rail networks and associated land 0 0 0
Port areas 0 0 0
Airports 0 0 0
Mineral extraction sites 0 0 0
Dump sites 0 0 0
Construction sites 0 0 0
Green urban areas 1 1 0
Sport and leisure facilities 0 1 0
Non-irrigated arable land 0 0 0
Permanently irrigated land 0 0 1
Rice fields 1 0 1
Vineyards 0 0 0
Fruit trees and berry plantations 0 0 0
Olive groves 0 1 0
Pastures 1 0 0
Annual crops associated with permanent crops 0 0 0
Complex cultivation patterns 0 0 0
Land principally occupied by agriculture, with significant areas of natural vegetation 1 1 1
Agro-forestry areas 1 1 1
Broad-leaved forest 1 1 0
Coniferous forest 0 0 0
Mixed forest 1 1 0
Natural grasslands 1 0 0
Moors and heathland 1 0 0
Sclerophyllous vegetation 0 0 0
Transitional woodland-shrub 0 1 0
Beaches, dunes, sands 0 0 0
Bare rocks 0 0 0
Sparsely vegetated areas 0 1 0
Burnt areas 0 0 0
Glaciers and perpetual snow 0 0 0
Inland marshes 1 0 1
Peat bogs 1 0 0
Salt marshes 0 0 0
Salines 0 0 0
Intertidal flats 0 0 0
Water courses 0 0 0
Water bodies 0 0 0
Coastal lagoons 0 0 0
Estuaries 1 0 1
Sea and ocean 0 0 0

Table 2

Reclassed values defining the Globcover suitability layers 1 = suitable and 0 = unsuitable. AEVE = Aedes vexans, ANPL = Anopheles plumbeus and CUMO = Culex modestus.


Post-flooding or irrigated croplands (or aquatic) 1 0 1
Rainfed croplands 0 0 0
Mosaic cropland (50–70%)/vegetation (grassland/shrubland/forest) (20–50%) 1 0 1
Mosaic vegetation (grassland/shrubland/forest) (50–70%)/cropland (20–50%) 1 1 1
Closed to open (>15%) broadleaved evergreen or semi-deciduous forest (>5m) 1 1 0
Closed (>40%) broadleaved deciduous forest (>5m) 1 1 0
Open (15–40%) broadleaved deciduous forest/woodland (>5m) 0 1 0
Closed (>40%) needleleaved evergreen forest (>5m) 0 0 0
Open (15–40%) needleleaved deciduous or evergreen forest (>5m) 1 1 0
Closed to open (>15%) mixed broadleaved and needleleaved forest (>5m) 1 1 0
Mosaic forest or shrubland (50–70%)/grassland (20–50%) 1 0 1
Mosaic grassland (50–70%)/forest or shrubland (20–50%) 1 0 1
Closed to open (>15%) (broadleaved or needleleaved, evergreen or deciduous) shrubland (<5m) 0 0 0
Closed to open (>15%) herbaceous vegetation (grassland, savannas or lichens/mosses) 1 0 0
Sparse (<15%) vegetation 0 0 0
Closed to open (>15%) broadleaved forest regularly flooded (semi-permanently or temporarily) – Fresh or brackish water 1 0 1
Closed (>40%) broadleaved forest or shrubland permanently flooded – Saline or brackish water 0 1 0
Closed to open (>15%) grassland or woody vegetation on regularly flooded or waterlogged soil – Fresh, brackish or saline water 1 0 1
Artificial surfaces and associated areas (Urban areas >50%) 0 1 0
Bare areas 0 0 0
Water bodies 0 0 1
Permanent snow and ice 0 0 0

Modelling procedure

A range of modelling techniques available in the VECMAP™ [3] system including Non Linear Discriminant Analysis [6], Logistic Regression [7] and Random Forests [8], using 10–25 repeated bootstraps per run, were used to provide a range of outputs for expert assessment.

The covariates offered to the modelling procedures were drawn from a standardised set of ecological parameters, and in particular a suite of Fourier processed MODIS satellite imagery [9] which provides a range of biologically interpretable variables related to levels and seasonality of temperature and vegetation related factors during the period 2000–2012. These are summarised in Table 3, and are all available to registered members of the VMerge/EDENext Data Website (www.vmergedata.com) [10].

Table 3

Covariates offered to modelling procedures.

1 ED1803A0: Middle infra-red mean 38 ED1814P2: NDVI phase 2
2 ED1803A1: Middle infra-red amplitude 1 39 ED1814P3: NDVI phase 3
3 ED1803A2: Middle infra-red amplitude 2 40 ED1814VR: NDVI variance 41 ED1815A0: EVI mean
4 ED1803A3: Middle infra-red amplitude 3 42 ED1815A1: EVI amplitude 1
5 ED1803MN: Middle infra-red minimum 43 ED1815A2: EVI amplitude 2
6 ED1803MX: Middle infra-red maximum 44 ED1815A3: EVI amplitude 3
7 ED1803P1: Middle infra-red phase 1 45 ED1815MN: EVI minimum
8 ED1803P2: Middle infra-red phase 2 46 ED1815MX: EVI maximum
9 ED1803P3: Middle infra-red phase 3 47 ED1815P1: EVI phase 1
10 ED1803VR: Middle infra-red variance 48 ED1815P2: EVI phase 2
11 ED1807A0: Daytime LST mean 49 ED1815P3: EVI phase 3
12 ED1807A1: Daytime LST amplitude 1 50 ED1815VR: EVI variance
13 ED1807A2: Daytime LST amplitude 2 51 EDBC2K12: BioClim Annual Precipitation
14 ED1807A3: Daytime LST amplitude 3 52 EDBC2K13: BioClim Precipitation of Wettest Month
15 ED1807MN: Daytime LST minimum 53 EDBC2K14: BioClim Precipitation of Driest Month
16 ED1807MX: Daytime LST maximum 54 EDBC2K15: BioClim Precipitation Seasonality (Coefficient of Variation)
17 ED1807P1: Daytime LST phase 1 55 EDBC2K16: BioClim Precipitation of Wettest Quarter
18 ED1807P2: Daytime LST phase 2 56 EDBC2K17: BioClim Precipitation of Driest Quarter
19 ED1807P3: Daytime LST phase 3 57 EDBC2K18: BioClim Precipitation of Warmest Quarter
20 ED1807VR: Daytime LST variance 58 EDBC2K19: BioClim Precipitation of Coldest Quarter
21 ED1808A0: Nighttime LST mean 59 EDV590AS: DEM (Aspect)
22 ED1808A1: Nighttime LST amplitude 1 60 EDV590EL: DEM (Elevation)
23 ED1808A2: Nighttime LST amplitude 2 61 EDV590RG: DEM (Ruggedness)
24 ED1808A3: Nighttime LST amplitude 3 62 EDWC57A0: WORLDCLIM precipitation mean
25 ED1808MN: Nighttime LST minimum 63 EDWC57A1: WORLDCLIM precipitation amplitude 1
26 ED1808MX: Nighttime LST maximum 64 EDWC57A2: WORLDCLIM precipitation amplitude 2
27 ED1808P1: Nighttime LST phase 1 65 EDWC57A3: WORLDCLIM precipitation amplitude 3
28 ED1808P2: Nighttime LST phase 2 66 EDWC57MN: WORLDCLIM precipitation minimum
29 ED1808P3: Nighttime LST phase 3 67 EDWC57MX: WORLDCLIM precipitation maximum
30 ED1808VR: Nighttime LST variance 68 EDWC57P1: WORLDCLIM precipitation phase 1
31 ED1814A0: NDVI mean 69 EDWC57P2: WORLDCLIM precipitation phase 2
32 ED1814A1: NDVI amplitude 1 70 EDWC57P3: WORLDCLIM precipitation phase 3
33 ED1814A2: NDVI amplitude 2 71 EDWC57VR: WORLDCLIM precipitation variance
34 ED1814A3: NDVI amplitude 3 72 EDXXGRPD: GRUMP Population density
35 ED1814MN: NDVI minimum 73 EDXXGRPW: GRUMP Population weighted
36 ED1814MX: NDVI maximum 74 EDXXJRCA: JRC Access
37 ED1814P1: NDVI phase 1 75 EDXXLPG1: Length of Growing Period LGP

LST = Land Surface Temperature. NDVI Normalised Difference vegetation Index; EVI Enhanced Vegetation Index. DEM Digital Elevation. All files starting with ED18 are Fourier processed MODIS Satellite Imagery produced by the TALA research Group Oxford [9].

Files with Bioclim and Worldclim in filename derived from WORLDCLIM datasets [11].

GRUMP derived from population layers produced by [12].

JRC Accessibility downloaded from [13].

Length of growing Period derived from data provided by FAO, Rome. Available from www.vmerge.com [10].

All layers extracted and standardised by ERGO for EDENext (www.edenextdata.com) [14].

Output layers

The suitability masked modelled outputs are produced in the form of probability maps at the pixel level with a resolution of 1 kilometre for each species. Quick view for each vector species is available in Appendix 2 available within this data package.

Sampling strategy

Training sample point data for the model was extracted as follows:

  • Random present points were created from any area within a NUTS 3 polygon recorded as present and where the suitability masked did not indicate unsuitability.
  • Random absence points were selected areas from identified in the mask as unsuitable.

Quality Control

The model outputs were initially evaluated using the standard, and extensive, accuracy metrics (e.g. R-squared, AIC, Kappa, Confusion matrices) provided by the VECMAP™ [3] software. Providing the accuracy metrics indicated sufficient statistical reliability.

The range of models were then sent to the relevant experts who were asked to choose from the selection provided. These included paper authors themselves and individuals listed in the Data Creator section of this paper. This feedback is critical as experts can comment further on how the maps compare to species prevalence on the ground. This can look very different from the presence/absence picture reported at NUTS 3 polygons by VBORNET. This is most obvious in areas such as central Spain where the hot arid environment means large areas may be unsuitable for certain vector species. But presence can be recorded from suitable microenvironments which registers a strong visual signature in that area on the VBORNET present/absence maps. On these occasions we use the expert opinion to validate where we set the environmental limits we refer to earlier in the paper.

In the first phase of modelling (Aedes vexans and Culex modestus) the best model selected by the experts was used as the final model for that species. During phase 2 of the modelling (Anopheles plumbeus), Ensembles of the different model techniques were preferred to attempt to iron out any inherent bias within individual modelling methods. Naturally if a model was not approved by the network experts it was not included in the ensemble.

Ground truthing has yet to be completed on these models although fieldwork has been subsequently sponsored by the VectorNet project which will visit areas which have been modelled, but currently have no data available. So retrospective quality assessments should be completed in the future.


There were no constraints in the data production.





4. Dataset description

Data type

Processed data; Interpretation of data.



Format names and versions


Creation dates


Dataset creators

The following table lists VBORNET contributors who directly contributed to the VBORNET Mosquito database that was used as training data in the models presented in this data paper. While Francis Schaffner’s extensive experience of research in the field of mosquitoes in and around Europe were extremely useful in the land cover suitability exercise and when assessing the maps and the success of the model outputs.

Contributor Affiliation

Albieri, Alessandro Centro Agricoltura Ambiente “Giorgio Nicoli”, Bologna, Italy
Alten, Bulent Hacettepe University, Ankara, Turkey
Alves, Maria Joao Minesterio da Saude, Lisbon, Portugal
Antunes, Ana Faculdade de Medicina Veterinária – Universidade de Lisboa, Lisbon, Portugal
Aranda, Carles Consell Comarcal del Baix Llobregat, Servei de Control de Mosquits, Barcelona, Spain
Beeuwkes, Jacob Laboratory of Entomology, Wageningen, The Netherlands
Bødker, Rene National Veterinary Institute (DTU), Frediksberg, Denmark
Bucher, Edith Biological Laboratory, Laives, Italy
Bueno Mari, Ruben Laboratorios Lokímica, Valencia, Spain
Collantes, Francisco Universidad de Murcia, Murcia, Spain
Dikolli, Enkelejda Institute of Public Health, Tirana, Albania
Eritja, Roger Consell Comarcal del Baix Llobregat – Servei de Control de Mosquits, Barcelona, Spain
Falcuta, Elena Cantacuzino Institute, Bucharest, Romania
Fontenille, Didier IRD/Directeur de l’Institut Pasteur du Cambodge, Cambodge
Gewehr, Sandra Ecodevelopment, Thessaloniki, Greece
Gunay, Filiz Hacettepe University, Ankara, Turkey
Hristovski, Slavco Faculty of Natural Sciences and Mathematics, Skopje, Macedonia
Hufnagl, Peter Austrian Agency for Health and Food Safety (AGES), Vienna, Austria
Ibañez-Justicia, Adolfo Centre for Monitoring of Vectors, Wageningen, the Netherlands
Kalan, Katja University of Primorska, Koper, Slovenia
Kampen, Helge Friedrich-Loeffler-Institut, Greifswald – Insel Riems, Germany
Kavur, Hakan Cukurova University, Dept of Medical Parasitology, Adana, Turkey
Klobucar, Ana Institute of public health “Dr. Andrija Stampar”, Zagreb, Croatia
Krüger, Andreas Berhard Nocht Institut für Tropenmedizin, Hamburg, Germany
Medlock, Jolyon Public Health England, Porton Down, UK
Miranda Chueca, Miguel Angel University of the Balearic Islands, Department of Biology, Palma de Mallorca, Mallorca
Montalvo, Tomas Agència de Salut Pública de Barcelona, Barcelona, Spain
Mosca, Andrea IPLA, Turin Area, Italy
Ognyan, Mikov National Centre of Infectious and Parasitic Diseases, Parasitology and Tropical Medicine, Sofia, Bulgaria
Pajovic, Igor University of Montenegro, Biotechnical Faculty, Montenegro
Perrin, Yvon Centre National d’Expertise sur les Vecteurs, Montpellier, France
Petrić, Dusan Faculty of Agriculture, University of Novi Sad, Serbia
Piazzi, Mauro IPLA, Turin Area, Italy
Plenge-Bönig, Anita Div. Hygiene and Infectious Diseases, Institute for Hygiene and Environment of the City of Hamburg, Hamburg, Germany
Prioteasa, Liviu Cantacuzino Institute, Bucharest, Romania
Regan, Eugenie National Biodiversity Data Centre, Ireland
Sousa, Carla A. Instituto de Higiene e Medicina Tropical, Lisbon, Portugal
Sulesco, Tatiana Academy of Sciences of Moldova, Chisinau, Moldova
Walder, Gernot Medizinische Universität Innsbruck, Division of Hygiene and Medical Microbiology, Innsbruck, Austria
Zamburlini, Renato University of Udine, Dept. of Agricultural and Environmental Science, Udine, Italy
Zygutiene, Milda Centre for Communicable diseases and AIDS, Vilnius, Lithuania



Programming language



The open licence under which the data has been deposited CC-BY.

Accessibility criteria

The data are distributed as GIS raster GeoTIFF formats. Which is a standard proprietary GIS raster format. To access and analyse the raster data directly GeoTIFFs can be read by most GIS software and some other software packages. These formats are compatible with proprietary (ESRI ArcGIS) and open source Quantum GIS (QGIS) or (R-project raster package). If the user has no suitable software already installed the authors suggest downloading the open source QGIS software free of charge from http://www.qgis.org to view these data.

A simple schematic of the data layers and directories found within this data package is shown below with descriptions where filenames are not self-explanatory:

  • Appendices – Directory containing the appendices for this document.
    • ohd_VBNMBD_SchaffnerEtAl_Appendix1.pdf
    • ohd_VBNMBD_SchaffnerEtAl_Appendix2.pdf
  • Quickview – Directory containing small JPEG files allowing the reader to view the data visually without specialist software.
    • appendix1mapsAEVE.jpg – VBORNET Status Aedes vexans
    • appendix1mapsANPL.jpg – VBORNET Status Anopheles plumbeus
    • appendix1mapsCUMO.jpg – VBORNET Culex modestus
    • appendix2mapsAEVE.jpg – VBORNET Status Aedes vexans
    • appendix2mapsANPL.jpg – Model output Anopheles plumbeus
    • appendix2mapsCUMO.jpg – Model output Culex modestus
  • Tiff – Directory containing model output data for display and interrogation within GIS and geostatistical software.*
    • aevemodelMsk.tif – Model output Aedes vexans
    • anplMskensNFL.tif – Model output Anopheles plumbeus
    • cumomodelMsk.tif – Model output Culex modestus

*Only the tif files within this directory are listed. Other file formats of the same name within the directory are ancillary files that provide additional data to the GIS software and as a rule should be copied along with the TIFF file of the same name if you are moving the data between directories.

Publication date


5. Reuse potential

These layers have been created in an attempt to identify probable areas of species distribution where there are currently no sample data. These maps therefore could be useful in identifying suitable areas for further sampling in an attempt to identify the true distribution of the species. The VectorNet project [2] plans to utilise these datasets in such a way.

The covariates of the models are also mainly climate orientated. A possible avenue of further work therefore could be to use the models to assess the potential change in distribution after a shift in climate parameters.