Modelling vector surveillance data of the nuisance mosquito species Anopheles plumbeus (Diptera: Culicidae) in the Netherlands

Abstract

Introduction Landscape modifications, urbanization or changes of use of rural-agricultural areas, can benefit adaptable mosquito species in such a way that they can become a nuisance for humans. In case these mosquito species are capable of transmitting viruses and parasites, this combination could potentially lead to mosquito-borne disease outbreaks. In 1970, the Netherlands was officially declared as malaria free (Plasmodium vivax). Nowadays, the main malaria mosquito vector, Anopheles atroparvus, is confined in coastal areas (presence of brackish water) and not abundant. In contrast, vector surveillance data revealed the mosquito species Anopheles plumbeus, a human biting laboratory vector of West Nile Virus and potential vector of malaria parasite (Plasmodium falciparum), as the 7th mosquito species most commonly found in the country. In addition, nuisance due to high number of individuals of this species is regularly reported. Here, we present a map showing the potential spatial distribution of Anopheles plumbeus in the Netherlands, using presence/absence mosquito data collected during the national vector surveillance and we validate the results with mosquito data obtained from other surveys.

Full text

Faculty
of Veterinary
Medicine
Modelling vector surveillance data of the nuisance
mosquito species Anopheles plumbeus (Diptera:
Culicidae) in the Netherlands
Adolfo Ibañez-Justicia1& Daniela Cianci2
1 Dutch National Centre for Monitoring of Vectors, Food and Consumer Product Safety Authority, Wageningen, The Netherlands
2 Faculty of Veterinary Medicine, Utrecht University, Utrecht, The Netherlands
Introduction
Landscape modifications, urbanization or changes of use of rural-agricultural areas, can benefit adaptable mosquito species in such a way
that they can become a nuisance for humans. In case these mosquito species are capable of transmitting viruses and parasites, this
combination could potentially lead to mosquito-borne disease outbreaks. In 1970, the Netherlands was officially declared as malaria free
(Plasmodium vivax). Nowadays, the main malaria mosquito vector, Anopheles atroparvus, is confined in coastal areas (presence of
brackish water) and not abundant. In contrast, vector surveillance data revealed the mosquito species Anopheles plumbeus, a human
biting laboratory vector of West Nile Virus and potential vector of malaria parasite (Plasmodium falciparum), as the 7th mosquito species
most commonly found in the country. In addition, nuisance due to high number of individuals of this species is regularly reported. Here,
we present a map showing the potential spatial distribution of Anopheles plumbeus in the Netherlands, using presence/absence mosquito
data collected during the national vector surveillance and we validate the results with mosquito data obtained from other surveys.
Methods
Input for the models:
• Mosquito field data: occurrence data collected during the cross-sectional national
mosquito survey program in the Netherlands carried out from April to October 2010-
2013 (778 locations) using Mosquito Magnet Liberty Plus MM3100 traps.
• Environmental data: Temporal Fourier transformed variables from MODIS sensor on
NASA’s Terra and Aqua satellites for 2000-2012 (middle infra-red, enhanced
vegetation index, normalised difference vegetation index, day-time and night-time
land surface temperature) and precipitation (CMORPH, WorldClim), MODIS digital
elevation model and population density in 2000.
Validation:
• Sensitivity (Se), specificity (Sp) and area under the curve (AUC) →predictions vs.
observations of the national mosquito survey.
• External validation (Error rate) →predictions vs. presence points (145), observed in
other surveys and nuisance notifications of An. plumbeus (2009-2014).
The analysis was performed with VecmapTM. Random forest was preferred to non-linear
discriminant analysis and generalised linear models because it was the model with the
best predictive skills (biggest AUC). Variable importance is given as mean decrease in
Gini index.
Results
Discussion
• Precipitation and vegetation indexes were found to be important for predicting habitat suitability for An. plumbeus. This could be
associated with the biology of this species, which typically breeds in tree holes containing rain water with high organic material content.
• Inland parts of the country are shown to be suitable for An. plumbeus. This is in agreement with the locations of reported nuisance (the
south-eastern provinces of Limburg and Brabant). The nuisance seems to be linked to the presence of abandoned pig stables with
uncleaned manure collecting pits.
• Given the results of the internal and external validation, surveillance methodology (cross-sectional, random stratified) seems to be
adequate for generating reliable mosquito data occurrence for input in this modelling technique.
• Low error rate, but the data used for the external validation provide only presence points.
• The reliable estimates in the Netherlands provided by the map could be used as input for risk assessment of VBD.
Correspondence: a.ibanezjusticia@nvwa.nl ACKNOWLEDGEMENTS: David Morley, Nienke Hartemink, Arjan Stroo,
Marian Dik, Wietse den Hartog, Veerle Versteirt, Neil Alexander
Accuracy measures
Sensitivity 0.93
Specificity 0.77
AUC 0.92
Error rate 0.19
Mean decrease GINI index
CMORPH precipitation P1
MODIS Digital Elevation Model
Norm. Dif. Vegetation Index P3
WORLDCLIM precipitation VR
CMORPH precipitation MX
Norm. Dif. Vegetation Index DA
WORLDCLIM precipitation D1
Middle Infra-Red P1
CMORPH precipitation P2
Norm. Dif. Vegetation Index D1