Content uploaded by Wouter Dekoninck
Author content
All content in this area was uploaded by Wouter Dekoninck on Feb 25, 2015
Content may be subject to copyright.
Mosquito vectors of disease: spatial biodiversity,
drivers of change, and risk
”MODIRISK”
V. VERSTEIRT, E. DE CLERCQ, W. DEKONINCK, D. DAMIENS,
A. AYRINHAC, F. JACOBS, W. VAN BORTEL
Biodiversity
SCIENCE FOR A SUSTAINABLE DEVELOPMENT
(SSD)
FINAL REPORT
Mosquito vectors of disease: spatial biodiversity, drivers of
change, and risk
”MODIRISK”
SD/BD/04
Promotors
Marc Coosemans
Institute of Tropical Medicine (ITM),
Department of Parasitology, Nationalestraat 155,
B-2000 Antwerpen
Guy Hendrickx
Avia-GIS, Risschotlei 33, B-2980 Zoersel
Patrick Grootaert
Royal Belgian Institute of Natural Sciences (RBINS),
Department of Entomology
Thierry Hance
Université Catholique de Louvain (UCL)
Unité d’écologie et de biogéographie,
centre de recherche sur la biodiversité
Willem Takken
Wageningen University and Research Centre (WUR)
Laboratory of Entomology,
The Netherlands
Authors
Veerle Versteirt (ITM)
Wim Van Bortel (ITM)
Eva De Clercq (Avia-Gis)
Wouter Dekoninck (RBINS)
David Damiens (UCL)
Audrey Ayrinhac (UCL)
Frans Jacobs (WUR)
D/2012/1191/30
Published in 2011 by the Belgian Science Policy
Avenue Louise 231
Louizalaan 231
B-1050 Brussels
Belgium
Tel: +32 (0)2 238 34 11 – Fax: +32 (0)2 230 59 12
http://www.belspo.be
Contact person: Aline Van der Werf
+32 (0)2 238 36 71
Neither the Belgian Science Policy nor any person acting on behalf of the Belgian Science Policy
is responsible for the use which might be made of the following information. The authors are
responsible for the content.
No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any
form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without
indicating the reference :
Versteirt V., De Clercq E., Dekoninck W., Damiens D., Ayrinhac A., Jacobs F. & Van Bortel W.
"Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk". Final Report.
Brussels : Belgian Science Policy 2009 –152 p. (Research Programme Science for a Sustainable
Development)
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 3
TABLE OF CONTENT
1 Introduction ............................................................................................................................................... 11
2 Biodiversity of mosquitoes in Belgium and The Netherlands ...................................................... 15
2.1 Sample design of the inventory of Belgian Culicidae .................................................................. 15
2.1.1 Tools developed to facilitate the field work ............................................................................... 17
2.2 Spatial distribution and biodiversity of indigenous and exotic mosquitoes in Belgium ........... 20
2.2.1 Mosquito sampling and morphological identification ............................................................... 20
2.2.2 Molecular identification of the Anopheles maculipennis complex ......................................... 20
2.2.3 Data analysis ................................................................................................................................. 21
2.2.4 Spatial distribution and mosquito diversity (2007 and 2008) ................................................. 21
2.3 The importance of museum collections to basic invertebrate inventories ................................ 25
2.3.1 Voucher specimens ...................................................................................................................... 25
2.3.2 Identification .................................................................................................................................. 25
2.3.3 Re-evaluation of RBINS mosquito collection ............................................................................ 25
2.3.4 Calculating trends in mosquito distribution in Belgium ........................................................... 26
2.4 Set up a molecular reference archive of Belgian mosquitoes based on the DNA
barcoding region ............................................................................................................................... 28
2.4.1 General molecular approach....................................................................................................... 28
2.4.2 DNA barcoding of Belgian mosquito species from MODIRISK ............................................. 29
2.4.3 A larval assay to rapidly and correctly identify mosquito species ......................................... 29
2.5 Longitudinal data of Dutch mosquitoes for validation of habitat characteristics, mosquito
species composition and abundance .......................................................................................................... 30
2.5.1 Methodology .................................................................................................................................. 30
2.5.2 Results ........................................................................................................................................... 31
3 Ecology and population dynamics of mosquitoes in Belgium .................................................... 37
3.1 Spatial modelling of indigenous mosquito species in Belgium; detection of hotspots and
implications for field sampling design and risk analysis in Europe ......................................................... 37
3.1.1 Methodology .................................................................................................................................. 37
3.1.2 Modelling outcomes ..................................................................................................................... 43
3.2 Anopheles plumbeus: shift of habitat and risk for autochtonous malaria in Belgium.............. 57
3.2.1 Methodology .................................................................................................................................. 57
3.2.2 Results ........................................................................................................................................... 59
3.3 Anopheles plumbeus nuisance in the Netherlands...................................................................... 64
3.3.1 Methodology .................................................................................................................................. 64
3.3.2 Results and discussion ................................................................................................................ 66
3.4 Repeated introduction of the exotic Aedes japonicus as suggested by microsatellite
markers ............................................................................................................................................... 77
3.4.1 Methodology .................................................................................................................................. 77
3.4.2 Results ........................................................................................................................................... 80
3.5 Competitive interactions between larvae of Aedes japonicus and Culex pipiens under
laboratory conditions ...................................................................................................................................... 84
3.5.1 Methodology .................................................................................................................................. 84
3.5.2 Results ........................................................................................................................................... 86
3.6 Arrival and acclimatisation of the exotic mosquito species Aedes koreicus in Belgium,
Europe ................................................................................................................................................ 89
3.6.1 Methodology .................................................................................................................................. 89
3.6.2 Results ........................................................................................................................................... 90
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 4
4 Biology and genetic population structure of mosquitoes ............................................................. 93
4.1 Influence of temperature on oviposition and larval development of Culex pipiens in
laboratory conditions ........................................................................................................................ 93
4.1.1 Methodology .................................................................................................................................. 93
4.1.2 Results ........................................................................................................................................... 94
4.2 First strong evidence of two genetic forms of Culex pipiens present in sympatric populations
in Europe ............................................................................................................................................ 98
4.2.1 Methodology .................................................................................................................................. 98
4.2.2 Results ........................................................................................................................................... 98
4.3 Genetic characterization of Anopheles claviger and Anopheles plumbeus, potential
autochtonous vectors of malaria ..................................................................................................... 99
4.3.1 Methodology .................................................................................................................................. 99
4.3.2 Results ........................................................................................................................................... 99
5 Surveillance system on mosquitoes ................................................................................................. 101
5.1 Accuracy of using a less intensive MODIRISK sampling strategy for surveillance ............... 101
5.1.1 Defining a cost-effective spatial sampling strategy ............................................................... 101
5.1.2 Results ......................................................................................................................................... 102
6 Policy support ......................................................................................................................................... 107
6.1 Towards predicting distribution patterns in a changing environment ...................................... 108
6.1.1 Aedes japonicus: increased risk? ............................................................................................. 108
6.1.2 How to deal with outbreaks of Anopheles plumbeus? .......................................................... 108
6.1.3 How to predict outbreaks of An. plumbeus in Flanders? ...................................................... 109
6.2 Exotic and endemic pest species ................................................................................................. 109
6.2.1 Aedes japonicus: need for a secure lab .................................................................................. 109
6.2.2 Aedes koreicus ........................................................................................................................... 110
6.2.3 Anopheles plumbeus.................................................................................................................. 110
6.2.4 Aedes albopictus ........................................................................................................................ 111
6.2.5 Measurements against exotic species .................................................................................... 111
7 Dissemination ......................................................................................................................................... 113
7.1 Public link to MoDiRIsk information system ................................................................................ 113
7.1.1 Aimed at the general public ...................................................................................................... 113
7.1.2 Aimed at the end-users.............................................................................................................. 113
7.2 Associated projects ........................................................................................................................ 114
8 Publications ............................................................................................................................................. 115
8.1 ITM .................................................................................................................................................... 115
Avia-GIS ........................................................................................................................................................ 116
RBINS ............................................................................................................................................................ 116
UCL ................................................................................................................................................................ 118
ACKNOWLEDGEMENTS ....................................................................................................................................... 121
REFERENCES ...................................................................................................................................................... 123
ANNEX 1: COPY OF THE PUBLICATIONS ............................................................................................................. 133
ANNEX 2: MINUTES OF THE FOLLOW-UP COMMITTEE ....................................................................................... 133
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 5
ACRONYMS
AVHRR
Advanced Very High Resolution Radiometer
COI
Cytochrome oxidase I
DNA
Deoxyribonucleic acid
GPS
Global Positioning System
IIS
Microsoft Internet Information Services
ITMA
Institute of Tropical Medicine (Antwerp)
ITS2
Internal Transcribed Spacer 2
LST
Land Surface Temperature
MaT
Maximum Apparent Temperature
MGRS
Military Grid Reference System
MODIS
Moderate Resolution Imaging Spectroradiometer
MS SQL
Microsoft Structured Query Language
mtDNA
Mitochondrial DNA
NOAA
National Oceanographic and Atmospheric Administration
PBS
Potential Breeding Site
PCR
Polymerase Chain Reaction
PDA
Personal Digital Assistant
PHP
Hypertext Preprocessor
RBINS
Royal Belgian Institute of Natural Sciences (Brussels)
RDA
Remote Data Access
rDNA
Ribosomal DNA
ROK
Republic of Korea
SBS
Small Business Server
UCL
Université Catholique de Louvain (Louvain)
UTM
Universal Transverse Mercator
WUR
Wageningen University and Research Centre
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 7
SUMMARY
Knowledge of the taxonomic and functional biodiversity of both endemic and invading vector
mosquito species as well as the factors driving change, is missing in Belgium. Acquiring this
knowledge is an essential step towards understanding current risk and preparing for future
treads. Therefore the objectives of the project MODIRISK (Mosquito vectors of disease:
spatial biodiversity, drivers of change, and risk) were (1) to inventory endemic and invading
mosquito species in Belgium considering environmental and taxonomic elements of
biodiversity, (2) to assess the population dynamics of endemic and invasive mosquito
species and their interrelationship (3) to model mosquito biodiversity distribution at a one km
resolution in the Benelux, and (4) to disseminate project outputs to the scientific community,
end users and the general public. During the first phase (years 2007-2008), the project
focused on the inventory activities; setting–up laboratory experiments for studying life history
traits of Culex pipiens in relation to temperature and the first selection of models based on
the field results. Whilst during the second phase of the project (years 2009-2010) the focus
was on the spatial model building and validation, on the longitudinal study and dynamics of
selected indigenous and exotic species that were found during the inventory of the first phase
and on more population genetic driven research.
Sampling design and biodiversity outcomes of the mosquito inventory.
The cross-sectional field survey was conducted in 2007 and 2008 by use of a network of
CO2-baited Mosquito Magnet Liberty Plus traps throughout Belgium in three key habitats.
These habitats (urban, agriculture and nature) were selected based on the Corine database.
Twenty seven traps operated simultaneously (nine per team, three teams leaded by three
partner institutes). Each trap operated seven days on one study site after which it was placed
on the next study site. During the inventory 936 randomly selected sites were selected of
which 97% were sampled. Additionally sites in import risk areas were sampled to evaluate
the presence of exotic mosquito species in Belgium. A MODIRISK website, a palm-to-web
tool and a database were developed which serve now as an example to implement a cluster
of spatial mosquito sampling and modelling projects in several European countries as part of
the IAP program of ESA (European Space Agency). After two years of intensive inventory
and based on as well morphological as molecular identification 23 Culicidae species
belonging to 5 genera were found. The number of species caught is close to the expected
number of species (about 27 species) possibly present in Belgium. Two exotic species were
found, Aedes japonicus japonicus in the province of Namur and Aedes (Finlaya) koreicus in
the province of Limburg. Biodiversity differed amongst the sampled habitats. Diversity indices
indicate the highest richness in natural habitats, although also urban areas score well (which
could have important implications on transmission risks). There were lower in rural areas.
At the Royal Belgian Institute of Natural Sciences (RBINS) about 1400 mosquito-specimens
from the Belgian collection of the Entomology Department were screened and if needed,
added to the collection. All these and previous Belgian records were added to a newly
established database CULIBEL. This database will be integrated into the Belgian Biodiversity
Platform and will be kept updated by RBINS. Both RBINS and MODIRISK collections were
used to compare recent and old data distributions (UTM 10x10km squares). A trend criterion
was made of well surveyed grid cells and a decline of diversity near larger cities could be
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 8
observed. An increase of distribution area was observed for several potential mosquito
vectors having the capacity to use artificial containers as breeding sites. For 23 species there
is a relative change in distribution area in 56 (10x10km) grid cells.
A molecular archive was constructed of all collected species based on the DNA barcoding
region at the ITMA. Moreover, a larval molecular identification assay was developed to
rapidly detect and identify possible invasive species.
Modeling indigenous mosquitoes
A spatial data archive of low resolution remote sensing data was developed. Eco-climatic
zones were identified using an unsupervised k-means clustering. In the first phase, the
objective was to determine whether the data extracted from the MODIS data-series was
useful for the prediction of mosquito distribution. The distribution models were tested on two
species namely Anopheles claviger and Aedes cinereus/geminus, for which a training
sample was selected. The explanatory variables, composed of 28 data layers were
standardised to facilitate model output interpretation. The stepwise regression procedure was
successful in ruling out a considerable number of explanatory variables without decreasing
the predictive value of the models. Using this model a distinct distribution over the study area
was obtained for both species showing that eco-climatic variables are paramount to explain
this variation. In the second phase models were generated for those species present in more
than 20 sampled sites. Fourteen species fit these conditions and were further analyzed using
random regression “forest” statistics. Finally, the model was used to make a map of
environmental suitability (ranging between 0 and 100 %) for the entire area of the BeNeLux.
Field validation in Belgium of the preliminary models was done according to the same
protocol as for the inventory during July-October 2009 in 73 sites (97% of planned sites) and
May-August 2010 in 74 sites (99% of planned sites). During the same period (in 2009 and
2010) a similar validation took place in the Netherlands in respectively 53 and 55 sites. The
north-east of the country showed a high environmental suitability for a majority of the
considered species, such as Aedes cinereus, Anopheles maculipennis, Ae. vexans, Culiseta
annulata, Coquillettidia richiardii, Ae. cantans, and Ae. punctor. In the southern part of the
country a wedge-shaped band of high environmental suitability was noticed.
Ecology, biology and population genetics of selected indigenous and exotic vectors
During the second phase of the project, longitudinal studies were conducted on selected
indigenous and exotic species found during the first phase of the project. RBINS followed a
severe nuisance problem of Anopheles plumbeus in Torhout after complaints to the
MODIRISK project of some of the inhabitants of that area. It was found to be present at very
high numbers, being aggressive to humans and breeding in old abandoned piggeries. This
problem of high nuisance has also been reported in other provinces (Antwerp and Liege). A
similar development of An. plumbeus was observed in the Netherlands (WUR). The
population of Aedes japonicus at Natoye was surveyed by UCL. The results of as well
morphological as molecular studies showed that the species is numerously abundant at the
second hand tyre company but can also colonise natural breeding sites, like tree holes,
nearby. Furthermore, this species competetes with Culex pipiens, delaying its development
untill Ae. japonicus starts to decline. Lab experiences performed by UCL using safe insectory
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 9
facilities of EID (Montpellier) confirmed the high competition of Ae. japonicus larvae in the
presence of Cx pipiens larvae for food. Molecular studies showed that in 2009 three separate
gene pools were present, evincing that probably three importation events occurred. However
in 2010, mixing of these genetic groups occurred. This may increase genetic diversity and so
increase the species invasive potential. Another exotic species, Aedes koreicus, found near
Maasmechelen was followed by ITMA This species has priovously never been reported
outside its natural region of Asia but seemed to be well established in Maasmechelen but still
not spreading. Ae. koreicus is not readily attracted to the currently used traps and was not
found to be aggressive to humans; however the species is, in the former USSR, suspect to
be a vector of Japanese Encephalitis. Furthermore the species colonises both natural and
artificial habitats which can encompass a risk of spreading in Belgium.
A laboratory colony of Culex pipiens was set-up at Université Catholique de Louvain (UCL) to
the study the impact of temperature on life history traits of the most wide spread mosquito
species from Belgium. Males emerged 1,2 to 5,4 days before females and time between
pupation and emergence increased with low temperatures. Although not excessive, larvae
reared at low temperatures (T15, T11) gave bigger adults than the larvae reared at high
temperatures (T28, T20, T35 and T40). Furthermore the results show that temperature
influence the time required to obtain copulation in Cx pipiens. The parameter temperature
has a great influence on development and mating activity of Cx pipiens and the results of
these laboratory tests will be included in transmission models.
Members of the Anopheles genus as well as the population structure of Culex pipiens were
genetically analyzed due their potential as indigenous vectors. Different molecular methods
were applied, tested and compared. For Anopheles species rDNA marker seem to work best,
which coincide with literature data; although some inconsistencies occurred. Using
microsatellite markers two genetic groups were observed in the Culex pipiens individuals
tested, indicating the possible presence of two described forms: Cx pipiens pipiens and Cx
pipiens molestus. Its presence as well in same habitat as time frame have implications on
vector monitoring and transmission risk particularly when hybrids are present acting as
potential bridging vectors between bird and human populations.
Overall, the results of the modeling indicate that a stratified random sample is a good
methodology for selection sampling locations for mosquito monitoring. However for future
monitoring this strategy is labor intensive and costly thus it was necessary to analyze how
the sampling size could be reduced without reducing the accuracy of the model outputs.
Based on a simulation of four indicator species, one trap per 300km² resolution (in stead of
one trap every 30km² used in present study) would be an appropriate for further monintoring.
Based on these four indicator species, hotspots can be identified for intensifying trapping.
These findings will make future monitoring more efficient and less expensive.
Key words: inventory, Belgium, The Netherlands, indigenous and exotic species, modeling,
sampling strategy, biodiversity, vector population studies, molecular identification, DNA
barcoding, Aedes japonicus, Aedes koreicus, Anopheles plumbeus, Anopheles maculipennis
s.l., Culex pipiens
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 11
1 INTRODUCTION
Increased globalization, landscape management and changing socio economic behavior
create suitable conditions for the (re)emergence of vector-borne diseases in Europe. It was
estimated that during the decade 1990-2000 nearly a third (29%) of the recorded events
related to emerging infectious diseases were due to vector-borne diseases (Hendrickx &
Lancelot 2010). Furthermore, Reiter (2010) stated that especially the importation and
establishment of vector-borne pathogens that have a relatively low profile in their current
habitat is a serious danger to Europe and throughout the world. Of these, mosquito-borne
diseases are prime candidates. Recent outbreaks of West Nile virus in Romania and Greece
(Papa et al. 2010, Sirbu et al. 2011) and autochthonous transmission of Dengue and
Chikungunya in southern France (La Ruche et al. 2010), indicate the existing danger of local
transmissions. A key factor in this increased susceptibility of Europe to vector borne diseases
is the rapidly expanding transportation of humans and goods. Due to this intensification of
worldwide trade and tourism which increases indeed the likelihood of human-mediated
introduction of organisms, posing a risk to biodiversity, economy and human health (Tatem et
al. 2006, Lines 2007). The probability of a country or region for bioinvasions appears
generally to be correlated to the extent of international trade it conducts (economic variable)
and to its national wealth and human population density (demographic variable) (Pysek et al.
2010). In the last decade, dramatic increase in traffic has been observed between eastern
Asia and Europe and North America (Tatem 2009). Due to the mix of economic and
demographic influences combined with the further increasing speed and pervasion of
modern transport networks, global trade, transport and tourism are becoming more and more
pivotal in the spread of vectors and the infectious diseases they transmit (Wilson 1995,.
Wilson et al. 2009, Pysek et al. 2010). There are three possible scenarios that may influence
the risk of transmission and outbreaks of arbovirusses: the import of an exotic species that
can transmit an arbovirus, the import of an arbovirus that is transmitted by an exotic
established mosquito, the import of an arbovirus that is transmitted by indigenous species.
Of all mosquitoes worldwide, Aedines seem to have the highest invasive potential as the
eggs of most species tolerate considerable periods of desiccation thus surviving long
transports (Reiter & Sprenger 1987). Furthermore, many Aedes species prefer small and
often man-made containers as oviposition sites, including used tyres and „lucky bamboo‟
containers (Eritja et al. 2005), which are frequently moved across international borders. In
southern Europe, past and recent accidental importations of mosquito vector species such as
Aedes (Stegomyia) aegypti (L.) and Aedes (Stegomyia) albopictus (Skuse) have created
suitable conditions for local transmission of arbovirusses (Christophers 1960, Eritja et al.
2005, Angelini et al. 2007, Fontenille et al. 2007, Angelini et al. 2008). Aedes species
imported into northern Europe include Ae. albopictus (Schaffner et al. 2004), Aedes (Finlaya)
japonicus japonicus (Theobald) (Schaffner et al. 2009, Versteirt et al. 2009), and Aedes
(Ochlerotatus) atropalpus (Coquillett) (Scholte et al. 2009). Besides these Aedes species, the
invasive potential of Culex and Anopheles species should not be underestimated, although it
seems that these introductions were more frequently in past eras. Culex pipiens was
introduced into the USA more than hundred years ago (Kesavaraju et al. 2011) and Culex
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 12
quinquefasciatus was introduced into Australia and New Zealand around 1848 (Lounibos
2002), both established successfully. Many Anopheles species were imported into North and
South America due to increased trade in early 1900. For example, the African malaria vector
Anopheles gambiae s.l arrived from West Africa in 1930 and was rapidly established and
spreading into northeastern Brazil (Lounibos 2002). Larvae or adults of this anopheline are
believed to have traveled by air or fastpassenger ship from Dakar, Senegal, to Natal, Brazil
(Lounibos 2002), where the first malaria epidemic attributable to the species occurred in
March–May, 1930. Although malaria was endemic in northeastern Brazil, the native
anopheline vectors were less efficient transmitters compared with the highly anthropophilic
and endophilic An. arabiensis. However Anopheline mosquitoes, among which are the
vectors of malaria, are less succesfull in colonization new regions.
Not only importation of a vector species can cause problems, also the introduction of
pathogens poses a risk for human and animal health. Global travel has grown continuously
during the last 2 centuries, with especially a massive increase in the last 50 years
(www.world-tourism.org). Worldwide tourism allows humans to interact with microbes and
spread pathogens to new locations and populations (Mavroidi 2008). This massive increase
in international travel has reduced or even eliminated natural geographic barriers and
increased the spread of infectious diseases worldwide. In addition, the world population
continues to grow rapidly causing more and more people to live closer together, which again
increases the risk of spreading diseases. Apart from humans travelling, the transport of
infected animals can cause the same problems. During the Bluetongue outbreaks in Europe,
the transport of sheep and cattle became very restricted by the European Commission to
avoid outbreaks in new locations. The outbreak of WNV in the USA in 1999 started in New
York City and it is assumed that the pathogen was imported along with an exotic bird in a zoo
nearby (Lanciotti et al. 1999). Globalization is therefore potentially a far greater challenge to
public health in Europe than any future changes in climate would be.
Besides abiotic risk factors and the hazards posed by invading mosquitoes, biotic factors
linked to indigenous and exotic species can also play a role in possible transmission and
outbreaks of mosquito borne diseases. For instance, spatial distribution of adult mosquitoes
is associated with various environmental and climatic factors. Alterations of these factors can
produce significant differences in the distribution of adult mosquito populations which may
have important implications for current risk on (re)emergence and epidemiology of vector-
borne diseases and the implementation of control strategies. Climate can influence the host,
pathogen and vector populations in several ways and thereby increase or decrease the
incidence or prevalence of arboviral diseases. For example, it can cause a shift in the
geographical distribution and the density of host and vector populations, potentially bringing
the two in closer contact and increasing the prevalence of an infection (Mills et al. 2010).
Recurring year to year fluctuating in day and night temperature and precipitation will have an
impact on oviposition strategies, larval development and overall larval and adult survival,
thus effecting directly population density of many vector species in Belgium. Temperatures
can also have profound effects on the development of pathogens and pathogen loads in
arthropod vectors (Sutherst 2004, Gage et al. 2008). However, knowledge of the taxonomic
and functional biodiversity of both indigenous and invading vector mosquito species as well
as the factors driving change is missing in Belgium. Except for a paper published in 2004,
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 13
recent information on the presence and distribution of indigenous and invasive mosquito
(Diptera: Culicidae) species in Belgium is lacking. Before that, the sole published information
relates to three studies. In the beginning of the previous century the first species list on
Belgian mosquitoes was published after extensive collection campaigns of insects all over
Belgium (Goetghebuer 1910). This list was extended and a new publication on Culicidae
present in Belgium appeared in 1925 (Goetghebuer). In the 1940s the presence of malaria
mosquitoes and associated malaria transmission risk was studied at several occasions in
Belgium (Rodhain and van Hoof 1942, Rodhain and van Hoof 1943, Rodhain and Van
Mechelen 1944) and during the early 1950‟s mosquito nuisance was investigated around the
city of Antwerp (Wanson 1952). Albeit a large collection is present at the Royal Belgian
Institute of Natural Sciences (RBINS) many of this collected material was never identified. In
1991 a checklist of Belgian Culicidae was created compromising 24 species present in the
RBINS collection (Gosseries and Goddeeris 1991) collected mainly between 1910 and 1960.
Moreover, records are sometimes clustered in space and time, as, for example, between
1940 and 1950 mosquitoes were mainly collected around Ghent and Brussels (Dekoninck et
al. 2011). Based on all this literature information, the identification key for European mosquito
species (Schaffner et al. 2001) record 26 species possibly present in Belgium, although a
total of 30 different species were mentioned in all previously published lists, some were
discarded due to the low probability of occurrence in Belgium.
Acquiring detailed knowledge on distribution and population dynamics of Belgian mosquitoes
is an essential step to anticipate, prevent or prepare for the establishment and spread of
vector-borne diseases. National presence-absence maps are the first step for understanding
current risk and preparing for future threats and one of the main objectives of this study was
to compile predictive modeling maps which give an overview of the current distribution of the
Belgian mosquito fauna. Next to distribution data, information on bionomics and population
dynamics of mosquito species is crucial in the evaluation of the possible risk and in the
preparedness to propose management politics of mosquito species (Parmesan and Yohe
2003, Zell 2004).
The inventory done in the framework of the MODIRISK project is based on a random
(statistical) approach that is designed for model building. This is unique in Europe (and even
in world) since most models are based on historical records. Based on the experience gained
during MODIRISK a cost-effective sampling strategy will be designed for use in follow-up and
similar studies. Modeling will mainly assist in defining the minimal field sample needed to
produce acceptable distribution maps, and how these samples are best distributed in space.
Furthermore it contributes to the development of state of the art scientific tools integrating
collection-based information technology at various resolutions with geographic mapping
efforts and remote sensing driven continuous distribution models. This enables to better
describe the spatial distribution of mosquito biodiversity, and to understand how these are
organized in communities and habitats.
In this report we describe the objectives and the applied methodology of the MODIRISK
(Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk) project and we
discuss main results and achievements to formulate in the end recommendations for
preparedness and responsiveness. The project directly contributes to discovering biodiversity
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 14
and monitoring/predicting its changes, and actively prepares to address issues such as the
assessment of impacts of biodiversity change with particular reference to new invasive
mosquito species and the risk to introduce new pathogens. An improved understanding of
the biodiversity of mosquito vectors is an essential step towards an improved understanding
of the ecology of the diseases they transmit. Looking at Europe, vector borne diseases are
becoming more and more a topical issue that we cannot simply ignore.
MODIRISK fits in the science plan of the global initiative Diversitas, which was one of the
main drivers of the „Research programme Science for a Sustainable Development (SSD)‟.
MODIRISK directly contributes to discovering biodiversity and monitoring/predicting its
changes, and actively prepares to address issues such as the assessment of impacts of
biodiversity change with particular reference to new invasive species and the risk to
introduce new pathogens (impact on health). These are two of the three key topics,
respectively addressed in the three „core projects‟ of Diversitas.
The filling of an essential knowledge gap in Europe, and the expansion of model outputs
through linking up with a project in The Netherlands, enables the project to produce more
robust results and to prepare better for later expansion of activities in Europe. MODIRISK
plays its role as interplay between newly gained insights and the end-users. MODIRISK
participated to the risk assessment group of the Scientific Institute of Public Health
concerning the presence of the exotic vector species Aedes. j. japonicus and Ae. koreicus
and participated to meetings of the European Centres for Disease Prevention and control
Control (ECDC). Links were made with the Belgium Forum on Invasive Species by
participating to the discussions on the „Guidelines for environmental impact assessment and
list classification of non-native organisms in Belgium” and by acting as member of the
scientific committee of the „Science Meeting Aliens” conference on biological invasion (11th of
May 2009). MODIRISK was consulted by the AGORA project on „set-up of monitoring of
potential effects of climate change on human health and on the health of animals. In January
2009 MODIRISK organised a workshop on vector control in Belgium bringing together
persons potentially involved in the decision making process on vector control, other
stakeholders and interested persons. A strong plea was made to set-up entomological
surveillance to follow the situation and to evaluate the spread of the exotic species in
Belgium. Moreover a flow chart of competence and responsibilities of the different authorities
potentially involved in vector control should be established. Members of the MODIRISK
project attended to several meetings and workshops on national and international level.
Moreover MODIRISK was asked to take part in a number of national (VIRORISK, CULIMON,
BC42W) as international (VBORNET, Vecmap) projects. MODIRISK has become the contact
for numerous questions and problems relating biting insects especially mosquitoes and biting
midges.
The report is subdivided into different important themes as biodiversity of indigenous and
exotic mosquitoes in Belgium and The Netherlands; ecology and population dynamics of
mosquitoes in Belgium; biology and genetic population structure of mosquitoes in Belgium
and application for surveillance systems in Belgium and Europe
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 15
2 BIODIVERSITY OF MOSQUITOES IN BELGIUM AND THE
NETHERLANDS
2.1 Sample design of the inventory of Belgian Culicidae
Cross-sectional field surveys were conducted during the first phase of the project to
inventorize Culicidae. CO2-baited traps were used throughout Belgium in a grid-based
sampling approach where different habitats in each grid were sampled. One trapping device,
Mosquito Magnet Liberty Plus which is a high performance CO2-baited trap was used. In a
recent study it outperformed both in number of specimens and number of genera collected
compared to seven other trap systems (Dennett et al. 2004). Furthermore it is the only
commercial available trap type that allows a certain autonomy which was necessary for our
trapping scheme This trap runs on propane gas which is converted into CO2, which is a good
attractant for most mosquitoes. This CO2 leaves the trap through a central tube with a
constant flow (500 ml/min). Attracted insects are drawn into the apparatus and trapped in a
small net. Mosquitoes are killed by placing the nets in the freezer at -20oC. This trap was also
employed in the Dutch studies.
Using the Corine Land Cover (2000) classification (NGI, 2004), potential mosquito habitats
were delineated. The Corine Land Cover Classes were regrouped in 6 classes. These
classes are shown in Figure 1.
This data layer was overlayed with the Military Grid Reference System (MGRS) which is
used internationally for species mapping such as mammals (Amori et al. 2002) and birds
(Hagemeir and Blair 1997). The MGRS is an extension of the UTM system. Across Belgium,
312 10x10km MGRS cells are identified. Per cell an average of three points is to be sampled,
thus the total number of sample points amounts to 936. Per aggregated class the number of
points assigned was proportional to its total surface (Table ) and each point received a
random set of X and Y coordinates.
A
B
Figure 1. (a) Corine Land Cover Classification and (b): Aggregated Land Cover Classes
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 16
Table I. Number of samples per stratum
Class
Pixels
Percentage
Samples
Urban
769723
17.7 %
173
Agriculture
2514014
57.7 %
564
Natural
888272
20.4 %
199
Specific
56479
1.3 %
13
Secondary
103411
2.4 %
23
Given the random location each point was assigned to a full address, i.e. street, house
number, and postal code using the geocoding functionality from ArcView3.2 and based on
the geocoding street network data layer (TeleAtlas MultiStreetNet). Each point was initially
linked to the nearest street segment (i.e. a segment of a street between cross roads) using a
spatial join. The house number was generated randomly within the range of house numbers
of that street. If there were no houses in the street segment, and the point belonged to the
urban category, the nearest street segment with houses was used. If the point belonged to
the nature or agriculture category, it only received a street name and no house number. The
survey was conducted over a period of two years with the first set of 250 points sampled in
the first half of the season (referred to as „spring‟) of 2007, the second set in the second half
of the season (referred to as „summer‟) in 2007 and the third and fourth set similarly spread
in 2008 (Figure 2). No sampling was conducted during winter. In 2009 and 2010 additional
samples were taken to validate model outputs (see 3.1).
Figure 2: Sampling scheme for the extensive survey of Belgium
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 17
2.1.1 Tools developed to facilitate the field work
Websites
A general website was designed providing information on the project and the project outputs
(www.modirisk.be). This website provides a link to a site composed of a public part as well as
a private part. User access security was implemented to limit the access to the private part.
The public part showed the general progress of the field work (http://modirisk.avia-gis.com/).
Each sample point is colour coded according to its status (not sampled, visited, processed).
The private website has two basic functionalities: determining the location of the sampling
point and filling out the different forms. Using the sampling point locator (Figure 3), the field
teams could identify each sampling point through the interactive map. This map allowed full
capabilities of zooming (Figure 4) and panning. The sampling points were colour coded in the
same manner as on the public web site. For each sampling point, the field teams could query
which team was responsible for trapping, and during which season the point had to be
sampled. The high resolution satellite imagery background (Figure 4) enabled a rapid
assessment of accessibility and contributed to efficient planning of the field visits.
Figure 3. Point locator on private website
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 18
Figure 4. Zooming capabilities
In the administration section, the field teams could complete the sampling forms. The data
from the first form was uploaded from the field teams‟ PDA. The other forms had to be
completed manually. Once the data was uploaded or entered, the data base was
automatically updated. The status of each sampling point on the web-maps automatically
reflected any change in the database. For validation purposes both the website and the
database were extended. On the website the listing page was updated in order to avoid
mixing of the original points and the validation points.
Software development for PDA
The Fujitsu Siemens LOOX N560 was selected (after comparison with 29 others) because of
its high quality and the respective performance of its screen, processor and memory. The
PDA was running Windows Mobile 5 and had a built-in GPS. The field form was implemented
on the PDA. All functionalities were individually tested. During the test phase, the software
was further refined and the user interface was optimized. A manual was written for the
software, and uploaded to the main private MODIRISK website. Permanent on-line help and
troubleshooting was available to all MODIRISK partners.The software downloaded the
coordinates from the central database server to the MS SQL Server Mobile database on the
PDA. The user could query and edit the data, and all changes were uploaded after the field
visit directly in the central MS SQL Server database.
Database setup
The database server uses Windows Server 2003 SBS R2 as operating system, and is
running IIS with PHP for site development, MS SQL Server for database development and
SQL Server Mobile Tools to allow remote access from a PDA. Three types of MODIRISK
forms were prepared by the MODIRISK coordinator and adapted during a group session at
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 19
ITM: (1) Field form, (2) Morphological identification form, (3) Mosquito storage form. Based
on these, relevant tables (Figure 5) were developed by Avia-GIS, implemented in the
database, and transferred to the web server.
Figure 5. Tables present in the MS SQL Server Database
The database was modified for validation accordingly: the points selected were entered in
sample point database and in order to distinguish them from the original data points the ID
numbering was in a different ID range.
During the validation phase, forms for identification and validation were edited and added to
the private website of MODIRISK. The identification forms allowed the users to fill out data
with respect to the morphological identification of species collected in the trap (Form 4), the
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 20
morphological identification of the tube content of the collected adults (Form 5) and larvae
(Form 6). The database was adapted to allow the storage of the data coming from these 3
new forms.
2.2 Spatial distribution and biodiversity of indigenous and exotic
mosquitoes in Belgium
2.2.1 Mosquito sampling and morphological identification
Mosquitoes were sampled from May till October 2007 and 2008 (inventory) and from August
till October 2009 and May till August 2010 (validation) according to the sampling design
explained above. Three teams were involved in the field collections namely, ITM, RBINS and
UCL. In order to standardize the field work a written standard protocol was elaborated by the
ITM in cooperation with the other field teams.
Before the implementation of the field work a short training in the use of the Mosquito Magnet
Liberty plus trap was organised from 3-4 April 2007 at Wageningen University. During the
field work, twenty seven traps operated simultaneously (9 by team). Each trap operated
seven days on one study site after which it was placed on the next study site. During the first
phase of the project (inventory), field work was done on Monday, Tuesday, and Wednesday:
each day three traps were emptied and replaced. The remaining days were used for the
organisation of the field work and the morphological identification of the collected
mosquitoes. During the second phase of the project (validation), each field team was free to
organise the field work as suited.
Morphological identification was done mainly using the electronic identification key of
Schaffner et al. (2001) and the paper key of Becker et al. (2003). Data were stored into the
web based data base as described above.
Project staff (2 persons ITM, 2 persons RBINS, 1 person UCL, 1 person WUR) was trained
during two sessions in state of the art taxonomic identification using morphological
techniques based on reference collection from an expert in European morphological
mosquito identification and field collected mosquitoes. Moreover, in July 2007 and January
2009 a quality control was done by this international expert. All teams performed well and
improved their identifications skills during the training (from 88% accuracy before the second
training to almost 100% at the end of the second training). Further verification of performed
identification (validation) was done internally.
2.2.2 Molecular identification of the Anopheles maculipennis complex
Members of the Anopheles maculipennis complex were identified to species level using a
specific designed PCR-RFLP method (Nicolescu et al. 2004) of the Internal Transcribed
Spacer 2 (ITS2) region. DNA extraction was performed using the protocol described by
Collins et al. (1987) after which the ITS2 region was amplified using primers described by
Collins and Paskewitz (1996). Consequently the positive amplification products were
digested using CfoI restriction enzyme (Roche Molecular Biochemicals Ltd, Sussex,
England). Restriction fragments were visualised on a 3% agarose gel. A selected number (5
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 21
of each species) of positive ITS2 PCR amplifications were send for sequencing
(Genoscreen, Lille, France).
2.2.3 Data analysis
Input data was analysed at the main Corine habitat. All traps used in sites belonging to one
of the five defined main habitat groups were treated as a single input point to test habitat
effect. Alpha diversity was calculated using three different indices. Simpson's index of
diversity, 1 − D = 1 − Σ[ni*(ni − 1)/N*(N − 1)], where ni is number of the ith species and N is
the number of individuals in the studied habitat, as a measurement for the probability that two
randomly selected individuals in an area belong to different species. The closer 1-D is to one,
the more diverse the habitat is. Shannon-Wiener index (H′ = − Σpi * lnpi, where pi is the
proportion of the ith species in the studied habitat) was used as a measure of community
heterogeneity (Krebs 1989); eveness (E‟ = S/ln(H‟), where S is the total number of species)
calculates how individuals are distributed among species habitat. Rarefaction based
estimates were calculated using EcoSim (Gotelli & Entsminger 2001) to estimate and
compare the relative abundance and the density of mosquito species among habitats. The
use of rarefaction allows comparison of the number of species in samples of different sizes
by limiting the sample to the smallest size in the set of populations and calculating the
species richness. Individual based rarefaction curves were created in GraphPad.
2.2.4 Spatial distribution and mosquito diversity (2007 and 2008)
1
Based on the above described sampling strategy 936 sites were randomly identified in three
key habitats (urban, agriculture and natural); 97% of these were effectively sampled.
Additionally 45 import risk areas were included in the inventory of which 27 areas were
located in natural habitat (IRA-nature,in total 37 traps) and 18 companies (IRA-industry, total
24 traps). The import risk areas for exotic mosquitoes included zoos, safari parks, second
hand tire import/storage companies, lucky bamboo importers, harbours and airports. Risk
areas for import of pathogens included protected areas involving presence of large numbers
of migratory birds. Table II summarizes the actual number of sites sampled in each category.
Table II Overview of the number of study sites.
YEAR 1 (2007)
YEAR 2 (2008)
Habitats
Selected
To be done
Done
To be done
Done
Urban
173
82
81
91
90
Agriculture
564
283
280
281
277
Natural
199
101
90
98
94
Import risk areas
42
21
26
21
37
1
Note: data gathered during field work validation in 2009 & 2010 has been used for model validation (see xxx) and is included in
this part of the report
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 22
Species diversity, abundance and eveness
A total of 26533 individuals, morphologically belonging to 22 species and 5 genera, were
collected and morphologically identified (Figure 6 ). The most species rich genus in Belgium
was Aedes whereas Coquillettidia was only represented by 1 species. The five most
abundant species were Culex pipiens (61.62%), Coquillettidia richiardii (15.43%), Aedes
cinereus/geminus (5.04%), Anopheles claviger (3.52%) and Aedes vexans (2.93%). The high
abundance of Cq richiardii is however due to one study site, a nature reserve in the harbour
of Antwerp, where more than 3700 specimens of this species were collected. The species
was caught only in 38 study sites. Interestingly is the general occurrence of Anopheles
species, mainly An. claviger and An. plumbeus. Two exotic mosquito species were collected.
Aedes japonicus japonicus
2
was found in a second hand tyre company (Versteirt et al. 2009)
whereas Aedes koreicus was found in a randomly selected site near an old sand mine in
nature restoration but close to a recycle company and industrial zone (Versteirt et al) (see
also 3.6.).
Eighty individuals of An. maculipennis s.l. were collected of which 67 were molecularly
identified using the protocol of Nicolescu et al (2004). Forty three individuals belonged to
An.maculipennis s.s. (64.18%), 24 to An. messae. This PCR-RFLP does not make the
differentiation between An. messae and the recently described and closely related An.
daciae (Nicolescu et al. 2004). Although positive ITS2 amplifications of An. messae were
sequenced no clear separation between the latter species could be observed. Both species
were especially present in the northern part of Belgium, in 6 sites they occurred sympatric, in
24 other positive sites either one of them occurred (7 sites positive for An. messae, 17 for
An.s maculipennis s.s.).
Combining morphological and molecular identification, a total of 23 species were collected
during the inventory. Seen individuals of Ae. geminus and Ae. cinereus can only be
morphological separated with certainty based on the shape of male genitalia and seen little is
known on the medical importance of these sibling species, individuals were not further
distinguished.
The taxonomic biodiversity differed among the three main habitats. A large number of
species has been found in the import risk areas whereas it only represented a small portion
(6-7%) of the sample sites. Most species (21) were caught in the IRA-nature habitat (4% of
the study sites) followed by rural (56% of the study sites), and nature habitat (20% of the
study sites) where respectively 20 and 19 species were collected. In sites classified as
urban, representing 17% of the study sites, 16 species were collected; lowest number of
species were collected in IRA_industry sites (Table III). Total abundance was highest in rural
sites, largely due to a high number of Cx pipiens individuals (>10 000); whilst the mean value
was actually the lowest (20,97 individuals per trap). Furthermore, individual based rarefaction
2
Reinert, (2000) divided the genus Aedes Meigen into genera Aedes and Ochlerotatus (Lynch Arribalizaga) on the basis of
“consistent primary characters” and supplemental features. Ochlerotatus was elevated to generic rank and was further divided
into two sections based on features of the fourth-instar larvae and pupae. However the controversy surrounding this separation
has left non-taxonomists in doubt. The authors would follow in this article the taxonomy of Reinert (2000) . Aedes japonicus
japonicus is one of morphologically similar subspecies originating from Japan, Korea, Taiwan, eastern China and Russia,bu
seen the other species do not occur here we will hereafter use Aedes japonicus to refer to this species.
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 23
curves suggest that, for equal sample size, species richness in rural habitats is only slightly
lower than in habitats with highest species richness (Figure 7). Species diversity (H‟ +
Simpson) and evenness (E‟) were both higher in nature and IRA_nature sites than in other,
which can also be observed by the rarefaction curve. Individual based rarefaction curves
suggest that for equal sample sizes, species richness might be slightly higher in nature and
IRA_nature than in urban and rural areas; lowest species richness could be expected in IRA-
industry although the rarefaction curve of this latter habitat is still far from reaching a plateau.
Half of the (morphological) identified species were shared amongst habitat.
Looking at the species occurring in all main Corine habitats, eight species seem very
common in Belgium: Ae. cantans, Ae. cinereus/geminus, An. claviger, An. plumbeus, Cq
richiardii, Cs annulata, Cx pipens and Cx torrentium Although these species are found over a
large range of spatial divisions, some species were more collected in one habitat. Anopheles
claviger was especially found in in sites classified as rural (44%) or IRA_nature (39%); An.
plumbeus was mostly present in rural (51%) habitats. Coquillettidia richiardii was found in
high abundances in IRA-nature habitats (98%). Other species, which are not exclusively
linked to one of the spatial levels, show a distinct and sometimes somewhat unexpected
preferences. Aedes vexans, for example, was especially found in sies classified as urban
habitat (78,5%).
Table III. Taxonomic diversity by main Corine habitat
Main Corine habitat
Urban
Rural
Nature
IRA_nature
IRA_industry
Number of sites sampled
153
568
183
37
27
Number of species collected
16
20
19
21
12
Total of specimens collected
3992
12100
2553
6857
1031
Mean number of specimens per trap
26,09
21,30
13,95
185,32
38,19
Species diversity (Simpson)
0,476
0,299
0,765
0,605
0,099
Species diversity (Shannon H')
1,107
0,830
1,951
1,354
0,274
Species Eveness ( Shannon E')
0,399
0,277
0,663
0,438
0,110
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 24
Number of individuals
0200 400 600 800 1000
Culex torrentium
Culex territans
Culex pipiens
Culiseta morsitans
Culiseta annulata
Coquillettidia richiardii
Anopheles plumbeus
Anopheles maculipennis s.l.
Anopheles claviger
Aedes vexans
Aedes sticticus
Aedes koreicus
Aedes rusticus
Aedes punctor
Aedes japonicus
Aedes geniculatus
Aedes detritus s.s.
Aedes communis
Aedes cinereus/geminus
Aedes caspius
Aedes cantans
Aedes annulipes
5000 1000015000
Figure 6 Species collected during the two years inventory study (2007-2008).
Figure 7: Species diversity in main Corine habitat based on Simpson (black) and Shannon (Grey)
diversity indices
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 25
2.3 The importance of museum collections to basic invertebrate
inventories
2.3.1 Voucher specimens
In the beginning of past century many mosquitoes were collected all over Belgium by
dipterologists as M. Goetghebuer and M. Bequaert who both built up the most representative
and rich collections of Belgian Diptera, preserved at RBINS (Grootaert et al., 1991). In the
Belgian Culicidae collection of RBINS four subcollections are present: a general collection,
two subcollections (Goetghebeur and Becquart), and a subcollection of unidentified
specimens i.e. the supplements. The subcollection Bequaert was mainly collected between
1912-1958 and counts 135 voucher specimens. The subcollection Goetghebuer was
collected between 1909-1946 (mainly between the period 1910-1930) and counts 269
specimens. In the general collection 241 specimens are present all of them collected
between 1878-1967 (mainly between 1880-1925). The supplements are the largest
subcollection with 737 specimens collected between 1892-2005 (mainly during 1920-1960).
2.3.2 Identification
All voucher specimens from the available collections were re-identified at the species level
using Schaffner et al., 2001.
2.3.3 Re-evaluation of RBINS mosquito collection
The most recent checklist of the Belgian Culicidae counted 24 species, which was the
number of identified species found in RBINS collection and additional species mentioned in
the card-indexes of RBINS (Gosseries and Goddeeris 1991). The latter authors suggested at
that time that the real number of species to be expected to occur in Belgium being
approximately 50. However since 1991 only a few mosquito species were added to the
Belgian fauna; Culex hortensis (Versteirt et al. 2009) and Culiseta ochroptera (Schaffner
pers. com.). All 1381 specimens (24 species) in RBINS collections were reidentified and
digitised. Most of the specimens (77%) were collected between 1910 and 1960 (Figure 8).
Most specimens were collected between 1940 and 1950. The intensity of research and
mosquito-sampling fluctuated during this period, as revealed by the number of voucher
specimens per decade (Figure 8). The oldest specimens (collected in 1878) are deposit in
the general collection. In this collection 16 species were discovered, in the subcollection
Bequaert, the subcollection Goetghebuer and in the supplements respectively 18 species, 21
species and 20 species were counted.
Culex pipiens and Culiseta annulata were the most abundant recorded species present in the
collection. Also many voucher specimens of Aedes punctor were present.
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 26
Number of specimens of Culicidae in RBINS collection
0
50
100
150
200
250
300
350
400
<1900
1900-1910
1910-1920
1920-1930
1930-1940
1940-1950
1950-1960
1960-1970
1970-1980
1980-1990
1990-2000
>2000
Figure 8: Number of identified voucher specimens for each considered period
Records of rare and interesting species
During the re-evaluation some so far seldomly recorded mosquito species were discovered
in the supplements. A female of Cx hortensis, only recently added to the Belgian fauna
(Versteirt et al. 2009), was collected by Bequaert in Aywaille, Nonceveux on 6/vii/1947. Other
remarkable records are retrieved from two sites were Goetghebuer collected Cs fumipennis:
on 1/vii/1920 and 11/vi/1914 in Hockai near Stavelot (Prov. Liège), probably in bogs and wet
heathland vegetations of the Hautes Fagnes. Later it was also collected in Lovendem,
Vinderhoute (Prov. Eastern-Flanders) on 9/5/1920. Two specimens of Cs subochrea were
found in the collection of BEQUAERT: one male collected at Destelbergen, Heusden (Prov.
Eastern-Flanders), 2.viii.1944 and one female collected in Blankenberge, (Prov. Western-
Flanders), 6.xi.1955. No evidence was found that introductions of exotic mosquitoes were
noticed or invasive specimens were collected prior to 2000.
Establishment of a MODIRISK-subcollection in RBINS mosquito collection
At RBINS a new Culicidae subcollection with specimens collected during the MODIRISK
project was generated. Of all species collected during the project specimens of at least 5-10
sites are stored (if the species was collected in more than 5-10 sites). Moreover all partners
involved in the morphological identification started their own reference collection with help
from the expert. Therefore 1-5 correctly identified specimens per species were pinned and
stored in insect boxes.
2.3.4 Calculating trends in mosquito distribution in Belgium
Both datasets were used to test whether or not trends in changing mosquito diversity were
observed for several regions in Belgium. Results of this large scaled survey provided us with
an estimated present observed mosquito diversity (hereafter called PresDiv). Estimated
former observed mosquito diversity was obtained from the revision of RBINS mosquito-
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 27
collection (hereafter called ForDiv). All present and former records (i.e. a species collected at
a given site (locality on the label) on a given date) were attributed to 10x10km grid cells of
the UTM (Universal Transverse Mercator) projection, hereafter called grid cells. To compare
the present diversity with that obtained form voucher specimens we restricted our analyses
to grid cells where at least two species were found (ForDiv and PresDiv ≥ 2; in most cases at
least Culex pipiens and Culiseta annulata present).
Methodology
For each of these grid cells a trend criterion (degree of decline or increase) was calculated
(Dufrêne and Desender 2007, Desender et al. 2008). The degree of change in diversity for
each grid cell (TREND) was calculated by TREND = [PresDiv – (ForDiv * COR)] / [(ForDiv *
COR) + PresDiv], where PresDiv and FormDiv are the number of mosquitoes observed per
grid cell and COR the correction factor to correct for difference in sampling intensity and
methods during both inventory systems. This correction factor was calculated by dividing the
sum of PresDiv by the sum of ForDiv for all well studied grid cells. Here the correction factor
was + 1.215. By calculating this TREND we generate a value between -1 (all diversity lost)
and +1 (all diversity new) for each well surveyed grid cell.
To obtain relative changes in distribution area for 23 mosquito species, a linear regression
was performed of the logit-transformed proportions from the recent inventory as a function of
the logit-transformed proportions retrieved from the revision of RBINS collection. Proportions
in both surveys were calculated as P=(x+1)/(n+1) where x is the number of recorded grid
cells for a given species and n is the total number of grid cells surveyed. The logit-
transformed proportions were calculated as logit(P)=ln[P/(1-P)] (Telfer et al. 2002).
The index of relative change in distribution area for each species was calculated by its
standardized residual from the fitted regression line (Van Landuyt et al. 2008). Species that
obtained positive values of this index had relatively increased their distribution area. Species
that obtained negative values of this index had relatively decreased their distribution area.
Values equal to zero indicate no relative change in distribution area for this mosquito species
between the two survey period
Results
Mainly near big cities mosquito diversity has declined because of the loss of undisturbed
natural habitats and their inhabiting mosquito species. On the other hand in the north of the
country and the Kempen region some grid cells show an increase in mosquito diversity
(Figure 9).
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 28
Figure 9: Trends in mosquito diversity for well studied grid cells based on mosquito diversity obtained
from voucher specimens and from the recent inventory.
An estimation of the relative changes in distribution area of most of the Belgian mosquito
species revealed that eight species have a positive index of relative change in distribution
area and might be considered as species that enlarged their relative distribution area
recently. Amongst them are two An. species and the An. maculipennis s.l. complex. Also Ae.
vexans and Ae. cinereus/geminus. have a positive relative index of change in distribution
area. The highest indexes of change in distribution area were retrieved for Cx torrentium, Cx
pipiens and An. plumbeus. that recently also use man-made habitats as larval breeding site
as used tires and waste waters, have enlarged their distribution area.
2.4 Set up a molecular reference archive of Belgian mosquitoes
based on the DNA barcoding region
2.4.1 General molecular approach
A general molecular identification framework based on the DNA-barcoding approach is
developed for the Culicidae from Belgium. This framework will allow verifying the species for
which the morphological identification is problematic or when sibling species are involved.
Furthermore DNA barcoding will be an essential tool for the fast and reliable identification of
possible invading species. The development of a comprehensive DNA barcodes dataset of
the Belgian species based on fresh field material will contribute to disentangle possible
taxonomic problems and improve further species identification. Based on the currently
available information markers other than COI, have been explored to improve the
differentiating power of the DNA barcoding like ND4 (for Anopheles and Aedes species) and
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 29
D3 (for Anopheles species). Two major regions of interest (ITS2-rDNA and COI-mtDNA)
have been PCR-amplified and sequenced. The PCR on ITS2 is done according to the
method described in Van Bortel et al. (2000) whilst for the COI PCR the universal DNA-
barcoding primers are used (Folmer et al. 1994). Protocols for the ND4 regions were done
based on Cameron et al (2010) for Aedes species and based on Fonseca et al. (2001) for
Anopheles species; whilst PCR on D3 was done according to the method of Sharpe et al
1999. The protocol for the microsatellite PCR was adapted from Keyghobadi et al. (2004)
whilst that for the CQ11 locus from Bahnck and Fonseca (2006).
2.4.2 DNA barcoding of Belgian mosquito species from MODIRISK
A total of 964 individuals (20 species) were molecularly identified, amplifying the COI mtDNA
barcoding region and the ITS2 rDNA region. Sequence results were good for the COI region.
The ITS2 showed different copies in the same individual with a returning deletion/insertion at
a fixed site. Especially Anopheles species showed this multi-copy phenomenon. The COI
sequences were assembled and aligned using ClustalW (BioEdit) and neighbour joining tree
was constructed using Mega4. Preliminary analysis (neighbour joining, Mega4) of the COI
sequence results confirms the utility of the COI region for species identification. Most
branches of the tree are supported by bootstraps values higher than 70% (for most species
even 90% and more). On genus level most branches are not supported, hence for more
phylogenetic studies other markers were tested as well from the mitochondrial as from the
ribosomal DNA (like for Anopheles species). It seems that depending on the studied genus,
other markers and methods should be employed. A well supported and integrated
identification system for field material will be created usable in a broader European context
(Biosurveillance and vector control programmes). Furthermore, as Sequences of all species
will be submitted to GenBank so they are freely available for researchers and students
worldwide these results will be available for worldwide mosquito taxonomy studies. This work
has been done in the cooperation with the JEMU (RBINS and RMCA) project in the
framework of their flagship project Barcodes for TwoWings (BC42W).
2.4.3 A larval assay to rapidly and correctly identify mosquito species
Correct identification of the vector is one of the important factors in the study of arboviral
diseases (Cook et al., 2005). In addition, the precise identification of the target species has
direct medical and practical implications, particularly in developing vector control strategies.
Identifying larvae is not easy and often problems are encountered in separating larvae of
endemic and exotic container mosquito species (Beebe et al. 2007). However some already
described tests work only on a specific targeted genus. Moreover, preliminary test results of
of Belgian mosquito larvae (based on the rDNA ITS2 region, and the mtDNA ND4 and COI
region) showed that DNA extracted from field larvae was highly contaminated by organic
matter competing in the PCR reactions.
After several tests assessing the best way in which larvae (lab and field material) could be
killed, preserved and DNA extracted to increase the percentage of positive results after PCR
amplification, an easy and efficient protocol is set up that will be generally applicable.
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 30
2.5 Longitudinal data of Dutch mosquitoes for validation of habitat
characteristics, mosquito species composition and abundance
2.5.1 Methodology
Longitudinal studies were conducted in 55 selected sites across The Netherlands in the late
summer of 2009 and the spring of 2010. The aim was to assess whether mosquito presence
and abundance in the selected sites were correlated to those found in Belgium. It was
assumed, though, that as the typical meadow landscapes as found in much of the North
West of The Netherlands are not present in Belgium, differences in mosquito species
composition and/or abundance might occur.
Mosquito sampling and identification:
Mosquitoes were sampled with CO2 traps used throughout the Modirisk project. Each site
was sampled for one week continuously, once between August and October 2009, and once
again between April and June 2010. In this way, spring and fall populations of mosquitoes
were expected to be found. Samples taken to Wageningen for taxonomic identification using
the key provided by Schaffner (Schaffner 1993). Members of the An. maculipennis complex
were further identified to species using a PCR of the ITS2 region (Marinucci et al. 1999).
Study sites
A total of 55 sites distributed across The Netherlands were studied (Figure 10). Study sites
included agricultural and urban sites, forest, dunes and marshes, as well as one harbour
(Rotterdam) and one airport site (Schiphol, Amsterdam) (Table IV). More than 56% of the
sites were classified as agricultural, and included mostly meadows, as well as 2
greenhouses. At greenhouses, traps were placed outdoors, in the vicinity of the main
entrance used for loading cargo.
Figure 10. Locations of LP CO2 traps in The Netherlands in 2009 and 2010
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 31
Table IV: Detailed overview landscapes sampled in The Netherlands (2009- 2010)
Type
Specification
No. locations
Perc.
1
Agricultural
Pastures
Cattle
12
21.8
Sheep
5
9.1
Horses
5
9.1
Stable
Chickens
2
3.6
Pigs
1
1.8
Crop production
4
7.3
2
Greenhouses
2
3.6
3
Urban
8
14.5
4
Forest
5
9.1
5
Dunes
2
3.6
6
Marsh
Peatbog (high)
2
3.6
Peatbog (low)
2
3.6
Freshwater
2
3.6
Brackish
1
1.8
7
Harbour
1
1.8
8
Airport
1
1.8
Total
55
100
2.5.2 Results
Twenty two species of Culicidae were found (Table V). Cx pipiens, An. plumbeus and Ae.
cantans were the most abundant species. There were marked differences in phenology
among the species collected. Species that only were present in the spring include Cx
torrentium, Ae. punctor while Ae. cantans was much more abundant in the spring than in the
late summer. No species were only present in the late summer, although Cx pipiens, Cs
annulata, Cq richardii and An. maculipennis were much more abundant in late summer than
in the spring. Table VI summarizes the distribution of the most common species found per
landscape type. Both Cx pipiens and Cx torrentium were common in industrial areas like the
airport. Cs annulata was equally common in agricultural, urban areas and marshes. Cs
morsitans prefered marshes but was also found in forests and in agricultural areas.
Coquillettidia richardii preferred marshes, but was also common in urban areas. Anopheles
maculipennis clearly preferred agricultural areas. Anopheles claviger was most common in
dunes and to a slightly lesser extent in marshes, although only two sites “dunes” were
sampled, while the number of marshes studied was higher. Anopheles plumbeus was by far
most common in agricultural areas. This species was abundant on only a few farms (in high
numbers) while on most other farms it was absent (see 3.3). Most Aedes-species preferred
forests and were to lesser extend found in marshes or agricultural areas. Only Ae. vexans
and Ae.cinereus appeared to prefer marshes and agricultural areas.
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 32
Table V: Total number of mosquitoes caught in The Netherlands in 2009 and 2010
Species
July, August,
Sept. 2009
April, May, June
2010
Total
Culex pipiens
1069
145
1214
Culex torrentium
0
33
33
Culex territans
3
5
8
Culex modestus
1
0
1
Culiseta annulata
92
12
104
Culiseta suborchea
0
1
1
Culiseta morsitans
2
12
14
Culiseta fumipennis
2
1
3
Coquillettidia richardii
104
9
113
Anopheles maculipennis
142
18
160
Anopheles claviger
20
39
59
Anopheles plumbeus
483
661
1144
Aedes vexans
0
6
6
Aedes cantans
72
1035
1107
Aedes riparius
20
57
77
Aedes annulipes
5
58
63
Aedes communis
0
8
8
Aedes cinereus/geminus
70
81
151
Aedes punctor
0
427
427
Aedes leucomelus
0
9
9
Aedes rusticus
0
20
20
Aedes geniculatus
0
1
1
Table VI: Distribution (%) of most common mosquitoes caught per landscape type per week (2009 & 2010).
Species
Agricultural
Forest
Urban
Marsh
Dunes
Harbour
Airport
Culex pipiens
15.5
2.6
9.2
2.8
2.7
24.0
43.2
Culex torrentium
2.2
0.0
0.0
0.0
5.1
92.7
0.0
Culiseta annulata
26.5
5.3
38.0
30.2
0.0
0.0
0.0
Culiseta morsitans
31.4
15.0
0.0
53.6
0.0
0.0
0.0
Coquillettidia richardii
8.8
3.3
37.2
50.7
0.0
0.0
0.0
Anopheles maculipennis
64.7
0.0
12.3
16.0
7.0
0.0
0.0
Anopheles claviger
1.5
3.1
9.6
28.4
57.4
0.0
0.0
Anopheles plumbeus
98.0
0.5
1.6
0.0
0.0
0.0
0.0
Aedes cantans
15.8
75.0
0.2
9.0
0.0
0.0
0.0
Aedes annulipes
43.3
42.1
3.3
11.3
0.0
0.0
0.0
Aedes riparius
4.2
90.2
1.0
4.6
0.0
0.0
0.0
Aedes punctor
2.6
85.9
0.9
10.7
0.0
0.0
0.0
Aedes rusticus
2.8
97.2
0.0
0.0
0.0
0.0
0.0
Aedes vexans
31.1
0.0
0.0
68.9
0.0
0.0
0.0
Aedes cinereus/geminus
0.3
6.5
7.0
86.2
0.0
0.0
0.0
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 33
Table VII shows the species composition of the Anopheles maculipennis complex in The
Netherlands: An. maculipennis s.s., An. messeae and An. atroparvus. The by far most
common species of this complex was An. messeae (95.1 %) and to a lesser extent the
species An. maculipennis s.s. (4.3 %) and An. atroparvus (0.6 %). Anopheles messeae was
found all over the country but clearly preferred the pastures on lowland peatbogs in the west
of The Netherlands. Anopheles atroparvus was found only once, in the natural reserve “The
Putten” near Camperduin in Noord Holland. This open brackish marshland consists of areas
with shallow open water and broad reed beds. Anopheles maculipennis s.s. was found
especially in agricultural areas in the East of the country. Species of the An. maculipennis
complex were found more in the late summer of 2009 than in the spring of 2010. From
previous studies in The Netherlands it was found that populations of An. maculipennis s.l.
develop slowly in the spring, to peak in late July/August.
Table VII:Speciescomposition of the Anopheles maculipennis complex (2009 & 2010).
2009
2010
Number
Percentage
Number
Percentage
Anopheles messeae
139
95.9
16
88.9
Anopheles atroparvus
1
0.7
0
0.0
Anopheles maculipennis s.s.
5
3.4
2
11.1
Total
145
100
18
100.0
A seasonal distribution of mosquitoes is shown in Table VIII, which includes the mean
number of mosquitoes caught per location per month. Culex pipiens appears to be most
common in late summer and the early autumn, while Cs annulata, An.maculipennis and Cq
richardii were most common in July. Anopheles claviger was most common in spring, while
Anopheles plumbeus was most common in June and July. Most Aedes species reached their
summit in numbers in spring. Mosquitoes that appeared earliest in spring are Culex pipiens,
Cx territans, Cx torrentium, Cs annulata, An.maculipennis and An. claviger. Of the Aedes
species, Aedes punctor was usually the first, followed by Ae. cantans, Ae. riparius, Ae.
annulipes and Ae. rusticus. Aedes cinereus and Ae. vexans appeared just a bit later.
Mosquitoes that were still active in September were Cx pipiens, Cs annulata, Cs morsitans,
Cs fumipennis, An. claviger, An. maculipennis, Cq richardi and Ae. cinereus. Only Cx pipiens
and Cs annulata remained active till late in the autumn (Oct, Nov). Culex pipiens was even
still caught at the beginning of December 2009 in some old forests in the Betuwe! It seems
that a number of the mosquito species that appear early in the spring were still active late in
the autumn
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 34
Table VIII: Mean number of caught mosquitoes per location per month
Species
2009
2010
July
August
September
April
May
June
Culex pipiens
12.1
23.4
35.2
0.6
0.8
4.6
Culex torentium
0
0
0
0.1
0.1
1.5
Culex territans
0.2
0
0
0.1
0.0
0.2
Culex modestus
0.1
0
0
0.0
0.0
0.0
Culiseta annulata
2.4
1.3
2
0.4
0.0
0.3
Culiseta suborchea
0
0
0
0.1
0.0
0.0
Culiseta morsitans
0
0
0.3
0.0
0.5
0.1
Culiseta fumipennis
0
0
0.2
0.0
0.0
0.1
Coquillettidia richardii
3.3
0.5
4
0.0
0.0
0.6
Anopheles maculipennis
5.8
1
0.5
0.3
0.2
0.5
Anopheles claviger
0.2
0.6
0.5
0.9
0.9
0.3
Anopheles plumbeus
23
1
0
0.0
0.0
14.5
Aedes vexans
0
0
0
0.0
0.2
0.1
Aedes cantans
3.4
0.2
0
0.4
27.3
30.3
Aedes riparius
1
0
0
0.1
0.4
3.1
Aedes annulipes
0.3
0
0
0.0
0.0
3.9
Aedes communis
0
0
0
0.0
0.1
0.4
Aedes cinereus
0
2.8
0.2
0.0
0.3
4.9
Aedes punctor
0
0
0
2.5
3.0
21.2
Aedes leucomelus
0
0
0
0.0
0.3
0.2
Aedes rusticus
0
0
0
0.0
0.1
1.1
Aedes geniculatus
0
0
0
0.0
0.0
0.1
By trapping mosquitoes in a comparable way during one week in the late spring (mid April to
mid June) as well as late summer (July till the beginning of September) on 55 sites evenly
spread over the country, a good impression of the distribution of the different mosquito
species in The Netherlands was obtained. Most locations have been chosen to verifiy the
model that predicts the presence or absence of mosquitoes on the basis of the habitat. If, by
chance, many farms would have been chosen with suitable breeding places for An.
plumbeus, their numbers found would be manyfold of what would have been found if farms
would have been chosen where breeding places for this species are absent. The three most
commonly found mosquitoes in this survey were Cx pipiens, An. plumbeus and Ae. cantans.
Mosquitoes that are typical for brackish water habitats were almost absent from the catches.
Only one specimen of An. atroparvus was found. Although this species was historically
present in high abundance (Swellengrebel and de Buck 1938), this species is rarely found in
recent years (Takken et al. 2002), presumably because of a change in habitat and significant
change in water quality, from being brackish to fresh. The sibling species An. messeae, by
contrast, is widely present in all ditches in agricultural areas. Other species, which are typical
for this habitat, like Ae. caspius, Ae. dorsalis and Ae. detritus, were not found at all. It is
unlikely that these species are absent in The Netherlands, although they were previously
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 35
found (Takken et al 2007). Probably the lack of places with a suitable habitat for these
species that were chosen for this research is the reason for their absence. By comparing the
mosquito species and abundance recorded in the present study with those obtained in the
Belgian study, using GIS referenced habitats, the data from The Netherlands can be scaled
up and related to the landscape types as mentioned above.
Distribution of Culex pipiens in The Netherlands
We focus here on the distribution of Cx pipiens, as this was the species that dominated the
collections and was found throughout the country (Figure 11 and 12). Distribution of other
species are available from the Laboratory of Entomology at Wageningen University. Figure
11 shows the distribution of Cx pipiens in late summer of 2009 and Figure 12 shows the
distribution of this species during late spring of 2010. The red dots present the places where
traps were present, but where no Cx pipiens were found, the light blue dots present the
number caught in a logarithmic scale. In late summer of 2009 Cx pipiens was much more
common than in the late spring of 2010, with spatial differences in abundance. In the spring
this species was in the southern part a bit more common than in the rest of the country
(Figure 12), which is in line with previous studies (Huijben et al. 2007). As with An.
maculipennis spp., populations of Cx pipiens are low in the spring and develop slowly until
they reach a population threshold level at the end of June, when population development
rises rapidly.
Several other species show a dishomogeneous distribution in the country. For example, An.
plumbeus, Cq richardii, Ae. cantans and Ae. punctor show a distinct segregated distribution
pattern, which can be linked to habitat type. Other species were more evenly distributed
across habitats/landscapes, or were too few in numbers to draw any conclusion about habitat
preference. These include Cx modestus, Cs fumipennis, Ae. vexans, Ae. communis, Ae.
leucomelus and Ae. geniculatus, of which <10 specimens were found. The near absence of
Ae. vexans is remarkable, given the ubiquitous distribution of this species elsewhere in
Europe, where it often reaches a high nuisance level (Becker 1997). However, this absence
has been observed before, and is possibly due to unsuitable vegetation and/or lack of
inundations at the time when the eggs need to hatch (Takken et al., 2007).
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 36
Figure 2. Number of Culex pipiens caught per week in the late summer
of 2009.
0 mosquitoes
1 –10 mosq.
100 –1000 mosq.
10 –100 mosq.
Figure 11: Logaritmic number and distribution of caught Culex pipiens during late summer of
2009
Figure 3. Number of Culex pipiens caught per week in the spring
of 2010.
1 –10 mosq.
10 –100 mosq.
100 –1000 mosq.
0 mosquitoes
Figure 12: Logaritmic number and distribution of caught Culex pipiens during early summer of
2010
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 37
3 ECOLOGY AND POPULATION DYNAMICS OF MOSQUITOES IN
BELGIUM
3.1 Spatial modelling of indigenous mosquito species in Belgium;
detection of hotspots and implications for field sampling
design and risk analysis in Europe
3.1.1 Methodology
3.1.1.1 Development of a spatial data archive
Low resolution remote sensing
NOAA AVHRR data has been archived from the following online archive:
http://www.class.noaa.gov/saa/products/. Each individual image has been visually inspected
for noise. If the noise level was too high, then the image was not included in the Avia-GIS
archive, this to avoid contamination of the archive and corruption of the processing chain and
the processed images. The NOAA data starting from 2000 were then processed using the
Avia-GIS software. To remove cloud contamination 10 day composites using the Maximum
Apparent Temperature (MaT) algorithm (Cihlar 1994) were created. These images also had
automatic geometric correction using the ground control points included with the satellite
image. The images were then subjected to further noise removal.
Images from the MODIS sensor were used to derive similar variables. MODIS data were
ordered through the following data gateway: http://elpdl03.cr.usgs.gov/pub/imswelcome/.
Images from the MODIS sensor were used to derive additional variables. Both Land Surface
Temperature (LST-day and -night) and Vegetation Indices (VI: EVI and NDVI) for the years
2004, 2005, 2006, 2007 and 2008 were used to detect yearly trends. After Fourier
transformation, the first three harmonics were included as predictor variables (Scharlemann
et all. 2008). The number of days with a mean temperature above (i) 0 °C, (i) 5 °C and (iii)
12,5 C were derived from the MODIS images for the year 2007 and 2008.
NOAA data has been used the most extensively in the past, mainly due to its extensive
historical archive. Since the beginning of 2000, MODIS imagery is an alternative to the
NOAA archive. Its main features are higher spectral resolution, and higher spatial resolution
for several bands. Moreover, ready-made derived products are also available to the general
public. The temperature profiles from NOAA and MODIS were compared to a limited ground
truth meteorological data set (August–September 2006). The meteorological data set
includes hourly observations of temperature. To allow for comparison, the time of recording
the land surface temperature (LST) from MODIS was extracted from the science data set
using the MODIS Reprojection Tool. The accuracy of the satellite derived temperature was
assessed through the calculation of the root mean square error and the bias. A good fit was
found between the profiles extracted from both NOAA and MODIS data.
In a paper by Scharlemann et al. (2008), a cubic spline interpolation technique was tested for
seasonality extraction to be used as input for species distribution modelling. This novel
algorithm of spline interpolation followed by regular resampling of the composited satellite
data was developed to produce a 5-day interval MODIS time series that could then be
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 38
subjected to standard temporal Fourier processing methods. This algorithm was found to
capture the input amplitude and phase information correctly. This algorithm was applied to all
MODIS image time series to remove missing data pixels. This includes the following
environmental parameters: NDVI, EVI and Land Surface Temperature (LST) from 2000 up to
2010.
Other spatial data sources
All relevant data layers needed to map mosquito habitats and plan standardized spatial
sampling of mosquitoes were collected. The following data layers have been archived: land
use and land cover classification, administrative boundaries, geocoding data layers (MultiNet
Street data). In January 2008 a session with the field teams was organised to create a
shortlist of data layers that can be useful for further analysis.
Information on land cover was derived from the CORINE dataset (JRC-IES 2005). The
percentage of the three Level-1 land cover classes, i.e. urban, agriculture and natural, within
a one-kilometre-pixel was computed. Within the forest class, cover percentage was also
computed for three forest types: broadleaved, coniferous, and mixed forest. Human
population pressure on the landscape was assessed by population density expressed by the
number of inhabitants/km² (GWPv3.0 by SEDAC 2000). Other environmental data layers
were the available water capacity of topsoil (JRC-IES 2009), distance to waterways (GfK
2009) and the GTOPO30 elevation (USGS 1996), mean of total yearly precipitation and
mean of monthly mean of precipitation were obtained from the WORLDCLIM dataset
(Hijmans et al. 2005).
All data layers were clipped to the extent of the study area and resampled to a 1- and 5-km
resolution. GIS manipulations were performed using ArcGIS9.3 (ESRI 2009).
Eco-climatic seasonality analysis
Fourier analysis is a family of mathematical techniques, all based on decomposing signals
into sinusoids. Through the use of a Fourier transform, any real world signal can be split into
basic sine/cosine waves, each at a different frequency. The more sinusoids included, the
better the approximation of the real-world signal. Each of the harmonic frequencies is defined
by a magnitude (amplitude) and a phase. The phase indicates how to shift the harmonic
before adding it to the sum.
Fourier analysis is ideally suited for summarizing seasonal variables (Rogers et al. 1996)
because seasonal activity is a driven factor for vegetative status, vector abundance etc. The
seasonal dynamics directly influence vector population dynamics. The importance of Fourier
derived products has been shown by Rogers et al. (1996) in prediction tse tse fly distribution¸
and by Baylis et al. (2002) and Purse et al. (2004) for predicting Culicoides imicola. All
satellite data has been processed using the Fourier analysis. For each of the environmental
parameters, the first three harmonics were retained for further analysis.
Eco-climatic zones were identified using an unsupervised k-means clustering. This clustering
was performed on the following variables:
- Altitude
- First three Fourier transforms on EVI, NDVI, LST(day) and LST(night)
- Precipitation : minimum, mean, and maximum values
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 39
- Land cover:
- Proportion of artificial surfaces,
- Proportion of agriculture areas,
- Proportion of natural vegetation
3.1.1.2 Development of spatial distribution models
Entomological data.
In the modeling approach, the trap was the sampling unit. Table IX gives an overview of the
field data, not in number of individuals, but in number of traps in which individuals of a given
species were found. In order to construct a robust and statistically sound model, a minimum
of 20 positive (containing the species) and 20 negative (not containing the species) traps
were needed. For some mosquito species, not enough positive traps were available. No
models were constructed for these „rare‟ species, e.g. Culex territans, Culiseta morsitans.
Other individuals were damaged to a degree that made identification impossible, which lead
to a species name „sp‟. No models were constructed for these species either. The list of
species excluded from analysis can be found in the last column of Table IX
Table IX: Summary of field data based on the trap as sample unit. Positive trap: a trap in which the
species is present
Species name
Positive traps
Individuals
Model
Aedes cinereus/geminus
74
1,328
Aedes cinereus/geminus
Aedes sp
3
8
Excluded
Aedes vexans
29
776
Aedes vexans
Anopheles claviger
185
935
Anopheles claviger
Anopheles maculipennis s.l.
45
80
Anopheles maculipennis s.l.
Anopheles plumbeus
114
391
Anopheles plumbeus
Anopheles sp
4
5
Excluded
Coquillettidia richiardii
38
4,095
Coquillettidia richiardii
Culex pipiens
698
16,338
Culex pipiens
Culex sp
1
5
Excluded
Culex territans
7
11
Excluded
Culex torrentium
75
139
Culex torrentium
Culiseta annulata
162
577
Culiseta annulata
Culiseta morsitans
17
92
Excluded
Culiseta sp
2
2
Excluded
Aedes annulipes
8
42
Aedes annulipes/cantans
Aedes cantans
63
348
Aedes annulipes/cantans
Aedes caspius
12
80
Excluded
Aedes communis
33
280
Aedes communis
Aedes detritus s.s.
4
48
Excluded
Aedes geniculatus
48
164
Aedes geniculatus
Aedes japonicus
1
3
Excluded
Aedes punctor
65
531
Aedes punctor
Aedes rusticus
33
178
Aedes rusticus
Aedes sp
12
37
Excluded
Aedes sticticus
13
63
Excluded
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 40
In Table X, summary information is given on the species that were most widespread and
most abundant in the traps.
Table X. Presence data for the most abundant and easy-to-catch species in Belgium. The
column „total‟ indicates the number of positive traps, also expressed as a percentage (%) of the
total number of traps. The column „N‟ gives the total number of individuals caught. The column
„N/+Tr‟ lists the average number of individuals per positive trap.
Species
Total
%
N
N/+Tr
Aedes vexans
29
2.97%
776
26.76
Coquillettidia richiardii
38
3.89%
4,095
107.76
Aedes cinereus/geminus
74
7.57%
1,328
17.95
Culex pipiens
698
71.37%
16,338
23.41
In Figure 133 the number of species found in each trap is visualized. The majority of the
traps contained only one or two species. Rarely more than five mosquito species were
recorded in one trap. This gives an idea how difficult it is to catch all mosquito species
present at a given location. The maximum number of mosquito species found in one and the
same trap was 12.
Figure 13. Species richness as observed in the traps.
Project SD/BD/04 - Mosquito vectors of disease: spatial biodiversity, drivers of change, and risk “MODIRISK”
SSD-Science for a Sustainable Development - Biodiversity 41
Explanatory variables
In order to predict the presence/absence of a given mosquito species, all data layers defined
under 2.1 were used as explanatory variables.
Modelling approach.
In order to obtain a balanced data set for the construction of the models, a subsample of the
available sample was taken. For each species, the number of traps reporting respectively a
presence or an absence of the species was counted. If the number of presences for a given
species was lower than the number of absences, all traps reporting a presence were retained
in the calibration data set. A sample of the same size was randomly drawn from all traps
reporting an absence of the species. If the number of presences was superior to the number
of absences, the procedure was inversed; all absences were retained in the data set, and the
presences were randomly drawn.
This calibration data set was then used to construct a model for each mosquito species.
Models were generated using Random Forests (RF) (Breiman 2001). Random Forests is a
robust ensemble learning technique, which can be applied to model probability maps,
expressing the probability of occurrence, through a random classification forest or
abundance maps through a random regression forest. The technique consistently
outperforms traditional modelling techniques such as logistic regression (Cutler et al. 2007,
Peters et al. 2007). Random classification forests have been used to assess if temperature
and precipitation affected the minimum infection rate of Culex species for West Nile Virus in
Illinois (Ruiz et al. 2010) and to model the current spatial distribution of Aedes albopictus in
Europe using a wide set of predictor variables (ECDC 2009) .
Random regression forests allow both internal and external validation through a
bootstrapping procedure. For each classification or regression tree, the full data set is
bootstrapped, i.e. a number of data points are sampled from the complete data set with
replacement. From the bootstrapped sample approximately one third of data are excluded.
This set of the excluded data is referred to as the out-of-bag (OOB) data set for the tree;
each tree will have a different out-of-bag data set. Since t