MODIRISK: Mosquito vectors of disease, collection, monitoring and longitudinal data from Belgium

The MODIRISK project studied mosquito biodiversity and monitored and predicted biodiversity changes, to actively prepare to address issues of biodiversity change, especially invasive species and new pathogen risks. This work is essential given continuing global changes that may create suitable conditions for invasive species spread and the (re-)emergence of vector-borne diseases in Europe. Key strengths of MODIRISK, in the context of sustainable development, were the links between biodiversity and health and the environment, and its contribution to the development of tools for describing the spatial distribution of mosquito biodiversity. MODIRISK addressed key topics of the global Diversitas initiative, which was a main driver of the Belspo ‘Science for a Sustainable Development’ research program. Three different MODIRISK datasets were published in the Global Biodiversity Information Facility (GBIF): the Collection dataset (the Culicidae collection of the Museum of Natural History in Brussels); the Inventory dataset (data from the MODIRISK inventory effort); and the Longitudinal dataset (experiment data used for risk assessments).

The three datasets, which originated during the MODIRISK project ( Figure 1), were standardized to Darwin Core [1] and published by the Royal Belgian Institute for Natural Sciences (RBINS) through the Integrated Publishing Toolkit (IPT) [2] of the Belgian node of the Global Biodiversity Information Facility (GBIF).

METHODS
During the first phase of the project (2007)(2008), the focus was on inventory activities; setting up laboratory experiments to study life history traits of Culex pipiens in relation to temperature, and the first selection of models based on the field results. During the second phase of the project (2009-2010), the focus was on spatial model building and validation, the longitudinal study, and the dynamics of selected indigenous and invasive species found during the first-phase inventory and on more population genetics-driven research.
Thus, three different datasets were developed during the MODIRISK project: the 'Collection' dataset [3,4], dealing with both historic and recent Culicidae specimens, the 'Inventory' dataset [5], dealing with the current inventory of Culicidae in Belgium, and the 'Longitudinal study' dataset [6], dealing with questions about risk assessment, outbreaks, and possible distribution.
These three datasets are closely linked, but are published as three different Darwin Core archives. The Inventory dataset 'MODIRISK: Monitoring of Mosquito Vectors of Disease (Inventory)' was first published in 2013 (occurrence core), while the Collection dataset (occurrence core) and the longitudinal study (event core) were published in 2017.
The project was coordinated by the Institute of Tropical Medicine, Antwerp, Belgium.

The Collection dataset [3, 4]
In the early 1900s, Maurice Goetghebuer and Michel Bequaert collected many mosquitos from all over Belgium and established on of the most representative and richest collections of Belgian Diptera. These are preserved at RBINS [7]. The RBINS Culicidae collection comprises four parts: a general Belgian collection, the Goetghebeur subcollection, the Becquart subcollection, and a subcollection of unidentified specimens known as the 'supplements' (Figure 2 All 1381 specimens (24 species) in the RBINS collections were re-identified and digitized during the MODIRISK project. Furthermore, all voucher specimens from the available collections were re-identified at the species level [8]. Seventy seven percent of the specimens were collected between 1910 and 1960, with most specimens 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 3). The oldest specimens (collected in 1878) are deposited in the general Belgian collection. In this collection, 16 species were discovered; in the Bequaert, Goetghebuer and in the supplements collections, respectively, 18, 21 and 20 species were counted. Culex pipiens and Culiseta annulata were the most abundant recorded species present in the collection (Figure 4), as were many voucher specimens of Aedes punctor. 1374 records were published on the GBIF repository. 7 Records were left out because those specimens originated from outside Belgium and missed coordinates.

The Inventory dataset
The Culicidae Inventory dataset [5] (Figure 5) was created during a cross-sectional field survey from May till October 2007 and 2008 (inventory) and from August till October 2009 and May till August 2010 (validation), using a network of CO 2 -baited Mosquito Magnet Liberty Plus (MMLP) traps (Woodstream Corp., Lancaster, PA, USA), throughout Belgium in three key habitats (urban, agriculture and nature) [11]. Each location was sampled only once for 1 week. Twenty-nine mosquito species were identified to the species level ( Figure 6).
Each locality was sampled with a minimum of six traps: three types, two of each type, namely MMLP traps, BG Sentinel (Biogents, Regensburg, Bayern, Germany) and CDC Gravid

The Maasmechelen (MA) study site
The area surveyed was an old sand quarry near Hoge Kempen national park and the Maasmechelen industrial park, where several recycling companies are located. One site, MA1, was the initial reference site, a small mixed forest fragment with birch, oak and pine next to the industrial zone. The other subsite, MA2, was on the opposite side of the road, in a narrow strip of mixed forest adjacent to a large nature reserve (heath). Land cover is largely mixed small forest, moorland, sand quarry, and a large industrial zone.

The Natoye (NT) study site
The population of Aedes japonicus at Natoye was surveyed by researchers from the Catholic University of Louvain (Université Catholique de Louvain; UCL). Two Belgian second-hand tire companies located in the village of Natoye (Namur) were surveyed. Sites were named Natoye1 (NT1) and Natoye 2 (NT2). The companies import mainly tires for trucks and heavy vehicles originating from various European countries. Tires are stacked outside, and many are exposed to rainfall, so often contain water and organic material such as decomposing leaves. Land cover around Natoye 1 consists largely of deciduous forests, gardens and cultivated fields, while around Natoye 2 it is mostly gardens, cultivated fields and meadows.

The Ruiselede (RL) and Torhout (TH) study sites
Mosquitos were sampled mosquitoes at two different localities in Western Flanders with the same ecoclimatic region, during one complete active season from May until October.

Taxonomic coverage of the three datasets
Morphological identification of Culicidae was done mainly using an electronic [8] and a paper identification key [18]. If a sample was too damaged for accurate morphological identification, sp. was used to describe the level of identification. In the period 2000-2009, substantial changes were proposed for the Aedini tribe taxonomy, which resulted in almost tripling the number of genera in the entire Culicidae family. A recent publication [19] proposed to return to the taxonomy from before 2000, restoring a classification system useful for the operational community. This dataset was built during 2000-2009, using the available taxonomies at that time, resulting in the consequent use of the genus Ochlerotatus (Now Aedes) in the dataset.

Geographic coverage of the three datasets
Belgium is a small country in Western Europe. To the west, its 70 km coastline fronts the North Sea; to the north lies the Netherlands; to the east, Germany, and to the south, France

Geographical method
Collection dataset . The Universal Transverse Mercator Projection (UTM), an adaptation of the standard Mercator projection, uses a two-dimensional Cartesian coordinate system to identify locations on the surface of the Earth [20].
The Inventory dataset (2007)(2008)(2009)(2010). The 971 selected sample sites have exact GPS coordinates. Cross-sectional field surveys were conducted during the first phase of the project to create an inventory of Culicidae. Using the NGI (National Geographic Institute) CORINE land cover (2000) map classification [21], three strata were aggregated: urban, rural and natural. In each stratum, random points were assigned, amounting to 971 selected sampling points. The number of points assigned for each Corine land cover aggregated class was proportional to its total surface in Belgium. For details of the sampling design, see Versteirt et al. 2011 [13] and 2013 [11].
The Longitudinal study (2009). All the selected sampling sites have exact GPS coordinates.
The overall distribution of the mosquito records over time is illustrated in Figure 11.

DATA VALIDATION AND QUALITY CONTROL
To assure the quality of morphological identifications, a random sample (10%) of the identified mosquitoes collected through the MODIRISK project (Inventory and Longitudinal datasets) was re-identified by an external expert. The re-identification of the RBINS collections done during the MODIRISK project was also checked by an external expert.
Members of the An. maculipennis complex were further identified to species level using PCR of the ITS2 region [22].

DATASET DESCRIPTION
Occurrence data from the MODIRISK database are extracted, standardized, and published as three separate Darwin Core archives: the Collection, the Inventory and the Longitudinal study. The main rationale behind this is that these different datasets are built for their own  specific purpose and differ in sampling protocols and methods. Together these datasets represent a complete overview of the data collected during the MODIRISK project. We made the data available to GBIF to make future use, such as Species Distribution Modelling, possible.
The Darwin Core terms in the dataset at the time of publication can be found here [23].

DATA AVAILABILITY
This data paper is linked with three MODIRISK mosquito related datasets: the Collection [4], Inventory [5], and Longitudinal study datasets [6]. The database server uses Windows Server 2003 SBS R2 as the 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: the field form, morphological identification form, and mosquito storage form. Based on these, relevant tables were developed by Avia-GIS [24], implemented in the database, and transferred to the web server.
The data are published under a Creative Commons CC0 waiver and we kindly ask you to notify the corresponding authors of the respective dataset if you use the data, especially for research purposes.

EDITOR'S NOTE
This paper is part of a series of Data Release articles working with GBIF and supported by the Special Programme for Research and Training in Tropical Diseases (TDR), hosted at the World Health Organization [25].

DECLARATIONS LIST OF ABBREVIATIONS
UTM: Universal Transverse Mercator.

ETHICAL APPROVAL
Not applicable.

CONSENT FOR PUBLICATION
Not applicable.